Market Barriers to the Development of Pharmacotherapies for the Treatment of Cocaine Abuse and Addiction: Final Report. Appendix B: Summaries of Primary Studies on Need for Cocaine Treatment


ABT ASSOCIATES (Rhodes et al. 1993, 1995)

Rhodes, W. (1993). Synthetic Estimation Applied to the Prevalence of Drug Use. The Journal of Drug Issues 23(2), 297-321.


ABT ASSOCIATES (Rhodes et al., 1993)

Estimate (in thousands)




At risk (not incarcerated)(1990)


Heavy users in prison and jail(1990)


Heavy users in high school(1990)


High school age but incarcerated(1990)


Heavy users who are high school dropouts(1990)


Heavy users in college(1990)


Additional heavy users in household(1990)



ABT ASSOCIATES (Rhodes et al., 1993)

Estimate (in thousands)




At risk (not incarcerated) (1990)


Heavy users in prison and jail(1990)


Heavy users in high school(1990)


High school age but incarcerated(1990)


Heavy users who are high school dropouts(1990)


Heavy users in college(1990)


Additional heavy users in household(1990)



Rhodes estimates the number of weekly cocaine and heroin users. When necessary, the response more than 10 times per month was treated as weekly. The categories estimated are:

  • At risk (not incarcerated)
  • Heavy users in prison and jail
  • Heavy users in high school
  • High school age but incarcerated
  • Heavy users who are high school dropouts
  • Heavy users in college
  • Additional heavy users in household


Criminal justice involvement was estimated using arrests from the FBI's Uniform Crime Reports. Drug Use Forecasting (DUF) data were used to estimate the percentage of arrestees who would have tested positive for cocaine. Data from the Bureau of Justice Statistics (1990, 1991) were used to estimate the number of weekly cocaine users in prison. National Household Survey on Drug Abuse (NHSDA) data were supplemented with data from the High School Senior Survey to better estimate the number of high school students using cocaine. Also, data from the Bureau of Juvenile Statistics were used to complete the picture of high school-aged cocaine users. Data from the NHSDA prior to 1991 did not include college students who live in dormitories and fraternities. Consequently, they assumed that college students used drugs at the same frequency as high school seniors and made adjustments to the NHSDA data. They assumed that the homeless were largely included because of their involvement with the criminal justice system. Drug use in the military was assumed to be insignificant as was drug use in therapeutic communities. Their final estimates were deflated based on overlap across sources.


Synthetic estimation is used to arrive at estimates of the numbers of weekly cocaine and heroin users for 1990 in various categories. Established relationships are used to infer drug use when direct measures are unavailable. Conceptually, cocaine and heroin users are comprised of intersecting groups of criminally involved persons, the homeless, high school students and drop-outs, college students, those in the military, those in residential treatment facilities, and those in households. The approach involves estimating the number of weekly users within each set, determining the overlap, summing across sets, and subtracting the overlap.


Rhodes, W., Scheiman, P., Pittayathikhum, T., Collins, L., and V. Tsarfaty. (1995). What America's Users Spend on Illegal Drugs, 1988-1993. Executive Office of the President. Office of National Drug Control Policy.














































* The NHSDA estimates of cocaine users are adjusted for 1988 and 1990 to account for the survey's limited coverage during those years.


  • Hard-core: defined in the NHSDA as one who uses cocaine at least 1 or 2 days a week every week for the year prior to the survey, or as one who used heroin on more than 10 days during the month prior to the survey.
  • Hard-core: defined in the DUF as those who admitted using cocaine or heroin for more than 10 days during the month before being arrested.
  • Occasional users: defined in NHSDA as those whose drug use was less than the criteria used for defining hard-core users.
  • Occasional users: cannot be estimated from DUF.


NHSDA supplemented with data from DUF and UCR.


DUF data were converted to estimates of hard-core drug users throughout the criminal justice system. A weighting scheme was devised to estimate the number who would be expected to test positive in DUF sites. Since they over-represent large city lock-ups, a model was used to infer percentages who would have tested positive in non-DUF sites.


NHSDA data were adjusted for 1988 and 1990 to account for the survey's limited coverage during those years. The adjustment adds an estimate of hard-core drug users who live in college dormitories to the estimate of hard-core users derived from the NHSDA. Students living in college dormitories are represented in the 1991 and later NHSDA data. The NHSDA was not administered in 1989. Estimates for 1989 are the averages for 1988 and 1990.

Committee on Judiciary, U.S. Senate


U.S. Congress. Senate. Committee on the Judiciary. (1990). Hard-Core Cocaine Addicts: Measuring-and Fighting-The Epidemic. A Staff Report Prepared for the Use of the Committee on the Judiciary.



Estimate (in thousands)

Committee on the Judiciary, U.S. Senate (1990)


Total Unduplicated Hard-Core Cocaine Addicts


Arrested (1988)


Treated (1988)


Homeless (1988)


Household (1988)



Hard-Core Cocaine Addicts are defined as those who abuse cocaine at least once per week. This definition is intended to correspond to that of the NHDSA.

  • Total Unduplicated Hard-Core Cocaine Addicts
  • Arrested
  • Treated
  • Homeless
  • Household


Data on those receiving treatment were obtained by contacting the National Association of State alcohol and Drug Abuse Directors (NASADAD). The treatment data were supplemented with data from TOPS for 1979 to 1981. Criminal justice estimates were derived from DUF data. The NHSDA was the principal source for household data.


Data from the individual states contained information on individuals receiving treatment for cocaine. From the TOPS data, it was estimated that 3 out of every 10 admissions were people who had been treated earlier. They therefore adjusted for multiple admissions. Furthermore, they assumed that 95 percent of admissions for cocaine were "hard-core" addicts, a conservative estimate according to the authors. They applied the overall estimate of hard-core addiction to a conservative estimate of the number of homeless to arrive at the number of homeless hard-core cocaine addicts. They used DUF data to estimate the drug use among arrestees. They applied the proportion testing positive for cocaine to the broader class of arrestees to arrive at the total number of cocaine users for the cities sampled. Then, they applied the average (47%) to cities not sampled. For small cities they assumed that 15% of arrestees would have tested positive for cocaine use. Finally, they adjusted based on overlap across data sources. They assumed that 40% of those arrested also sought treatment. They assumed that 10% of the homeless were not counted elsewhere. They left the number of arrestees intact because of adjustments from other sources. They concluded that 30% of arrestees were picked up by the household survey. Also, they assumed a 10% overlap between their treatment population and the NHSDA.

Homer (1993)


Homer, Jack B. (1993). A System Dynamics Model for Cocaine Prevalence Estimation and Trend Projection. In The Journal of Drug Issues 23(2), 251-279.



Past Month Total


Compulsive Prefer Crack(1990)


Compulsive Prefer Powder(1990)


Casual Prefer Crack(1990)


Casual Prefer Powder(1990)



Broad Categories:

  • used past month: used in NHSDA.
  • used past year: used in NHSDA.
  • used in lifetime: used in NHSDA.

Sub-categories within each of the Broad Categories:

    • Compulsive, prefer crack.
    • Compulsive, prefer powder.
    • Casual, prefer crack.
    • Casual, prefer powder.
    • Compulsive (heavy) refers to users who have used every week for the past year (average of 8 grams per month). Corresponds to NHSDA definition of weekly use.
    • Casual (light) users average one-half gram per month.


    NHSDA data for the years 1976, 1977, 1979, 1982, 1985, 1988, and 1990 were supplemented with data from the High School Senior Survey (HSSS) annual reports 1976-1990, Drug Abuse Warning Network (DAWN) 1976-1989, the Drug Use Forecasting (DUF) 1988-1989, Uniform Crime Reports (UCR) 1977-1989, Offender-Based Transaction Statistics (OBTS) 1983-1987, National Narcotics Intelligence Consumers Committee (NNICC) Reports 1977-1989, and the System to Retrieve Drug Evidence (STRIDE) 1977-1990.


    A system dynamics model was developed to generate estimates and projections on the national prevalence of cocaine use. Casual hypotheses/relationships were modeled or translated into equations and attempts to explain historical data resulting in the rejection of some models and refinement of others. Emphasis was placed on modeling endogenous (feedback) relationships and internal factors rather than exogenous (external) influences (model diagram is pictured on page 259 of article). Endogenous factors/variables included cocaine user population (dependent variable(s)), reported cocaine use prevalence, social exposure to cocaine, perceived health risks, morbidity and mortality, perceived legal risks, drug law incarceration and arrests, and several factors related to the cocaine market. Exogenous factors/variables included marijuana use prevalence, introduction of crack cocaine, seizure fraction, and arrest rate and incarceration fraction.

    Institute of Medicine


    Gerstein, D., and H. Harwood. eds., 1990. Treating Drug Problems. Volume 1: A Study of the Evolution, Effectiveness, and Financing of Public and Private Drug Treatment Systems, National Academy Press.


    IOM METHODOLOGY (Gerstein and Harwood 1990)

    Numbers (in thousands)

    (Harwood et al., 1993) Total Estimate of Treatment Need(1991)


    Total Estimate of Treatment Need(1987-1988)



    Household Population: Clear need (1987-1988)


    Household Population: Probable need (1987-1988)


    Homeless(sheltered. street, and transient)


    Correctional Custody


    Probation and parole


    Pregnancies (live births)


    Less overlaps


    Need for Treatment:

    Estimates from the NHSDA based on a combination of:

      • frequency and intensity of drug use;
      • number of symptoms of dependence in past year reported in NHSDA; and
      • number of problems from drug use in past year reported in NHSDA.

      All those using their "primary" drug on 9 or more days per month were classified as in clear or probable need for treatment, based on whether they self-reported 3 or more, or fewer symptoms/problems, respectively. Use of a primary drug 2 to 8 days per month plus 3 or more symptoms/problems were equivalent to probable need for treatment.

      Estimates for other populations and surveys based on measures of frequency/intensity of use and/or reported symptoms/problems associated with drug use that are indicative of a need for treatment. The various surveys use quite different definitions and items.


      • Clear need was defined in terms of exceeding thresholds on the following 3 criteria: illicit drug consumption at least three times weekly, at least one explicit symptom of dependence, and at least one other kind of functional problem attributed to drug use.
      • Probable need was assigned if level of consumption, number of symptoms, or number of problems fell below one threshold value but exceed the threshold on others.
      • Possible need was assigned if there was at least some monthly use and some indication of symptoms or problems.
      • Unlikely need was assigned to all others.


      The principal source of data was the National Household Survey of Drug Abuse (NHSDA). DUF data were used to estimate the need for treatment among arrestees with extrapolations being used to extend the estimates to the national level. Estimates based on State prison surveys for 1986 revealed that 43 percent were in need to treatment according to the criteria for drug dependence. This percentage was applied to the 1987 state and federal prison census for 1987. The same percentage was applied to the parolees population. The homeless population estimates of prevalence rates ranged from 10 to 33.5 percent. The median value (20%) was applied to the estimate of the homeless population to derive the number in need to treatment. The 1988 NIDA survey provided an estimate of the number of women in high fertility age brackets who used illicit drugs. The overall birth rate for that age group was applied and an estimate of the number of fetal exposures was generated.

      National Comorbidity Survey


      Warner, L.A., Kessler, R.C., Hughes, M., Anthony, J.C., and C.B. Nelson (1995). Prevalence and Correlates of Drug Use and Dependence in the United States: Results from the National Comorbidity Survey. Archives of General Psychiatry, Vol. 52: 219-229.

      National Comorbidity Survey

      • Past 12 month drug use meeting the DSM-III-R criteria for drug dependence or abuse in a national probability sample of 10,000 individuals.

      Office of Applied Studies/SAMHSA


      Woodward, A., Epstein, J., Gfroerer, J., Melnick, D., Thoreson, R., and D. Wilson. (in press). The Drug Abuse Treatment Gap: Recent Estimates. Office of Applied Studies, Substance Abuse and Mental Health Services Administration. U.S. Department of Health and Human Services.


      SAMHSA (Woodward et al., in press)

      Estimate (in thousands)



      Level 1 Need(1994)


      Level 2 Need(1994)



      Total Treatment Need is defined as someone meeting at least one of the following four conditions:

      1. Drug Dependence: A person is defined as dependent for a specific drug in 1995 if they indicate that they have used that drug in both the core and the dependence sections of the questionnaire and they meet 3 of the 6 DSM-IV criteria by the 1995 dependence questions.

      2. Heavy Drug Use. Any of the following in the past year:

      a. Used heroin at least once in the past year.

      b. Used marijuana daily.

      c. Frequent use (52 + days/weekly) of some other illicit drug.

      3. IV Drug Use: Used heroin, cocaine or stimulants with a needle in the past year.

      4. Treated for Drug Abuse: Received treatment for any illicit drug in the past year.

      Level 2 Treatment Need is defined as a person with at least one of the following four conditions:

      1. Drug Dependence: dependence on any illicit drug except marijuana in the past year.

      2. Heavy Drug Use: Any of the following in the past year:

      a. Used heroin at least once in the past year.

      b. Used marijuana daily AND dependent on marijuana.

      c. Frequent use (52 + days/weekly) of some other illicit drug

      d. Daily use of any illicit drug except marijuana

      3. IV Drug Use: Used heroin, cocaine or stimulants with a needle in the past year.

      • Treated for Drug Abuse: Received treatment for any illicit drug at a specialty facility

      in the past year.

      Level 1 Treatment Need is defined as those meeting the conditions for total treatment but not meeting the criteria for Level 2 treatment need.


      The principal source of data is NHSDA data with ratio estimation to account for under-reporting and undercoverage. Supplemental data sources include National Drug and Alcohol Treatment Unit Survey (NDATUS) now called the Uniform Facility Data Set, the Drug Services Research Survey (DSRS), and the Uniform Crime Report (UCR).


      The National Household Survey of Drug Abuse covers non-institutionalized populations aged 12 and over. The sampling frame under-estimates drug abusers because it does not cover institutionalized populations, the homeless, and those in treatment. Also, adjustments for under-reporting are likely to be an issue. The authors use ratio estimation and data from supplemental sources to adjust the NHSDA estimates of substance abuse. Ratio estimation is built on the idea that better estimates are possible if there is a known population estimate of a related variable. The ratio estimation procedure uses both the in-treatment and arrest counts to obtain corrected national counts of 1) those arrested and treated, 2) treated but not arrested, 3) arrested but not treated, and 4) not arrested and not treated. The adjusted estimates of total treatment need, and Level I and Level II need are presented above.

      Older OAS/SAMHSA definitions:

      The DEP definition attempts to approximate the DSM-III-R definition.

      The DSM-III-R defines a person as dependent if they meet 3 of 9 criteria for dependence.

      The NHDSA survey contains questions that are combined to approximate 5 of the 9 DSM categories. They include: tolerance, withdrawal, inability to stop or control substance use, giving up or reducing social occupational or recreational activities, and continued substance use despite knowing consequences.

      The DEP definition classifies someone as dependent if they respond positively to 2 of the 5 items.

      RAND Corp.


      Everingham, S.C., and C. P. Rydell 1994. Modeling the Demand for Cocaine. Drug Policy Research Center. RAND Corporation.



      Heavy Cocaine Users (1992)

      1.72 million

      Light Cocaine Users (1992)

      5.60 million

      US Cocaine Users Observed (1991)

      7.27 million

      US Cocaine Users Modeled (1991)

      8.02 million


      • Heavy users: people who used cocaine weekly over the course of the last year (NHSDA).
      • Light users: all others.


      NHSDA data were supplemented with data from the Housing and Urban Development (HUD), National Bureau of Economic Research, ICF, and 1986 Survey of Inmates of State Correctional Facilities.


      A Markovian two-state, four parameter model was used. The two states were heavy and light users. The four parameters were 1) light users to non-users, 2) light users to heavy users, 3) heavy users back to light users, and 4) heavy users that flow out of cocaine use. The numbers of light and heavy users were year-dependent.


      Estimates were based on the NHSDA and supplemented with data to account for under-counting of the homeless and incarcerated. Therefore, while they adjust for incarcerated and homeless populations, they make no adjustment for under-reporting.