Substance Abuse by Older Adults:  Estimates of Future Impact on the Treatment System

Chapter 6. Substance Abuse Among Older Adults in 2020: Projections Using the Life Table Approach and the National Household Survey on Drug Abuse

Albert Woodward,* Ph.D.

Introduction

Several methods could be used to estimate substance abuse among older adults in the future. The approach considered here employs a life table, which tracks the mortality or morbidity experience of a group, and the National Household Survey on Drug Abuse (NHSDA). The NHSDA is a rich data source on the prevalence of psychoactive and nonmedically used psychotherapeutic substances among older adults (Office of Applied Studies [OAS], 2001a). Psychoactive substances include marijuana, cocaine, heroin, hallucinogens, inhalants, and alcohol. Psychotherapeutic substances include the nonmedical use of prescription-type pain relievers, tranquilizers, stimulants, and sedatives. Since 1990, the NHSDA has annually surveyed the civilian, noninstitutionalized population of the United States aged 12 or older. Between 1971, the first year of the survey, and 1990, nine NHSDAs were fielded intermittently (i.e., in 1988, 1985, 1982, 1979, 1977, 1976, 1974, 1972, and 1971). In 1999, the sample size of the NHSDA was increased almost fourfold from prior years-to nearly 70,000 persons-with a concomitant increase in the number of older adults surveyed. The larger sample size of older adults in the NHSDA makes this survey a potential data source for a life table approach to project substance abuse among older adults.

Current prevalence rates for older adults reported in the NHSDA can be applied to population projections to extrapolate the expected future number of older persons with substance abuse problems. When averaged NHSDA rates of past year drug dependence for persons aged 50 or older are multiplied by 2020 population estimates of persons aged 50 or older (U.S. Bureau of the Census, 2000), the number of past year alcohol- or drug-dependent³ persons is expected to increase from 500,000 to 700,000 between 1999 and 2020. The number of those who used illicit drugs or drank heavily⁴ is expected to increase from 930,000 to 1.1 million for the same time period. Simply applying an extrapolation of current rates to population projections, however, is inadequate to estimate the number of older persons with substance dependence or to measure prevalence beyond dependence and abuse because the NHSDA reports almost no information on recovery and none on death attributable to substance use. The various influences on substance use of older persons have been described in other sections of this publication. For example, there is growing evidence that the "baby boom" generation (i.e., those born between 1946 and 1964) will have an unprecedented level of substance-related health problems as it ages (see the first chapter by Korper and Raskin in this monograph). A simple extrapolation cannot account for these influences. A life table approach or other more sophisticated approach is needed.

This chapter reviews the NHSDA data and suggests what might be added to the NHSDA to make it more useful in a life table approach to project substance abuse at older ages. The analysis uses a comprehensive substance use categorization scheme (see the chapter by Ray in this monograph), then examines how these categories might be used in a projection. The proposed life table approach uses these categories to estimate the extent of substance abuse problems among older persons. If the NHSDA were to obtain more detailed information on substance users, this information could be used to project expected future substance use among older persons.

Life Table Approach

Life tables have numerous applications, such as determining the mortality or longevity of a population or ascertaining the significance of differences in mortality, longevity, or morbidity among groups. Because they are used to track morbidity in populations over time, life tables can be the basis for estimating the extent of future substance abuse problems among older persons.

Kuzma (1984) and Selvin (1996) described three basic types of life tables: (a) the current life table; (b) the cohort, or generation, life table; and (c) the follow-up, or modified, life table. The current life table shows the effect of age-specific death rates on a population. The cohort life table shows the historical record from a point in time until the last person in the group has died. The follow-up life table provides the probability of survival of patients in a group following treatment or exposure to a disease. The follow-up type also offers the best fit for tracking a cohort of substance users (e.g., baby boomers) as they move into old age by 2020 or 2030.

To construct a follow-up life table, the starting and end points (years) have to be clearly specified. Given a specific time period and population categorized into age groups, it is possible to tally how many survive, how many die, how many enter into the group, and how many exit the group. An illustrative follow-up life table is shown in Table  1 for a hypothetical substance and starting with a population of 1,000 under the age of 5 years.

In the life table approach, cohorts are followed for a given period of time to determine their various outcomes. In the table, one cohort's drug use is followed as the cohort ages. The first value in column 1 and all values in column 2 for those initiating use and columns 4 and 5 for those who died or discontinued use are exogenous to the table; that is, they are determined independently from the table and are not calculated from it. Values for the other columns are calculated as noted in the formulas.⁵ This approach is similar to a survivor analysis.

Each annual NHSDA survey provides a cross-section of the household population for only a year and does not follow individuals over time, but the survey series could be used to create cohorts. Conceptually, for example, NHSDA data could be used to create several cohorts of 5-year age groups. Two such NHSDA cohorts drawn from 5-year age group data could be created for two different years. The NHSDA would need to have a sufficiently large sample to provide cohort estimates precise enough for statistical comparison. Because NHSDA data have been collected for more than three decades, in theory several cohorts could be created, provided that their death rates and recovery rates from drug use can be estimated. An age group of 30 to 34 year olds in 1995 becomes the age group of 35 to 39 year olds in 2000 and so on. Even though the NHSDA does not currently track a cohort of 30 to 34 year olds from 1995 through 2000 when they would be 35 to 39 years of age, the two age groups in 1995 and 2000 are comparable in that they can be thought of as similar to a cohort.

Based on a known decline in substance use prevalence as age increases, one expects to see a proportionate decline in the estimates of substance abuse cohorts between the 2 years (i.e., 1995 and 2000). If one assumes that the expected declines across cohorts continue into the next two decades, the estimates can be extrapolated forward to the year 2020.⁶ Thus, the original age group of 30- to 34-year-old substance-dependent persons would become the group aged 55 to 59 years in 2020,⁷ and the group aged 30 to 39 years would become the group aged 60 to 64 years in 2020. The product of the projected rate difference between 1995 and 2000 and the projected population is the number of older persons dependent on substances in 2020.

Prevalence changes between age groups for particular years (e.g., 30 to 34 and 35 to 39 in 1995 and 2000) also can be compared. The expected declines are a measure of the change that can be extrapolated to 2020. This comparison does not have the same validity as an approach that tracks cohorts for a period of time and "trends" it forward, but it can be used to provide a quantitative estimate of future prevalence estimates. However, it does not measure those who begin substance use later in life.

NHSDA data cannot now be used for the exogenous values for columns 2, 3, and 4 in the life table example shown in Table  1. Moreover, the life table approach uses the number of new cases in a period of time (an incidence approach). Although the NHSDA is mostly used for prevalence information, it can be used for incidence analysis. For example, age of first use could be useful in the life table to estimate the number who began drug use during an interval. Therefore, the life table approach requires some modification when NHSDA data are used; it may also require a larger sample of older adults for categorizing into the life table columns. In place of drug use or drug-free, the following NHSDA data terms would be used:

• Used illicit drugs or drank heavily: Applying the NHSDA, illicit drug use includes at least one use of any of the following listed drugs: marijuana/hashish, cocaine (including crack), inhalants, hallucinogens (including PCP and LSD), heroin, or any prescription-type psychotherapeutic used nonmedically. Heavy alcohol use is defined as having five or more drinks on the same occasion on each of 5 or more days in the past 30 days.
• Substance dependent: In the NHSDA, dependence is based on seven DSM-IV criteria (APA, 1994). The number who are dependent is a subset of the number who have used substances (alcohol and illicit drugs).
• Not substance dependent: Estimates of the population who are determined to be not dependent in a given year are calculated from the total population aged 12 or older minus the number who are substance dependent.
• Total population 12 years old or older: The NHSDA draws data from the civilian, noninstitutionalized, resident population aged 12 or older, which includes almost 98 percent of the total U.S. population aged 12 or older. Excluded from the NHSDA are active military personnel, persons living in institutional group quarters, homeless persons not living in a shelter on the survey date, and U.S. citizens residing abroad. The NHSDA is adjusted to the civilian, noninstitutionalized population as of July 1^st of the given year. The Census estimate is calculated from a demographic components of change model that incorporates information on natural change (births minus deaths) and net migration (net domestic migration and net movement from abroad) that has occurred in the area since the date of the 1990 Census.

The NHSDA can be used to produce reasonable estimates for the groups who used illicit drugs or drank heavily or who were substance dependent. These estimates are slightly underestimated because the NHSDA does not collect data on persons living in institutional group quarters or homeless persons, two groups known to experience proportionately more substance abuse problems than the general civilian, noninstitutionalized resident population. This underestimate applies particularly to younger ages when most institutionalized persons are in prison, but it is offset among younger ages because the NHSDA excludes military personnel, who have much lower rates of substance problems (Bray et al., 1999).⁸ For those aged 50 or older, institutionalized persons probably have lower rates of substance problems.

NHSDA estimates of illicit drug use, heavy drinking, or substance dependence could be used by comparing cohorts over two time periods to provide a measure of new substance use (e.g., column 3 in Table  1). The estimates of those not dependent (or even using drugs) (e.g., column 2 in the table) could also be calculated.

Limitations of the NHSDA

The NHSDA has several limitations that preclude fully applying the life table approach, even if it is modified. The NHSDA does not collect information on certain types of substance users (e.g., those who are in recovery or who died, making it difficult to measure column 4 in Table  1). Also, changes to the NHSDA questionnaire limit the comparability of cohorts across years.

The NHSDA does not adequately identify those persons "in recovery." These can only be estimated from the NHSDA by determining the number who are no longer substance abuse dependent, using illicit drugs, or drinking heavily. This number is derived by counting those respondents who indicated that they had treatment for substance abuse but were not categorized as substance dependent, using illicit drugs, or drinking heavily. This determination is likely to be incomplete because the NHSDA does not report on individuals who had a problem with substance use/abuse and "recovered" without treatment.

Each NHSDA incorporates the total population who died during a year because the survey is adjusted to Census totals that include births and deaths. The survey, however, cannot measure the comparable item needed for a life table (e.g., the number of persons who had died and had used drugs). Alternatively, differential death rates for current and former heavy users or abusers could be compared with rates for nonusers or trivial users. Death rates for substance abusers can be estimated, but this determination is likely to be incomplete.

Changes to the questionnaire and sample sizes limit comparisons across years. If two recent NHSDA years are used (e.g., 1999 and 1994-1995),⁹ differences in the questionnaires for 1999 and 1994-1995 permit only very approximate comparability for cohorts across the 2 years. The computer-assisted interviewing (CAI) estimates are different from comparable year paper-and-pencil-interviewing (PAPI) results (OAS, 2001a); no simple adjustments can be done to make PAPI and CAI estimates comparable or to compensate for differences in reporting due to the method of interview implementation. There could be a higher or lower reporting of dependence, and so on, depending on whether CAI or PAPI was used. In short, when a dependency variable for 1994-1995 PAPI is defined, estimates of the population who are dependent or not dependent are not directly comparable.

Because of the limited sample sizes for older populations in the NHSDA, the standard errors of the estimates for these measures by 5-year or 10-year groupings are sufficiently large so that cohort differences between two periods may not be calculated with statistical precision. The inability to calculate differences within accepted confidence intervals further impedes using a life table approach with NHSDA data.

Possible Changes to the NHSDA

Several large, national databases have been used to describe the substance use and related problems of older persons. Each database has limitations that prevent it from forecasting future substance use and related problems among older persons in two decades or so. The National Longitudinal Alcohol Epidemiologic Survey (NLAES) was conducted only in 1992 and has no information on the frequency or quantity of nonalcoholic drugs (Stinson et al., 1998). The National Health and Nutrition Examination Survey (NHANES), although it had four waves of data from 1972 to 1992, collected only limited alcohol data (i.e., mean weekly drinking levels for these four waves) (see the chapter by Blow, Barry, Fuller, and Booth in this monograph). The Treatment Episode Data Set (TEDS) (OAS, 2001b) and the Department of Veterans Affairs' utilization data (see the chapter by Booth and Blow in this monograph), both large and statistically powerful datasets, represent demand for substance abuse treatment among special, not general, populations. Only the NHSDA, even with its limits, has enough historical data on incidence rates at younger ages and later problems and treatment from which projections can be made.

What would be needed to make the NHSDA more useful for the life table approach? The survey was never designed to be a national longitudinal survey to measure substance use, abuse, and health indicators across the life span. It could, however, ask selected questions of an expanded sample of older respondents. Using examples referred to above, an expansion of the cohorts who are in their 30s now would be in the 50 and older age groups in 2020. Their responses to these questions could be used in the life table.

Selected questions could be added as a special supplement to the NHSDA, as has been done for other special topics of public health concern. Questions could cover such topics as (a) permanent recovery to appropriate use or abstinence by age during the survey period, (b) incidence rates for major substances of abuse at older ages, (c) a measure of overlap among drugs, (d) current psychoactive prescription medications, and (e) first use of psychoactives by age. These questions would fill the gaps in information needed for the life table categories. For example, permanent recovery could be used to estimate those who are drug-free during the time interval. As another example, age of first use could be useful in the life table to estimate the number who began drug use during an interval.

There will still remain one limitation. The NHSDA cannot provide death rates for users by age because it accounts for these deaths as part of the total population who have died during a year, as the survey is adjusted to Census totals that account for births and deaths. Of course, more than 1 year of such a supplement could provide the data necessary for projecting use and problems among older persons in the next two or three decades.

The data could be applied to more than a life table approach (e.g., to a more sophisticated model). With more comprehensive and more years of data, it may be possible to develop projection models of the complexity and utility of demographic forecast models. Until such data are available, the life table approach will remain a framework for creating future estimates.

References

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author.

Bray, R. M., Sanchez, R. P., Ornstein, M. L., Lentine, D., Vincus, A. A., Baird, T. U., Walker, J. A., Wheeless, S. C., Guess, L. L., Kroutil, L. A., & Iannacchione, V. G. (1999, March). 1998 Department of Defense Survey of Health Related Behaviors Among Military Personnel: Final report (RTI/7034/006-FR, prepared for the Assistant Secretary of Defense [Health Affairs], U.S. Department of Defense, Cooperative Agreement No. DAMD17-96-2-6021). Research Triangle Park, NC: Research Triangle Institute.

Kuzma, J. W. (1984). Basic statistics for the health sciences. Palo Alto, CA: Mayfield.

National Institute on Alcohol Abuse and Alcoholism. (1995, March 17; updated 2000, October) NIAAA releases new estimates of alcohol abuse and dependence. Retrieved November 13, 20001, from http://www.niaaa.nih.gov/press/1995/nlaes.htm

Office of Applied Studies. (2001a). Summary of findings from the 2000 National Household Survey on Drug Abuse (DHHS Publication No. SMA 01-3549, NHSDA Series H-13, also /nsduh.htm). Rockville, MD: Substance Abuse and Mental Health Services Administration.

Office of Applied Studies. (2001b). Treatment Episode Data Set (TEDS): 1994-1999: National admissions to substance abuse treatment services (DHHS Publication No. SMA 01-3550, Drug and Alcohol Services Information System Series S-14, also /dasis.htm#teds2). Rockville, MD: Substance Abuse and Mental Health Services Administration.

Selvin, S. (1996). Statistical analysis of epidemiologic data (2nd ed.). New York: Oxford University Press.

Stinson, F. S., Yi, H., Grant, B. F., Chou, P., Dawson, D. A., & Pickering, R. (1998). Drinking in the United States: Main findings from the 1992 National Longitudinal Alcohol Epidemiologic Survey (NLAES) (NIH Publication No. 99-3519, also http://www.niaaa.nih.gov/publications/Nlaesdrm.pdf). Bethesda, MD: National Institute on Alcohol Abuse and Alcoholism.

Stoto, M. A., & Durch, J. S. (1993). Introduction. In M. A. Stoto, & J. S. Durch (Eds.), Forecasting survival, health and disability: Workshop summary, March 9-10, 1992. Washington, DC: National Academy Press.

U.S. Bureau of the Census. (2000). National population projections: II. Detailed files. Total population by age, sex, race, Hispanic origin, and nativity: (NP-D1-A) Annual projections of the resident population by age, sex, race, and Hispanic origin: Lowest, middle, highest series and zero international migration series, 1999 to 2100. Retrieved November 13, 2001, from http://www.census.gov/population/www/projections/natdet-D1A.html

________

* To whom correspondence should be sent at Office of Applied Studies, SAMHSA/DHHS, Room 16-105, 5600 Fishers Lane, Rockville, MD, 20857. Telephone: 301-443-6255. E-mail: awoodwar@samhsa.gov.

³ "Dependent on alcohol or any illicit drug during past year" is derived from an algorithm based on criteria in the Diagnostic and Statistical Manual of Mental Disorders (4^th ed., DSM-IV) (American Psychiatric Association [APA], 1994).

⁴ "Illicit drug use" and "heavy alcohol use" are defined in the subsequent narrative.

⁵ Note that the proportion drug-free at the end of the interval (column 8) is made up of the combined products of the probabilities of being drug-free at the end of the interval and all prior intervals: .

⁶ Of course, this assumption may be unrealistic because the baby boom generation may have different substance use patterns from those of previous generations.

⁷ A simple mathematical extrapolation is as follows: Let X_t = cohort 30 to 34 years in 1994; then X_t+5 = cohort 35 to 39 in 1999. The decline, - = (X_t+5 - X_t)/X_t. For the cohort in 2004, X_t+10 = - * X_t+5, and so on to the cohort at year 2020. Of course, such an extrapolation does not account for differences among generations (e.g., the baby boom generation).

⁸ The military has a "zero-tolerance" policy for illicit drug use and routinely tests for these drugs; those who test positive are discharged (Bray et al., 1999).

⁹ Data for 1995 are included with 1994 because the 1994 sample size is too small to produce precise estimates for detailed categories by age group.

This page was last updated on June 16, 2008.

SAMHSA, an agency in the Department of Health and Human Services, is the Federal Government's lead agency for improving the quality and availability of substance abuse prevention, addiction treatment, and mental health services in the United States.

Site Map | Contact Us | Accessibility Privacy Policy |  Freedom of Information Act |  Disclaimer  |  Department of Health and Human Services |  SAMHSA |  White House |  USA.gov

* Adobe™ PDF and MS Office™ formatted files require software viewer programs to properly read them. Click here to download these FREE programs now

What's New

Highlights

Topics

Data

Drugs

Pubs

Short Reports

Treatment

Help

Mail

OAS