An Integrated Substance Abuse Treatment
Needs Assessment for North Dakota

Final Report

Prepared by
William McAuliffe, Ph.D.
Ryan P. Dunn, B.A.
Caroline Zhang, M.A.

North Charles Research and Planning Group
North Charles, Inc.
875 Massachusetts Avenue
Cambridge, MA 02139
September 2, 2002
Phone: (617) 864-9115
wmcauliffe@ntc.org

North Dakota Department of Human Services (NDDHS)
Carol K. Olson, Executive Director
Division of Mental Health and Substance Abuse Services (DMHSAS)
Karen Romig Larson
600 South Second Street, Suite 1E
Bismark, ND 58504
Phone: (701) 328-8921
Toll Free: (800) 755-2719
TTY: (710) 328-8969
dhsmhsas@nd.gov

CSAT CONTRACT # 270-98-7064

Acknowledgements

The authors wish to acknowledge the contributions of others that made this study possible. The Center for Substance Abuse Treatment (CSAT) provided funding, administrative, and technical support for this study. Debra Fulcher was the CSAT project officer. Without CSAT's support for the State Treatment Needs Assessment Program (STNAP) this study and the studies upon which it drew would not have been possible. The entire field has advanced greatly as a result of the STNAP initiative.

The study's North Dakota project officer, Sue Tohm, helped out in more ways than we can mention here and showed great patience as the authors worked on the study. Karen Larson and Lauren Sauer of the Division of Mental Health and Substance Abuse Services offered indispensable insights on the treatment system. Kerry Wicks of the State Hospital generously shared his knowledge of the treatment needs of the homeless. Al Lick of the Department of Juvenile Services enthusiastically provided information on the juvenile corrections population and their treatment needs. Mike Froemke lent his knowledge of the needs of prisoners and their treatment system. Girish Budhwar was closely involved in implementation of the social indicator model.

The Gallup Organization conducted the telephone survey that was critical to completion of this integrated study. Many former North Charles staffers worked on the assessment of the treatment needs of recently incarcerated prisoners and the homeless, and the social indicator study; they included Richard LaBrie, Eric Sevigny, Ryan Woodworth, Jamie Mellitt, Stephen Haddad and Timothy Stablein. Athena Kazantzas provided administrative support while the Integration Study was being conducted. Earlier work by Stephanie Geller of the National Technical Center for Substance Abuse Needs Assessment created the foundation for the integrated analysis. Gary Houle, North Charles's Executive Director, has assisted in numerous ways over the years.

Authors

William E. McAuliffe is an associate professor in the Department of Psychiatry, Harvard Medical School at Cambridge Hospital. He has been a professor at Harvard since he received his doctorate in sociology from the Johns Hopkins University in 1972. His research has focused on drug abuse research, quality of medical care, and services planning. One of his studies earned him the Socio-Psychological Prize awarded by the American Association for the Advancement of Science in 1974. He developed a relapse prevention program for heroin and cocaine addiction, which is described in Recovery Training and Self Help: Relapse Prevention and Aftercare for Drug Addicts (Rockville, MD: National Institute on Drug Abuse, 1993). Dr. McAuliffe is Director of the National Technical Center for Substance Abuse Needs Assessment and the North Charles Research and Planning Group.

Caroline (Hui) Zhang is a Research Associate/Programmer at North Charles Research and Planning Group. She received her master's degree in economics from Tufts University in 2001. Before joining North Charles , she was a Research Assistant at Tufts where she worked on the Chinese Household Health & Nutrition Survey analysis and the Tobacco Use Survey analysis.

Ryan Dunn is a research assistant at North Charles Research and Planning Group. He received his bachelor's degree in economics from Vassar College in 2001.

Table of Contents

Executive Summary

• Introduction
• Methods
• National Trends
• Interstate Comparisons
• Households With Telephones
• Households Without Telephones
• American Indians
• Prisoners
• Homeless
• Gap Analysis
• Intrastate Variations
• Levels of Care, Obstacles, and Desired Treatment
• Summary and Conclusions

Executive Summary

This final report describes the integrated results of a family of studies of the substance use disorder treatment needs of North Dakota's citizens, especially those who are most in need of services. Employing funds from the Center for Substance Abuse Treatment (CSAT), State officials contracted with the National Technical Center (NTC) of the North Charles Research and Planning Group (NCRPG) to conduct this study. The problems and issues that were addressed included the answers to three basic planning questions:

• How many people are in need of treatment in the State? The goal was to have an adequate supply of services to meet the absolute level of demand that these cases would produce.
• Where should services be located? The goal was to locate where services are needed most.
• What mix of treatment modalities do these clients need and want? The goal is to match additional treatment services to the needs and desires of those who need and want them in order to achieve maximal effectiveness and efficiency.

North Dakota's Needs Assessment Studies

North Dakota conducted two rounds of needs assessment studies. The first round of studies included a household telephone survey, a survey of American Indians on reservations, an integrated study, and a social indicator study. The second round of studies included a social indicator study and the present integrated needs assessment.

Integrated Analysis

The integrated analysis presented in this report employed a series of methodologies to estimate the overall level of treatment needs in the State. First, the study examined trends in the past decade with regard to the need for treatment and the supply of services nationwide. Second, the study compared the State with other states to assess the comparative level of needs and services. Third, the analysis developed estimates of the past-year treatment needs of components of the State's population. The study integrated estimates of treatment need and services received for residents aged 12 and older of households with and without telephones, the homeless, and recently incarcerated prisoners. The sum of these estimates was a statewide estimate of the number of people who had a substance use disorder in the past year, how many of them have not received treatment, and how many would seek treatment if it were readily available. Analysis of the survey data assessed the levels of care needed to fill unmet demand and obstacles preventing those in need from obtaining treatment. Finally, the analysis used the State's social indicator data to determine where additional services were needed most.

National Trends. Analysis of a series of indicators of need and treatment services revealed that over the past decade the gap between the number of people in need and the amount of treatment services provided to them appeared to have been widening. Indicator trends were somewhat mixed. While alcohol use, alcohol mortality, DUI arrests, and alcohol treatment clients and admissions have declined, survey estimates of alcohol dependence and liquor law violation arrests have increased in the last decade. Some drug need indicators (e.g., positive drug tests among employees) suggested a long-term decline, other indicators (e.g., drug dependence rates, treatment measures) have been mixed or relatively stable, while yet other indicators (e.g., mortality, emergency room episodes, arrests, and survey reports of use) suggested increases, especially among young people, in the second half of the last decade. Cocaine use has declined, but use of stimulants and club drugs (e.g., Ecstasy) has increased.

To measure the relative gaps between the measures of treatment need and services, the authors divided the service rates by need indicators. Regardless of which measures of need (dependence, mortality or arrests) or services (survey, UFDS or TEDS) were considered, the gap between alcohol need and treatment increased over the decade of the 1990s. The alcohol treatment gap appeared to widen because there was a sharper decline in the number of persons receiving treatment than in the indicators of need that declined, and some of the indicators of need increased. Depending on which drug need and service indicators were used, the analysis suggested that the drug treatment gap widened or at least stabilized. Thus, over the past decade the amount of treatment per need for drugs and alcohol combined decreased.

Interstate Comparisons. To measure the adequacy of the State's treatment services relative to other states, the authors created a series of composite treatment need indexes. The Drug Need Index (DNI) consisted of the sum of standardized mean rates per 100,000 of explicit-mention drug mortality and drug possession/sale arrests. Similarly structured, the Alcohol Need Index (ANI) consisted of the sum of explicit-mention alcohol mortality rates and arrests rates for driving under the influence (DUI) and liquor law violations. The Substance Need Index (SNI) combined standardized explicit-mention drug and alcohol mortality rates and the sum of the drug and alcohol arrest rates.

North Dakota's biggest substance use problem is alcoholism. Its alcohol treatment need as measured by the ANI (55) ranked 14^th highest in the country in 1994-1996 . North Dakota's alcohol mortality rate was the 18^th highest in the country, and its alcohol arrest rate ranked 13^th highest. North Dakota ranked 3^rd on the BRFSS's measure of driving after drinking too much, and 23 ^rd on the alcohol-related traffic fatality rate . The State's alcohol treatment services were ranked slightly lower than one would expect based on the need indicators . That is, while the State's alcohol treatment needs were in the second highest quintile in the country according to the index, its treatment services were in the middle quintile according to the UFDS alcohol-only treatment client rate.

North Dakota's controlled drug treatment needs were the lowest in the nation. The State's DNI score of 9 was half that of the next lowest score (West Virginia and Vermont both scored 18) , but North Dakota ranked 35^th in the nation according to the NHSDA's 1999 household survey estimates of drug dependence. Unlike the DNI, the NHSDA's dependence measure consists mostly of cases of marijuana dependence. Consistent with North Dakota's NHSDA dependence measure, North Dakota had the highest percentages of marijuana arrests and marijuana treatment admissions. Compared to other states, North Dakota had the lowest drug mortality mean rate between 1994 and 1996 (0.31 per 100,000) and the second lowest mean drug arrest rate between 1994 and 1996 (119 per 100,000) . North Dakota had the lowest rate of drug-only treatment clients (15 per 100,000) in the UFDS between 1994 and 1996 and the second lowest rate of primary drug admissions (47 per 100,000) among the 41 states that reported to TEDS. The State's low treatment rate matches its low level of need for drug treatment.

North Dakota's Substance Abuse Need Index (SNI) ranking was 24 ^th in the country, clearly attributable to its high level of alcohol treatment needs. The State's combined UFDS substance abuse client rate (alcohol-only, drug-only, and drug plus alcohol) ranked 32 ^nd in the country for 1994-1996. By this measure, North Dakota's treatment services were again one quintile below its moderate overall treatment needs.

Trends in North Dakota. North Dakota's alcohol arrest (DUI, disorderly conduct, and liquor law violations) rates and explicit-mention alcohol mortality rates increased from 1993 to 1998. The other alcohol need indicators were stable over that period. Alcohol treatment admissions (TEDS data) declined slightly over the years between 1994 to 1998. It appears that the rate of admissions compared to rates of arrests and deaths, a proxy for the proportion in need who received treatment, has decreased slightly over time, suggesting a widening of the treatment gap. Whereas the alcohol indicators increased only slightly, the controlled drug indicators in North Dakota increased more sharply during the period from 1993 to 1998 . While drug admissions and drug clients increased as well, the increases in drug mortality and arrests appeared to be sharper. These trends suggest a widening of the drug treatment gap.

Statewide Treatment Need Estimate. To estimate the absolute number of persons in North Dakota who had a past-year substance use disorder, the study combined prevalence and population estimates of treatment need for adults (18 and over) in households with telephones, adolescents in households with telephones, persons 12 and older living in households without telephones, recently incarcerated state prisoners and training school inmates, and homeless people. Applying these estimates to population statistics from the 2000 Census count resulted in an estimated total of 30,880 people with a substance use disorder in North Dakota during the past year.

Whenever possible, the authors made conservative assumptions. It would be reasonable therefore to assume that there were at least 30,880 people with current substance use disorders in the State. If these individuals sought treatment, they would meet the minimum medical necessity criteria employed by treatment programs and managed care organizations.

Although residents of households with telephones account for the largest proportion of cases in the total, generalizing the prevalence rate for that group (5.2%) to the rest of the population would have produced an underestimate of the total number of people in need. In its report of the household survey, the Gallup Organization (1998) applied the telephone survey estimate to the entire population aged 18 and older rather than just the population of adults in households with telephones. After the present authors took the prevalence estimates for the other groups, the estimated total state prevalence rate for persons 12 and older increased to 5.9%. Each of the population subgroups not covered by the telephone survey (residents of households without telephones, recently incarcerated prisoners, and homeless, adolescents in households with telephones, and training school inmates) had a higher estimated prevalence of substance use disorders than the adults in households with telephones (11.1%, 62.5%, 47%, 8.5%, and 62.5% respectively). Although the prisoners and homeless had the highest estimated prevalence rates, they were small populations and therefore contributed relatively few cases to the total population in need. Persons 12 and older in households without telephones had a prevalence rate that was a little more than twice as high the prevalence rate of the adults in households with telephones. Because adolescents in households with telephones was a relatively large subpopulation, they contributed the most (4,820 cases or 15.6%) to the overall increase in the estimate of the total need for treatment. By estimating the rates for the groups other than those covered by the telephone survey rather than generalizing the prevalence rate from the telephone survey, the present integrated analysis arrived at an estimated number of people in need that was higher by 3,707 people.

Treatment Gap. There were clearly many people in North Dakota with a substance use disorder who did not obtain treatment in the past year. In 2000, an estimated 2,826 North Dakota residents received treatment for a substance use disorder. This number equals 9.2% of the estimated 30,880 people in need of treatment that year. These figures are probably the most reliable measure of the treatment gap in North Dakota.

Unmet Demand for Treatment. Even if treatment were readily available to all who needed it, only a portion of those in need would seek care in a given year. The study's surveys asked respondents who had a substance use disorder but who had not obtained treatment whether they thought they needed treatment and would have sought it had it been readily available. The integrated analysis estimated that 4.3% of the persons with a current disorder that did not obtain treatment in the past year said they thought they needed treatment and would have sought it if it were more readily available. Compared to several other states, that percentage was relatively low. Applied to the state's population, the study estimated that 1,204 North Dakota residents needed and wanted treatment in the past year but did not obtain it. This number would be a reasonable target for providing additional services, if the State sought to provide treatment on demand. Experience in other states suggests that survey estimates of unmet demand have successfully predicted the utilization of new substance abuse treatment services. This success appeared to depend on the type of treatment and location of the services in areas that clearly had relatively high levels of unmet need. If the State increased the number of people in need who obtained treatment by 1,204, the number who received treatment would increase by 43%. The total number who would receive care (4,030) would be 13.1% of the 30,880 who needed it.

Analysis of the telephone survey data showed that about one in five of the subjects who needed treatment and had not obtained it but wanted it should receive residential or hospital care in accordance with the patient placement criteria of the American Society for Addiction Medicine (ASAM). The remaining subjects should receive intensive outpatient treatment.

When asked what prevented them from obtaining treatment, North Dakota telephone survey respondents were most likely to cite lack of insurance, facilities being located too far away, programs being full, and lack of ancillary services such as child care or medical care.

Location of Treatment Needs

The authors compared the average annual treatment admissions rate for 1994-1998 (State data) with the SNI to determine how well the observed regional treatment admissions rates compared to the rates predicted by the Substance Abuse Need Index. In general, the existing distribution of treatment resources in North Dakota reflects relative need among regions reasonably well. Forty-seven percent of the variation in client rates among regions was explained by the SNI scores. Region V, the most populous region in North Dakota, had the largest gap (the observed average annual admissions minus the average admissions expected on the basis of need).

To allocate services geographically to meet the needs of the 1,204 persons with unmet demand, the authors used the SNI to ensure that all regions would have a treatment service rate consistent with its level of need . Because the current regional admissions rates already matched need reasonably well and because serving 1,204 more persons represents a substantial increase in the number of people served statewide, the analysis allocated some additional services to all regions.

The authors recommend that the State consider using the results of this analysis as one part of its decision making process for allocating services if additional funds become available. Although the authors found that the indicator data at the regional level to were reliable and valid, no single measure should be relied on in isolation. Accordingly, the estimates should be used along with other qualitative and quantitative information (e.g., knowledge of waiting lists in specific areas or concerns by other medical personnel or social agencies regarding the availability of specific services). Responses of local providers to the reasonableness of the estimates should also be considered. The social indicator methodology has been developed over a period of years, and has been used in other states. Whenever it is employed in a new state for the first time, there is always the possibility that modification must be made to refine the indexes.

Conclusions

The results of the needs assessment suggest that North Dakota would be justified in expanding its treatment services. The analysis of national, interstate, longitudinal data, and crossectional survey data produced evidence that a substantial number of the State's residents had an active addictive disease in the past year, but only a small percentage of them received treatment in the past year. While many of those individuals would probably not seek treatment immediately if the supply of services were increased, an estimated 1,204 people indicated that they wanted treatment even though they did not obtain it. Only experience will show how many of even that group will seek care, but the number is sufficiently large to suggest that an increase in the number of facilities would be reasonable. Recent statistics suggested that the treatment gap, especially regarding drugs, has been widening, and a reversal of that trend appears to be in order.

The analysis suggested that the State may wish to consider programming (e.g., outreach) directed towards increasing the proportion of persons in need who actually seek treatment. The persons who said that they wanted treatment was relatively small, and this group, especially in high-risk groups such as prisoners-to-be and homeless people, appeared to need relatively high levels of care, mostly residential and hospital treatment at the onset of treatment. Many of the household residents who wanted treatment appear to need intensive outpatient treatment to initiate treatment. Research suggests that location of future services in accordance with the indicators of unmet need, especially in rural areas, may be a key step for increasing the demand for treatment. Several administrative changes, such as reducing red tape, could make a difference. To increase access to treatment in rural areas, especially for youth, the State may wish to investigate the feasibility and efficacy of online counseling, assessment, and referral. Analysis of survey data from other states indicated that adolescents obtain a large proportion of treatment services from nonspecialty providers (e.g., clergy, school health counselors, general psychological counselors, and social workers). An important consideration for youth and residents of small towns and rural areas is the stigma attached to obtaining treatment services from specialty providers. A recent report by the National Center for Addiction and Substance Abuse suggests that relatively few of some nonspecialty providers such as clergy have received substance abuse training. Of course, attention to cultural issues and identification is important for American Indians.

The integrated analysis indicated several areas for which additional needs assessment research should be considered. The need indexes developed for the study should be kept up to date and refined. A commitment to ongoing data collection and updating of the social indicator data each year could provide the State with timely data for future planning. The study had to estimate the treatment needs of homeless and adolescents from studies conducted in other states. Those are two groups the state may consider studying in future rounds of the State Treatment Needs Assessment Program.

Introduction

This final report describes the integrated results of a family of studies of the substance use disorder treatment needs of North Dakota's citizens, especially those who are most in need of services. Employing funds from the Center for Substance Abuse Treatment (CSAT), State officials contracted with the National Technical Center (NTC) of the North Charles Research and Planning Group (NCRPG) to conduct this study.

Purpose of the Study

A primary objective of the study was to provide the State with the data it needs for its planning process. To assist the State in obtaining essential needs assessment data in a form that is most useful for the planning process, the integrated analysis made use of the needs assessment and resource data collected by the State in two rounds of needs assessment studies. The analysis can serve as a model for a systematic assessment of the adequacy of the current population treatment needs and services.

Problems and Issues

When conducting the comprehensive needs assessment for treatment of substance abuse, the authors sought to answer three basic planning questions:

• a. How many people were in need of treatment in the State during the past year? The goal was to have an adequate supply of services to meet the absolute level of demand that these cases would produce.
• b. Where should services be located? The goal is to locate the services where they are needed most.
• c. What mix of treatment modalities do these clients need and want? The goal is to have the optimal mix of treatment services to achieve maximal effectiveness and efficiency.

The study team examined data on treatment needs. It compared those needs with current resources in amount, type and location. It also examined special service delivery issues, such as the barriers to treatment service delivery in the large rural areas of the State and the service needs of such special populations such as women, American Indians, prisoners, and the homeless. The analysis used this information as a basis for recommendations regarding the gaps between treatment need and utilization.

Background

The roots of this comprehensive population-based study can be found in the recommendations of the Institute of Medicine's landmark study, Treating Drug Problems (Gerstein and Harwood 1990). The study recommended that each state conduct studies that produce objective estimates of need and use the resulting data to prepare a plan that should be the basis of the Substance Abuse Prevention and Treatment Block Grant application. This recommendation stemmed from a growing body of literature on substance abuse and mental health planning (Ford 1997; Frank 1985; Ingram 1988; Kimmel 1993; McKillip 1992; Maddock et al. 1988; NIAAA 1981; Goldsmith et al. 1992; Richards 1985; Ryan 1984-85; Schlesinger et al. 1994; Shapiro et al. 1985; Simeone et al. 1993; Soriano 1995; Wallack 1994; Warheit et al. 1977; Wilson and Hearne 1986; Cochran et al. 1997; Lo and Stephens 2000; SAMHSA 1997; CSAT 1999). Moreover, there is increasing recognition in the social sciences of the importance of needs assessment in education, health, and social services (Witkin and Altshuld 1995; Soriano 1995). These important works describe the basic epidemiological needs assessment methodologies to be used in this comprehensive study.

The basic needs assessment research model was described by the NTC in its telephone survey monograph (McAuliffe et al. 1995). New York pioneered this approach to treatment needs assessment (Frank 1985; Simeone et al. 1993; Welte and Barnes 1995). North Charles developed a similar statewide treatment services plan for Rhode Island a decade ago (McAuliffe et al. 1991). Rhode Island continues to use the plan for a broad range of policy and planning purposes, and Rhode Island currently has one of the most adequate supplies of treatment services in the country (McAuliffe et al. 1999a).

North Dakota's Needs Assessment Studies

North Dakota conducted two rounds of needs assessment studies. The first round of studies included a household telephone survey, a social indicator study, a survey of American Indians on reservations, and an integrated analysis. The Adult Household Survey (Gallup 1998a) interviewed 6,814 North Dakota residents aged 18 and older. The survey revealed that North Dakota adults had a low rate of substance use disorders relative to other states. Only 5.2% of the sample had a current substance use disorder (abuse or dependence). While 7.8% of those with a current treatment need received some in the past year, only 2% of those with a need that did not receive treatment in the past year expressed a desire for treatment. North Dakota's second Round One study was a social indicator study (Kraft 2000). The study used common indicators of substance abuse and treatment data to identify which service areas were in greatest need. The third Round One study was a face-to-face survey of American Indian adults on reservations that used the same instrument as the household survey (Gallup 1998b). The Round One studies also included an integrated analysis, however, the only portion of the need estimate from that study that had a diagnostic assessment was the household portion (Johnson, Bassin, and Shaw 1999).

The Round Two studies included a social indicator study, and the present integrated needs assessment. The social indicator study showed that North Dakotans suffered the lowest rate of social problems due to drug abuse in the United States (McAuliffe et al. 1999b). North Dakota ranked 14^th nationally in an index of alcohol abuse indicators. In a combined substance treatment need index, North Dakota ranked 24^th nationally. Analysis of the indicators within the state revealed that Human Service Regions I and III had the highest scores in an alcohol treatment need index.

The second Round Two study is this integrated treatment needs assessment. This final report seeks to combine previous estimates of treatment need for mutually exclusive population subgroups and to develop estimates for any groups not covered by one of the State's needs assessment studies in order to obtain a comprehensive, statewide estimate of need. This report will describe the estimates of need for each population subgroup including adults in households with telephones, people aged 12 and older in households without telephones, prisoners and juvenile detainees, the homeless, and adolescents. The report combines the needs of these groups to estimate total need and assesses the performance of the State's treatment system based on an analysis of the gap between need for treatment and treatment delivered (see chapter on gap analysis).

Organization of the Final Report

The next part of the report includes a review of the study's overall methodology, a chapter on historical trends in substance abuse needs and treatment services at the national level, and an analysis of how North Dakota's treatment needs compare to the treatment needs of other states. Those chapters are followed by a series of chapters that focus on the treatment needs of non-overlapping segments of the State's population. The first chapter presents the results for people in households with telephones, and the next chapter estimates the needs of residents of households without telephones. Another chapter addresses the needs of North Dakota's American Indian population (this population is not a component of the overall need estimate). The remaining chapters in the series address the needs of prisoners and the homeless. The next chapter combines the results for all subpopulations. The analysis examines the integrated statewide past-year need for treatment, the percentage of people in need who received treatment, and the unmet demand for treatment. The integrated estimate of treatment need for the entire state population will be used to identify the gap between treatment need and services provided at the statewide level. The North Dakota substance abuse indicator model is then used to distribute the needs across the State's regions. The penultimate chapter describes the results for the levels of care of those who need and want treatment, and it describes their own preferences for additional services. The last chapter of the report summarizes the results and presents a series of recommendations based on them.

References

Center for Substance Abuse Treatment (CSAT) and Substance Abuse and Mental Health Services Administration (SAMHSA). (1999). Proceedings of the 1999 Annual State Needs Assessment Program. Rockville, MD: U.S. Department of Health and Human Services.

Cochran, Deborah C., Sonia A. Alemagno, Thomas E. Feucht, Richard C. Stephens, Stephanie A. Wolfe, and John M. Butts. (1997). Ohio Needs Assessment Study Special Population Study: Homeless Shelter Pilot Study Report. Cleveland, OH: NOVA Research Company.

Ford, William E. (1997). "Perspective on the Integration of Substance User Needs Assessment and Treatment Planning." Substance Use and Misuse. 32(3): 343-349.

Frank, Blanche. (1985). "Telephone Surveying for Drug Abuse: Methodological Issues and An Application." Self-Report Methods of Estimating Drug Use: Meeting Current Challenges to Validity. Eds., Beatrice A. Rouse, Nicholas J. Kozel, and Louise G. Richards. Rockville, MD: National Institute on Drug Abuse. 71-82.

The Gallup Organization. (1998a). North Dakota Adult Household Survey. North Dakota Division of Mental Health and Substance Abuse Services.

The Gallup Organization. (1998b). Demand and Needs Assessment Study of Alcohol and Other Drugs Among Native American Indians Living on Reservations in North Dakota. North Dakota Department of Human Services, Division of Mental Health and Substance Abuse Services.

Gerstein, Dean R. and Henrick J. Harwood, Eds. (1990). Treating Drug Problems: A Study of the Evolution, Effectiveness, and Financing of Public and Private Drug Treatment Systems. Washington, D.C.: National Academy Press.

Goldsmith, Harold F., Roger A. Bell, and George J. Warheit. (1992). "Indirect Needs Assessment for Mental Health Services Planning: Introduction to This Special Issue." Evaluation and Program Planning. 15(2): 111-113.

Ingram, Jerry J. (1988). "Alcoholism Treatment Demand Estimation." Health Marketing Quarterly. 6(1-3): 195-205.

Johnson, Bassin, and Shaw, Inc. (1999). North Dakota Integrated Analysis Report. North Dakota Department of human Services, Division of Mental Health and Substance Abuse Services.

Kimmel, Wayne A. (1993). Need, Demand, and Problem Assessment for Substance Abuse Services. Rockville, MD: U.S. Department of Health and Human Services. 93-1741.

Kraft, Kathy, Robert Woodle, and Sue Tohm. (2000). North Dakota Substance Abuse Indicator Study. North Dakota Department of Human Services, Division of Mental Health and Substance Abuse Services.

Lo, Cecilia C. and Richard C. Stephens. (2000). "Drugs and Prisoners: Treatment Needs on Entering Prison." American Journal of Drug and Alcohol Abuse. 26(2): 229-245.

Maddock, John M., Dennis C. Daley, and Howard B. Moss. (1988). "A Practical Approach to Needs Assessment for Chemical Dependency Programs." Journal of Substance Abuse Treatment. 5(2): 105-111.

McAuliffe, William E., Paul Breer, Nancy White Ahmadifar, and Cathie Spino. (1991). "Assessment of Drug Abuser Treatment Needs in Rhode Island." American Journal of Public Health. 81(3): 365-371.

McAuliffe, William E., Richard A. LaBrie, Norah Mulvaney, Howard J. Shaffer, Stephanie Geller, Elizabeth A. Fournier, Eliot B. Levine, Qiuyun Wang, Susan M. Wortman, and Kathleen A. Miller. (1995). Assessment of Substance Dependence Treatment Needs A Telephone Survey Manual and Questionnaire, Revised Edition. Cambridge, MA: National Technical Center on Substance Abuse Needs Assessment.

McAuliffe, William E., Richard A. LaBrie, Nicoletta Lomuto, Rebecca Betjemann, and Elizabeth A. Fournier. (1999a). " Measuring Interstate Variations in Drug Problems." Drug and Alcohol Dependence. 53(2): 125-145.
McAuliffe, William E., Richard LaBrie, and Ryan Woodworth. (1999b). An Interstate Substance Abuse Indicator Chartbook. Cambridge, MA: North Charles Research and Planning Group.

McKillip, J. (1992). "On Defining Need." Needs Assessment Notes. 4(1): 2-3.

National Institute on Alcohol Abuse and Alcoholism. (1981). Current Practices in Alcoholism Treatment Needs Estimation. Rockville, MD: National Institute on Alcohol Abuse and Alcoholism.

Richards, Louise G. (1985). Drug Abuse Epidemiologic and Needs Assessment Approaches by States and Small Areas: A State of the Art Review. Rockville, MD: National Institute on Drug Abuse.

Ryan, Keith. (1984-1985). "Assessment of Need for Alcoholism Treatment Services: Planning Procedures." Alcohol Health and Research World. 9(2): 37-44.
Schlesinger, Mark, Robert A. Dorwart, Sherrie Epstein, and Jonathan Sablone. (1994). "The Mismeasure of Need: The Social Costs of Substance Abuse." Unpublished manuscript.
Shapiro, Sam, E. A. Skinner, Morton Kramer, D. M. Steinwachs, Darrel A. Regier. (1985). "Measuring Need for Mental Health Services in a General Population." Medical Care. 9: 1033-1043.
Simeone, Ronald S., Blanche Frank, and Zahra Aryan. (1993). "Needs Assessment in Substance Misuse: A Comparison of Approaches and Case Study." International Journal of the Addictions. 28(8): 767-792.

Soriano, Fernando I. (1995). Conducting Needs Assessments: A Multidisciplinary Approach. London: Sage Publications.

Substance Abuse and Mental Health Services Administration (SAMHSA), Office of Applied Studies and Research Triangle Institute. (1997). National Household Survey on Drug Abuse: Population Estimates 1996. Rockville, MD: U.S. Department of Health and Human Services.

Wallack, Stanley S. (1994). Resource Materials for State Needs Assessment Studies. Rockville, MD: U. S. Department of Health and Human Services.

Warheit, George J., Roger A. Bell, and John J. Schwab. (1977). Needs Assessment Approaches: Concepts and Methods. Washington, D.C.: Department of Health, Education, and Welfare.

Welte, John W. and Grace M. Barnes. (1995). Alcohol and Other Drug Use Among Hispanics in New York State. Alcoholism Clinical and Experimental Research. 19 (4):1061-1066.

Wilson, Robert A. and Barbara E. Hearne. (1986). "An Assessment of the State of the Art in Drug Abuse and Alcoholism Treatment Needs Estimation Methods." Treating the Drug User: Selected Planning Models, Issues, Parameters, and Programs. Ed. Stanley Einstein. Danbury, CT: Sandoz Publications. 186-221.

Witkin, Belle R. and James W. Altschuld. (1995). Planning and Conducting Needs Assessments: A Practical Guide. London: Sage Publications.

Methods

This chapter describes the study's research methodology. Development of an integrated needs assessment and plan requires a broad understanding of the treatment needs of the population (results of the family of studies), the resources currently available (from the utilization studies), and the policy and technical contexts in which the treatment system will be functioning in the coming years. Bringing together these three elements was the object of data collection and analysis for this study. To obtain this information, the study team used a series of methodologies, each selected to address a specific component in the process. The steps described below include: a review of the recent developments in substance abuse epidemiology, analysis of how North Dakota's services per unit of need compare to other states, background interviews with State substance abuse officials, a review of other relevant North Dakota studies and substance abuse literature, analyses of needs assessment data and findings from treatment utilization statistics, analyses of census data and prevalence statistics on special populations from the first round of studies and the literature, gap analyses comparing need and resource data, and analysis of the State's treatment system. In applying each of these methodologies, the study team attempted to determine the overall statewide level of need, substate area needs, and the appropriate treatment allocation. The specific questions addressed by each of the methods are described in Table 2.1. In fashioning plans to address the needs that have developed, the study team sought a clear picture of North Dakota's substance abuse treatment goals, what has been tried in the past, and how the treatment system will change in the future to address existing needs.

Description of Specific Methods

This comprehensive study will include a series of substudies described in Table 2.1. This section describes the methods employed in each of these study components:

A conference call with key officials

Review of prior related studies

Review of relevant literature and reports to analyze the substance abuse context: epidemics, long-term trends in treatment, and financing

Analysis of social indicator data

Analysis of interstate data on needs and services comparing North Dakota to other states

Integration of estimates from non-overlapping segments of the population

Analysis of needs and resources to identify gaps in amount of services, types of services, and location of services.

Any research, including a treatment needs assessment, should begin with hypotheses about what the study is likely to find or what the key questions are. Only then can one be sure that all of the essential data will be collected in the required form and all irrelevant data, no matter how "interesting" from an academic perspective, will be excluded. The hypotheses also inform the analysts of precisely what questions to ask of the data once it has been collected. The "Background Studies" in Table 2.1 are designed to be hypothesis generating.

Background Interviews

The study team conducted telephone interviews with agency officials who are knowledgeable about North Dakota's substance abuse treatment system and about groups that utilize the system (e.g. prisoners, homeless). Their perspectives and day-to-day experiences working within or with the treatment system were crucial to understanding how the system functions. Learning their ideas for change was critical for formulating recommendations. The interviews were semi-structured, with a series of questions devised by the study team prior to the telephone calls.

The calls covered the officials' roles in the system, perceptions of the system's performance in meeting the needs of North Dakotans, perspectives and recommendations for change, coordination issues, and service gaps. The objective of the calls was to develop specific recommendations about the treatment system's functioning.

Review of Prior Studies and Documents

North Dakota has conducted a number of previous studies and produced planning reports that served as background for the study. NCRPG obtained these studies and used them for this report.

Literature Review

Although every state is unique to some degree, the substance abuse problem is national in scope. Identifying relevant trends can help organize a range of facts. This understanding is critical when using the resulting data to develop a treatment needs assessment and plan for the State. Current trends that were important include the up-tick in use of drugs and the decline in alcohol use, increases in substance use disorders among prisoners, and widening of the treatment gap.

Social Indicator Study

NCRPG conducted a social indicator analysis to guide allocation of resources over substate regions for which the State's survey estimates are too imprecise (see McAuliffe et al. 1987, 1991 and Folsom et al. 1996 for a description of this use). Social indicator studies complement other studies using different methodologies (thus increasing validity) and thereby serve to strengthen a state's credibility when attempting to allocate substance abuse treatment resources equitably. Social indicator analyses are especially effective when used in conjunction with survey data. Because social indicator analyses provide relative differences, rather than absolute counts, household survey data can be used to provide a baseline from which to calibrate estimates for actual counts of need. Due to the small number of cases in substate areas, the combined telephone and supplementary surveys which comprise most of the family of studies are less reliable at the substate level than they are at the State level. As a result, the survey data may not be as effective for distributing resources over subareas as they are for estimating statewide need. Consequently, the social indicator study will be employed to supplement survey estimates and to estimate substate level estimates more reliably.

The study began with the selection of a small number of measures of need from different data sources or systems (i.e., substance-abuse related deaths and arrests). Data on substance-abuse related treatment admissions were also collected. Indicator selection is based on the measures' theoretical relationships with substance abuse treatment need, the results of previous validation studies, and data availability. County-level data were obtained for the selected variables. The data were cleaned, entered into a database, and then subjected to rigorous empirical reliability and validity testing.

The rationale for the initial selection of a small number of variables was that they are more manageable (e.g., can be cleaned and validated individually), explain most of the relevant variance in treatment service needs, and are more easily interpreted by both investigators as well as State officials and the public (Dembling et al. 1993; McAuliffe et al.1999). Experience in Rhode Island clearly indicates that the public will be interested in the validity of the index when the resulting resource allocations affect the relative availability of services in the State's substate areas (Breer et al. 1996).

Empirical research by the NTC (McAuliffe et al.1999) and by other states (Aktan, Calkins et al. 1997) has shown that the geographic location of alcohol problems often differs from the location of other drug problems. Analyzing individual indicators of both alcohol and other drugs together can obscure important information for planning purposes (e.g., where to locate drug-specific programs versus where to locate alcohol detox facilities).

Interstate Comparisons

It is useful to compare the State's treatment needs with those of other states in order to have a basis for evaluating the adequacy of its treatment services. Recent examples of interstate alcohol and drug treatment comparisons include McAuliffe et al.(1999; 2000). Following the analysis reported by Dayhoff et al. (1994) with regard to alcohol treatment services for all states, this study analyzed interstate differences with regard to both alcohol and drug treatment.

Uncovered Population Estimates

NCRPG has estimated that 90% of all people with substance use disorders nationwide live in households with telephones (Geller 1995). As a result, the backbone of the NCRPG model family of studies is the specially designed household telephone survey conducted by North Dakota in its first round of studies. Nevertheless, there are important treatment populations in which the percentage living in households with telephones is smaller. A recent face-to-face survey of household and nonhousehold residents of the Washington, DC metropolitan area revealed that failing to include nonhousehold populations would have little impact on the overall rates of illicit drug use, but it would result in a 20% underestimation of past year heroin and cocaine users (Gfroerer 1996).The Epidemiologic Catchment Area Survey (ECA) studies had previously included studies of nonhousehold populations to respond to this problem.

North Dakota surveyed American Indians living on reservations in order to obtain estimates for that group that were not biased due to low telephone coverage. One objective of the integration study is to employ data from such studies in order to reinforce the more comprehensive estimates of treatment need outlined below. Because all relevant groups were not covered in the North Dakota family of studies, the current study team developed estimates of the State's need estimates by obtaining statistics on the prevalence rates for the omitted groups and obtaining estimates of the size of the groups from census data (Geller et al. 1997; McAuliffe et al. 1998). The data to be used for this estimation process include estimates of numbers of adults in households without phones, adolescents, and the homeless. In addition to its own surveys of special populations such as the homeless (recently completed in Rhode Island), the study team conducted a series of literature reviews to obtain estimates of the prevalence of substance abuse treatment need for these groups from a literature search of such major databases as MEDLINE, PSYCHINFO, and HEALTHSTAR, adjusting the estimates so that they capture the correct age range and both past-year and lifetime prevalence. The number of persons in need of substance abuse treatment in each uncovered population was estimated by multiplying the adjusted prevalence rates for each of these populations by the number of members of these populations in North Dakota.

Gap Analysis

A central feature of this study was to bring together the State's data on treatment needs and resources. Many important questions about subgroups, substate areas, and the State as a whole will be addressed concerning the amount, type, and location of treatment needs and services. For example, what percentage of the population needed treatment, and what proportion of the people in need wanted to obtain services? What reasons did they cite that prevented them from obtaining treatment? Finally, the analysis identified the gaps in services by comparing the need and demand for services with the services that were available.

Conclusions and Recommendations

The report includes proposals regarding how the treatment system should respond to fill the gaps in treatment and meet the challenges of the coming years. The recommendations stemmed from analysis of the estimates of treatment gaps, the State's service mix, and the perspectives of State officials.

References

Aktan, Georgia B., Richard F. Calkins, Rafa M. Kasim, Sandra Kimball, and Karen Schrock. (1997). Social Indicators Modeling for Substance Abuse Treatment Needs Assessment Substance Abuse Need Index (SANI). Michigan Department of Community Health.

Breer, Paul, William E. McAuliffe, and Eliot B. Levine. (1996). "Statewide substance abuse prevention planning." Evaluation Review. 20(5): 596-618.

Dayhoff, D.A., G.C. Pope, and J.H. Huber. (1994). "State Variations in Public and Private Alcoholism Treatment at Specialty Substance Abuse Treatment Facilities." Journal of Studies on Alcohol. 55(5): 549-560.

Dembling, Bruce. (1993). Distribution of Consumer Demand for a Public Psychiatric Hospital: Application of a Geographic Information System. National Conference on Mental Health Statistics.

Folsom, Ralph E., Judith T. Lessler, Michael B. Witt, Joseph C. Gfroerer, Douglas A. Wright, and Joseph Gustin. (1996). Substance Abuse in States and Metropolitan Areas: Model Based Estimates from the 1991-1993 National Household Surveys on Drug Abuse, Summary Report. Substance Abuse and Mental Health Services Administration (SAMHSA), Office of Applied Studies (OAS). Rockville, MD: U.S. Department of Health and Human Services.

Geller, Stephanie. (1995). "Supplementary Studies: Introduction" Proceedings of the 1995 Newly Funded States Workshop: Measuring Need for Substance Abuse Treatment and the Implications of Managed Care. Cambridge, MA: National Technical Center (NTC).

Geller, Stephanie, W.E. McAuliffe, and P.A. Minugh. (1997). Bulletin #14: Combining Subpopulation Estimates to Develop Comprehensive Estimates of Treatment Need. Cambridge, MA: National Technical Center (NTC).

Gfroerer, J.C. (1996). "Special Populations, Sensitive Issues, and The Use of Computer-assisted Interviewing in Surveys." Health Survey Research Methods: Conference Proceedings. Hyattsville, MD: U.S. Department of Health and Human Services. 177-180.

McAuliffe, W.E., P. Breer, N. White, C. Spino, L. Goldsmith, S. Robel, and L. Byam. (1987). A Drug Abuse Treatment and Intervention Plan for Rhode Island: Review Copy. Cranston, Rhode Island: Rhode Island Department of Mental Health, Retardation and Hospitals, Division of Substance Abuse.

McAuliffe, W.E., P. Breer, N.W. Ahmadifar, and C. Spino. (1991). "Assessment of Drug Abuser Treatment Needs in Rhode Island." American Journal of Public Health. 81(3): 365-371.

McAuliffe, W.E., S. Geller, R.A. LaBrie, S.B.F. Paletz, and E.A. Fournier. (1998). "Are Telephone Surveys Suitable for Studying Substance Abuse Epidemiology? Cost, Administration, Coverage and Response Rate Issues." Journal of Drug Issues. 28(2): 455-482.

McAuliffe, W.E., R.A. LaBrie, N. Pollock, N. Lomuto, E.A. Fournier, R. Betjemann. (1999). "Measuring Interstate Variations in Drug Abuse." Drug and Alcohol Dependence. 53(2): 125-45.

McAuliffe, William E., Richard A. LaBrie, Nicoletta A. Lomuto, Nancy E. Pollock, Rebecca Betjemann, and Elizabeth Fournier. (2000). "Measuring Interstate Variations in Problems Related to Alcohol Use Disorders." Eds. Robert A. Wilson and Mary C. Dufour. The Epidemiology of Alcohol Problems in Small Geographic Areas. First ed. Bethesda, MD: National Institute on Alcohol Abuse and Alcoholism (NIAAA). 213-244.

National Trends

This chapter examines epidemiological evidence regarding national trends that affect individual states. As the next chapter will show, every state has its unique history and profile with regard to alcohol and drug abuse. However, history reveals convincingly that most substance abuse problems have a national or regional basis (Hunt 1974). The national heroin addiction epidemic of the 1950s got its start in Harlem (Brown 1965), and the psychedelic epidemic of the 1960s and 1970s began at Harvard and in California. Cocaine gained in popularity as a national media phenomena. The AIDS epidemic among injection drug users was first detected in New York, New Jersey, and Maryland (McAuliffe and Ackerman 1991). Only several years later did the prevalence of HIV become painfully evident in other states away from those epicenters. In the past decade, similar national trends in adolescent drug use and use of "club drugs," especially "ecstasy," have spread across the country (Community Epidemiology Work Group 2001). While regions and states may vary in their vulnerability to specific epidemiological trends, states can ill afford to ignore these developments. In the present context, these national trends help explain the current treatment needs in individual states. The remaining sections will describe national trends in alcohol and drug abuse and services over the past decade. The Data Sources appendix presents the citations for the sources of the statistics presented in this chapter.

Alcohol Treatment Needs

Indicators of alcohol use, dependence, and adverse consequences over the last decade suggest that the need for alcohol treatment services has not increased, except possibly among adolescents. Measures of national consumption of alcohol peaked at the beginning of the 1980s, and then diminished thereafter (Office of Applied Studies 2000c; Greenfield et al. 2000). Per capita alcohol consumption (gallons of ethanol per person 14 and older) declined from 1990 to 1995, and it remained more or less stable thereafter (Figure 3.1; Nephew et al. 2000). As shown in Figure 3.2, the percentage of National Household Survey on Drug Abuse (NHSDA) respondents who drank alcohol in the past year also tapered off throughout the late 1980s to 1998. An even steeper decline occurred among the survey's adolescent respondents.

Starting in 1985, survey estimates of heavy drinking in the past month (Figure 3.3) declined gradually, although that rate increased somewhat in 1998 (Office of Applied Studies 1999a, p. 34; 2000c, p. 31). Heavy drinking by the NHSDA's adolescent respondents increased slightly between 1994 and 1995, after a sharp decline in the late 1980s and early 1990s (Office of Applied Studies 2000c).

The rate of alcohol dependence symptoms in the 1984, 1990, and 1995 National Alcoholism Surveys increased slightly over time but there were no significant differences between it and the NHSDA (Figure 3.4; Midanik & Greenfield 2000). Because the NHSDA Main Findings include only rates of alcohol dependence symptoms, the authors analyzed NHSDA public-use data sets in order to find a dependence diagnosis rate. The rates of persons having three symptoms of dependence were higher than the diagnosis rates, but both held to the same upward trend between 1995 and 1998, the years for which the public-use data sets are available. Because the survey questions on symptoms changed between 1994 and 1995 and because the 1999 Main Findings focused on the dependence diagnosis rate rather than symptoms, Figure 3.4 does not include NHSDA estimates for years earlier than 1995 or later than 1998. The apparent differences between the National Alcoholism Survey and the NHSDA estimates, which is not unlike differences in the prevalence rates reported by the Epidemiological Catchment Area study and the National Comorbidity Survey, raises questions about how useful these survey estimates of alcohol dependence are for trend analysis and policy making (Regier et al. 1998).

Among the arrest categories that may be referred to as "alcohol-defined," the rates of arrests for driving under the influence (DUI) and for drunkenness showed the sharpest decreases (24%) between 1991 and 1998 (Figure 3.5). Disorderly conduct arrest rates, which often include arrests for drunkenness ( "drunk and disorderly"), were also down (9%) in that period. The one exception to this trend was the rate of arrests for liquor law violations, which increased 10% between 1991 and 1998. Between 1993 and 1996 liquor law violation arrests increased 32%. Liquor law arrests in part reflect underage drinking, which surveys suggested began increasing in 1995.

Figure 3.6 shows that alcohol mortality rates fell between 1989 and 1999. Three different measures displayed this pattern. A clear reduction is evident in the Centers for Disease Control's (CDC) "alcohol-induced" underlying cause mortality rates. Similarly, the National Institute on Alcohol Abuse and Alcoholism's (NIAAA) age-adjusted mortality rates with explicit-mention of alcohol as an underlying cause, which employs the same set of diagnoses as the CDC measure, sunk from 7.3 per 100,000 to 6.4 per 100,000 over the same period. While the age-adjusted rates are consistently lower than the crude rates, they follow virtually the same trend, suggesting that age alone cannot account for the drop. Finally, alcohol-related motor vehicle deaths (defined as alcohol-related by the investigating officer's judgment or a positive reading from any one of several blood tests), reported by the NIAAA from Fatal Accident Reporting System (FARS) data, dropped even more sharply than the other two indicators (Yi et al. 2000). Accidental deaths are not included in the explicit-mention indicators. These results are consistent with the sharp drop in DUI arrests just described.

Alcohol Treatment

Private spending for alcohol treatment went up 250% in the decade between 1979 and 1989 (Dayhoff et al. 2000). Across the country, the percentage of employees at large and medium-sized firms who had coverage for hospital detoxification increased two and one-half times from 37% in 1982 to 96% in 1989 (Dayhoff et al. 2000). Dayhoff et al. (2000) reported that national public spending for alcohol services increased 60% in the decade between 1979 and 1989. In the early 1990s, total public spending for alcohol and drug specialty treatment services continued increasing by 5%, largely due to the broadening of Medicaid coverage of substance abuse treatment and a 25% increase in the federal Alcohol, Drug and Mental Health Services Block Grant. However, individual state and local government spending for substance abuse services declined (16% and 31% respectively) in the first four years of the 1990s, thus dampening the continuing trend towards greater availability of public treatment services.

Three measures of the amount of alcohol treatment services from the OAS indicated that national alcohol treatment rates descended steadily from 1992 to 1999 (Figure 3.7). The three measures were primary alcohol treatment admissions (public specialty) from the Treatment Episode Data Set (TEDS), alcohol-only treatment clients (public and private specialty) from the National Drug and Alcohol Treatment Unit Survey (NDATUS)/Uniform Facilities Data Set (UFDS), and the percentage of NHSDA respondents that reported receiving treatment for alcohol use in the past year. The context in which the NHSDA's treatment question is asked in the national survey's interview suggests that "treatment" may include the specialty treatment measured in the UFDS and TEDS estimates as well as treatment provided in an emergency room, private doctor's office, prison, or self help group.

While the TEDS admission rate and the NDATUS/UFDS alcohol-only client rate fell steadily throughout the period, the NHSDA treatment statistics described a much steeper decline. It is also noteworthy that the estimated number of clients in the NHSDA series was much higher than the NDATUS/UFDS series that includes both private and public services.

Alcohol Treatment Gap

Although the alcohol treatment need and service indicators were not always perfectly in step, they mostly indicated stabilization or decline in alcohol use, problems, and services. To measure the relative changes between the measures of treatment need and treatment services, the authors divided the service rates by the need indicators. The ratio of treatment to the percent alcohol dependent in the NHSDA declined from 1995 to 1998 regardless of the measure of services, although the extent of the widening of the treatment gap depended greatly on which measure of services the analysis employed (Figure 3.8). The ratio of the percentage of NHSDA respondents who received alcohol treatment in the past year relative to the number who had a current alcohol dependence diagnosis dropped from a high of .383 in 1996 to .216 two years later. The last number suggested that less than one respondent received treatment in the past year for every four respondents who had a current alcohol dependence diagnosis. Note that because of the way these gap ratios were designed - the treatment rates are in the denominator while the need indicators are in the denominator - a smaller ratio represents a wider treatment gap. So, in the gap analysis graphs (for example, Figure 3.8), a line that goes down over time shows a widening in the treatment gap. The present analysis does not indicate whether the respondents who received treatment were the same ones that had an alcohol dependence diagnosis. It is also noteworthy that the NHSDA's dependence measure is a point prevalence referring to the percentage who had a diagnosis at the time of the interview. It does not indicate how many people had a diagnosis at any time during the year. By contrast, the treatment measure describes the percentage of respondents who received treatment for alcohol at any time during the past year. Thus, this ratio is conservative and may underestimate the true alcohol treatment gap, but it seems likely that the trend is reasonably accurate.

When the measure of treatment was the rate of admissions for publicly-funded specialty alcohol treatment according to the TEDS database, the decline was from .103 in 1995 to .074 in 1998. The variation between these three measures suggests that part of the task in measuring the national gap in alcohol treatment is to determine how best to measure the amount of treatment services that were provided each year.

When the measure of treatment in the numerator of the ratio was the UFDS/NDATUS alcohol-only rate, the decline relative to the percent dependent was more gradual, going from a peak of .037 in 1995 to .024 in 1998. The alcohol-only NDATUS/UFDS statistics describe a point prevalence of both public and private specialty treatment, but this measure excludes people who received treatment for both alcohol and drug problems. In some years, the excluded persons with both drug and alcohol treatment needs were nearly equal to the number of alcohol-only clients, while in other years they were nearly twice as numerous as the alcohol-only clients.

When assessing the gap between treatment and dependence, it is important to consider the amount of treatment services delivered to those in need of treatment for alcohol dependence. One of the ratios of services to dependence cited above measures all services reported by NHSDA respondents relative to all respondents that had a dependence. That is why the ratio for NHSDA treatment services to NHSDA dependence is the highest. The other treatment measures, the UFDS and the TEDS, capture only clients at specialty treatment facilities where all clients are likely to have a diagnosis, whereas the NHSDA data capture people in non-specialty treatment that may or may not have had a diagnosis in the past year.

As shown in Figure 3.9, the percentage of those with a past year diagnosis for alcohol dependence that received treatment in the past year declined steadily between 1995 and 1997, then dropped by more than three percentage points over the next year. The percentage of respondents with a past year alcohol dependence that received treatment in the past year in 1998 (6.58%) was less than two-thirds the rate in 1995 (10.72%). That trend was consistent with the trend shown in Figure 3.8 that compared treatment rates with the dependence rate regardless of whether those who received treatment were dependent.

Comparison of the ratio between the NHSDA past-year alcohol treatment and NHSDA diagnosed past-year alcohol dependence with the rate of treatment among those with a dependence diagnosis revealed that there were many respondents in each year that did not have a dependence diagnosis, but received treatment for alcohol use. Consider that the gap ratio indicated that the number that received treatment was about 37% of the number with a dependence diagnosis in 1995, while the treatment rate among NHSDA respondents with a diagnosis was less than 11% in that year. There are several reasons for this disparity. First, about 71% of those who said they received treatment for alcohol in 1995 did not have a diagnosed alcohol dependence. The NHSDA data do not permit diagnoses of non-dependent abuse. Those people are considered to be in need of treatment and it is likely that many of those without a diagnosis for dependence that received treatment would have met the criteria for abuse. Second, the treatment question in the survey is designed to capture a broad range of treatment types including non-specialty modes such as, participation in self-help groups, counseling, and consultation with a private doctor. As such, the treatment group may include respondents that had a lifetime diagnosis of abuse or dependence, but were in full remission at the time of the survey. Third, a diagnosis is for the whole year, but the survey only counts from the time of the survey itself. For example, a respondent might have had a diagnosis 13 months prior to the survey and still manifested symptoms during the time of the survey, but had only one or two symptoms and would therefore, not be diagnosed as dependent when he should have been.

For those reasons, it is difficult to state with certainty whether treatment among those with a diagnosis is the estimate most relevant to needs assessment or if a comparison of treatment against need is the correct measure. The correct estimate probably lies somewhere in between. A correct treatment need estimate measures both dependence and abuse and evaluates respondents for a diagnosis over the whole period, rather than gauging only symptoms over the past year. To confirm the inferences derived from using the NHSDA measure of dependence in the denominator, the authors replaced it with arrest and mortality indicators. The first analysis combined alcohol arrest rates (DUI, drunkenness, and liquor law violation arrests) and divided the measure into treatment as measured by the NHSDA, the NDATUS/UFDS client survey, and the TEDS admissions data (Figure 3.10). As before, the extent of the increase in the treatment gap depended on which measure the analysis used. The gap grew sharply when treatment was measured by the NHSDA estimate of the number of persons who received treatment in the past year. When the numerator was the NDATUS/UFDS estimate of people in treatment on a given day and the denominator was the number of people who were arrested in the past year, the gap widened only gradually. When the TEDS treatment measure was used as the numerator of the gap ratio, the gap increased gradually starting in 1994 and stabilized in 1997.

When the CDC's alcohol induced crude mortality rate in a year was the denominator, the trends remained basically the same as for when the survey dependence estimates or arrest rates were in the denominator of the ratio of treatment to need (Figure 3.11). The major determinant appears to be how treatment is measured rather than how need is measured. With NHSDA estimates of the number of persons who received treatment in the past year in the numerator, the treatment gap widened sharply over the period. With the NDATUS/UFDS survey estimates of static capacity in the numerator and alcohol-induced mortality in the denominator, the treatment gap widened more gradually over the period. The ratio of the TEDS treatment admissions rate to the alcohol-related mortality rate indicated a trend in the treatment gap very similar to the trend indicated by the gap measure that employed the NDATUS/UFDS treatment estimate.

Drug Abuse Treatment Needs

National surveys found that use of most illicit drugs decreased sharply during the 1980s, but then drug use rates (including marijuana, cocaine, and heroin) increased during the early 1990s, especially among youths (OAS 2000c, p. 30). The consumption of cocaine in particular declined to much lower levels since the mid 1980s (OAS 1999a, p. 32). For example, the percentage of persons aged 18 to 25 who reported using cocaine in the past year declined from 14% in 1985 to 4% in 1997. That age cohort routinely had the highest rates of controlled drug use. Past-month use of marijuana by this age group declined from 22% in 1985 to 11% in 1992. Beginning in 1993, past-month use of marijuana increased slightly until 1997, when it declined slightly to a rate (13%) that was still well below the peak year of 1979 (37%) and below the 1985 rate (22%). Statistics from the Monitoring the Future survey of youth also found that marijuana use began increasing between 1993 and 1997 before tapering off in 1998 and 1999 (Office of National Drug Control Policy [ONDCP] 2000). Cocaine initiations also turned upward in early 1990s. Statistics from the NHSDA indicated that past-month heroin use increased from 1990 to 1997 to levels that exceeded 1985 before dropping back down to 1985 levels in 1998 (Epstein and Gfroerer 2001; ONDCP 2000). With some exceptions, current levels of nonmedical controlled drug use are lower than 1985 levels despite increases since 1992 (ONDCP 2000).

The survey self-reports of dropping drug use were confirmed by urine test results from employee testing. Figure 3.13 shows that the percentage of positive urine testing conducted by employers around the country dropped steadily from 13.6% in 1988 to 4.6% in 1999 (Quest Diagnostics, Inc. 2000). The increases in drug use, most evident among adolescents, has apparently not yet begun to affect employee testing data significantly, although some flattening of the rate of decline is apparent in 1997-1999.

Drug Dependence

The authors analyzed NHSDA public-use data sets to obtain rates of drug dependence for the years 1995 through 1998. In order to illustrate the extent to which that dependence measure is overwhelmingly composed of marijuana and cocaine users, the drug dependence measure is graphed alongside NHSDA estimates of the rate of the population 12 and older that experienced at least three problems due to use that are components of a dependence diagnosis (Figure 3.14). The questions on problems stemming from use changed between 1994 and 1995 and, as such, fluctuations between those years may be exaggerated. Clearly, the overall drug dependence rate rose between 1995 and 1996, then remained constant at 1,770 per 100,000 in the population thereafter. The rate of having three problems due to cocaine or marijuana use remained reasonably steady, as well.

The total number of drug-related emergency room episodes recorded in the Drug Abuse Warning Network (DAWN) has increased from 1990 (Figure 3.15; Caulkins 2001; OAS 2001; ONDCP 2000). The number of emergency room visits for drug-related episodes declined briefly in 1995, but then increased steadily from 1996 to 1999. The most common drug mentions continued to be cocaine-related problems (Caulkins 2001), apparently resulting from long-term consequences of chronic use according to the ONDCP (2000). In the mid 1990s, the consequences of growing use of "club drugs" (MDMA or "ecstasy" and GHB) as well as prescription opiates (hydrocodone and oxycodone) began to show up in DAWN emergency room statistics (Substance Abuse and Mental Health Services Administration [SAMHSA] 2001). The sharpest increases in mentions of the club drugs occurred in the period between 1998 and 2000.

Arrests for drug abuse violations rose nationally in the past 15 years. From 1985 to 1995, drug abuse violation arrests rose 82% (Federal Bureau of Investigation [FBI] 1996, p.280), and by 1999 drug arrests were 7% higher than the same statistics in 1995. In 1999, drug arrests were 36% higher than they were in 1990. By contrast, arrests were down 1% for all offenses, 9% for violent crimes, and 26% for property crimes during the same period (FBI 2000, p.211). As shown in Figure 3.16, Uniform Crime Reports (UCR) drug arrest rates increased steadily from 1991 to 1998, with a noticeable increase in 1994, which may be a methodological artifact resulting from the change over of many states to the FBI's new reporting system that permits tabulation of drug arrests that were not the most serious crime in the episode.

Drug Mortality

Drug mortality rates from three data sources suggest the same trend: up (Figure 3.17). The rate of explicit-mention drug deaths using the present authors' (North Charles Research and Planning Group's [NCRPG]) set of diagnoses from multiple cause data supplied by the National Center for Health Statistics (NCHS) of the CDC closely parallels the "drug-induced" mortality underlying cause rate published by the CDC. The primary difference between the two rates is the use of multiple causes in the NCRPG measure and inclusion of overdoses due to suicide, assault, and undetermined external causes for some drugs in the drug-induced measure ( CDC 1993). The authors also included a measure based on the DAWN medical examiner statistics and the total population. Readers are cautioned not to interpret the absolute level of the DAWN rates as comparable to the other rates, since the DAWN deaths come mainly from urban areas. Lacking the relevant population base for the DAWN statistics, the authors used the total population as a proxy for indicating changes in population over time. The DAWN medical examiner data stem from the investigations of a panel of medical examiners (138 up to 1995, and 134 thereafter) (see Data Sources). In 1996, estimates were available for both sets of panels, and the results did not differ when rounded to one decimal. The medical examiner data include deaths due to suicide (ONDCP 2000). The obvious correlation between the three series suggests the basic validity of all three measures, which is important given the limited geographic coverage of the DAWN medical examiners data.

It is noteworthy that the average purity of retail heroin increased in 1992 and 1993 and has remained at high levels ever since, and the price per gram declined steadily after 1991 (ONDCP 2000). It is likely that increases in purity and decreases in price contribute to the increase in the prevalence of use and to the prevalence of drug overdoses by users.

Summary of Drug Treatment Need Indicators. The drug indicators revealed some inconsistency, but the overall trends were reasonably clear. Employee-testing and self-reported overall drug use by persons 12 and older went down steadily during the 1980s and the first few years of the 1990s, but the past month drug use by adolescents, drug mortality, emergency room episodes, and arrest statistics for the entire population increased in the decade of the 1990s; especially during the middle of the decade.

Drug Abuse Treatment

Contrasting trends were apparent in the measures of treatment services received during the 1990s. All three measures are maintained by the OAS and are shown in Figure 3.18. Whereas the TEDS primary drug treatment admissions reported by states and the NDATU/UFDS drug-only treatment clients increased over the period, the NHSDA's estimates of the number of persons who received treatment in the past year declined after a spike upward in 1996. Federal funding of treatment increased by 35% between FY1996 and FY2001 (ONDCP 2000).

It is noteworthy that the absolute rates among the NDATUS/UFDS clients and the NHSDA's estimates of the number of clients differed by between seven and ten-fold in the period. This difference contrasts with the assertion by the NHSDA's top analysts that the "NHSDA undercounts . . . drug clients treated" (Woodward et al. 1997, p. 8). In previous research, the present authors found that the interstate NDATUS rates for 1991-1993 did not correlate with the NHSDA's estimates of treatment received in the same years (McAuliffe et al. 1999). The lack of correlation among the indicators raises questions about the planned use of the NHSDA to measure the drug treatment gap (Woodward et al. 1997).

Drug Treatment Gap

Due to the inconsistency between the measures of treatment services, the measures of treatment relative to indicators of treatment need were also inconsistent. Because the authors were unable to obtain published NHSDA estimates of dependence on any controlled drug for more than two years, this study uses measures created from the NHSDA's public-use databases for 1995-1998.

The ratio of the NHSDA's estimates of the percentage of respondents who received treatment in the past year to the NHSDA's estimate of the percentage of respondents who had a drug dependence diagnosis at the time of the interview indicated an increasing drug treatment gap. That is, the proportion of persons in need that received treatment declined. Recall that in the alcohol treatment gap analysis, a smaller ratio of treatment to need indicators indicated a larger treatment gap. Hence, a graph line that shows a treatment to need indicator ratio decreasing over time shows a widening of the treatment gap. The greatest dip was in 1998 for which there was a published drug dependence estimate. By contrast, the ratio of the NDATUS/UFDS drug-only client rate to the percent with a drug dependence between 1995 and 1998 and the TEDS primary drug treatment admission rate to the percent who were drug dependent between 1996 and 1998 indicated a slightly decreasing gap in treatment relative to need. The TEDS-based measure declined only slightly between 1997 and 1998 and the NDATUS/UFDS-based measure increased slightly between 1997 and 1998, but the NHSDA-based series declined sharply from a peak in 1996 (Figure 3.19). In most years the NHSDA-based series moved in the opposite direction from the TEDS measure. Because Woodward et al. (1997) indicated their own concerns about the NHSDA's estimates of treatment, readers may prefer to place greater reliance on the other two measures.

It is important to examine the extent to which those with a current dependence on illicit drugs receive treatment. In the treatment need literature this is referred to as "met demand" for treatment. The treatment data from the NHSDA used above do not discriminate between those that received treatment for a dependence in the past year and those that received treatment, but had no past-year dependence. The UFDS and TEDS treatment data are assumed to capture mostly people with a substance abuse problem because those data sets are collected from specialty treatment facilities. The question on past-year treatment in the NHSDA is designed to capture a variety of treatment modes including self-help groups, all types of counseling, and consultation with a private physician. As shown in Figure 3.20, the percentage of those with a drug dependence that received some treatment in the past year peaked in 1997 at almost 19%. That was up by over five percentage points from 1995. The rate tumbled by almost six and a half percentage points the next year. The proportion of those with a drug dependence that got treatment fell by a little over 1 percentage point over the period 1995 to 1998 and the rate fell by about a third between 1997 and 1998. This decline in services among those with a diagnosed dependence is similar to the decline in services among those with a diagnosed alcohol dependence, but it started several years later.

Comparison of the gap ratio of reported drug treatment to drug dependence with the treatment rate among those with a diagnosed dependence, prompts similar concerns as was the case with the measures of the alcohol treatment gap. The reasons for the difference are similar, as well. The treatment measure used in the gap ratio measures all treatment while the met demand measure captures only treatment among those in need. In 1995, 73% of those that reported receiving drug treatment in the past year had no past-year dependence diagnosis. The survey question on treatment is designed to encompass a variety of treatment modes, so the measure may capture people in full sustained remission. It is also likely that the treatment measure includes people who would meet the criteria for a drug abuse diagnosis, which the NHSDA does not measure. Further, the survey only measures dependence at the time of the survey. Therefore, the dependence measure fails to capture those who had a dependence or abuse diagnosis over the past year, but had less than three dependence symptoms at the time of the survey. It is impossible to determine from the NHSDA data how many of those that received treatment in the past year should have been diagnosed as dependent or as an abuser but were missed by the survey. Further, it is impossible to determine how many were in full sustained remission and received treatment. Finally, people in long term outpatient treatment, such as methadone maintenance, would be counted as having received treatment, but were in partial remission and not counted as dependent. Because the rate of treatment need includes more than just those with a dependence and because the NHSDA fails to identify some of those with a treatment need, the gap ratio is included here.

Replacing the NHSDA drug dependence estimates with drug arrests as the measure of need eliminated some of the differences between the treatment need and supply measures, but inconsistency between the NHSDA's measure of treatment and the other two indicators of treatment persisted. The NHSDA-based treatment measure suggests a substantial increase in the treatment gap over time, while the other measures suggest a more stable relationship between services and need over time (Figure 3.21). While both the NDATUS/UFDS-based and TEDS-based measures indicated a broadening of the treatment gap, the amount was small in both cases.

Using the drug mortality rate as the measure of drug treatment need, the authors found that the basic findings remained unchanged. For this analysis, the authors used the crude drug-induced underlying cause mortality rate from the Vital Statistics program of the CDC (1999) because the variable had more data points than the other measures of drug mortality had. As shown in Figure 3.22, all three ratios indicated a broadening drug treatment gap, but the amount was much greater for the ratio using the NHSDA treatment measure than the ratios using the other two measures (TEDS and UFDS/NDATUS).

Summary

Analysis of survey, arrest, mortality, emergency room, and treatment statistics revealed several national trends. Alcohol treatment needs declined during the 1990s as part of a long-term trend, but the alcohol treatment gap appeared to widen somewhat because the number of persons receiving treatment declined more rapidly than the indicators of need declined. Drug use declined sharply in the 1980s and the very beginning of the 1990s, but increased thereafter, especially in the middle of the decade. Adolescent drug use increased notably, just as heavy alcohol use by adolescents increased. Emergency room episodes followed a similar pattern. Notable increases occurred as a result of use of club drugs. Drug arrests and mortality increased steadily during the decade. The percentage of persons who were dependent on drugs was stable from 1994 to 1998. The drug treatment gap appeared to be widening in the late 1990s, substantially according to NHSDA treatment statistics, but much more moderately according to treatment admission and client statistics. Thus, there was some indication of a widening of the treatment gap for both drugs and alcohol in the late 1990s, even though alcohol treatment needs were declining while drug treatment needs were growing. The total number of clients in treatment for both drug and alcohol according to UFDS statistics remained essentially constant between 1992 and 1997 (ONDCP 2000). A substantial increase occurred in 1998.

Appendix: Data Sources

Alcohol and Drug Mortality Rates

There are four measures of mortality used in this chapter. For the multiple cause of death data, the NCHS of the CDC provided the authors with disks containing death certificate data for 1991-1996. NCHS nosologists, using special software, coded each cause of death from death certificates according to the International Classification of Diseases, Ninth Edition (ICD-9) (NCHS, 1998; Hopkins et al. 1989).

Each death record contained information on the decedent's demographics, residence, a code for the underlying cause of death, and the contributing conditions, including as many as 20 "entity-axis" and 20 "record-axis" codes (NCHS, 1998). The authors extracted all records that included at least one cause-of- death code with explicit mention of alcohol or controlled drugs in any of the 41 cause-of-death fields. It was possible for a death that was counted as having a drug diagnosis also to have one or more explicit-mention-of-alcohol diagnoses as well; similarly, a death counted as having an alcohol diagnosis could also have a explicit-mention drug-related diagnosis.

The explicit-mention drug codes used in this measure included drug psychoses (292.0, 292.1, 292.2, 292.8, 292.9), drug dependence (304.0 to 304.9), nondependent abuse of drugs (305.2 to 305.7, 305.9), and accidental poisoning. The ICD-9 codes for accidental poisoning are supposed to include combinations of poisoning nature ("N") codes for controlled drugs and external ("E") codes which indicated that the poisoning was accidental. As explained in the next paragraph, the authors also included deaths that had E-codes indicating undetermined external causes for some drugs of abuse, but did not include deaths due to suicide, therapeutic accidents, or assault. The N codes included deaths associated with ingestion of opiates (965.0), other specified analgesics (965.8), sedatives (967.0 to 967.9), other gaseous, intravenous, and surface anesthetics (968.2, 968.3, 968.5), benzodiazepines (969.4), other tranquilizers (, 969.5), hallucinogens (969.6), psychostimulants (N969.7), parasympatholytics (971.1), and dietetics (977.0). The relevant accidental or undetermined intent E codes included E850.0-E850.2, E850.8, E851, E852, E853.2, E853.8, E854.1, E854.2, E855.1, E855.2, E855.4, E858.8, E980.0, and E980.4.

The combinations of N and E codes varied depending on the drug. All deaths with the relevant N and accidental E codes were considered cases. In some instances, the records did not have both an N code for poisoning and an E code. Deaths that lacked an N code but had an accidental poisoning E code for a drug of abuse were counted as relevant cases. Deaths due to undetermined external causes for opiates, cocaine, and dietetics were also considered cases because the large majority of deaths associated with these substances had accidental E codes in national 1994-1996 mortality data. Deaths with an N code for opiates or cocaine (surface and infiltration anesthetics), but no E code, were treated the same as cases that had an undetermined E code. A full count of all drug deaths was divided by the Census Bureau's total population estimate (Census 2000), then multiplied by 100,000 to obtain a rate of drug deaths for each year, 1991-96.

Annually, the CDC publishes alcohol-induced and drug-induced death rates in its "Vital Statistics" series (CDC 1999, 2001). These rates reflect the underlying cause of death. The age-adjusted rates are published alongside the crude rates. These drug data differ from the explicit-mention rates developed by the authors from mortality files in that the CDC measure includes E codes for suicides, assaults and undetermined causes for some drugs not included in explicit-mention counts. With the possible exception of combinations of N and E codes for alcohol poisoning deaths, the CDC's alcohol-induced mortality rate uses the same diagnostic categories as the present authors used in developing their rates. The 1999 CDC deaths were coded using ICD-10 codes, whereas the previous years reflect ICD-9 coding. The CDC provided a crosswalk in the appendix of the 1999 preliminary report that makes the differences in coding clear (CDC 2001). While the CDC cautioned against comparisons across ICD-9 and ICD-10 codes, review of the trends presented in this chapter did not reveal any obvious discontinuity in the trends from previous years.

Alcohol mortality data from the NIAAA (Stinson et al., 1996) included any case with the underlying cause of death listed as one of the widely used explicit-mention alcohol diagnoses (e.g., McAuliffe et al., 2000; Stinson et al., 1994, 1996). The explicit-mention alcohol ICD-9 codes were 291 (alcoholic psychoses), 303 (alcohol dependence), 305.0 (alcohol abuse), 357.5 (Alcohol Polyneuropathy), 425.5 (Alcohol Cardiomyopathy), 535.3 (Alcohol Gastritis), 571.0 (Alcoholic fatty liver), 571.1(Acute Alcoholic Hepatitis), 571.2 (Alcoholic Cirrhosis of the Liver), 571.3 (Alcoholic Liver Damage, Unspecified), 790.3 (Excessive Blood Level of Alcohol), E860.0 (Accidental Poisoning by Ethyl Alcohol Beverages), and E860.1(Accidental Poisoning by Ethyl Alcohol). The differences between the NIAAA rates and the present authors' explicit-mention rates were the use of multiple-cause versus underlying cause data and the combination of N and E codes for poisoning. Whereas the present authors use specific combinations of codes, the description of the methods used in the NIAAA rates suggests that any case with a relevant code was included.

The OAS publishes annual DAWN Medical Examiner (ME) reports (OAS 1998, 2000a). These reports contain data submitted by medical examiners in metropolitan areas nationwide. The medical examiners report on the incidence of drug use among decedents. A "consistent panel" of examiners who report most consistently is employed to facilitate comparisons among different years. That is, the data are compiled from the same group of medical examiners in order to discern differences among years. This chapter drew on two such consistent panels; one from 1993-96 and the other from 1996-99. The former had 138 medical examiners, and the latter 134 (OAS 1998, 2000a). When rounded to one decimal and graphed, these separate lines appear to form a single continuous line.

For purely methodological purposes, the authors included these ME statistics because of the assumed accuracy of drug-related diagnoses in medical examiner data when compared to mortality statistics for all deaths, which have been questioned by some authors (see McAuliffe et al. 2001 for a review). In order to compare the trends in the ME data to the other drug mortality rates, the authors created rates per 100,000 using total US population estimates from the Census Bureau (2000). The authors assumed that the population changes over time in the DAWN areas would closely reflect changes in the national population. Because the DAWN data reflect urban populations rather than the total national population, readers should not interpret the absolute level of the rates as estimates of the national mortality rates.

Treatment Client Rates

With the exception of 1994, the OAS (1995a,b, 1996a, 1997b,c, 1999b, 2000g) has conducted an annual survey of treatment facilities to assess the number of persons in specialty substance abuse treatment at one point in time. The study has been known by two names over the relevant years: NDATUS from 1987 to 1994 and the UFDS thereafter. The survey was not conducted in 1994; instead the values for that year were estimates from an econometric model incorporating prior years, changes in state treatment funding levels, unemployment, population changes, and changes in food stamp costs (OAS 1996a). In the years used in this chapter (1992-1998), the sampling frame was a listing called the National Facility Register that OAS updates continuously. The total number of facilities ranged from 11,316 in 1992 to 19,174 in 1998. The facilities included specialty providers of substance abuse treatment, including public and private free-standing units and units in multi-purpose institutions. The facilities were owned by private for-profit and non-profits, as well as federal, state, local, and tribal governments. The federal facilities included the Veterans Administration, Defense Department, Bureau of Prisons, and the Indian Health Service. Identified mostly by state and federal agencies, these providers completed a questionnaire about all active clients in treatment on a specific reference day in the previous year (e.g., September 30, 1992, and October 1 or in the first week in October in 1993 to 1998).

The NDATUS/UFDS clients are categorized into three groups: Alcohol only (35% nationally), alcohol and drugs (40%), and drug only (25%) in 1991-1993. On the basis of theoretical and empirical analysis, the authors selected the drug-only and alcohol-only clients as measures rather than the combined drug-only plus drug-and-alcohol-treatment clients or the combined alcohol-only plus drug-and-alcohol treatment clients. Of course, the primary reason for not including the drug-and-alcohol treatment clients in both measures was to avoid counting the same clients in both measures, and the authors believed that the selected measures would be more valid indicators of the supply of the two different types of services. The authors' review of NDATUS/UFDS statistics revealed that providers in many states (especially Massachusetts, Alaska, New Hampshire, Nebraska and Texas) were more likely to utilize the drug-and-alcohol-treatment clients measure than the drug-only measure, but providers in other states (Alabama, Arizona, New Mexico, New York, Rhode Island, and California) were more likely to use the drug-only category. Years of clinical experience working in several of these states caused us to hypothesize that many of the clients in the drug-and-alcohol category may have had alcohol use disorders but were only users of illicit drugs rather than persons who met standard criteria for drug abuse or dependence. Although NDATUS/UFDS defines this category adequately in the glossary that accompanies its survey questionnaire, even experts sometimes use the term "drug abuse" to refer to the use of illicit drugs, whereas they reserve the terms "alcohol abuse" for excessive use that results in symptoms and is likely to require treatment (see McAuliffe et al. 1999).

The empirical behavior of the drug-only client measure and the combined client measure for 1991-1993 data mostly supported the authors' hypothesis. Although the two measures correlated substantially with each other (.79, p<.05), the correlation of drug arrest rates with drug-only client rates (.63) was significantly greater than the correlation of drug arrest rates with the combined client rates (.42). Drug mortality rates correlated significantly more with the drug-only client rate (.89) than with the combined client rates (.70), and IV-AIDS statistics correlated significantly more with the drug-only client rate than with the combined client rate (.76 versus .57). The drug-only client rate correlated slightly more strongly with NHSDA model estimates of past year drug treatment than did the combined measure (.28 versus .23), but the difference was not significant. The drug-only client rates correlated significantly less than the combined client measure with National Association of State Alcohol and Drug Abuse Directors (NASADAD) drug-related admissions (.47 versus .62).

Treatment Admission Rates

The OAS collects administrative data on treatment admissions from all states. The data include admissions from only those programs that are licensed by the state substance abuse agency and required to report data to it. Inclusion of private, hospital, and prison treatment varies from state to state depending on the state substance abuse agency's licensing and reporting regulations. Mostly, the admissions represent admissions to publicly-funded treatment. OAS publishes reports called the Treatment Episode Data Set (TEDS)using those data (OAS and Synectics for Management Decisions, Inc., 2000e). The report includes rates per 100,000 in the total population that are involved in substance abuse treatment at a facility of the type described above. Because the data are published on an admissions rather than a client basis, an individual may be counted more than once if he/she was admitted for treatment more than once within the same year. Rates are reported for total admissions as well as for specific categories of substances (such as alcohol, cocaine, marijuana or hashish, etc.).

Survey-Reported Treatment

The NHSDA (OAS 1996b, 1997a, 1999a, 2000c, d, f) reports the percentage of respondents who said they received substance abuse treatment in the previous 12 months. The data break down into two categories: primary alcohol treatment, and primary drug treatment. For the purposes of this chapter, the authors expressed the percentages as rates per 100,000 in the population in order to compare those statistics with the rates per 100,000 used for the other treatment measures.

The NHSDA treatment rates are higher than the rates indicated by the other treatment measures used in this chapter. There are several reasons for this. First, the questions on past-year treatment in the NHSDA are designed to capture a broader group than the other data sets. Whereas, the UFDS and the TEDS data are collected at specialty treatment facilities, the NHSDA captures participation in self-help groups, consultation with a doctor, and all types of counseling for substance-related problems. Second, the NHSDA and the TEDS data set capture the whole year, whereas the UFDS is a one-day census of treatment facilities. Third, the NHSDA does not require a respondent to have had a dependence in the past-year to have received treatment for substance use-related issues in the past year. Although the UFDS and the TEDS data sets do not require a diagnosis for a client to be counted, if an individual did not have a current, diagnosable substance-related condition, it is unlikely they would be at a treatment facility.

Some of the treatment gap variables used in this chapter express all treatment reported in the NHSDA as a proportion of dependence. The current authors attempted to create a measure of the treatment gap that would more accurately reflect the gap between need for treatment as represented by the past-year substance use disorder rate and treatment received among those in need as represented by the NHSDA treatment estimate using NHSDA public-use data sets. Unfortunately, limitations posed by the NHSDA data prevented creation of a gap estimate that was clearly better than the one used in this chapter. The authors' initial concern was that including all reported cases of treatment overestimated the proportion of those with a past-year dependence that received treatment. One possible solution was to use only those treatment cases where there was also a dependence diagnosis. However, dependence alone may not be an adequate measure of treatment need. For example, the NHSDA does not permit diagnoses of abuse. Therefore, excluding those treatment cases without a diagnosis might understate the amount of treatment delivered to those with a current treatment need. Another possible solution was to include all cases where there was a need for treatment, but no past-year dependence diagnosis in the denominator of the treatment gap ratio. Because the NHSDA does not permit lifetime diagnoses of dependence or abuse, the authors examined the perceived need for treatment in the past 12 months as a possible proxy for a lifetime diagnosis rate. Because the perceived need variable had a small number of cases and correlated with the past-year dependence variable significantly for both alcohol and drugs (.173 p<.01 for alcohol and .229 p<.01 for drugs), it was decided that it was not an adequate proxy for lifetime treatment need, nor did it necessarily capture past-year abusers without a diagnosis.

As such, the authors decided to use the ratio of past-year treatment to past-year dependence despite its imperfections. The authors corroborated the trends in the treatment gap using social indicators such as substance-related arrests and substance-related mortalities as the denominators in ratio estimates of the treatment gap. Further, the authors included trends in treatment among those with a past-year dependence diagnosis, the "met demand" for treatment, in order to confirm the trend in the gap between diagnosed treatment need and treatment delivered.

Alcohol- and Drug-Defined Arrest Rates

The FBI's Uniform Crime Reporting (UCR)(FBI 1996, 1997, 1998, 1999) system reports statistics on arrests for violations of state and local laws that are associated with alcohol abuse, including DUI arrests, drunkenness, disorderly conduct, and liquor-law violations. An arrest occurs when police detain a person and a record is made of the detention. The arrest statistics in the early part of the decade counted only those cases in which the drug or alcohol charge was the most serious (General Accounting Office, 1990), but states that have converted to the UCR's new National Incident-Based Reporting System (NIBRS) coding system count any substance-defined arrest (FBI 1992).

The authors downloaded the UCR "Persons Arrested" data from www.FBI.gov . The reports include rates of arrest for specific crimes per 100,000 in the population based on their estimate of the population that is covered by agencies reportingcomplete data for a given year. The FBI also publishes estimates of nationwide total arrests for each crime, the rates used in this chapter were based on the actual number of offences reported and the covered population estimate.

Population Estimates

The population bases for specific indicators vary depending on the decisions of the agencies that supplied the rates. The rates created by the present authors used denominators based on annual Census estimates of the entire population. The 1993 through 1999 estimates were obtained on-line at census.gov/population/estimates/nation/popclockest.txt. (Census 2000). The authors did not use age-adjustments because the crude rates are more relevant for determining the amount of services that are needed. Age-adjustments are relevant only when the analyst wishes to determine when changes in rates occur independently of the effects of aging, but aging is a relevant variable in determining the need for treatment.

Alcohol-Related Traffic Fatalities

The annual rates per 100,000 in the population that died in an alcohol-related traffic accident were drawn from "Surveillance Report #53: Trends in Alcohol-Related Fatal Traffic Crashes, United States, 1977-98"(Yi et al. 2000). The authors of that report drew the data from the Fatal Accident Reporting System (FARS) of the National Highway Traffic Safety Administration (NHTSA). The authors of the report defined "alcohol-related" fatalities somewhat differently than they are defined in the FARS reports. Yi et al. (2000) defined a fatality as alcohol related when either the blood alcohol content of the driver involved in a crash was known to be above zero or when the reporting officer indicated that alcohol was involved in the accident.

Alcohol and Drug Dependence

The NHSDA Main Findings (OAS 1996b, 1997a, 1999a, 2000c, d, f) reports the percentage of respondents with three symptoms of dependence on alcohol, marijuana, and cocaine for persons 12 and older. Because those estimates do not reflect a rate of dependence diagnoses, the authors analyzed NHSDA public-use data sets to determine dependence rates for alcohol and drugs. The authors expressed the figures as the rate per 100,000 for purposes of comparability.

Several factors affected the utility of NHSDA for trend analysis. In 1999, the NHSDA switched from a paper and pencil interview (PAPI) method of administration to a computer-assisted method, which combined a computer-assisted personal interview conducted by the interviewer (CAPI) and a computer-assisted self interview (audio computer assisted self interview [ACASI]). In a effort to assess the effect of the new interview mode, the NHSDA also included a supplemental sample of 13,809 cases using the same PAPI methodology employed in the previous year. However, analysis of the resulting data suggested that the 1999 PAPI data were not comparable to the previous year. There were "unusually large and unexpected increases in various drug measures " (OAS 2000b, p. D-1). Because the NHSDA had also expanded the size of its sample to include interviews from all states, the survey contractor had to hire a large number of new interviewers. The analysis showed that the prevalence of drug use declined as the new interviewers became more experienced during the year, and interviewers from prior years obtained lower levels of prevalence than the new interviewers obtained. As a result, the NHSDA developed adjusted estimates for some of its indicators in 1999 (OAS 2000b, p. D-11). The only 1999 NHSDA estimates this report employed were treatment rates.

The NHSDA Main Findings for years prior to 1999 did not report dependence rates for alcohol or drugs, but instead reported rates of respondents that had three symptoms of dependence. Those rates are higher than actual rates of dependence. Further, Main Findings reports prior to 1998 reported rates of respondents that had three symptoms for cocaine and marijuana, but not other drugs. If a respondent had three symptoms for both cocaine and marijuana dependence, he was counted for both substances. The 1998 data includes variables for drug dependence and alcohol dependence. The authors replicated those variables for 1995 through 1997 using public-use data sets. The authors were able to replicate the results for the alcohol dependence variable in 1998, but the rate of drug dependence obtained by the authors was lower than the result published in the 1998 Main Findings. As such, the rates drug dependence rates for the previous years are likely lower as well. Although the rates of respondents that had three problems due to use of substances probably captured some non-dependent abusers that were also in need of treatment, the authors decided to use the dependence rates for this report because the group captured by that variable is well defined.

References

Brown, Claude. (1965). Manchild in the Promised Land. New York. Signet Books.

Caulkins, Jonathan P. (2001). "Drug Prices and Emergency Department Mentions for Cocaine and Heroin." American Journal of Public Health. 91(9): 1446-1148.

Centers for Disease Control and Prevention (CDC). (1993). "Advance Report of Final Mortality Statistics, 1991." Monthly Vital Statistics Report. 42(2, Supplement): 56-61.

Centers for Disease Control and Prevention. (1999). "Deaths. Final Data for 1997." National Vital Statistics Report. 47(19): 75-102.

Centers for Disease Control and Prevention. (2001). "Deaths: Preliminary Data for 1999." National Vital Statistics Report. 49(3): 14-17.

Community Epidemiology Work Group (CEWG). (2001). Epidemiologic Trends in Drug Abuse. Volume I: Proceeding of the Community Epidemiology Work Group. Highlights and Executive Summary. Rockville, MD: National Institute on Drug Abuse (NIDA).

Dayhoff, Debra A., Carol J. Urato, and Gregory C. Pope. (2000). "Trends in Funding and Use of Alcohol and Drug Abuse Treatment of Specialty Facilities, 1990-1994." American Journal of Public Health. 90(1):109-11.

Epstein, Joan F. and Joseph C. Gfroerer. (2001). "Heroin Abuse in the United States." In Analyses of Substance Abuse and Treatment Need Issues. Downloaded 8/22/01 from www.samhsa.gov/oas/nhsda/treatan/httoc.htm.

Federal Bureau of Investigation (FBI). (1992). Uniform Crime Reporting Handbook, National Incident-Based Reporting System Edition. Washington, D.C.: U.S. Department of Justice.

Federal Bureau of Investigation (FBI). (1996). Uniform Crime Reports for the United States 1995. Washington, DC.: U.S. Department of Justice.

Federal Bureau of Investigation (FBI). (1997). Uniform Crime Reports for the United States 1996. Washington, DC.: U.S. Department of Justice.

Federal Bureau of Investigation (FBI). (1998). Uniform Crime Reports for the United States 1997. Washington, DC.: U.S. Government Printing Office.

Federal Bureau of Investigation (FBI). (1999). Uniform Crime Reports for the United States 1998. Washington, DC.: U.S. Government Printing Office.

Federal Bureau of Investigation (FBI). (2000). Uniform Crime Reports for the United States 1999. Washington, DC.: U.S. Government Printing Office.

Greenfield, Thomas K., Lorraine T. Midanik, and John D. Rogers. (2000). "A ten-year national trend study of alcohol consumption, 1984-1995: Is the Period of Declining Drinking Over?" American Journal of Public Health. 90(1): 47-52.

Hopkins, David D., Joyce A. Grant-Worley, and Terrie L. Bollinger. (1989). "Survey of Cause-of-Death Query Criteria Used by State Vital Statistics Programs in the US and the Efficacy of the Criteria Used by the Oregon Vital Statistics Program." American Journal of Public Health. 79(5): 570-574.

Hunt, Leon Gibson. (1974). Recent Spread of Heroin Use in the United States: Unanswered Questions. Washington, DC.: Drug Abuse Council, Inc.

McAuliffe, William E. and Kathy Ackerman. (1991). Health Care Policy Issues in the Drug Abuser Treatment Field. In Health Policy and the Disadvantaged. Brown, L. D., Editor. Durham, NC: Duke University Press. 105-142.

McAuliffe, William E., Richard A. LaBrie, Nicoletta Lomuto, Rebecca Betjemann, and Elizabeth A. Fournier. (1999). "Measuring Interstate Variations in Drug Problems." Drug and Alcohol Dependence. 53(2): 125-145.

McAuliffe, William E., Richard A. LaBrie, Nicoletta A. Lomuto, Nancy E. Pollock, Rebecca Betjemann, and Elizabeth Fournier. (2000). "Measuring Interstate Variations in Problems Related to Alcohol Use Disorders." In The Epidemiology of Alcohol Problems in Small Geographic Areas, First ed. vol. 36. Eds. Robert A. Wilson and Mary C. Dufour. Bethesda, MD: National Institute on Alcohol Abuse and Alcoholism (NIAAA). 213-44.

McAuliffe, William E., Richard A. LaBrie, and Ryan Woodworth. (2001). Measuring Interstate Variations in Substance Abuse Treatment Needs. Unpublished manuscript.

Midanik, Lorraine T. and Thomas K. Greenfield. (2000). "Trends in Social Consequences and Dependence Symptoms in the United States: The National Alcohol Surveys." American Journal of Public Health. 90(1): 53-56.

National Center for Health Statistics (NCHS). (1998). 1994 Multiple Cause-of-Death File. Hyattsville, MD: Centers for Disease Control and Prevention.

National Institute on Alcohol Abuse and Alcoholism (NIAAA). (1999). Number of deaths and age-adjusted rates per 100,000 population for categories of alcohol-related (A-R) mortality, United States and States, 1979-96. Downloaded from http://www.niaaa.nih.gov/databases/qf.htm on 8-13-01. Table based on F.S. Stinson, T.M. Nephew, M.C. Dufour, and B.F. Grant. (1996). State Trends in Alcohol-Related Mortality, 1979-92. U.S. Alcohol Epidemiologic Data Reference Manual, Volume 5, First Edition. Bethesda, MD: NIAAA, National Institutes of Health, and unpublished data from the Alcohol Epidemiologic Data System (AEDS).

Nephew, T.M., G.D. Williams, F.S. Stinson, K. Nguyen, and M.C. Dufour. (2000). Apparent Per Capita Alcohol Consumption: National, State and Regional Trends, 1970-98. Surveillance Report #55. Rockville, MD: National Institute on Alcohol Abuse and Alcoholism, Division of Biometry and Epidemiology.

Office of Applied Studies (OAS). (1995a). Overview of the National Drug and Alcoholism Treatment Unit Survey 1992 and 1980-1992. Rockville, MD: Substance Abuse and Mental Health Services Administration (SAMHSA).

Office of Applied Studies. (1995b). Overview of the FY94 National Drug and Alcoholism Treatment Unit Survey Data from 1993 and 1980-1993. Rockville, MD: Substance Abuse and Mental Health Services Administration.

Office of Applied Studies and Macro International Inc. (1996a). National Drug and Alcoholism Treatment Unit Survey Data for 1994 and 1980-1994. Rockville, MD: Substance Abuse and Mental Health Services Administration.

Office of Applied Studies and National Opinion Research Center. (1996b). National Household Survey on Drug Abuse Main Findings 1994. Rockville, MD: Substance Abuse and Mental Health Services Administration.

Office of Applied Studies and National Opinion Research Center. (1997a). National Household Survey on Drug Abuse Main Findings 1995. Rockville, MD: Substance Abuse and Mental Health Services Administration.

Office of Applied Studies. (1997b). Uniform Facility Data Set Data for 1995 and 1980-1995. Rockville, MD: Substance Abuse and Mental Health Services Administration.

Office of Applied Studies. (1997c). Uniform Facility Data Set Data for 1996 and 1980-1996. Rockville, MD: Substance Abuse and Mental Health Services Administration.

Office of Applied Studies. (1998). Drug Abuse Warning Network (DAWN) Annual Medical Examiner Data 1996. Rockville, MD: Substance Abuse and Mental Health Services Administration.

Office of Applied Studies and National Opinion Research Center. (1999a). National Household Survey on Drug Abuse: Main Findings 1997. Rockville, MD: Substance Abuse and Mental Health Services Administration.

Office of Applied Studies and Synectics for Management Decisions, Inc. (1999b). Uniform Facility Data Set 1997 Data on Substance Abuse Treatment Facilities. Rockville, MD: Substance Abuse and Mental Health Services Administration.

Office of Applied Studies. (2000a). Drug Abuse Warning Network Annual Medical Examiner Data, (DAWN) 1999. Rockville, MD: U.S. Department of Health and Human Services, Substance Abuse and Mental Health Services Administration.

Office of Applied Studies. (2000b). Drug Abuse Warning Network: Advance Report #17. Downloaded 8/22/01 from www.samhsa.gov/oas/dawn/ar17_toc.htm.

Office of Applied Studies and Research Triangle Institute. (2000c). National Household Survey on Drug Abuse: Main Findings, 1998. Rockville, MD: Substance Abuse and Mental Health Administration.

Office of Applied Studies. (2000d). Summary of Findings from the 1999 National Household Survey on Drug Abuse. Rockville, MD: Substance Abuse and Mental Health Services Administration.

Office of Applied Studies. (2000e). Treatment Episode Data Set (TEDS)1993-1998, National Admissions to Substance Abuse Treatment Services. Rockville, MD: National Clearinghouse for Alcohol and Drug Information (NCADI).

Office of Applied Studies and National Opinion Research Center. (2000f). National Household Survey on Drug Abuse Main Findings 1996. Rockville, MD: U.S. Department of Health and Human Services.

Office of Applied Studies. (2000g). Uniform Facility Data Set 1998. Rockville, MD: Substance Abuse and Mental Health Services Administration.

Office of Applied Studies. (2001). Mid-Year 2000 Preliminary Emergency Department Data from the Drug Abuse Warning Network. Rockville, MD: Substance Abuse and Mental Health Services Administration .

Office of National Drug Control Policy (ONDCP). (2000). Drug Facts and Stats: Data Snapshot: Drug Abuse in America"“Office of National Drug Control Policy, June 2000. Downloaded on 10/6/00 from www.whitehousedrugpolicy.gov/drugfact/presentations.html.

Quest Diagnostics, Incorporated. June 20, 2000. Positive Drug Test Results in 1999 Decline to Record Low in Quest Diagnostics' Workplace Drug Testing Index. Teterboro, New Jersey. www.questdiagnostics.com/corporatehealth/news/dti.htm. Downloaded on 11/30/2000.

Regier, Darrel A., Kaelber, Charles T., Rae, Donald S., Farmer, Mary E., Knauper, Barbel, Kessler, Ronald C., and Norquist, G. S. (1998). "Limitations of Diagnostic Criteria and Assessment Instruments for Mental Disorders: Implications for Research and Policy." Archives of General Psychiatry. 55(2): 109-115.

Stinson, Frederick, Ann H. Proudfit, Thomas C. Harford, Mary C. Dufour, and John A. Noble. (1994). County Alcohol Problem Indicators 1986-1990. NIH Publication No.94-3747. Fourth Edition ed. Rockville, MD: National Institute on Alcohol Abuse and Alcoholism.

Stinson, F.S.,T.M. Nephew, M.C. Dufour, and B.F. Grant. (1996). State Trends in Alcohol-Related Mortality, 1979-92. U.S. Alcohol Epidemiologic Data Reference Manual, Volume 5, First Edition. Bethesda, MD: National Institute on Alcohol Abuse and Alcoholism, National Institutes of Health.

Substance Abuse and Mental Health Services Administration (SAMHSA). (2001). Year-end 2000 Emergency Department Data from the Drug Abuse Warning Network. Adapted by the CESAR FAX. 10(33):1.

United States Census Bureau. (2000). Population Estimates Program. Population Division. Downloaded from http://www.census.gov/population/estimates/nation/popclockest.txt.

United States General Accounting Office (GAO). (1990). Rural Drug Abuse: Prevalence, Relation to Crime, and Programs. Washington, DC.

Woodward, Albert, Joan F. Epstein, Joseph C. Gfroerer, Daniel Melnick, Richard Thoreson, and Douglas Willson. (1997). "The Drug Abuse Treatment Gap: Recent Estimates." Health Care Financing Review. 18 (3): 5-17.

Yi, Hsiao-ye, Frederick S. Stinson, Gerald D. Williams, and Mary C. Dufour. (2000). Trends in Alcohol-Related Fatal Traffic Crashes, United States, 1977-98. Surveillance Report #53. Bethesda, MD: National Institute on Alcohol Abuse and Alcoholism.

Interstate Comparisons

This chapter describes how the State compares to other states with regard to its need for treatment and its supply of treatment services. The analysis employs existing indicators of drug, alcohol, and combined substance abuse to describe the distribution of differences among states in the need and supply of treatment services. The indicators were average annual rates per 100,000 in the population in the three years between 1994 and 1996. The analysis identifies treatment service gaps by comparing state treatment needs and services. The ultimate objective of the study is to determine whether the State's provision of treatment services is on par with other states.

Methods

The authors created a series of composite treatment need indexes. The Drug Need Index (DNI) consists of the sum of standard (z) scores of mean rates of explicit-mention drug mortality and drug possession or sale arrests. The denominators for each annual rate were state population projections for the relevant years. The combined z-scores were scaled to range from 0 to 100. The state with the most severe drug or alcohol abuse problem received the highest score and was ranked 1^st, and the state with the least severe problem received the lowest score and was ranked 50 ^th . The drug mortality indicator counted only deaths with codes that explicitly mentioned drugs of abuse as one of the causes listed on the death certificate. The diagnostic codes included accidental drug overdoses, drug dependence, nondependent drug abuse, and drug psychoses (including drug withdrawal syndrome). An index score of 0 reflects no arrests or deaths in the period, while a score of 100 reflects the sum of the highest observed mean rates of both components. Similarly structured, the Alcohol Need Index (ANI) consists of the sum of explicit-mention alcohol mortality rates and arrests rates for driving under the influence (DUI) and liquor law violations. The alcohol mortality measure employed 12 explicit-mention diagnoses widely used as measures of alcohol abuse. Examples were alcohol dependence, non-dependent alcohol abuse, alcohol psychoses, alcoholic cirrhosis of the liver, and alcohol cardiomyopathy. The Substance Abuse Need Index (SNI) combines all explicit-mention drug and alcohol mortality rates and the sum of the drug and alcohol arrest rates. The authors conducted analyses that established that the three indexes had a reasonably high degree of reliability and validity (McAuliffe et al. 1999, 2000, 2001).

The Office of Applied Studies (OAS) conducts a national survey of substance abuse treatment facilities annually. The name of the survey changed between 1994 and 1995. Formerly, it was called the National Drug and Alcohol Treatment Unit Survey (NDATUS), and it is now called the Uniform Facility Data Set (UFDS). The survey counts treatment clients on a single day (usually September 30), and the clients are categorized as drug-only, alcohol-only, and drug and alcohol. The present study used the drug-only client rates for the DNI and alcohol-only client rates for the ANI in order to avoid double counting clients. The SNI used all clients.

A second measure of interstate treatment services is the Treatment Episode Data Set (TEDS), also managed by the Office of Applied Studies. TEDS collects admissions statistics from the states. It includes only state-funded treatment. The present study categorized admissions by primary substance: alcohol or drugs.

Because of the differences between the TEDS primary alcohol admission estimates and UFDS alcohol-only client estimates, the two available national measures of treatment for drugs and alcohol are imperfectly correlated (.57 for drugs and .49 for alcohol). The authors tried other combinations (e.g., alcohol-only and alcohol-and-drug client rate and any mention of alcohol rate), but the correlation did not increase. Based on experience with several states, the authors suspect that the composition of the alcohol-and-drug clients and any-mention-of-alcohol admissions measures depends upon the state. In states where alcohol use disorders are the predominate substance abuse problem, the combined categories consist mostly of people with primary alcohol problems who also use marijuana. In states where drug abuse problems predominate, the category consists of primary drug abusers (mostly opiate and cocaine) who also drink. Because of the imperfect correlation among the two measures, the analysis uses both client and admissions measures and notes differences between them. The client measure was complete while the admissions measures has nine missing states. Therefore, the client measure is used in the graphs that compare need and treatment.

Data Sources

This study employs existing substance abuse indicator data that the study team gathered from national sources (see Appendix). Before using them, the analysts examined each data set for the presence of outliers and other sources of error. An example of an outlier would be an annual count that is many times higher (lower) than the previous (subsequent) year for the state, especially when the annual change is not consistent with the usual annual variations in the data for that state and other states. In such cases, the study team alerted State officials who contacted the responsible officials or agency about the outlying values. If corrected values or comparable figures were available, the authors used them. As a general protection against undetected or uncorrected random errors, the study combined multiple years of data, 1994-1996, to obtain more reliable composite indicators.Experience with the 1991-1993 data indicated the three-year mean rates were reliable.

Analysis

The analysis begins with the drug indicators, and then turns to the controlled alcohol indicators. The analysis then describes the combined substance abuse indicators. In the presentation of results for each substance, the report begins with the composite index, and then it describes the components of the index and other supplementary indicators.

RESULTS

Drug Treatment Needs

Drug Need Index. The map of the DNI (Figure 4.1) reveals the wide variation among states in problems related to drug use disorders. The states with more severe drug treatment needs were New York (100), California (81.4), New Jersey (73.8),and Maryland (72.8)"“all states long known to have acute drug problems. They were the same four states with the highest scores on a similar 1991-1993 index described by McAuliffe et al. (1999).

The most severe drug problems were found in two clusters: the Northeastern urban states and the West Coast and Southwest. A member of the Northeast cluster, Rhode Island ranked 5^th in the country in 1991-1993 (McAuliffe et al. 1999), but its rank dropped tothe 18^th most severe drug abuse treatment needs in the nation during 1994 to 1996. Neighboring Connecticut (7^th) and Massachusetts (15 ^th ) also had high DNI scores in that period.

Colorado ranked 13^th in the country, just at the cutoff for the states in the most severe quartile. Colorado's DNI (49.8) was substantially higher than the scores of other plains and mountain states that border it, although it was lower than New Mexico's (67.7). Colorado appears to be in a transitional area with neighboring states having both higher and lower rankings. Among the states with moderately severe controlled drug treatment needs were Alaska with a DNI of 36 (ranked 31 ^st ) and Montana with a DNI of 34.4 (34^th).

North Dakota, Vermont, West Virginia, Iowa, and South Dakota had the lowest drug index scores in the country.All are rural states. North Dakota's DNI (8.5) reflected their rural nature. North Dakota is surrounded by states in the lowest quartile on this composite measure of drug treatment need.

Drug Mortality. A key component of the DNI counted deaths for which at least one of the multiple causes included drug dependence, nondependent drug abuse, accidental poisoning or drug psychoses (e.g., withdrawal) associated with controlled drugs. As Figure 4.2 shows, New York's average annual drug-mortality in 1991-1993 (12 per 100,000) was the highest in the country and was 40 times as great as North Dakota's (0.3 per 100,000), which was the lowest. The mortality rates and ranks for those two states in 1994-1996 were virtually unchanged from the earlier period. North Dakota's rate was one half as large as the next highest state (Vermont). Rhode Island's 1991-1993 rate (4.4 per 100,000) ranked 7 ^th most severe in the country. Three years later the State's drug mortality rate increased to 5.3 per 100,000, but its rank dropped to 15 ^th .The 1991-1993 rate was 3.2 for Colorado (15^th), but in 1994-1996 its rate increased to 5.6 per 100,000 (12 ^th in the country). Montana's drug mortality rate increased from 1.5 per 100,000 (ranking 29^th) to 1.9 (ranking 36 ^th). Alaska's rate increased from 1.4 to 4.3 per 100,000, while its rank increased from 31 ^st in 1991-1993 to 18 ^th in 1994-1996.

Drug-Defined Arrests. In 1994-1996, New York also had the highest drug arrest rate, while Vermont had the lowest (Figure 4.3). Montana had the lowest rate in 1991-1993, while North Dakota had the third lowest in the country (57 and 77 per 100,000, respectively). North Dakota's rank dropped to 49^th in the nation in the 1994-1996 period. Alaska's 1991-1993 rate (157 per 100,000) ranked 41 ^st in the nation and rose to 38^th (313 per 100,000) in the 1994-1996 period.

In contrast with Rhode Island's high drug mortality rate, its 1991-1993 average annual arrest rate of 293 per 100,000 was in the middle of the national distribution (25^th most serious in the country). Three years later Rhode Island's average annual drug arrest rate had increased to 409 per 100,000, but its rank dropped to 30 ^th in the country. Rhode Island's neighboring states, such as New York (845) and Connecticut (635) had substantially higher drug-defined arrest rates, but Massachusetts' rate (428) was similar to Rhode Island's.

In contrast to drug mortality, the composition of the drug arrests rates represents somewhat different aspects of drug treatment need, which may vary from one region of the country to another. Nationally, 41 % of all drug-defined arrests were associated with marijuana possession and sales, which is the same percentage associated with cocaine and heroin arrests combined (41%). Much smaller percentages of arrests are associated with use of synthetic narcotics (2%) and non-narcotic drugs other than marijuana (16%). Figure 4.4 shows that in rural states a majority of the arrests involved marijuana arrests, whereas in urban states a majority of the arrests involved cocaine and opiates. The percentage of arrests due to marijuana varied from 82% in North Dakota to 19% in California. Marijuana arrests were predominate in the rural and mountain states of the West (including North and South Dakota, Alaska, and Colorado) and northern New England (Maine, Vermont). Comparable arrest data were not available for Montana. Arrests for possession and sale of hard drugs (combined arrests for cocaine/opiates and synthetic narcotics) ranged from 60% in New York to six percent in Idaho. Hard drug arrests were predominate in the urban East Coast states of Connecticut, Pennsylvania, Maryland, New York, New Jersey, Massachusetts, and Delaware. Rhode Island was in the middle of the distribution with regard to soft versus hard drugs. Rhode Island's arrests for drugs other than marijuana, cocaine, and opiates were among the lowest in the country. High arrest rates for drugs such as amphetamines were evident in the West and Midwest.

Drug Incarceration Rates. Rhode Island ranked 14^th in the country (40 states had data) with regard to the rate of incarcerations related to drug offenses in 1994 (Figure 4.5). Colorado and Montana ranked 37 ^th and 38 ^th respectively. Drug incarceration data were not available for North Dakota or Alaska.

Surveys of Drug Abuse. There was surprisingly little survey data available on drug dependence among the 50 states. The National Household Survey on Drug Abuse (NHSDA) expanded its sample to include all states in 1999. Because small states had samples of only 750 interviews, the NHSDA estimates are supplemented by census and social indicator data (deaths, arrests, and treatment statistics). Although based on some of the same indicator data, the DNI and the NHSDA drug dependence estimates were not significantly correlated with each other in 1991-1993 versions (McAuliffe et al. 1999) and were correlated only moderately (.46) in 1994-1996. One key difference between the DNI and NHSDA estimates is that the NHSDA's drug dependence survey measurements overwhelmingly (78% in 1995-1998) reflect marijuana use disorders rather than other drugs of abuse such as cocaine and opiate use disorders. Thus, the NHSDA estimates are likely to be good predictors of treatment needs only in rural states, where treatment for marijuana abuse is predominant. Thus, Alaska (ranked 1 ^st in the country), Montana (15^th ), and North Dakota (35 ^th ) have much higher ranks on the NHSDA's estimate of drug dependence than on the other measures of drug treatment need (Figure 4.6). No doubt because of the high rate of marijuana use and marijuana arrests in Rhode Island, it ranked 8^th in the country on the NHSDA's past-year drug dependence measure.

Drug Treatment Clients and Admissions. In 1991-1993, New York had the country's highest mean NDATUS drug-only client rate (269 per 100,000), and the state retained the same position in the 1994-1996 UFDS drug-only client rates with a rate of 271 per 100,000 (Figure 4.7). Rhode Island ranked second in the nation behind New York in 1991-1993 (192 per 100,000) as well as in 1994-1996 (229 per 100,000). Colorado's 1991-1993 rate was 91 (11^th), and its 1994-1996 rate declined to 79 (16^th). Alaska's 1991-1993 rate was 30 (40^th), and its 1994-1996 rate was 44 (34 ^th ). Montana's rates were 14 in 1991-1991 and 20 in 1994-1996; both ranked 47^th in the country. North Dakota had the lowest drug-only client mean rate in the country for both three-year periods: 9 per 100,000 in 1991-1993 and 15 per 100,000 in 1994-1996.

The TEDS admissions rates shown in Figure 4.8 revealed similar patterns. Rhode Island ranked 6^th with regard to its drug admissions rate, and Colorado's drug admissions rate was next (12^th). Alaska and Montana have very similar drug admissions rates (Montana was ranked 18^th and Alaska 19 ^th). North Dakota's rate was next to the lowest in the country.

As was true for the arrest statistics, rural states had higher rates of persons receiving treatment for a marijuana use disorder, whereas urban states had higher rates of persons receiving treatment for heroin use disorders (Figure 4.9). As a proportion of treatment, North Dakota had the highest percentage of marijuana admissions in the country, while California had the lowest percentage of marijuana admissions. Montana (8th), Alaska (11^th), and Colorado (13^th; it ranked 12^th in marijuana arrests) also had high percentages of admissions for marijuana use disorders; in some cases more admissions for marijuana use disorders than for any other drug use disorder in 1994-1996. Alaska treated more cocaine dependent people, but marijuana was a close second. Cocaine treatment was a close second in Colorado. Despite the large number of marijuana arrests in Rhode Island and its high ranking on the marijuana-dominated NHSDA drug dependence measure, the state had relatively few admissions for marijuana dependence. It is noteworthy that a large percentage of marijuana treatment admissions are for adolescent abusers.

The states with the highest percentage of their admissions for heroin and other opiates were California, and the eastern states of Rhode Island, Massachusetts, Connecticut, New Jersey, and Maryland. Rhode Island provided treatment to a higher percentage of people with opiate use disorders than any other state in the country. New Mexico, Washington, Oregon, Nevada, and Colorado were western states that treated a moderately high proportion of opiate addicts (about half as many as marijuana and cocaine). Montana, Alaska, and North Dakota had relatively few admissions of people with primary opiate use disorders.

Comparison of Drug Treatment Need Index and Treatment Client Rate

The DNI was compared to the UFDS drug-only client rate to determine how states compared with regard to the supply of services relative to need as illustrated in Figure 4.10. For this analysis the authors divided the two measures into quintiles. Rhode Island had a relatively high level of need (18^th nationally) and was in the highest quintile of the client treatment rate (2^nd, also 6^th in drug treatment admissions). Alaska (31^st need, 34 ^th treatment client), Colorado (13^th, 16 ^th ) and North Dakota (50 ^th , 50^th) matched the levels of drug need and drug-only treatment client rates. Montana had a low level of need (34 ^th ), but its treatment client rate was in the lowest quintile (47 ^th ). Montana's drug treatment admissions ranked much higher (18 ^th out of 41 states). Consequently, Montana's position in the matrix may reflect a measurement artifact of the UFDS survey.

It is important to recognize that survey results suggest that there is a lot of unmet need for treatment, and therefore relative differences among states may say little regarding their absolute ability to meet demand for treatment services.

Alcohol Treatment Needs

Alcohol Need Index (ANI). According to the Alcohol Need Index (based on deaths with explicit mention of alcohol, driving under the influence arrests, and liquor law violation arrests), New Mexico had the nation's highest alcohol treatment needs, while Hawaii had the lowest alcohol treatment needs (Figure 4.11). In contrast to the states with the highest drug treatment needs, the states in the highest quartile were rural and in the West, except for North Carolina. Alaska'salcohol needs ranked second in the country, while Colorado's ranked 5^th, Montana's ranked 7^th, and North Dakota ranked 14^th. Rhode Island ranked 34^th.

Causes of Death with Explicit Mention of Alcohol.A key component of the ANI, alcohol mortality has long been used as an indicator of the severity of an area's alcohol problems. The authors' own research suggests that deaths with explicit mention of alcohol is the best available indicator of alcohol abuse (McAuliffe et al. 1999, 2000). Inspection of the distribution of this indicator reveals that Alaska had the highest rate in the country, and Colorado (ranked 3^rd ) and Montana (7 ^th )were closely behind it (Figure 4.12). Rhode Island ranked 13^th with regard to the alcohol mortality indicator, and North Dakota was also high in the distribution of deaths with explicit mention of alcohol (18^th). Hawaii had the lowest explicit alcohol mortality rate in the country. Alaska's alcohol mortality rate was nearly four times higher than Hawaii's.

Alcohol Defined Arrests. As shown in Figure 4.13, there was wide variation among states in the rate of alcohol-defined arrests, including drunk driving and liquor law violation arrests.For the alcohol need index, the authors wanted to include a measure of alcohol-defined arrests and decided to focus on drunk-driving arrests and liquor law violations. Drunk driving is the most common cause of arrest in the United States, and a large percentage of drunk drivers have been found to have alcohol use disorders according to a variety of standardized measurements (see review in McAuliffe et al. 1999). Liquor law violation arrests are often arrests for underage drinking (in the median state, 25% of arrestees for liquor law violations were adolescents). Other alcohol-related arrest statistics, such as drunkenness, were not used because of the inconsistency among states with regard to these offenses (e.g., 21 states report drunkenness arrests, the rest do not). South Dakota had the highest rate of alcohol-defined arrests, while Rhode Island had the lowest rate. Colorado (5^th ), Alaska (7 ^th), Montana (12^th ), and North Dakota (13 ^th ) had high rates of alcohol arrests.

DUI's and Driving After Drinking Too Much. Survey data and arrest rates on drinking and driving were inconsistent indicators in some states (Figure 4.14). With regard to the Centers for Disease Control's Behavioral Risk Factor Surveillance Survey (BRFSS) statistics on the percentage of the population that reported driving after drinking too much, the highest percentage was reported in Nevada, and the lowest percentage was in Maine. The highest rate of Uniform Crime Report driving under the influence (DUI) arrests in 1994-1996 was in North Carolina, and the next highest rate was in New Mexico. North Dakota ranked 3^rd nationally in the BRFSS for driving after drinking to much, but 30^th on the DUI rate . Colorado ranked 10^th in driving after drinking too much and 4 ^th on DUIs per 100,000. Alaska ranked 31^st in the BRFSS driving and drinking measure but 6 ^th in the DUI arrest rate . Montana ranked 12^th in the BRFSS measure and 16^th in the DUI rate . Even though Rhode Island ranked 21^st nationally according to the BRFSS measure, its drunk-driving arrest statistics ranked dead last in 1994-1996 and in 1991-1993. Because of the wording of the BRFSS question, it is possible that the survey and arrest statistics focus on different parts of the drinking and driving picture.

Another explicit-mention alcohol mortality measure is the state rates of motor vehicle deaths in which the driver or a non-occupant had a blood alcohol concentration equaling or exceeding .10, a level exceeding the official drunk driving standard. New Mexico's rate was the highest in the country, which comes as no surprise since its ANI was also the highest (Figure 4.15). Montana's alcohol-related motor vehicle fatality rate was 5^th highest in the country, substantially higher than its DUI rate. Alaska, Colorado, and North Dakota had rates that were above average. Rhode Island's relatively low DUI statistics were confirmed by the State's ranking of 49^th lowest with regard to motor vehicle fatalities in which the blood alcohol content was at least .10 percent. Rhode Island's overall motor vehicle fatality rate was 50 ^th in the country for 1993.

Summary: Alcohol Treatment Needs. Analysis of the Alcohol Need Index and related measures showed that Alaska, Colorado, Montana, and North Dakota had relatively severe rates of alcohol problems. Rhode Island's level of need was slightly below average (34^th nationally). These index scores were generally confirmed by other alcohol-related indicators.

Alcohol Treatment Clients and Admissions

Colorado had the highest rate of TEDS alcohol treatment admissions in 1994-1996, and it ranked second with regard to the estimated rate of alcohol-only treatment clients reported by the UFDS survey (Figures 4.16 and 4.17). Colorado ranked 4^th in the country in the NDATUS alcohol-only measure in 1991-1993. Alaska ranked 2^nd in the nation in the TEDS measure, but 11^th in the UFDS measure. A similar divergence is apparent for Montana. There the alcohol-only client rate ranked 32^nd nationally, and the TEDS primary alcohol treatment rate ranked 9^th nationally. North Dakota's values for both measures ranked nearly the same: 22 ^nd in alcohol-only treatment clients and 23^rd in TEDS alcohol treatment admissions. In 1991-1993, North Dakota ranked 21 ^st in the NDATUS alcohol-only client measure. Rhode Island ranked the same (17^th) in both the 1994-1996 TEDS and UFDS measures. It ranked 13^th in the 1991-1993 NDATUS measure.

Comparison of Alcohol Treatment Need and Services

Figure 4.18 compares states with regard to the Alcohol Treatment Need Index and the number of people per 100,000 who were in treatment for alcohol only on a single day. For ease of inspection, the measures are in quintiles. The analysis describes how states compare to each other in meeting their residents' alcohol treatment needs, but it does not describe the absolute level at which states are meeting those needs. Research suggests that all states have been unable to meet the need and demand for alcohol treatment.

Even when persons who are being treated for both alcohol and drugs are included, Nevada stood out as the state that did not have as many alcohol treatment clients as one might expect based on its alcohol mortality and arrests. Nevada's alcohol-only client rate ranked 43 ^rd in the nation despite the State's relatively high alcohol mortality rate, maternal drinking, motor vehicle fatalities associated with drunkenness, chronic and binge drinking, and DUI rates. Nevada ranked 10 ^th in the country with regard to alcohol treatment need. New York (35 ^th in need, 11^th in treatment admissions) and Kentucky (38 ^th in need, 1 ^st in treatment clients) had the highest levels of clients compared to the states' levels of need.

Rhode Island appeared to be exceeding most states in its treatment of alcohol problems when need is taken into account. Rhode Island ranked 34^th with regard to alcohol needs and 17 ^th with regard to alcohol-only treatment clients per 100,000 residents. As with drug problems, Rhode Island has done a relatively good job providing treatment services to its residents.

Colorado's has also done a good job relative to other states. Colorado ranked 5^th with regard to the Alcohol Need Index and second with regard to treatment clients. It was first with regard to alcohol treatment admissions.

Alaska's relative performance depends on which measure of treatment supply one uses. Alaska ranked second in the country with regard to treatment needs. The State also ranked second with regard to the primary alcohol treatment admissions per 100,000, but it ranked 11 ^th with regard to alcohol-only treatment clients. Even when one includes treatment clients that were treated for both alcohol and drugs, the State's client rate is slightly below its treatment need rank. North Dakota's services were slightly below what would be expected on the basis of its residents' treatment needs. The State ranked 14 ^th in need and 22 ^nd with regard to clients in treatment.

Montana ranked 7^th in the Alcohol Need Index, but it ranked 32^nd in the alcohol-treatment client rate. However, the State ranked 9^th in the primary alcohol treatment admissions rate. Consequently, the State's position on the grid may overstate the gap between alcohol need and treatment services relative to other states.

Summary

Overall, the analysis indicated that Rhode Island provided services relative to need at an above-average level among the states for both alcohol and drugs. Colorado's services matched its rank on drug needs and was above average with regard to alcohol. Alaska's drug services were on par with its level of drug treatment need. Alaska's performance regarding alcohol treatment depended on the treatment measure that was used. The State's needs and alcohol treatment admissions matched perfectly, but the alcohol-only treatment client rate was below its treatment needs. North Dakota provided services that matched it low drug need, but its alcohol services were below average relative to its citizens' needs. Montana's success in meeting its residents' needs also depended on which measure was used. For both drugs and alcohol, Montana's treatment admissions rate was much more favorable than its treatment client rate.

Substance Abuse Treatment Needs and Services

Substance Abuse Need Index. Because this index combines the previously presented drug and alcohol need indexes, there should be few surprises. Most notable is the extent to which alcohol needs dominate drug needs because the number of deaths and arrests associated with alcohol is greater than the number of deaths and arrests associated with controlled drugs. As Figure 4.19 shows, New Mexico (99) had the highest SNI score, while Pennsylvania had the lowest (35). While New Mexico's score indicates that it had among the highest rates of both combined substance related mortality and arrests, Pennsylvania's score indicates that no state is even close to being free of substance abuse deaths and arrests.

Colorado and Alaska ranked 2 ^nd and 3 ^rd in the country respectively. Both had very high alcohol need scores (87 and 86 respectively). Montana's SNI ranking was 11 ^th, most clearly attributable to its high alcohol treatment needs. North Dakota ranked 24 ^th in the country, entirely a function of its severe alcohol treatment needs. Finally, Rhode Island ranked 35 ^th in the country, despite its significant drug treatment needs.

Substance Abuse Treatment Clients. New York State had the highest total treatment client rate in the country for 1994-1996, and Washington had the second highest treatment client rate (Figure 4.20). Minnesota and Arkansas had the lowest rates in the country. Rhode Island ranked 4^th and Colorado ranked 5 ^th among states. Alaska ranked 14^th. North Dakota ranked 32 ^nd , while Montana ranked 42^nd.

Figure 4.21 Substance Abuse Treatment Needs and Services
Substance Abuse Treatment Client Rate (Ranks)	Substance Abuse Need (SNI) Ranks
	Lowest (41-50)	Low (31-40)	Moderate (21-30)	High (11-20)	Highest (1-10)
Highest (1-10)		RI, KY	MI, ME	NY, MD WA, DE	OR,CO
High (11-20)	OH, MA	CT, IL	UT, FL		NM, AK, WY, CA
Moderate (21-30)	VA, VT, IN, PA	LA	NE, KS, NJ	SC, WI
Low (31-40)	WV	MO, NH	OK, ND	ID	SD, NC, AZ, NV
Lowest (41-50)	TX, AL, HI	TN, AR, IA	GA	MT, MN, MS

Comparison of Substance Abuse Treatment Needs and Treatment Clients

As Figure 4.21 shows there was wide variation among states in the extent to which the total treatment client rate was consistent with the SNI scores. Alaska's overall need for substance abuse services was among the highest in the country, while its services measured by the UFDS treatment clients ranked in the second quintile (11^th ). That placed the State slightly below average when compared to other states. This position was due to the State's very high alcohol treatment needs and above average treatment clients. The State's drug need and clients were in the same quintiles. Also, Alaska's treatment admissions rate was second in the country, so by that measure its needs and services matched closely. Colorado ranked in the top ten states with regard to both substance abuse treatment needs and treatment clients. North Dakota was ranked in the middle on both need and supply of services. Montana's treatment services were several quintiles below its high treatment needs. As noted earlier, Montana's TEDS treatment admissions rates ranked higher than its UFDS treatment client rates, and therefore Montana's apparent underperformance may be an artifact of the measurement method. Rhode Island's treatment client rate was among the highest in the country, while its treatment needs were below average, due mainly to a relatively low rank on the alcohol measures.

The analysis of Alaska's and Montana's relative rankings is a clear example of why measuring the adequacy of a state's treatment needs requires valid measurement of both treatment need and supply. If the treatment admissions rather than the treatment clients had been the focus of analysis, the states' performance would have improved substantially. The authors have previously conducted research on the TEDS and UFDS measures, and the findings indicated a slight edge in favor of the UFDS measure. While the two measures correlate with each other reasonably, they are not identical. One measures only public treatment paid by the states but covers all admissions in a year, while the other covers both public and other treatment (e.g., private, tribal, Indian Health Service, etc.) clients on a single day. As explained in the methods section, the use of the UFDS measure in the comparisons with need was dictated by completeness rather than obvious superiority. Accordingly, the authors believe that concern regarding a state's performance in this analysis is warranted only if both measures clearly indicate below average performance.

Appendix

References

McAuliffe, W.E., R.A. LaBrie, N. Pollock, N. Lomuto, E.A. Fournier, and R. Betjemann. (1999). "Measuring Interstate Variations in Drug Problems." Drug and Alcohol Dependence. 53 (2): 125-145.

McAuliffe, William E., Richard A. LaBrie, Nicoletta A. Lomuto, Nancy E. Pollock, Rebecca Betjemann, and Elizabeth Fournier. (2000). "Measuring Interstate Variations in Problems Related to Alcohol Use Disorders." The Epidemiology of Alcohol Problems in Small Geographic Areas . vol. 36. Eds. Robert A. Wilson and Mary C. Dufour. Bethesda, MD: National Institute on Alcohol Abuse and Alcoholism. 213-44.

McAuliffe, William E., Richard A. LaBrie , and Ryan Woodworth. (2001).Measuring Interstate Variations in Substance Abuse Treatment Needs . Cambridge, MA: National Technical Center for Substance Abuse Needs Assessment.

Households With Telephones

Introduction

This chapter summarizes the results of the household telephone survey conducted in North Dakota by The Gallup Organization in 1996 and 1997. The estimate of treatment need from the household survey is by far the most important component of the integration study because it captures the largest share of the overall population. In 2000, 95.8% of households in North Dakota had telephones (Belinfante 2001), and only a small percentage of the population does not live in a household. The treatment need estimate from this survey, combined with treatment need estimates from subsequent studies of smaller population subgroups will provide an overall estimate of the substance use disorder treatment needs of North Dakotans. The telephone survey also provided an estimate how many of those in need of treatment would like to receive treatment. Comparing that figure with an estimate of how many people received treatment in North Dakota in the past year will make clear the extent of the gap between treatment demand and services.

Methods

The Adult Household Telephone Survey was administered by The Gallup Organization in partnership with the Division of Mental Health and Substance Abuse Services of the North Dakota Department of Human Services. The survey was implemented using a computer-aided telephone interview (CATI) system and a random-digit-dialing sampling procedure designed to avoid unproductive calls to non-household telephone lines (Casady & Lepkowski 1993). The sample consisted of 6,814 North Dakotans ages 18 and over that lived in households with at least one telephone at the time of the survey. The sample was designed to provide substance use disorder prevalence estimates for each of eight planning regions; referred to as Regions 1 through 8. The survey employed the National Technical Center's (NTC) procedure for optimal allocation of the sample. The 18-45 age bracket was oversampled as well in order to obtain estimates for illicit drug use. The survey data provided prevalence estimates of substance use disorders and demand for treatment among North Dakota's American Indians, along with women with dependent children and injection drug users. Response rates for this study were computed using the Council of American Survey Research Organizations' (CASRO) (1982) method.

The telephone survey used the National Technical Center's Needs Assessment Questionnaire (McAuliffe, LaBrie, et al.1995) which measured demographics, alcohol and drug use, DSM-III-R diagnoses of substance use disorders, treatment history, unmet demand for services, treatment service mix, and obstacles to obtaining treatment services. Designed for telephone administration, the instrument included measures drawn from the questionnaires of the Behavioral Risk-factor Surveillance System (BRFSS) (Remington, Smith, et al. 1988; Gentry & CDC, 1989; Bradstock, Forman, et al.1988) and the National Household Survey on Drug Abuse (NHSDA) (SAMHSA & Office of Applied Studies, 1997), as well as the Diagnostic Interview Survey (DIS) (Robins, Cottler, et al. 1990, 1995; Robins, Helzer, et al. 1981, 1989).

The instrument enabled the survey researchers to assess the DSM-III-R's clinical criteria for a substance use disorder for each case (American Psychiatric Association 1994). Several investigators have published results indicating that the diagnoses have reliability and validity (Aktan, Calkins, et al. 1997;Alemagno, Cochran, et al. 1996). Using the data, the authors calculated the overall percentage of North Dakota adults that had a substance use disorder (dependence or abuse) in the past year. Following the recommendations of the National Technical Center, the study assumed that a person with a current substance use disorder was in need of some form of treatment. Further, the study estimated the four levels of care (hospital inpatient, residential, day treatment, outpatient) required by those in need. The level of care estimates were based on the patient placement criteria of the American Society of Addiction Medicine (ASAM) (Hoffmann et al. 1991). This chapter employed the prevalence rates produced by the household telephone survey and population estimates from the 2000 Census (2001). In order to provide perspective for North Dakota's data, the chapter will present household survey data from similar telephone surveys in Montana and Alaska (Table 5.1) (Haynes, et al. 2001; The Gallup Org. 1998b).

Results

The North Dakota Adult Household Survey was conducted between November of 1996 and August of 1997. It interviewed 6,814 non-institutionalized individuals living in households with telephones in North Dakota. Interviewers achieved a 74% percent response rate for the general statewide sample and more than 68% in each of the substate planning regions. That statewide response rate was greater than the rate obtained in Alaska, but less than the rate in Montana. All of the surveys met the NTC's recommended 70% response rate (Table 5.1).

Demographics

The survey respondents from the three states had similar demographic characteristics in most respects (Table 5.2). The average adult respondents were in their early 40's, and slightly more than half were females. While whites comprised the largest racial group in the samples (95% in North Dakota), the Alaska and Montana samples were more than one-fifth American Indians or Alaska Natives. There were few African Americans, Hispanics, or Asian Americans in any of the three states. The adult samples had similar educational levels and employment rates, although North Dakota's employment rate was slightly lower than the rates in Alaska and Montana. In subsequent analyses, the data have been weighted to adjust for oversampling.

Substance Use Rates

Among the three states, Montana had slightly higher levels of alcohol use, while Alaska had the highest drug use rates for most time periods (Table 5.3). North Dakota had the lowest drug use rates; Montana's drug use rates were consistently between the other two. As is typical for such statistics, the patterns for marijuana use closely followed estimates of any illicit drug use. It is noteworthy that the state studies estimated past month alcohol use at the same level as the BRFSS telephone survey. Both surveys used the same item to obtain those estimates.

Lifetime Substance Use Disorder Rate

The North Dakota household telephone survey estimated that 9.6% of North Dakota adults aged 18 and older had a lifetime substance use disorder diagnosis. The percentage for respondents aged 18 to 65 was 10.9% (Table 5.4). Alcohol use disorders were most common: 13.2% of the men and 4.0 % of the women surveyed had a lifetime alcohol use disorder (Gallup 1998). In the current study, having a substance use disorder (either abuse or dependence) implies a need for treatment, but only a past year substance use disorder implies a current treatment need. According to the report, North Dakotan men experienced a lifetime need for alcohol treatment that was three times the needs of women, and a need for drug treatment that was four times the women's rate. Dependence was more common than abuse in both the case of drugs and alcohol (Gallup 1998). North Dakota's lifetime substance use disorder rate for 18 to 65-year-olds was lower that Alaska and Montana's.

Current Substance Use Disorders: Need for Treatment

Based on data from the household telephone survey, NCRPG estimated that 5.2% of North Dakotans over the age of 18 living in households with telephones had a current substance use disorder. In 2000, that estimate represented over 23,000 people (Table 5.5). North Dakota's rate was less than Montana's and Alaska's for the 18 to 65 age group. The lion's share of the residents in need of treatment in all three states consisted of people with alcohol use disorders. Of those between the ages of 18 and 65 surveyed in North Dakota, 5.9% had a past year alcohol use disorder while less than 1% had a past year drug use disorder. Among all persons ages 18 and older, there was a 0.1% rate of marijuana dependence (Gallup 1998a). Statewide estimates of dependence and abuse for all other drugs were 0%. There were some non-zero estimates for drugs other than marijuana at the regional level, but only one where zero was not within the standard error of the point estimate (a 0.2% rate of stimulant dependence in the Northwest region with a standard error of 0.1)(Gallup 1998) . The preponderance of marijuana problems over other drug use disorders is consistent with the high proportions of marijuana arrests and treatment admissions demonstrated in the previous chapter (the interstate chapter).

*Includes hallucinogens, cocaine, and opiates in all states.

Comparison with 1999 National Household Survey on Drug Abuse

Interestingly, the 1999 National Household Survey on Drug Abuse (NHSDA) produced estimates of substance dependence for North Dakota and Alaskathat were nearly as high as the total substance use disorder estimates from the telephone surveys in those states (Table 5.4) (OAS 2000). The past year substance dependence estimate from the NHSDA for Montana was four percentage points lower than the substance use disorder estimate from the telephone survey. Consequently, the NHSDA would rank the states differently than the telephone surveys in terms of relative need for treatment. According to the NHSDA, Alaska had the highest past year substance dependence rate among the three (7.3%), while Montana had the lowest rate (OAS 2000). The NHSDA uses the 12 and older population as its base, whereas, the estimates cited above use the 18 to 65 population as a base. Therefore, it is difficult to know how to interpret these differences. In most studies, adolescents have higher dependence rates than adults, especially when older adults are included. It is reasonable, therefore, that the NHSDA dependence estimates are somewhat higher than the telephone survey dependence estimates. The NHSDA indicated that adults in Alaska continued to experience relatively high rates of substance dependence after the age of 26 (6.3%), and because that is the largest population group in the NHSDA sample, the rate for the State was higher. The NHSDA's sample also included residents of households without telephones, whereas the telephone survey does not. As shown in the next chapter, that difference alone could account for the relatively high NHSDA estimate.

Table 5.5 Need for Treatment, Met and Unmet Demand Among North Dakota Adults Living in Households with Telephones (18 and older)
	Current Substance Use Disorder		Lifetime Substance Use Disorder
	%	Number	%	Number
Need for Treatment	7.4	29,739	14.3	57,468
Met Treatment Demand Among Those In Need of Treatment	14.7	4,372	-	-
Unmet Treatment Demand Among Those In Need of Treatment	1.8	457	-	-
The lifetime and current numbers in need of treatment were estimated using the 18 and older population count from the 2000 Census.

Met Demand for Treatment

Those who received treatment for a substance use disorder in the past year may be thought of as having a "met demand" for treatment, though some may not have received the amount or type of treatment they wanted and some may have been ordered to obtain to treatment by the courts. The North Dakota Adult Household Survey revealed that 7.8% of those with a current diagnosis for a substance use disorder received treatment in the past year (Table 5.5). That was 1,796 North Dakotans in 2000. That estimate included both specialty and non-specialty treatment, as reported by respondents in the telephone survey. Men with a past-year need were more likely to receive treatment than women.

Unmet Demand for Treatment

Those who needed treatment in the past year and wanted it, but did not receive any can be considered to have an "unmet demand" for treatment . Among the respondents in need of treatment who did not receive it during the past year, less than 2% reported that they would have sought treatment if it had been readily available. That estimate implied that only 425 North Dakota adults who needed treatment in the past year but did not get any, perceived they had a need for treatment and wanted it. However, 15.2% of those with a current diagnosis who did receive treatment in the past year reported that they would have liked additional or more intensive treatment than they received. That group's needs must be taken into account when assessing the need for new services. Women with a past-year need were far more likely than men to express an unmet demand for treatment.

The Treatment Gap

Because the telephone survey estimate of need comprises the largest part of total statewide estimated need for treatment, it is informative to examine the gap between reported need and reported treatment services from the telephone survey. Whereas 23,031 of those surveyed needed treatment in the past year, only 1,796 received treatment for a substance use disorder. Hence, 7.8% of total need for treatment among adults in households with telephones was accommodated to at least some degree. If the State were to provide treatment services to all of the persons estimated to need it, services would have to increase by almost 13 times the current level. Of course, the fact that a person was diagnosed as having an addictive disease does not imply that the individual would seek treatment if it were readily available. Given the rate of unmet demand for treatment in North Dakota, treatment services would have to be increased by 23.7% of their current level in order to serve everyone with a demand.

Summary

There is a moderate level of need for treatment for substance use disorders in North Dakota. Over 5% of the adult household population had a need for treatment in the past year. Less than 8% of those with a need for treatment received some form of treatment in the past year. Among those in need of treatment who did not receive treatment in the past year, only 2% expressed a desire to get treatment. Because the percentage of those that expressed an unmet demand was low, the State may wish to encourage demand for treatment among those in need by employing outreach and other techniques designed to remove obstacles to obtaining treatment. Potential target groups for such an effort include American Indians and people between the ages of 18 and 25.

References

Aktan, Georgia B., Richard F. Calkins, Kurt M. Ribisl, Alice Kroliczak, and Rafa M. Kasim. (1997). "Test-Retest Reliability of Psychoactive Substance Abuse and Dependence Diagnoses in Telephone Interviews Using a Modified Diagnostic Interview Schedule - Substance Abuse Module." American Journal of Drug and Alcohol Abuse. 23(2):229-48.

Alemagno, Sonia A., Deborah C. Cochran, Thomas E. Feucht, Richard C. Stephens, John M. Butts, and Stephanie A. Wolfe. (1996). "Assessing Substance Abuse Treatment Needs Among the Homeless: a Telephone-Based Interactive Voice Response System." American Journal of Public Health. 86(11):1626-28.

American Psychiatric Association. (1994). Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition): DSM-IV. Washington, DC: American Psychiatric Association.

Belinfante, Alexander. (2001).Telephone Subscribership in the United States (Data Through November 2000). Washington, D.C.: Federal Communications Comission.

Bradstock, M. K., Michele R. Forman, Nancy J. Binkin, Eileen M. Gentry, Gary C. Hogelin, David F. Williamson, and Frederick L. Trowbridge. (1988). "Alcohol Use and Health Behavior Lifestyles Among U.S. Women: the Behavioral Risk Factor Surveys." Addictive Behaviors. 13(1):61-71.

Casady, Robert J. and James M. Lepkowski. (1993). "Stratified Telephone Survey Designs." Survey Methodology. 19(1):103-13.

Council of American Survey Research Organizations (CASRO). (1982).On the Definition of Response Rates: A Special Report of the CASRO Task Force on Completion Rates. Port Jefferson, NY: Council of American Survey Research Organizations.

The Gallup Organization. North Dakota Adult Household Needs Assessment Survey, North Dakota Adult Household Telephone Survey Statewide and Substate Planning Regions. (1998a). North Dakota Department of Health and Social Services, Division of Alcoholism and Drug Abuse.

The Gallup Organization.North Dakota Adult Household Telephone Survey . 1998b. North Dakota Division of Mental Health and Substance Abuse Services.

Gentry, Eileen M. and CDC. 1989. Behavioral Risk Factor Surveillance System: Operations Manual . Atlanta, GA: Office of Surveillance and Analysis, CDC.

Hoffmann, Norman G., James A. Halikas, David Mee-Lee, and Richard D. Weedman. (1991).Patient Placement Criteria for the Treatment of Psychoactive Substance Use Disorders. Washington, DC: American Society of Addiction Medicine.

Kish, Leslie. (1965).Survey Sampling. New York, NY: John Wiley & Sons.

McAuliffe, William E., Richard A. LaBrie, Norah Mulvaney, Howard J. Shaffer, Stephanie Geller, Elizabeth A. Fournier, Eliot B. Levine, Qiuyun Wang, Susan M. Wortman, and Kathleen A. Miller. (1995). Assessment of Substance Dependence Treatment Needs: A Telephone Survey Manual and Questionnaire, Revised Edition . Cambridge, MA: National Technical Center on Substance Abuse Needs Assessment.

Office of Applied Studies and Research Triangle Institute. (2000). Summary of Findings from the 1999 National Household Survey on Drug Abuse. Substance Abuse and Mental Health Services Administration.

Remington, Patrick L., Meredith Y. Smith, David F. Williamson, Robert F. Anda, Eileen M. Gentry, and Gary C. Hogelin. (1988). "Design, Characteristics, and Usefulness of State-Based Behavioral Risk Factor Surveillance: 1981-87." Public Health Reports. 103(4): 366-75.

Riley, William, George Haynes, and Vincent Smith. (1997). Montana Adult Household Substance Abuse Treatment Analysis. Montana Department of Public Health and Human Services.

Robins, Lee N., Linda B. Cottler, and Thomas F. Babor. (1990). Diagnostic Interview Schedule-- Substance Abuse Module (DIS-SAM). St. Louis: Department of Psychiatry, Washington University School of Medicine.

Robins, Lee N., Linda B. Cottler, Kathleen K. Bucholz, and Wilson M. Compton III. (1995). Diagnostic Interview Schedule for DSM-IV (DIS-IV). St. Louis, MO: National Institute of Mental Health.

Robins, Lee N., John E. Helzer, Linda B. Cottler, and E. Goldring. (1989). NIMH Diagnostic Interview Schedule Version III Revised (DIS-III-R). St. Louis: Department of Psychiatry, Washington University School of Medicine.

Robins, Lee N., John E. Helzer, J. L. Croughan, and K. S. Ratcliff. (1981). "The NIMH Diagnostic Interview Schedule: Its History, Characteristics and Validity." Archives of General Psychiatry . 38:381-89.

Substance Abuse and Mental Health Services Administration (SAMHSA), Office of Applied Studies and Research Triangle Institute. (1997). National Household Survey on Drug Abuse: Population Estimates 1996. Rockville, MD: U.S. Department of Health and Human Services.

United States Census Bureau. (2001). State and County QuickFacts : Population Estimates, 2000 Census of Population andHousing. www.quickfacts.census.gov/qfd/states/30000.html.

Households Without Telephones

This chapter estimates the amount of potential noncoverage bias in estimates of the need for treatment that may result because telephone substance abuse surveys fail to include households without telephones. Of the populations missed by a telephone survey, the largest group consists of residents of households without telephones (McAuliffe et al. 1998).If the substance use disorder rate of residents of households without telephones differs from the rate of residents of all households, failing to adjust for this difference would result in bias if a telephone survey's estimates were generalized to the entire household population. This shortcoming of telephone surveys is of course widely recognized by scientists and lay persons, but few researchers can afford to eliminate it by conducting a face-to-face survey or a multi-mode survey that covers households without telephones (McAuliffe et al. 1998). To address the issue, authors routinely caution readers in publications of the results of telephone substance abuse surveys.

These caveats provide readers with little guidance to interpret the survey's results. The amount of bias depends upon the proportion of households without telephones and the size of the difference between rates for households with and without telephones. In the last decade, the national rate of households without telephones was between 5% and 6%. The median state rate was 5% in 1990 and 5.4% in 2000 (Bureau of the Census 1994; Belinfante 2001). The Census Bureau's most recent national estimate of the percentage of households without telephone coverage was 5.6% in 2000(Belinfante 2001). The range for individual states was narrow: only two states have rates in excess of 10% (Mississippi 10.2% and Arkansas 11.4%). Consequently, the size of the difference between households with and wit hout telephones would have to be substantial to cause a meaningful amount of bias for most purposes. For example, if a state's telephone noncoverage rate were 5% and the substance use disorder rate for the residents of households with telephones were 10%, the rate for residents of households with telephones would have to be 30%, or three times greater, in order to increase the estimated rate for residents of all households by one percent, that is, to 11%. If the difference in the substance abuse rate between households with and without telephones were smaller than 30%, even though the difference was statistically significant, the amount of bias may be no more than a few tenths of a percent. Removing bias of this magnitude may not be important enough for many substantive questions to increase the survey cost significantly. Many publications report results rounded to whole percentage points or to one decimal place (e.g., National Houshold Survey on Drug Abuse). However, many readers do not know that the bias is likely to be that size. It is therefore important to estimate the likely impact of excluding households without telephones so that investigators may adjust their results or provide readers with more precise guidance regarding the accuracy of their results (McAuliffe et al. 1998).

Telephone Noncoverage Bias in Prior Research

Unfortunately, published research on this issue is too limited and the results are too inconsistent to provide the estimates needed for this purpose from existing statistics. Several studies have found that failure to include households without telephones had only modest effects on most health indicators (Anderson et al. 1998; Ford 1998 ), but some authors have argued that substance use may be an exception (Aquilino 1992;Gfroerer & Hughes 1991; Pearson et al. 1994). Using data from the 1994 National Health Interview Survey (n=12,035 to 83,719), Anderson et al. (1998) found that one of the largest differences between all households and households with telephones concerned current smoking (25.4% vs 24.4%, a difference of 1%), but the difference was greater for households in poverty (36.5% vs 33.3%). Using data from the National Health and Nutrition and Examination Survey conducted between 1988 and 1991 (n=10,120), Ford (1998) found that participants with and without telephones did not differ significantly with regard to alcohol consumption, but the trend suggested more drinking by people in households with telephones. Aquilino (1992) reported that a face-to-face survey of 1,002 New Jersey household residents conducted in 1986 found that a higher percentage of people with telephones than people without telephones reported use of alcohol in the last month and in the last year. Aquilino (1992, Table 4) also found the same effect for ever use of marijuana, but an opposite effect for the amount of alcohol use in the last month, marijuana use in the last month, and cocaine use in the last month, year, and ever. Only one of these differences was statistically significant. Moreover, Aquilino reported his results as whole percentages, and therefore the impact of these latter differences on the percentages for all households versus households with telephones were 1% for past month marijuana and cocaine use and 0% for ever use and past month use of marijuana and cocaine.

Gfroerer & Hughes (1991, Table 3)reported the potential for telephone noncoverage bias in analysis of the 1985 and 1988 National Household Survey on Drug Abuse (n= 5,751 in 1985 and 8,814 in 1988). They found that people in households without telephones reported greater lifetime and past month use of marijuana and cocaine than people in households with telephones. The percentage differences between all households and households with telephones ranged from 0.3% for past year cocaine use and 1.3% for lifetime marijuana use in 1988; the differences were smaller in 1985, although in the same direction. Pearson et al. (1994) reported significant differences between households with and without telephones in a face-to-face survey of 410 young adult American Indians; the odds ratio for past month marijuana use was 3.5. The present authors could find no study that reported data on rates of substance use disorders or treatment utilization. No studies reported the effect for Hispanic populations. Overall, the number of studies is small, the findings were incomplete and somewhat inconsistent, the geographical coverage and sample sizes for several of the studies were limited, and most of the results for drugs are now somewhat out of date.

The present study used a large sample of recent data to test the hypothesis that excluding households without phones results in noncoverage bias in estimates of substance abuse and treatment needs if the results are generalized to all households, but the size of the bias is relatively small at the national level. The effects vary depending on the substance and the demographic composition of the target population.

Methods

The study employed public use data collected in the 1995, 1996, 1997, and 1998 versions of the National Household Survey on Drug Abuse (NHSDA). The accompanying public use codebooks downloaded from the Office of Applied Studies' web page provided the variable names and labels, value codes and labels, and composite variable definitions. The authors used the NHSDA's composite definitions as much as possible in order to produce results that are comparable to the findings reported in the codebooks and in published NHSDA documents. The NHSDA's target sample is the civilian, noninstitutionalized, 12 and older residents of households and group quarters. Interviews were conducted in person, with a mix of interviewer- and respondent-recorded (answer sheets) questions. The sampling design is a multi-stage, area probability sample with an in-household selection procedure to ensure that desired race/ethnicity and age subsample sizes are achieved (see SAMHSA & Office of Applied Studies 1999for details). The overall (screening x interview) response rates ranged from 72% to 76% for the four years (Office of Applied Studies 1996,1997a,b, 1999, 2000). Individual responses were weighted to adjust for probability of selection, household and person-level nonresponse, and to replicate census demographic distributions.

The present study analyzed the NHSDA interview data from the 84,796residents living in households and for whom the NHSDA were able to determine telephone subscribership. Beginning in 1991, the National Household Survey on Drug Abuse added a small, heterogenous subsample of persons living in noninstitutional group quarters (college dorms, shelters, etc.). Because persons in group quarters are often excluded from telephone household surveys, the study excluded them and persons lacking data (refusals, item nonresponse) on telephone subscribership from the analysis. The percentage of subjects living in households without at least one telephone for the combined four years of survey data was 6.8% for the unweighted NHSDA sample of respondents.

Results

Compared to respondents living in households with telephones, those without telephones reported higher rates of drug use, dependence, need for treatment, and having received treatment (Table 6.1). All of the differences between residents of houses with and without telephones were statistically significant at the .01 level. The median relative risk for residents of households without telephones was 2.13, with most risk ratios between 1.00 and 2.00 (6) and between 2.00 and 3.00 (10). Only one exceeded 3.00. Consistent with previous studies, persons in households without telephones reported lower rates of alcohol use in the past month and year than persons in households with phones.

Table 6.1 Effect of Excluding Households Without Telephones on Substance Use, Dependence, and Treatment Estimates
Substance	Measure	A. Households without Telephones (n=5,822)	B. Households with Telephones (n=78,974)	C. Relative Risk (A/B)	D. All Households (n=84,796)	% Difference Between All Households and Households with Telephones (D-B)	Ratio of All Households and Households with Telephones (D/B)
Alcohol or Illicit Drugs	% Dependent Past Year	10.16	4.78	2.13	4.99	0.21	1.04
	% Received Treatment in Past Year*	2.14	0.99	2.16	1.04	0.05	1.05
	% Received or Perceived Need for Treatment in Past Year*	4.6	1.6	2.87	1.7	0.1	1.07
	% Received Treatment Lifetime*	7.5	3.3	2.27	3.5	0.2	1.05
Alcohol	% Used Past Month	46.5	51.7	0.90	51.5	-0.2	.996
	% Used Past Year	56.9	64.8	0.88	64.5	-0.3	.995
	% DependentPast Year	8.3	4.0	2.09	4.2	0.2	1.04
	% Received TreatmentPast Year	1.34	0.78	1.72	0.81	0.03	1.04
Any Illicit Drug	% Used Ever	39.47	34.80	1.13	34.98	0.18	1.005
	% Used Past Year	17.95	10.37	1.73	10.67	0.30	1.03
	% Dependent Past Year	4.44	1.57	2.83	1.68	0.11	1.07
	% Received Treatment in Past Year	1.01	0.40	2.53	0.43	.03	1.08
Marijuana	% Used Ever	35.8	32.0	1.12	32.2	0.2	1.005
	% Used Past Year	14.5	8.3	1.76	8.5	0.2	1.03
	% Dependent Past Year	3.15	1.28	2.46	1.35	0.07	1.05
Cocaine	% Used Ever	13.4	10.2	1.31	10.4	0.1	1.01
	% Used Past Year	4.3	1.7	2.57	1.8	0.1	1.06
	% Dependent Past Year	1.25	0.25	5.00	0.29	0.04	1.16
Heroin /opiates	% Used Ever*	2.7	1.0	2.79	1.1	0.1	1.1
Note: The table presents the unweighted sample size and the weighted percentages. * The sample sizes are slightly smaller for these variables because of missing data, blanks, and item refusals.

Comparing the prevalence rates for the entire household sample with the rates of persons in households with telephones, the analysis showed that the impact of omitting households without telephones was generally small. The percentage difference between all households and households with telephones ranged from between -0.3% and 0.3%, with a mean of 0.09, median of 0.1%, and mode of 0.1%. The ratio of the rate for all households to the rate for households with telephones ranged from .995 to 1.16, with a mean and median of 1.04. In other words, the percentages for all households were 4% higher on average than the percentages for households with telephones alone.

There was lessbias for the alcohol than for illicit drugs, and there was less bias for ever use than past year dependence. The smallest indication of noncoverage bias for a specific drug was for ever use of marijuana, while the largest was for past-year cocaine dependence.

Bias and Demographic Characteristics

The amount of bias resulting from telephone noncoverage depends on the percentage of people without telephones and the relative risk of not having a telephone. As expected from previous research (e.g., Aquilino, 1992; Belinfante, 2001), the highest proportions of respondents living in households without telephones were among American Indians/Alaskan Natives (19.2% ) and African Americans (9.2%), p oor people (12.3%), Hispanics (8.6%), males (4.3%), and adolescents (4.1%) (Table 6.2). Significant variations in the amount of telephone noncoverage bias in dependence or treatment estimates were evident across each of these major demographic variables. There was a significant interaction between race and phone subscribership on past-year drug dependence. The greatest amount of bias (2.18 ratio of all households to households with telephones) was a mong American Indians/Alaska Natives (19.2% were without phones and the relative risk of drug dependence for people without a telephone was 6.8).

Lower income persons had more telephone noncoverage bias than people with higher incomeswith respect to the dependence measures, even though the relative risk of not having a telephone was significantly greater for the higher income group. Few respondents with incomes over $20,000 lacked telephones, although those who reported no phone had significantly higher rates of drug, alcohol, and substance use disorders than those with telephones. Adolescents had significantly lower rates of telephone noncoverage bias than adults had with regard to the dependence measures, an effect due entirely to the adolescents' lower relative risks. Hispanics had higher proportions of households without phones than non-Hispanics had, but Hispanics had much lower relative risks. As a result, Hispanics had relatively smaller differences than non-Hispanics between all households and households with telephones for all four of the measures in Table 6.2. Females had slightly more evidence of telephone noncoverage bias than males had for drug dependence and treatment, but females were less likely than males to live in households without telephones. The males and females had the same relative risks for alcohol dependence and substance dependence (i.e., the interaction between gender and telephone subscribership was not significant for those two variables).

Discussion

Analyzing data from four years of the National Household Survey on Drug Abuse, the study found that residents of households with telephones consistently reported less drug use, less dependence on drugs and alcohol, and less substance treatment, but more alcohol use than residents of households without telephones. The percentage differences between respondents with telephones and all respondents averaged one tenth of a percent, and ratios of all households to households with telephones averaged 1.04. Thus, adjusting the rates for households with telephones to estimate noncoverage bias would likely result in an estimate that is about 4% larger for the entire household population.

The amount of bias appeared to vary by substance, timeframe, use versus dependence, and demographic characteristics. Whereas alcohol use was higher in households with telephones, drug use was higher in households without telephones. Ever use of marijuana had less bias than past year cocaine dependence. Race and income had the greatest variations in the amount of potential bias that was evident.

Comparison With Previous Studies. The present study found smaller differences between all households and households with telephones in use of marijuana and cocaine than reported by Gfroerer and Hughes (1991). As Anderson et al. (1998) found with other health indicators, household drug use estimates from a telephone survey would not generally be increased substantially by noncoverage of households without telephones.

The present study was the first to report estimates of the effect of noncoverage of households without telephones on dependence and treatment estimates. Most likely due to their small samples, the previously published studies examined use of alcohol, marijuana and cocaine, but none of them considered dependence or treatment. The percentage difference between all household and households with telephones was 0.2% for dependence on alcohol or controlled drugs. Adding in households without telephones increased the estimated prevalence by a small proportion (0.04). Estimates of bias were larger for drug dependence than for alcohol dependence and were larger for cocaine dependence than for marijuana dependence.

Race and Ethnicity.In the present study, American Indians/Alaskan Natives had the greatest amount of telephone noncoverage bias, and the amount for African Americans was second. Asian and Pacific Islander respondents had the lowest evidence of noncoverage bias. Pearson et al. (1994) found significant odds ratios among Native Americans between residents with and without telephones, but his small study presented no information that permitted an investigator to estimate the amount of bias that these differences might produce. Because Native Americans have the highest rates of households without telephones of any racial group, the evidence in the present study should be especially useful for interpreting telephone survey results for that racial group. Aquilino (1992) found higher rates of noncoverage effects among African Americans than among the full sample, but his sample of African Americans was small (21 without telephones and 139 with). Findings from a much larger study reported by Gfroerer and Hughes (1991, Table 2) for African Americans and whites were inconsistent with Aquilino's findings: the differences in percentages who used a substance between households with and without telephones were smaller for African Americans than for whites in every comparison. The present study found that Hispanics had lower evidence of noncoverage bias than persons who were not Hispanics.

Age and Income. In the present study, adolescents had lower rates of telephone noncoverage bias than adults.In the only previous study that addressed the issue, Gfroerer & Hughes (1991) found that lifetime and past year rates of marijuana and cocaine use were higher for adults in nontelephone households than in telephone households in 1985 and 1988, but for adolescents in the telephone households respondents reported higher rates of drug use in 1985 and the reverse in 1988. Consequently, telephone surveys of adolescents appear to be less vulnerable to noncoverage bias. The present study found that low-income persons had more telephone noncoverage bias than people with higher incomes. This result was consistent with Anderson et al.'s (1998) finding with regard to tobacco.

Use of the Results. How much concern researchers and policy makers should have regarding these levels of potential bias depends upon the nature of their study or their intended use of the data. The relevance of these differences between subsamples of face-to-face survey respondents may generalize imperfectly to the differences between telephone and face-to-face surveys or the parts of a multi-mode study that used telephone interviewing for residents of households with telephones and face to face interviews for residents of households without telephones, but it is likely that the present results provide an indication of what is likely to be found.

In some research situations even very small systematic errors of estimation translate into large numbers of people. However, a review of many reports of substance abuse surveys in scholarly journals and state needs assessment study reports reveals that rates of substance use and use disorders were generally reported as whole percentages or at most to one decimal place. In those cases, a potential bias of 0.1% to .3% is likely to have relatively little or no effect on the study's results or conclusions.

From a scientific standpoint, it is also reasonable to argue that the NHSDA, the nation's premier substance abuse survey, should be as accurate as possible and should include groups, such as people in households without telephones and group quarters, that may be ignored by most studies that have limited budgets. Otherwise, important subgroups would be systematically missing from the nation's primary data on substance abuse, and of course there would be no data available for studies such as the current one.

For many state and local studies of the general population, however, it is often only financially feasible to conduct telephone surveys or perhaps mixed-mode surveys. The cost savings of a telephone survey is substantial compared to a face-to-face survey (McAuliffe et al. 1998), and the bias reported in the present study is within the level of accuracy required by many state and local survey users. For the same amount of money, a face-to-face survey would have a fraction of the number of interviews and would therefore have much more random error of estimation than a telephone survey would have. The tradeoff is therefore between systematic error and random error. For studies that require a high level of precision but lack the resources to conduct face-to-face interviews, estimates from the current study can be used with estimates of the number of people in households with telephones to adjust the results for noncoverage bias. In any case, researchers can use the present results to provide readers with more informed guidance regarding their study's accuracy.

State Estimates

Using the information from the analysis, the authors estimated the percentage of residents in all households who had a current substance use disorder (treatment need). The estimate was imputed by constructing a weighted sum of the treatment need prevalence of people living in households with telephones from the state's telephone survey plus the treatment need rate of residents in need living in households without telephones. For example, in North Dakota's the calculation first obtained the percent of the household population who were people in need in households with telephones (telephone survey estimate of the percentage with a substance use disorder diagnosis x the proportion of people in households with telephones= 5.2% x .958= 4.982%) and then imputes the estimate of the prevalence rate for people in households without telephones by multiplying the telephone prevalence rate times the relative risk estimated in Table 1 above (telephone rate x relative risk= 5.2 x 2.13= 11.076). That imputed estimate is then weighted by the proportion of people in households without telephones (11.076 x .042= .465). In this instance, the authors assumed that the percentage of people living in households without telephones was equal to the percentage of households without telephones. The two weighted prevalences were then added together (4.982 + .465= 5.4). The result is an imputed estimate of the percentage of people in need who live in all households. These estimates will be incorporated into the integration study to adjust for noncoverage of households without telephones.

Table Appendix 6.1 NHSDA's response rates
Year	Sample Size	Screening	Interviewing	Overall Response Rate	Source
1995	17,747	94.2%	80.6%	75.9%	Office of Applied Studies 1996, 1997a
1996	18,269	92.7%	78.6%	72.9%	Office of Applied Studies 1997b}
1997	24,505	92.7%	78.3%	72.5%	Office of Applied Studies 1999
1998	25,500	93%	77%	71.6%	Office of Applied Studies 2000
Total	103,830

Table Appendix 6.2 Survey Access Groups
Persons Living In:	1995	1996	1997	1998	Total	%
Households with telephones	16,226	16,765	22,571	23,412	78,974	91.8
Households without telephones	1,300	1,273	1,605	1,644	5,822	6.8
Household Sample	17,526	18,038	24,176	25,056	84,796	98.6
Group Quarters	141	135	203	267	746	0.9
NHSDA's Analyzed sample	17,667	18,173	24,379	25,323	85,542	99.4
Missing, Don't Know, or Refused for Telephone Coverage	80	96	126	177	479	0.6
Total NHSDA sample	17,747	18,269	24,505	25,500	86,021	100
Note: These sample counts are the raw, unweighted number of subjects that were interviewed.

References

Anderson, John E., David E. Nelson, and Ronald W . Wilson. (1998). "Telephone Coverage and Measurement of Health Risk Indicators Data from the National Health Interview Survey." American Journal of Public Health. 88(9): 1392-1395.

Aquilino, William S. (1992). "Telephone Versus Face-to-Face Interviewing for Household Drug Use Surveys." International Journal of the Addictions. 27(1): 71-91.

Belinfante, Alexander. (1998). FCC Releases New Telphone Subscribership Report. Washington, D.C.: Federal Communications Commission.

Belinfante, Alexander. (2001). Telephone Subscribership in the United States. Washington, D.C.: Federal Communications Commission.

Ford, Earl S. (1998). "Characteristics of Survey Participants With and Without a Telephone: Findings from the Third National Health and Nutrition Examination Survey." Journal of Clinical Epidemiology. 51(1): 55-60.

Gfroerer, Joseph C. and Arthur L. Hughes. (1991). " The Feasibility of Collecting Drug Abuse Data by Telephone." Public Health Reports. 106(4): 384-393.

McAuliffe, William E., Stephanie Geller, Richard A. LaBrie, Susannah B. F. Paletz, and Elizabeth A. Fournier. (1998). "Are Telephone Surveys Suitable for Studying Substance Abuse Epidemiology? Cost, Administration, Coverage and Response Rate Issues." Journal of Drug Issues. 28(2): 455-482.

Office of Applied Studies and Research Triangle Institute. (1996). National Household Survey on Drug Abuse: Population Estimates 1995. Rockville, MD: U.S. Department of Health and Human Services.

Office of Applied Studies and National Opinion Research Center. (1997a). National Household Survey on Drug Abuse: Main Findings 1995. Rockville, MD: U.S. Department of Health and Human Services.

Office of Applied Studies, Substance Abuse and Mental Health Services Administration (SAMHSA), and Research Triangle Institute. (1997b). National Household Survey on Drug Abuse Population Estimates 1996. Rockville, MD: U.S. Department of Health and Human Services.

Office of Applied Studies (OAS), Substance Abuse and Mental Health Services Administration (SAMHSA) and Research Triangle Institute. (1999). National Household Survey on Drug Abuse Main Findings 1997. Rockville, MD: U.S. Department of Health and Human Services.

Office of Applied Studies and Research Triangle Institute. National Household Survey on Drug Abuse Main Findings, 1998. (2000). Rockville, MD: U.S. Department of Health and Human Services.

Pearson, David, Allen Cheadle, Edward Wagner, Robert Tonsberg, and Bruce M. Psaty. (1994). "Differences in Sociodemographic, Health Status, and Lifestyle Characteristics Among American Indians by Telephone Coverage." Preventive Medicine. 23: 461-464.

U.S. Bureau of the Census. (1994). Phoneless in America . Washington, DC: U.S. Department of Commerce.

American Indians

This chapter describes several diagnostic studies of the treatment needs of American Indians. It includes a summary of the studies conducted on American Indian reservations in North Dakota, Montana, and South Dakota. Although American Indians on reservations will not comprise a separate component of the total treatment need estimate developed in the chapter on gaps in treatment, this chapter is meant to bring attention to this group's specific needs. Further, American Indians have the lowest rates of telephone coverage among all of the Census Bureau's race categories (Census 1994). Given the high rates of substance use disorders found in North Dakota's face-to-face survey of American Indian adults on reservations described below, the telephone survey probably underestimated need. In other reports, the authors have estimated the needs of recently incarcerated prisoners, households without telephones, and homeless people. These subgroups typically face a high risk of substance abuse and do not intersect with each other or the population represented in the telephone survey. Because some of those captured by the estimate from the face-to-face survey were also captured by the telephone survey, it would be incorrect to integrate the two.

While there have been few diagnostic studies of American Indians, there is a substantial body of literature focusing on other measures of substance abuse problems affecting the American Indian population. Those studies describe behavior such as binge drinking (Beauvais 1998), women drinking while pregnant (May 1989), and drunk driving (Tippetts and Voas 1999). This literature has brought much attention to the adverse consequences of those behaviors, such as a high rate of alcohol-related motor vehicle fatalities (Tippetts and Voas 1999), a high rate of substance abuse related deaths (Beauvais 1985, May 1989), and a high rate of fetal alcohol syndrome among some segments of the American Indian population (May et al. 1983). There are also indications that American Indians are less likely to want treatment for their problems (Bhardwaj et al.1998). While estimates from the National Household Survey of Drug Abuse (NHSDA) show American Indians were the most likely to have received treatment in the prior year, the gap between past year treatment and past year dependence was also the widest for that group (OAS 2000).

Review Criteria and Methods

In an effort to gather all of the materials with substance abuse prevalence estimates for American Indians, the authors conducted a general literature search. Those materials provided background for this report, especially in regards to the nature of substance use among American Indians. The team used basic internet search methods to find references including MedLine, the National Clearinghouse for Alcohol and Drug Information (NCADI), and the NIAAA's Alcohol and Alcohol Problems Science Database (ETOH). The team also searched North Charles Research and Planning Group's own library of literature. These materials were compiled in a file on substance abuse and treatment among American Indians developed throughout the course of the State Treatment Needs Assessment Program. Based on knowledge that several states did surveys on reservations with CSAT funding, the study team contacted officials in those states to obtain copies of their reports. The authors obtained three state studies (North Dakota, Montana, and South Dakota), which are the focus of the present review. Each of these reports contained substance use disorder prevalence rates based on the National Technical Center's adaptation of the Diagnostic Interview Schedule's (DIS) Substance Abuse Module (Robins et al. 1990). Besides the three State Treatment Needs Assessment Program studies, the team did not find any population-based diagnostic studies of American Indians. Manson et al. (1992) reported encouraging results regarding the use of the DIS to measure alcohol use disorders in limited samples of three tribes in the Southwest, the Pacific Northwest, and the Norther Midwest Plains.

Each of the three studies included in this review were conducted on reservations within the respective states. Therefore, the studies refer to generally similar populations, though there were a number of different tribes involved in each study. The size of the samples ranged from 998 to 2,249, and each sample was restricted to adults 18 and older. The Montana study had an upper age limit of 65. The interviews were conducted in person, and the North and South Dakota studies used a computer-assisted format.

Results

Overview

Each of these studies found high substance abuse rates on reservations (Table 7.1). Further, each found that alcohol was the primary problem. The studies found that men on the reservation experienced higher rates of substance use disorders than women, often by a wide margin. The reports also suggested that treatment utilization was low. The study done in North Dakota showed that few of those with an untreated need wanted treatment. The Montana study measured obstacles to treatment and showed many with a need that did not get treatment cited family pressures, transportation issues, and administrative barriers as reasons for not receiving any. The surveys also found that many American Indian women of childbearing age were heavy drinkers, putting children at a higher risk for Fetal Alcohol Syndrome (FAS).

Table 7.1 Substance Use Disorders Among American Indians on Reservations
Reference	Data Collection Location and Date	Study Population	Diagnostic Measurement	% Alcohol or Drug Use Disorder
				Current		Lifetime
Bhardwaj et al. 1998	North Dakota and Montana 1997	998 adults on 5 reservations; ages 18 and over	NTC version of DIS-III-R modified for face-to-face computer-assisted interviewing	Past Year	Alcohol: 19 Drugs: 2.6	27.1
Bray et al. 1999	South Dakota 1996-1997	Random sample of 2,449 adults on 9 reservations; ages 18 and over	NTC telephone survey based on DIS-III-R modified for face-to-face computer-assisted interviewing	Past Year	30	-
Haynes et al. 2001	Montana 2000	1,821 adults on 6 reservations; ages 18 to 65	NTC survey based on DIS-III-R modified for paper and pencil interviewing	Past Year	28.4	-

Generalizing the Results

In order for a study to be included in this literature review, it had to include a population-based diagnostic estimate of substance use disorder prevalence among American Indian on reservations. Because applying diagnostic criteria within a survey is a relatively new practice (dating back to 1980-1984), there have been only a few such studies of the reservation population. Because the studies included here were done in states with a distinct geographic proximity to each other and each uses the same diagnostic scale, the results of each may comfortably be compared. Previous studies of substance use and substance use-related problems show that these issues can be similar across American Indian groups in different regions (Manson 1992). Therefore, the South Dakota and Montana studies are used to corroborate the results of the North Dakota study.

Other studies that did not include substance abuse prevalence estimates often looked at rates within narrow subsets of the American Indian population, such as youths or clinical cases. Some studies included American Indians living off the reservations. Many of the studies focused only on alcohol because the problems American Indians face with alcohol abuse are more widely known. Beauvais and LaBoueff addressed the problem of inference by noting, "Other studies which purport to measure Indian drug and alcohol use have such a skewed sample of Indians as to make their conclusions suspect"(Beauvais 1985). This criticism points not only to the problem of attempting to generalize from a narrowly defined sample, but also to the problem of interpolating results across American Indian tribes that can have dramatically different cultures and problems. Because the present study used only population-based diagnostic estimates, the authors assumed that some broader similarities do exist. Fortunately, North Dakota did a study of adults on the reservations. Therefore, an estimate exists for that specific population and the findings of other studies can be used to corroborate the findings of the North Dakota study.

The North Dakota study reported current diagnoses for drugs and alcohol separately, while it offered more complete findings of lifetime diagnoses including demographic breakdowns. It is not possible to simply add up the current rates for drugs and alcohol, as it is likely that some people are included in both. The Montana and South Dakota reports contained estimates of the current substance use disorder rate of American Indian adults on the reservations.

North Dakota's Study on American Indian Adults on Reservations

Bhardwaj et al.(1998) surveyed a sample of 998 adult American Indians on 5 reservations in North Dakota. Just over 54% of the respondents were women. The study used a modified version of the National Technical Center's DIS-III-R instrument for the survey. This study found that 27.1% of those over the age of 18 on reservations had a lifetime substance use disorder diagnosis (Bhardwaj et al. 1998). The larger part of those with a diagnosis were alcohol abusers: 19% were dependent and 7.2% were abusers. Further, 19% of adults on the reservation had a current alcohol use disorder: 14.2% were dependent and 4.8% were abusers. Although alcohol was the most common problem, 3.5% had a lifetime dependence or abuse diagnosis for marijuana while 1.2% had a lifetime methamphetamine dependence. There were some who had a diagnosis for both marijuana and alcohol (Bhardwaj et al. 1998). Less than 1% had a lifetime diagnosis for drugs only, while 3.1% had a lifetime diagnosis for both drugs and alcohol. Less than 2% had a current diagnosis of dependence or abuse for marijuana while 0.6% had a current methamphetamine dependence.

Despite a high need for treatment among American Indians on reservations in North Dakota, the survey found that few of those surveyed wanted treatment. Only 1.4% of those with a diagnosis who did not receive treatment in the past year said they wanted some (Bhardwaj 1998). However, many of those who wanted treatment in the past year, but did not get it complained of physical handicaps and not getting the type of treatment they wanted as obstacles to getting treatment. However, the sample size was low for those estimates, so the results should be treated as indicative of obstacles at best. Reservations are typically rural and convenient access to treatment facilities is a problem. The report from the Gallup Organization estimated the survey population to be 13,163. That implies that 2,501 American Indian adults on reservations had an alcohol treatment need. There were 224 in need of treatment for marijuana, but it is likely that many of those were counted as being in need of treatment for alcohol as well. Those totals are not unsubstantial when compared with the statewide need total estimated in the chapter on gaps in treatment. As noted previously, though, some cases were captured by both surveys.

To get a notion of the magnitude of potential underestimation due to low telephone coverage on the reservations, consider results from North Dakota's household telephone survey. That survey indicated that American Indian adults had a current treatment need rate of 9.3%. Although that estimate includes American Indians living off the reservations, the difference of almost 10 percentage points between it and the alcohol-only estimate (19%) from the face-to-face survey is indicative of the large difference that arises when those in households without telephones are included in the sample.

Other Studies of American Indians on Reservations

Montana

Haynes et al.(2001) surveyed 1,821 adults between the ages of 18 and 65 living on reservations in Montana using a paper and pencil interview. Along with the necessity of using a face-to-face interview to evade telephone coverage bias, the authors emphasized the use of interviewers recruited from the reservations. The authors argued that doing so was critical to elicit optimum honesty from respondents. This survey found that 28.4% of American Indian adults on reservations in Montana had a past-year substance use disorder. A large proportion of those surveyed were dependent on drugs or alcohol (15.3%); men were almost twice as likely as women to be dependent on a substance. The reservation's population had a greater need for alcohol treatment than for drug treatment according to this study, but the authors emphasized in their report that the rate of drug dependence was nonetheless far greater than that of the general United States population (Haynes et al. 2001). Frequently, subjects demonstrated a need for both drug and alcohol treatment (6.6%).
Almost 13% of those diagnosed with a current substance use disorder had a demand for treatment for their disorder. Compared to the rate of unmet demand in North Dakota, that rate is high. Those who expressed a desire for treatment, but did not get it in the past year cited a variety of obstacles to treatment. Many specified administrative factors such as, red tape and waiting lists while lack of insurance and adequate transportation also presented barriers.

The difference between the household survey results for American Indians and the face-to-face interview results in Montana is also illustrative of potential underestimation due to low telephone coverage. Riley at al. (1997) reported specifically on the substance use disorder rates of American Indians based on the Montana Adult Household Survey. That telephone survey was designed to procure statistically significant estimates for American Indians by oversampling in 8 census tracts on reservations. However, the sample they reported on did not distinguish between American Indians on and off the reservations, though it was largely comprised of those from reservations. The report on the household telephone survey showed that 13.7% of American Indian adults in Montana were in need of treatment for a substance use disorder (Riley et al. 1997). Whereas the rate for the adult population from the telephone survey was 8.8%, it is clear that American Indians had a substantially higher rate of current substance use disorders. Lifetime substance use disorder estimates for American Indian adults from the telephone survey were high. For the whole American Indian population, 26% had a lifetime alcohol use disorder while 8.9% had a lifetime diagnosis of a drug use disorder.

However, the rate of current substance use disorders for the adult American Indian population from the survey did not approach the estimated rate for adult American Indians on reservations in Montana from the more recent survey of that population. As noted above, that population had a current treatment need rate of over 28%. This difference is primarily attributable to bias from low telephone coverage on reservations (Riley et al. 1997, Haynes et al. 2001).

South Dakota

Bray et al. (1999) surveyed the largest sample among the studies reviewed here. They interviewed 2,449 adult Native Americans aged 18 and over on nine reservations in South Dakota. The study found that about 30% of adult American Indians living on reservations needed treatment for a current substance use disorder. Consistent with estimates from the other surveys covered here, this study found that a greater proportion of men were in need of treatment and that alcohol was the predominant substance of abuse. The report stated, " Nearly one-quarter of adults in this population were in need of treatment specifically for alcohol (with or without the need for treatment for problems related to use of other drugs). In comparison, about 1 in 10 Native Americans was in need of treatment for the other drugs covered in the survey" (Bray et al. 1999). Like those surveyed in the studies described above, American Indians on reservations in South Dakota experienced high rates of alcohol use disorders accompanied by a greater incidence of drug problems than in the general population. Furthermore, some individuals demonstrated both drug and alcohol use disorders. This survey found that less than one-third of those in need of treatment in the sample received some in the past year. The report emphasized lack of insurance coverage as a significant barrier to obtaining treatment (Bray et al. 1999).

Methodological Problems of Integrating the Estimates

Unfortunately, there is no way to integrate an estimate of need for those American Indian adults on reservations into the overall estimate. The American Indian population is implicitly included in all of the population subgroups covered in the gap analysis done in this study. Each adult component of the overall need estimate overlaps with the target population of the reservation survey. The current analysis is meant only to bring attention to the high rate of treatment need found in this group.

Conclusions

Though there are few studies available that offer estimates of substance use disorder rates for American Indians, those that do exist have consistent results. Alcohol use disorders are common among adults on the reservations, as previously inferred form high rates of alcohol-related problems such as, high motor-vehicle fatality rates, high FAS rates, and high alcohol-related mortality rates. In North Dakota, the problem is apparent in the rates of substance use disorders reported in the face-to-face survey. Over 27% of the adults surveyed had a lifetime diagnosis for substance abuse or dependence; the overwhelming majority of those had alcohol problems. The group had a low rate of unmet demand for treatment (1.4%) and among those that did receive treatment in the past year there was little desire for additional services (0.8%). The State may consider ways to get those individuals interested in treatment including more convenient access to services, increasing awareness of the dangers of substance abuse, developing programs specifically targeted at American Indians, and increasing the variety of services available.

References

Beauvais, Fred, and Steve LaBoueff. (1985). "Drug and Alcohol Abuse Intervention in American Indian Communities." The International Journal of the Addictions. 20(1).

Beauvais, Fred. (1998). "American Indians and Alcohol." Alcohol Health and Research World. 22(4).

Bhardwaj, Ajay, David Moore, and Max D. Larsen. (1998). Demand and Needs Assesment Study: Alcohol and Other Drugs of Native American Indians on Reservations in North Dakota. North Dakota Department of Human Services, Division of Mental Health and Substance Abuse Services.

Bray, Robert M., Barbara T. Dalberth, Mindy Herman-Stahl, June A. Walker, and Rebecca P. Sanchez. (1999). Substance Use and Need for Treatment: Findings from the 1996-97 South Dakota Native American Survey. South Dakota Department of Human Services, Division of Alcohol and Drug Abuse.

Haynes, George, Vincent Smith, and Nathaniel R. St. Pierre. (2001). Need for Substance Abuse Treatment on Montana's Native American Reservations. Montana Department of Public Health and Human Services, Addictive and Mental Disorders Division.

Manson, Spero M., james H. Shore, Anna E. Baron, Lynn Ackerson, and Gordon Neligh. (1992). "Alcohol Abuse and Dependence Among American Indians." in Alcoholism in North America, Europe, and Asia. (1992). John E. Helzer and Glorisa J. Canino.

May, Philip A., Karen J. Hymbaugh, Jon M. Aase, and Jonathan M. Samet. (1983). "Epidemiology of fetal Alcohol Syndrome Among American Indians of the Southwest." Social Biology. 30(4).

May, Philip A. (1989). "Alcohol Abuse and Alcoholism Among American Indians: An Overview." in Alcoholism in Minority Populations. (1989). Thomas D. Watts and Roosevelt Wright, Jr.

Office of Applied Studies(OAS), Substance Abuse and Mental Health Services Administration, and the National Opinion Research Center. (1998). Prevalence Substance Use Among Racial and Ethnic Subgroups in the United States 1991-1993. Substance Abuse and Mental Health Services Administration .

Office of Applied Studies(OAS), Substance Abuse and Mental Health Services Administration, and Research Triangle Institute. (2000). Summary of Findings from the 1999 National Household Survey on Drug Abuse. National Clearinghouse for Alcohol and Drug Information .

Riley, William, George Haynes, and Vincent Smith. (1997). Montana Adult Household Substance Abuse Treatment Analysis. Montana Department of Public Health and Human Services, Addictive and Mental Disorders Division.

Robins, Lee N.; Cottler, Linda B., and Babor, Thomas F. (1990) "Diagnostic Interview Schedule-- Substance Abuse Module (DIS-SAM)." St. Louis Department of Psychiatry, Washington University School of Medicine.

Tippetts, Scott, and Robert B. Voas. (1999). Ethnicity and Alcohol-Related Fatalities: 1990 to 1994. Center for Alcohol Policy Analysis, Pacific Institute of Research and Evaluation.

U.S. Bureau of the Census. (1994). Phoneless in America. Washington, DC. U.S. Department of Commerce.

Prisoners

This chapter estimates the treatment needs of prisoners in North Dakota. Because the State did not conduct a survey of recently incarcerated prisoners as part of the State Treatment Needs Assessment Program (STNAP) , the authors obtained an estimate through examination of other prisoner studies. This substudy estimated the number of persons who were on the outside and needed treatment during the past year but whom the household survey would have missed because they were incarcerated when the household survey was underway. Prisoners represent the largest and most high-risk institutionalized population who are missed by the household survey (McAuliffe et al. 1998; Anthony 1991).

Other Prisoner Studies

In order to gather materials that included population-based diagnostic substance use disorder estimates for prisoners, the study team reviewed the existing literature on the treatment needs of incarcerated populations dating back to the early 1980s. The study team searched the NIAAA's Alcohol and Alcohol Problems Science Database (ETOH), the National Library on Medicine, the National Criminal Justice Reference Service's abstract database, the Addiction Research Foundation's bibliographic database, and a series of internet search engines (e.g., Google) for unpublished papers. The team made use of the North Charles Research and Planning Group's own extensive needs assessment library and database that included reports collected from states that conducted prison studies as part of the STNAP. The authors examined the bibliographies of all of the articles for any studies not found by computer search routines.

The search focused on prisoner studies conducted in the United States and Canada. Relevant studies included efforts to estimate the percentage of recently incarcerated prisoners who had a substance use disorder diagnosis that met the clinical criteria of the American Psychiatric Association's (1994) Diagnostic and Statistical Manual of Mental Disorders (DSM). The telephone survey estimated the DSM-III-R or DSM-IV. Although the primary objective was to obtain prevalence estimates for the prior year when the prisoners were on the outside, estimates that approximate that time period were included because of the paucity of estimates available in the literature. The study team also focused on studies that used structured instruments, especially investigations that used the Diagnostic Interview Schedule. Because changes in mandatory sentencing laws have led to a substantial increase in the number of persons in prison for drug-related crimes (Schiraldi et al. 2000), the study limited its review to the 1980s and more recently.

Table 8.1 Published Studies of Substance Use Disorders among Prisoners
Authors	Data Collection Location/ Date	Study Population	Diagnostic Measurement	% Alcohol or Drug Use Disorder
				CurrentTime Frame	Current Estimate	Lifetime Estimate
Regier et al. 1990	Baltimore,Durham,New Haven, St. Louis, and Los Angeles, 1980-84	715 combined ECA prisoner sample	DIS-III			72
Teplin 1994	Chicago, 1983-84	728 males at Cook County Department of Corrections	DIS-III	Past 2 weeks	29	61
Bland et al. 1990	Edmonton, 1986-87	222 males,18-44 years of age in two correctional centers	DIS-III	Past six months	62	87
Teplin et al. 1996	Cook County, 1991-1993	1,272 female arrestees awaiting trial at Cook County Department of Corrections	DIS-III-R	Past six months	60	70
Hudik et al. 1994	Iowa, 1993	242 males and 132 females newly admitted to Iowa's prisons. Estimates reweighted to prison gender proportions (8% female)	Substance abuse counselors using local assessment scale	Time of interview	80	85
Farabee 1994, 1995	Texas 1993, 1994	1,030 entering adult males (except gang members and known homosexuals); 500 females	DIS-III-R items. Asked if had 10+ drinks or used illicit drugs in past year. Dependence=3+ Sx, abuse=1 or 2 Sx.	Past year	63
Illinois Dept of Corrections 1995	Illinois, 1994	630 (105 females)new inmates in four intake sites. Estimates reweighted to prison gender proportions(7% female)	DIS-III-R	Past year	35	57
Peters et al. 1998	Huntsville, TX 1996	400 male inmates in Holliday Transfer Facility where they are processed during the first 14-60 days	DSM-IV (SCID-IV)	Past month	56	74
Farabee et al. 1997	Kentucky, 1997	600 randomly selected inmates (567 male/ 33 female) from 15 prisons	Items based on DSM-IV. Qualify for items if ever used alcohol or ever felt addicted to drugs.			59
Kerber 2000	Texas, 1998	792 males newly admitted to 4 intake facilities	DIS-III-R	Past year	64
Kerber and Harris 2001	Texas, 1998	658 females newly admitted to two intake facilities	DIS-III-R	Past year	60
Kerber 2001	Texas, 1998	498 males newly admitted to six jail intake facilities	DIS-III-R	Past year	63
Studies from the State Treatment Needs Assessment Program (STNAP)
Riley et al. 1997	Montana, 1997	150 recently incarcerated adult male inmates at Montana State Prison and 44 female inmates at Women's Correctional Facility. Estimates reweighted to prison gender proportions.	DIS-III-R	Past Year	58	77
McAuliffe et al. 2000	Rhode Island, 1999	198 adult and 118 adolescent inmates recently incarcerated in Adult Correctional Institution or Intake Service Center. Estimates reweighted to prison gender proportions	DIS-IV	During Past Year When on Outside	adult: 82 adol: 81	adult: 85 adol: 81
McAuliffe et al. 2001	Alaska, 2000	208 adult inmates recently incarcerated in three Alaskan prisons. Estimates reweighted to prison gender proportions (9%).	DIS-IV	During the Past Year When on Outside	79	91

The study team found twelve published studies that fit its criteria (Table 8.1). Most (7 of 12) of the studies were not found in journals. The study periods range from the early 1980s (Regier et al. 1990) to the very recent studies in Montana, Texas, Alaska and Rhode Island (Kerber 2000, McAuliffe et al. 2000; McAuliffe et al. 2001; Riley et al. 1997). Consequently, the summary described in Table 1 is the most up-to-date and complete compilation of data on this issue that is currently available.

The pioneering Epidemiological Catchment Area (ECA) Study used the Diagnostic Interview Schedule (DIS-III) in 1980-1984 to assess the prevalence of substance use disorders of a combined sample of 715 prisoners from five sites across the country (Baltimore, New Haven, St. Louis, Los Angeles, and Durham) (Regier et al. 1990; Robins & Regier 1991). The DIS-III permitted lay survey interviewers to determine whether respondents met clinical criteria for alcohol and drug use disorders according to the Diagnostic and Statistical Manual, Third Edition (DSM-III) of the American Psychiatric Association (1980). Seventy-two percent of the inmates in that study had a history of substance abuse or dependence (Regier et al. 1990). This rate was the highest of any of the major population groups (household residents, patients in psychiatric hospitals, homeless, and nursing home residents) surveyed in the classic epidemiologic study (Robins and Regier 1991, p.109). The prisoners had a one-year alcohol use disorder prevalence of 26.4%. Furthermore, 55% of the males and 70% of the females had a lifetime diagnosis of drug abuse or dependence (Anthony 1991, p. 67).

In a 1983-84 study, Teplin (1994) evaluated a sample of 728 male jail detainees in Cook County using the DIS-III. She found that 29% had a current substance use disorder (19% alcohol and 15% drugs), and 61% had a lifetime substance use disorder. Similar results were obtained ten years later in another study of Illinois inmates using the revised version of the NIMH instrument, the DIS-III-R (Illinois Department of Corrections 1995). It found that 56.6% of 526 male prisoners and 62.5% of 104 female prisoners (57% in the overall weighted sample) had a diagnosis of abuse or dependence on one or more drugs during their lifetime. The Illinois study used the National Technical Center's DSM-III-R instrument that all states used in their household surveys to measure the revised criteria of the American Psychiatric Association (American Psychiatric Association 1987).

Subsequent studies conducted elsewhere using similar methods and measurement scales have found similar rates of substance use disorders among prisoners in most cases. Using the DIS-III, Bland et al. (1990) found that nearly nine out of ten male inmates (87%), 18 to 44 years of age, in Canada had a lifetime substance use disorder. In an unpublished study conducted in Iowa, experienced substance abuse counselors conducted clinical assessments of newly incarcerated prisoners and found that 81% of the males and 71% of the females had symptoms indicating a current diagnosis of substance abuse or dependence (Hudik et al. 1994). It was the only study in Table 1 that did not use a structured instrument. Two unpublished Texas studies, one conducted with male inmates in 1993 and one conducted with female inmates in 1994 found that 63% of the subjects (both males and females) had a current substance use disorder diagnosis (Farabee 1994, 1995). An unpublished survey of males entering prison in Kentucky by the same author found that 59% had a lifetime diagnosis of a substance use disorder according to the American Psychiatric Association's DSM-IV criteria (Farabee 1997). Peters et al. (1998) found that 56% of a sample of Texas inmates had a current substance use disorder, and an unpublished study (Kerber 2000) found that 64% of a Texas prisoner sample had a current substance use disorder. Additional studies done by Kerber at women's intake facilities and men's state jail intake facilities (Kerber and Harris 2001; Kerber 2001) yielded very similar results for current substance use disorder rates.

State Needs Assessment Program (STNAP) Prisoner Studies

Montana, Alaska, and Rhode Island recently conducted prisoner studies as part of the STNAP program. A review of findings from the Montana prisoner study may simulate a North Dakota study most closely because of the geographic proximity and similar population characteristics of the two states. The Montana study employed a survey instrument that provided for a diagnostic assessment based on the DIS as adapted by the present authors. Riley et al. (1997) conducted a treatment need study of incoming male inmates of the Montana State Prison and female inmates of the Women's Correctional Facility. Using a modified version of the National Technical Center's telephone survey questionnaire, the study team conducted face-to-face interviews with 150 male inmates and 44 female inmates. The resulting information provided estimates of the prevalence of substance use disorders and levels of care estimates.

The study found that 77% of in-coming male inmates and 89% of female inmates had a lifetime substance use disorder (Riley et al. 1997, Table 1). For males, the breakdown by substance was: 64% had an alcohol use disorder, and 49% had a drug use diso rder. Among women, 73% had an alcohol use disorder while 75% had a lifetime drug use disorder. In both cases, a large percentage had both drug and alcohol use disorders. As for current substance use disorders, Riley et al. found that 58% of in-coming male inmates had a drug or alcohol disorder in the past year and 64% of inmates of the Women's Correctional Facility had a current use disorder diagnosis. Among male inmates, 53% had a current alcohol use disorder, and 20% had a current drug use disorder. Among female inmates, 52% had a current alcohol use disorder, and 36% had a current drug use disorder. The estimates for lifetime and current substance use disorder prevalence for men and women prisoners can be combined as a weighted average, but b ecause there are nearly 20 times as many male inmates as female inmates the rates are no different from the male rates when rounded to the nearest whole number.

The authors concluded that alcohol and drug use prevalence rates were significantly higher for prison inmates than for the general population. These inmates, especially the females, were also far more likely to demand treatment than the general population. The prisoners mostly wanted in-patient services. Fifty-nine percent of women inmates had a current demand for treatment while 35% of men at the Montana State Prison had a current demand for treatment (Riley at al. 1997).

In a study conducted between July and October of 2000, the current authors interviewed 208 recently incarcerated male and female inmates of three correctional facilities in Alaska: Hiland Mountain Correctional Center, Palmer Correctional Center, and Wildwood Correctional Center. The study was designed to capture those who were on the outside prisons for at least a month during the past year but were incarcerated. It was assumed that those people would not be captured by the telephone survey. The study team achieved a 77% overall response rate. The sample was weighted to accurately reflect the gender proportions of the prison population, and the racial composition of the sample was similar to that of the general prison population in Alaska. Although the study sampled only three prisons, analysis suggested that the results may be generalized to the population of inmates that were incarcerated in State prisoners during the previous 11 months. The study estimated that number to be 610 inmates.

The study found that substance use disorders were widespread among prisoners in Alaska. Among recently incarcerated Alaska prisoners, 79% needed treatment for a substance use disorder during the past year when they were on the outside. Over half were dependent on alcohol, while 39% were dependent on drugs. More than a quarter (27%) were dependent on cocaine. Among those with a treatment need, dependence was more common than abuse for all categories of drugs and alcohol except hallucinogens. Two percent of the sample were dependent on hallucinogens, and 2% were abusers of hallucinogens. Dependence was over five times more common than abuse when all substances were considered. More than nine out of ten prisoners had a lifetime diagnosis for a substance use disorder. More than half had a lifetime drug dependence, while almost 60% had a lifetime alcohol dependence. Prisoners had a 38% lifetime rate of cocaine use disorders, and the same percentage had a lifetime marijuana use disorder. More than eight out of ten had a lifetime alcohol use disorder.

Male prisoners had a slightly higher lifetime rate of substance use disorders than female prisoners; 91% versus 85%. However, the rate of past year treatment need was essentially the same; 79% for men and 78% for women. Alaska Natives/American Indians had higher lifetime rates and past year rates of substance use disorders than Whites and African Americans. Alaska Native and American Indian prisoners had a 97% lifetime substance use disorder rate, whereas whites and African Americans had 90% and 83% lifetime substance abuse rates, respectively. Rates varied somewhat from facility to facility, but the differences were small. The rates of drug treatment need at Hiland Mountain were higher than the other facilities because that is the women's facility. Female prisoners in this study had higher rates of drug use disorders than men, especially for cocaine and stimulants. A recent study of the suitability for women of substance abuse treatment programs in prisons that were originally designed for men showed that female prisoners used drugs more frequently and used harder drugs than male prisoners (Langan et al. 2001). Male prisoners had higher rates of alcohol use disorders, and accordingly the rates of alcohol treatment need reported at Wildwood and Palmer were higher than the rates at Hiland Mountain.

Of interviewed prisoners, 53% reported receiving some type of treatment for their substance use disorder in the past year while outside prison. Overall, 79% reported receiving some type of treatment in their lifetime. A little more than a quarter of those who received treatment in the past year on the outside received specialty treatment, while the largest portion of the sample (37%) received "self help" treatment, which included Alcoholics Anonymous (AA) and Narcotics Anonymous (NA). A third of the subjects received treatment from a nonspecialty source, which included a range of sources such as counseling with clergy or a health advisor, consulting a family doctor, or drunk-driving classes. Most of the study respondents received some combination of specialty and non-specialty services. A majority of these prisoners received specialty treatment and self help treatment in their lifetime. More than half reported receiving treatment for an alcohol use disorder while in prison at some point in their lives, though only a third of those who reported receiving alcohol treatment were women. Among those who reported receiving alcohol treatment while in prison, 48% received some during their current incarceration.

Among those with a need for treatment (that is, had a substance use disorder) in the past year, 57% received treatment while outside prison in the past year. Among those diagnosed with a current treatment need that did not receive it in the past year, 37% reported that they wanted treatment. This percentage is referred to as the amount of "unmet demand" for treatment. A slight majority (54%) of the prisoners with an unmet demand for treatment said that they took steps to obtain treatment, but failed. These individuals sought a variety of treatment types including detox, residential care, outpatient treatment, and self-help groups. No specific type of treatment was the most common response. Among those that did receive treatment, 49% reported that they wanted additional treatment. These individuals reported demand for multiple types of service. More than ninety percent wanted treatment outside of a formal program and outpatient care. Eighty-eight percent wanted to attend self-help groups such as AA or NA. More than three quarters wanted residential or inpatient rehabilitation and the almost as many wanted treatment at a halfway house or group home. Over half wanted additional detox services. Clearly, the additional treatment demanded by prisoners that received some in the past year were extensive.

Alaskan prisoners with an unmet demand for treatment cited a variety of obstacles to obtaining it. Among those with an unmet demand for treatment and those who had a demand for additional treatment, a majority reported lack of insurance as a barrier to obtaining treatment. A majority of all prisoners with an unmet demand reported that bureaucracy, or "red tape", was an impediment. A majority (51%) of those who wanted more treatment said that treatment facilities were full, and 51% of those who pursued treatment but did not receive it cited a lack of adequate transportation. The other reasons that were cited, albeit less frequently, included not wanting their problem to be public, distance from facilities, and failure to address cultural and gender issues such as language barriers, and programs' inattentiveness to the special needs of women.

Finally, the Rhode Island prisoner study used the research design and instrument as employed in the Alaska prisoner study, and the Rhode Island study included a sample of 198 adults and 118 adolescents. The Rhode Island survey used the same assessment instrument as the authors used in the Alaska prisoner study. The results showed that 82% of the adults and 81% of the adolescents had a DSM-IV substance use disorder diagnosis in the past year when they were on the outside.

Lifetime and Past Year Estimates of Treatment Need

Although nearly all of the published prisoner studies used a version of the DIS and sampled recently incarcerated inmates, the studies reported a wide range of results. The surveys occurred at different times, used three different versions of the DSM and DIS, used different sampling strategies (e.g., persons incarcerated in the last year versus persons just entering prison), and defined "current" differently (time of interview, last month, last year). The variation in the percentages of lifetime diagnoses was not as great as the variation in the percentages with current diagnoses. Several of the studies disproportionately sampled female inmates, but the results were reweighted to reflect the proportion of female inmates in the prisoner population.

Alaska's study found that 91% of the sample had a lifetime substance use disorder, the highest of the nine published studies with lifetime estimates of the substance use disorder prevalence among prisoners. Those 10 studies had an unweighted mean rate of 72.9%. The weighted mean rate was 69%. The median rate among those nine studies was 74% (Figure 1). The lifetime estimates ranged from 57% to 87%.

The unweighted mean rate for the current disorder rates among the 10 studies with a current diagnosis was 58.8%, the weighted mean rate was 56%, and the median rate was 62% (Figure 2). The estimates of current use disorders ranged from 29% to 84%. The lowest rates were the two Illinois studies. The authors decided to use the median rate as the estimate for North Dakota's newly incarcerated prisoners.

References

American Psychiatric Association. (1980). Diagnostic and Statistical Manual of Mental Disorders. Washington, D.C.: American Psychiatric Association.

American Psychiatric Association. (1987). Diagnostic and Statistical Manual of Mental Disorders (Third Edition- Revised): DSM-III-R. Washington, DC: American Psychiatric Association.

American Psychiatric Association . (1994). Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition): DSM-IV. Washington, DC: American Psychiatric Association.

Anthony, James C. (1991). "The Epidemiology of Drug Addiction." in Comprehensive Handbook of Drug and Alcohol Addiction. Ed. Norman S. Miller. New York: Marcel Dekker, Inc. 55-86.

Bland, Roger C., Stephen C. Newman, Ronald J. Dyck, and Helene T. Orn. (1990). "Prevalence of Psychiatric Disorders and Suicide Attempts in a Prison Population." Canadian Journal of Psychiatry. 35:407-13.

Farabee, David. (1994). Substance Use Among Male Inmates Entering the Texas Department of Criminal Justice--Institutional Division: 1993. Austin, Texas: Texas Commission on Alcohol and Drug Abuse.

Farabee, David. (1995). Substance Use Among Female Inmates Entering the Texas Department of Criminal Justice--Institutional Division: 1994. Austin, TX: Texas Commission on Alcohol and Drug Abuse.

Farabee, David, Carl G. Leukefeld, Deena D. Watson, Michael Townsend, Hugh Spalding, and Richard Purvis. (1997). Substance Abuse Treatment Needs Among Kentucky Prison Inmates. Kentucky: University of Kentucky Center on Drug and Alcohol Research, Kentucky Division of Substance Abuse, and Kentucky Department of Corrections.

Hudik, Terry L., David Huff, Lettie Prell, Laura Roeder, Becky Bell, and Kim Porter. (1994). An Assessment of the Substance Abuse Treatment Needs of the Inmates of Iowa's Correctional Institutions. Iowa: Iowa Department of Human Rights.

Illinois Department of Corrections. (1995). Adult Prison Study: Draft Version. Illinois Department of Alcoholism and Substance Abuse.

Kerber, Lisa. (2000). Substance Use Among Male Inmates: Texas Department of Criminal Justice--Institutional Division: 1998. Austin, Texas: Texas Commission on Alcohol and Drug Abuse.

Kerber, Lisa and Rand Harris. (2001). Substance Use Among Female Inmates: Texas Department of Criminal Justice, Institutional Division: 1998. Texas Commission on Alcohol and Drug Abuse.

Kerber, Lisa. (2001). Substance Use Among Male State Jail Inmates: Texas Department of Criminal Justice, State Jail Division: 1998. Texas Commission on Alcohol and Drug Abuse.

Langan, Neal P. and Bernadette M.M. Pelissier. (2001). "Gender differences among prisoners in drug treatment." Journal of Substance Abuse. 13 (3): 291-301.

McAuliffe, William E., Stephanie Geller, Richard A. LaBrie, Susannah B. F. Paletz, and Elizabeth A. Fournier. (1998). "Are Telephone Surveys Suitable for Studying Substance Abuse Epidemiology? Cost, Administration, Coverage and Response Rate Issues." Journal of Drug Issues. 28(2):455-82.

McAuliffe, William E., Richard LaBrie, Stephen Haddad, Eric Sevigny, Scott Ronis, Ryan Woodworth, Sarah Williamson, JoAnn Scherer-Gonzales, and Jocelyn Leary. (2000). The Substance Abuse Treatment Needs of Rhode Island's Prisoner and Training School Populations . Cambridge, MA: North Charles Research and Planning Group.

McAuliffe, William E., Richard LaBrie, Eric Sevigny, Ryan Woodworth, Jaime Mellitt, Timothy Stablein. (2001).The Substance Abuse Treatment Needs of Alaska's Newly Incarcerated Prisoner Population: Final Report . Cambridge, MA: North Charles Research and Planning Group.

Peters, Roger H., Paul E. Greenbaum, and John F. Edens. (1998). "Prevalence of DSM-IV Substance Abuse and Dependence Disorders among Prison Inmates." American Journal of Drug and Alcohol Abuse. 24(4): 573-87.

Regier, Darrel A., Mary E. Farmer, Donald S. Rae, Ben Z. Locke, Samuel J. Keith, Lewis L. Judd, and Frederick K. Goodwin. (1990). "Comorbidity of Mental Disorders With Alcohol and Other Drug Abuse: Results From the Epidemiologic Catchment Area (ECA) Study." Journal of the American Medical Association. 264(19): 2511-18.

Riley, William, George Haynes, and Vincent Smith. (1997).Montana State Prison and Women's Correctional Facility Substance Abuse Treatment Demand Analysis. Montana Department of Public Health and Human Services, Addictive and Mental Disorders Division.

Robins, Lee N. and Darrel A. Regier, Eds. (1991). Psychiatric Disorders in America: The Epidemiologic Catchment Area Study. New York: The Free Press.

Schiraldi, Vincent, Barry Holman, and Philip Beatty. (2000). Poor Prescription: The Costs of Imprisoning Drug Offenders in the United States. Washington, DC: Justice Policy Institute.

Teplin, Linda A. (1994). "Psychiatric and Substance Abuse Disorders among Male Urban Jail Detainees." American Journal of Public Health. 84(2): 290-293.

Teplin, Linda A., Karen M. Abram, and Gary M. McClelland. (1996 ). Prevalence of Psychiatric Disorders Among Incarcerated Women. pretrial jail detainees. Archives of General Psychiatry . 53(6): 505-512.

Homeless

In a recently released report on the adequacy of substance abuse epidemiology, the National Academy of Sciences (2001) criticized the National Household Survey on Drug Abuse for failing to include estimates of drug use by nonhousehold high-risk populations. The report stated, "The committee recommends that methods be developed to supplement the data collected in the National Household Survey on Drug Abuse . . . in order to obtain adequate coverage of subpopulations with high rates of drug use" (p. 3-8). A large number of studies have documented the high rates of drug and alcohol use among the homeless (e.g., Research Triangle Institute 1994; see Table 1 below). This chapter describes the results of a literature review designed to estimate the need for treatment (i.e., the prevalence of current and lifetime substance use disorders) in homeless populations. The overall estimate of the need for treatment in the State combines estimates of need for non-overlapping populations of households with telephones, households without telephones, and non-household high-risk groups: homeless people and prisoners incarcerated during the past year. Using the National Technical Center's (NTC) needs assessment questionnaire, the State previously obtained a statewide estimate of the number of people who needed substance abuse treatment by interviewing the population living in households with telephones. In other chapters, the authors estimated the needs of recently incarcerated prisoners and households without telephones . The present substudy will add to that statewide needs assessment by estimating the number of persons who were homeless that needed treatment when the household survey was underway.

The Homeless in North Dakota

Because the authors could not identify any diagnostic studies of the substance use disorder prevalence rate of homeless North Dakotans, they drew upon the composite rate from the literature review described below to develop a synthetic estimate of treatment needs of the homeless. However, the homeless population in each state has nuances that can affect the needs of the homeless and their treatment system. In order to gain a clearer notion of homelessness in North Dakota, the authors had a conference call with state officials in the Division of Mental Health and Substance Abuse Services and the State Hospital who were knowledgeable about the characteristics of the homeless and how the treatment system handles those individuals. The state officials addressed a variety of issues including a description of the homeless population in general as well as their treatment options and needs. Although the population at any one point in time is estimated to be small (see chapter on gap analysis) (North Dakota Division of Community Services and North Dakota Coalition for Homeless People, Inc. 2000) relative to other states, one official stated that there are many "couch homeless" who are essentially transients that divide their time between staying with friends and family and being homeless. Although the household survey was designed to capture those individuals, they use the same treatment system as the shelter and street homeless. The July 2000 survey of the homeless done through service providers confirmed that there were nearly as many couch homeless as shelter homeless (211 versus 237). The state officials stated that the treatment system does not serve nearly all the needs of the homeless population and that the number of available slots in residential programs were inadequate. They explained that because North Dakota is a rural state, it is difficult to have a full continuum of treatment available in all areas. Treatment is provided at the Human Service Centers which are located in urban areas. As such, it is difficult for those outside of the cities to access treatment. Further, because the population is so spread out, it is difficult economically to provide a full complement of specialty services where there are fewer potential clients. When asked what types of additional services are needed for the homeless, there was a strong consensus that wrap-around services including mental health care, habilitation services, and housing are necessary to effectively treat homeless clients.

Homeless Population Estimate. Because they could not identify an estimate of the number of North Dakota residents who were homeless, the authors developed an estimate based on the 2000 Census count of the sheltered homeless using an adjustment described in the literature for estimating the number of nonsheltered homeless (Smith et al. 2001, Jencks 1994). The Census count captures only the sheltered population, and it was necessary to use an estimate of the ratio of the number of homeless on the street to the number in shelters in order to obtain a more accurate estimate of North Dakota's homeless population. That procedure yielded an estimate just shy of 315 homeless. The ratio estimation procedure is described in more detail in the chapter on gap analysis. The authors considered using the count from the July 2000 survey of service providers, but determined that it was not possible to separate out children accompanied by adults, who are not part of the target population (North Dakota Division of Community Services and North Dakota Coalition for Homeless People, Inc. 2000). Further, the count from service providers includes the couch homeless who are part of the household telephone survey population.

Literature Review

Because virtually every study done elsewhere found that the homeless have higher rates of substance use disorders than the general population, generalizing from the literature is better than either leaving the homeless out altogether or using the rate observed in the general population (which is done implicitly when telephone survey results are generalized to the entire target population).

Review Criteria and Methods

In order to collect the results of published estimates of the treatment needs of incarcerated populations, the study team reviewed the existing literature. The study team searched the NIAAA's Alcohol and Alcohol Problems Science Database (ETOH), the National Library on Medicine, the National Criminal Justice Reference Service's abstract database, the Addiction Research Foundation's bibiographic database, and a series of internet search engines for unpublished reports. The team made use of the North Charles Research and Planning Group's own extensive needs assessment library and database that included reports collected from states that conducted prison studies as part of the State Treatment Needs Assessment Program (STNAP). The authors examined the bibliographies of all of the studies for any publications not found by computer search routines. Earlier reviews of the literature on substance use and abuse by the homeless (Fischer 1991; Johnson and Barrett 1995) were examined as well.

Table 9.1 Substance Use Disorders in Homeless Samples
Reference	Data Collection Location and Date	Study Population	Definition of Homeless	Diagnostic Measurement	% Alcohol or Drug Use Disorder
					Current		Lifetime
Breakey et al. 1989	Baltimore, Maryland 1986-1987	203 adults interviewed in missions, shelters, and jails. Oversampled females: 57% male.	Prionsers were homeless at time of arrest.	Psychiatrists Evaluated DSM-III Criteria			Alc: Males: 68 Females: 32 Drugs: Males: 22 females: 17
North and Smith 1993	St. Louis, Missouri 1989-1990	600 males and 300 females randomly selected from all night shelters and day shelters in St. Louis and sampled from street corners. Age 18 and over.	Stayed prior night in a shelter or on street, car, abandoned building; no permanent home.	DIS-III-R	Past year	Alc: 37 drugs: 16
Koegel et al. 1988	Los Angeles, California Date Not Reported	328 adults interviewed in missions, skid row hotels, shelters, soup kitchens, and drop-in centers in Los Angeles. Ages range from 18 to 69. 95 % male.	Not reported	DIS-III	Past six months	31	69
Robertson et al. 1997	Alameda County, California 1991	564 adults (age 18 and over) living in shelters, soup kitchens, and drop-in shelters in Alameda County. 68% male.	Spent previous night in shelter or on street	DIS-III-R	Past year	52	69
Alemagno et al. 1996	Ohio 1994	207 adults (age 18 and over) were interviewed in shelters , soup kitchens, and a health care clinic. 76% male.	Not reported	DIS-III-R			Alc: 57; Cocaine: 45
OASAS, RIA. 1995	New York 1994	354 adults (age 18 and over) were interviewed in NYC and upstate shelters (n = 333) and in single room occupancies (n =21) in New York City. 64% male.	Not reported	DIS-III-R	Past year	48	51
Peuquet et al. 1997	Delawaare 1994-1995	147 adults (age 18 and over) were interviewed by telephone from 16 shelters. % male not reported.	Residing in an emergency shelter	DIS-III-R	Past year	46
Kline et al. 1997	New Jersey 1996	540 residents of 32 shelters. Aged 18 and older. 66% male.	Not reported	DIS-III-R	Past year	46
Caton et al. 2000	New York 1996 and 1997	200 adults with no history of a psychiatric disorder were interviewed in drop-in shelters in New York City. Ages ranged from 18 to 65. 50% male.	Literally homeless: no fixed abode, sleeps in shelter, a public place or shelter	Structured Clinical Interview for DSM-IV			Alc: Males: 51 Females: 40 Drugs: Males: 55 Females: 59
McAuliffe et al.1999	Rhode Island 1999	198 adults from shelters, drop-in center, soup kitchen, and transitional housing. Aged 18 and over. 70% male.	Resided in shelter, living on the street, or resident of transitional housing with history of homelessness	DIS-IV	Past year	66	78
Women Only
Bassuk et al. 1998	Worcester, Massachusetts 1992-1995	220 single mothers were interviewed in shelters in Worcester. Aged 15 to 58.	Not reported	Structured Clinical Interview for DSM-III-R-Non-Patient (SCID-NP)	Past month	5	41
Smith et al. 1993 ; Same as North and Smith 1993 above	St. Louis, Missouri 1989-1990	300 females randomly selected from all night shelters and day shelters in St. Louis and sampled from street corners. Aged 18 and over.	Stayed prior night in a shelter or on street, car, abandoned building; no permanent home.	DIS-III-R	Past six months	18	31
Adolescents and Young Adults
Warheit and Biafora 1991	Miami, Florida, date unknown	100 homeless (74 males, 26 females) in a short-term shelter, Aged 18-21.