Skip To Content
Click for DHHS Home Page
Click for the SAMHSA Home Page
Click for the OAS Drug Abuse Statistics Home Page
Click for What's New
Click for Recent Reports and HighlightsClick for Information by Topic Click for OAS Data Systems and more Pubs Click for Data on Specific Drugs of Use Click for Short Reports and Facts Click for Frequently Asked Questions Click for Publications Click to send OAS Comments, Questions and Requests Click for OAS Home Page Click for Substance Abuse and Mental Health Services Administration Home Page Click to Search Our Site

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

Table Of Contents

Chapter 2. A Conceptual Model for Measuring Substance Misuse and Abuse Through the Life Cycle: The Importance of Recovery and Death Rates

Barbara A. Ray,* Ph.D.

Abstract: Clinical reports of substance-related health problems among older adults speak to the dangers of overdose, dangerous combinations of therapeutic drugs, and misdiagnosis of drug-induced mental confusion as early dementia. Misdiagnosis of drug-induced health problems may trigger prescribing of still more drugs. To date, there are no population-based estimates of the size of this problem, but there are increasing indications that drug-related health problems will be at unprecedented levels in the baby boom generation (born from 1946 to 1964) as it begins to reach Medicare eligibility in the year 2012. Recent population estimates for Medicare beneficiaries show that in 1998, nearly 42 percent of drug expenditures were by seniors, who were only 13 percent of the population. Six years earlier, in 1992, the average number of prescriptions per elderly person (including refills) was over 19 and was projected to increase to over 38 by 2010. That would mean nearly 40 prescriptions per person, likely to include antidepressants and other psychoactive drugs. Considering the rising cost of prescription drugs, expected to triple by 2010 from current levels, a potential financial crisis lies ahead as increasing numbers of seniors require health care that they cannot afford. Methods are needed for estimating this future impact that account for drug abuse/misuse (a) incidence, (b) prevalence, (c) recovery, and (d) death throughout the life cycle.



The U.S. and global populations are aging. As a result, this age shift has focused attention on the coming need for health care services for older adults, particularly the generation of baby boomers born between 1946 and 1964 who are now entering their fifth decade of life. It is well known that this generation was exposed to an unprecedented array of psychoactive substances, both during the youth drug culture of the 1960s and 1970s, and subsequently to an unprecedented array of psychoactive medications available by prescription and over the counter. This trend continues in subsequent generations; psychoactive drugs are now prescribed to preschoolers at a rapidly increasing rate (Zito et al., 2000).

There is growing evidence that this generation will have an unprecedented level of substance-related health problems in their older years (Edgell, Kunik, Molinari, Hale, & Orengo, 2000; Hanlon, Fillenbaum, Schmader, Kuchibhatla, & Horner, 2000; Rice & Duncan, 1995; Steinberg et al., 2000; U.S. Department of Health and Human Services [DHHS], 1999). Inappropriate use of medications is reported among community-dwelling older adults, nonalcohol-related use disorders are found among geropsychiatric patients, and national admissions to substance abuse treatment repeatedly show that an alcohol-plus-drug problem is the leading problem at admission. The separation between alcohol use and drug use is becoming blurred. If the expected increase in drug problems among elderly people materializes, it will mean a shift in health care services and costs because prescription drug expenditures have grown at double-digit rates every year since 1980 (Employee Benefits Research Institute [EBRI], 1999). Misuse of psychoactive medications, for example, can contribute to physical problems, including liver and heart disease. New kinds of treatment may be required to deal with complications from the many psychoactive substances now available that can compromise the health of older adults through mental confusion leading to a misdiagnosis of severe memory loss or dementia. The risk of inappropriate medication is currently quite high, according to a recent report from the National Academy of Sciences (Kohn, Corrigan, & Donaldson, 2000). The report estimates that medication error accounts for 7,000 deaths annually in the United States. These deaths are among the conservatively estimated 44,000 deaths attributable to medical errors of all kinds, ahead of auto accidents (43,458), breast cancer (42,297), and AIDS (16,516), according to data from the 1999 National Vital Statistics System (National Center for Health Statistics [NCHS], 2001). Among elderly homebound individuals, 40 percent have been found to have atleast one inappropriate prescription medication (Golden et al., 1999). Among elderly outpatients in 1995, prescriptions for psychoactive medications were greatest in those over age 84 (Aparasu, Mort, & Sitzman, 1998).

By the time a baby boomer has survived to an older age, a complex history of drug use combined with new life stresses, such as the loss of a loved one and retirement, can trigger late-onset abuse of alcohol and other drugs. Distinguishing appropriate and health-enhancing drug use from debilitating overuse is difficult but essential to measure if the quality of later life is to be preserved in the baby boom and future generations.

To provide a rational basis for future health care needs and costs, some estimate of the past and future use of psychoactive drugs is necessary in order to also estimate the associated residual health disorders, both physical and mental. The key variables for predicting future problems are as follows:

These three variables are the basic elements in the conceptual model for estimating the future impact of psychoactive substance use on the health of older adults between now and 2030.

Because older persons have not been a consistently high national priority, and because locating and interviewing infirm or frail older adults is both expensive and intrusive, national health statistics are disproportionately based on young and middle-aged persons. It is recognized, therefore, that national statistics may carry a large estimation error for older adults. The data used here suffer from these same sampling problems but are used to begin the estimation process and to highlight the need for more accurate measures of the incidence and prevalence of substance use among older adults.

The model is designed to estimate level of use of psychoactive substances by age to the year 2030. Whether any given level of substance use constitutes "abuse" depends on operative clinical standards.



For the purposes of this chapter, the following definitions apply.

"Use" is operationally defined by the quantity of substance consumed, which can then be categorized as "low," "medium," or "high" according to current clinical standards.

"Permanent" recovery is ideally defined as lifetime abstinence from problematic substances. Due to the limitations of available data, retrospective reports of any prior treatment by age are used here to model lifetime recovery. Because recovery is a complex concept and subject to measurement error, to define it as abstinence from all psychoactive substances for the rest of one's life is not realistic when psychoactives are the recommended treatment for an array of medical diagnoses. Moreover, to define it as abstinence from illegal substances is not always useful because legal substances (such as alcohol, tobacco, and some medications) are the largest contributors to the Nation's substance-related problems.

"Recovery," as used here, refers to abstinence or near abstinence from substances that have previously created problems. National survey data are available describing the number of prior treatment episodes by age, and these data have been used to estimate national recovery rates. Defining recovery must take into account these four possibilities: (a) recovery is permanent, meaning lifetime, for all problematic psychoactive substances (this is the ideal); (b) recovery is short term and relapse quickly ensues; (c) recovery from one substance is replaced by use or abuse of another; and (d) recovery is counterfeit with continued drug use successfully concealed.

"Death rate" is defined as the national poisoning rate attributable to psychoactive substances (Fingerhut & Cox, 1998). This is a conservative estimate based on medical examiner, coroner, and physician opinions on the cause of death. To date, it is the only comprehensive analysis of deaths for psychoactives. It must be noted that all substances with abuse liability are psychoactive, but not all psychoactives have abuse liability. For example, it is difficult to maintain persons on antipsychotic medications due to their unpleasant side effects.


Developing the Model

Because psychoactive substances can induce aberrant cognitive and emotional behavior, the question of which came first, the drug or the mental disorder, is important for both diagnosis and treatment. The model, however, requires no assumption about which came first because it addresses only the level of psychoactive substance use.

When psychoactive substances are used to excess, mental problems are inevitable (Figure 1). Mental confusion, perceptual distortion, even hallucination can follow directly from psychoactive substance misuse. With prolonged lifelong use and abuse, physical problems also begin, involving almost every organ of the body. The consequences of drug use, therefore, can masquerade as almost anything. Unless the drug problem is recognized and successfully treated (recovery), life will be shortened by probable accidents and accumulated health problems (early death).

When a preexisting mental disorder leads to the use of psychoactive substances, either by self-medication or medically monitored prescription, the risk of excessive use is present (Figure 2). The quality of dosage monitoring determines whether psychoactive medication has a normalizing influence (recovery) or adds substance use problems to the mental ones (early death). Anxiety is the most prevalent mental diagnosis in persons aged 55 or older, according to the 1999 Surgeon General's report on mental health (DHHS, 1999). Anxiety and depression symptoms coexist to the extent that this comorbidity is the rule and not the exception. The potential for medication complications in persons older than 55 is raised by the many coexisting mental and physical complaints that can lead to psychoactive medication.



Figure 1 Substance Abuse Lifeline




Figure 2 Preexisting Mental Disorder Lifeline


In order to estimate future health requirements stemming from current or past substance abuse, the model includes a measure of the residual physical problems stemming from prior use or abuse even among recovered individuals. Studies of comorbidity between substances used and chronic physical disorders are the sources for this information. A summary measure of lifelong residuals stemming from typical patterns of substance abuse is beyond the scope of this chapter, but it will be necessary as estimates are revised and refined. A cradle-to-grave picture of all possible lifelines involving drug or mental problems is shown in Figure 3.



Figure 3 Possible Lifelines Involving Drug and Mental Problems


The model accounts for those individuals who exit from the population of psychoactive drug users by recovery or death. Deaths attributed to psychoactive drugs (Figure 4) accounted for 46 percent of poisoning deaths in 1995, considerably more than half of all poisonings in the year. This makes psychoactives a significant predictor of the death rate. Considering that national surveys of older adults must rely on living survivors, the impact of substance use in prior years may be greatly underestimated by retrospective data.



Figure 4 Percentage of All Reported Poisoning Deaths Using Death Certificate ICD-9 E-Codes


Source: Fingerhut and Cox (1998).

To separate psychoactive substances from other poisoning agents, all death certificate International Classification of Disease codes - 9th revision (Central Office on ICD-9-CM, 1997) pertaining to psychoactives were identified and grouped from those listed in a report on 1995 poisoning deaths (Fingerhut & Cox, 1998). These ICD-9 codes are shown in Table  1.

Recovery rates are another predictor of future health status and substance use. A pattern of repetitive recovery and relapse is not considered a significant predictor of health in the older years, but permanent lifelong recovery should have a strong positive influence on later health. Unpublished data from the Alcohol and Drug Services Study (ADSS) show that 50 percent or more of those treated for substance abuse reported no prior treatment. Of those with no prior treatment, about half required repeated treatment within the following 5 years. The age-specific percentages from the ADSS can be used in the model as a first approximation of recovery rate across ages from 10 years to 84 years. The number of persons in treatment drops starting about age 40 to 45, and estimates of recovery rate are more accurate below this age.



Table 1 ICD-9 Codes, by Category


ICD-9 codes


All Psychoactives

All Drugs

All Substances



E850.0, E851-E855,


E850-E858, E860-E866, E867-E869

Nondependent Abuse


305.0, 305.1-305.9

305.0, 305.1-305.9






E950.0-E950.9, E951-E952








E890.0, E890.3



Source: Fingerhut and Cox (1998).

The model assumes that the first significant use of a drug at any age continues until recovery or death. Temporary periods of abstinence followed by relapse are considered continued use by the model. Incidence rates for major substances of abuse are available from the National Household Survey on Drug Abuse (NHSDA), but that sample is not adequate to cover incidence at older ages (Office of Applied Studies [OAS], 2000). Nevertheless, the incidence rates at younger ages can be used in the model, which assumes that use continues until recovery or death. Because one person will use more than one drug, a measure of overlap among drugs is needed in the model.

A complication of measuring substance use prevalence is the problem of multiple use, what in the model is called "overlap." One person may use four or five psychoactive substances, such as alcohol, marijuana, cigarettes, over-the-counter diet pills, and prescribed antianxiety drugs. A national estimate of the average overlap among drugs by age is needed for the model. The Treatment Episode Data Set (TEDS), which tracks national admissions to treatment by substance, provides enough information about multiple substance use to begin to measure overlap. The overlap between alcohol and drugs is reported annually in this dataset, and further analysis may provide an estimate of overlap among alcohol, sedatives, stimulants, tranquilizers, and hallucinogens. TEDS captures illicit and licit substance use once it has become a problem requiring treatment. Knowing the extent of "use" of each substance by individuals at every age would improve the accuracy of the model in estimating future health requirements.


Conceptual Model

The conceptual model is as follows:

Users = (Cumulative incidence ÷ Overlap) – DeathsPermanent recoveries,


The definition of misuse and abuse is not constant but changes with time, depending on current clinical judgment, the law, and national data standards. At any given time, the current standard is applied to the national population at a given age as follows:

Misusers = Users × Age-specific rate of problem use.

For older adults, the term "misuse" is preferred to the term "abuse" because problems with recommended and legal medications predominate among older adults.

If the necessary population-based longitudinal data were available, the following formula could be applied:

This Chapter 2 formula, which is to be used with longitudinal data, presents the author's conceptual model for estimating the number of users of any psychoactive drug at age i as a function of the cumulative incidence of use of each psychoactive category, the average number of psychoactive categories per person, the cumulative deaths for users of any psychoactive drug, and the number of continuing living recoverers by age i.


Because the lifelong rate of change (slope) for incidence may be significantly different from the lifelong rate of change in number of categories per person, the formula does not conceptually reduce to a linear regression.

This is the conceptual model in search of numbers. As a first attempt, national datasets have been selected that offer crude estimates of the concepts involved. The critical review of this first attempt, it is hoped, will be the stimulus for improvements in both data selection and data collection.


Data Quality for Older Adults


Recovery Rate Estimating

Persons in substance abuse treatment have, by definition, encountered problems due to substance use, and their history of prior treatment can give a picture of recovery status by age. Error is introduced to the estimate of recovery rate to the degree that persons referred to treatment are unable to obtain it, also known as the "treatment gap." The data used to estimate recovery in this chapter came from the ADSS, which is not yet published and is based on a nationally representative sample of treatment programs. The age distribution of the ADSS sample shows the treatment rate increasing until about 40 years old and then declining rapidly (Figure 5). In contrast, the distribution of any prior treatment by age is flatter, with the rate rising slightly to age 40 and then holding steady until age 60. The rates of prior treatment in age groups above 60 cannot be considered reliable as they are based on too small a sample (e.g., the 100 percent rate in the 80 to 84 age group is based on one person).



Figure 5 ADSS Sample Distribution, by Age (Unweighted Data)


Source: OAS, unpublished data from ADSS.

For persons up to age 65 (above this age the samples are too small to give meaningful results), the rates of prior treatment are remarkably stable by age (Figure 6). Between 40 and 60 percent are entering treatment for the first time (i.e., zero prior treatment episodes) over the entire age range up to age 65. About 20 percent have had one prior treatment episode. About 10 percent have had two prior episodes, and 5 percent have had three prior episodes. This means that of the approximately 50 percent in treatment for the first time, half do not return. Of those in treatment for the second time, half again do not return. And so on, until a small number of individuals remain who are treated more than three times. Are all those who do not return to treatment permanently recovered? If so, the permanent recovery rate is about 87 percent for the treated population over the life span. More likely, those who do not return are recovered, or dead, or maintaining a drug problem without treatment. Until more detailed research is done on the lifetime use of psychoactive substances, it is impossible to know whether substance abuse problems are maintained, reappear, or appear for the first time at these older ages. The literature suggests that all of these occur.



Figure 6 Prior Treatment, by Age (ADSS Weighted Data)


Source: OAS, unpublished data from ADSS.


Death Rate Estimation

The sample for the death rate information is the U.S. population, excluding territories and protectorates. Death certificates (completed by physicians, medical examiners, or coroners) are the source of national information on cause of death. The information is forwarded to the NCHS by the 50 States and the District of Columbia. Since 1979, the cause of death has been coded according to ICD-9 codes, which show that in 1995 poisoning was the third leading cause of death (18,549 persons) after motor vehicle-traffic and firearm causes (Fingerhut & Cox, 1998). Of these deaths by poisoning, 81 percent were due to drugs and 46 percent to psychoactive substances.

Fingerhut and Cox (1998) separated out the codes for death by opiates and cocaine from the overall death-by-drug codes. As shown in Figure 7, the distribution of deaths by drugs across age shows a similar pattern whether for all drugs, opiates, or cocaine. The highest death rates occurred between the ages of 35 and 45. Because these distributions are similar in shape, the model will calculate deaths due to psychoactives as 56.7 percent of the deaths due to all drugs (45.9 to 80.9 percent; Figure 4).



Figure 7 Deaths per 100,000 by Poisoning, 1995


Source: Fingerhut and Cox (1998).


Overlap Estimation

The incidence of drug use by age is reported by the NHSDA as "year of first use" for marijuana, cocaine/crack, inhalants, hallucinogens, heroin, other illicit, alcohol, cigarettes, and any "nonmedical" use of psychotherapeutics, including pain relievers, tranquilizers, stimulants, and sedatives. Medical use and substance combinations are not reported, although overlap might be estimated by a special analysis of the data. The category of "psychotherapeutics" omits substances that are psychoactive but medically used for other purposes (e.g., beta-blockers prescribed for cardiovascular problems). The overall medical use of psychoactive drugs among elderly people does not appear to be measured by any Federal national dataset. Medicare's source of information about prescription drugs derives from the Medicare Current Beneficiary Survey begun in 1992 that asks about prescription drugs. The NHSDA sample, as with many other national datasets, is biased toward younger persons (Figure 8), which is consistent with its focus on incidence and prevalence of illicit drug use. When distributed across the age range of interest, 0 to 105 years, the NHSDA has a high proportion of 12 to 26 year olds but drops dramatically for those in the age range older than 35. The NHSDA would be an ideal source of numbers to fill in the model if it covered all psychoactive substances, measured lifetime recovery rates, were adjusted for death rates by users and better represented older adults. This is a tall order for a survey focused on illicit drugs.

TEDS reports on psychoactives used by those admitted to publicly funded treatment. The strengths and weaknesses of TEDS, as well as its significant findings, are described elsewhere in this report (see Henderson's chapter). TEDS omits antidepressants and antipsychotic categories from its catalogue of substances, but does report on stimulants (includes crack, cocaine), opiates, sedatives, tranquilizers, hallucinogens, inhalants, alcohol, and over-the-counter drugs. TEDS measures national admissions to publicly funded substance abuse treatment as reported by States, territories, and the District of Columbia. Because a person can enter treatment more than once in a given year, TEDS provides a count of admissions, not people.



Figure 8 1999 NHSDA as a Percentage of the U.S. Population


Sources: U.S. Bureau of the Census' 1999 U.S. population middle estimate and 1999 NHSDA special analysis runs.


There were 1,479,203 admissions to treatment in 1997. The age distribution ranged from age 10 to persons older than 90, and the rate of admissions declined after age 40 (Figure 9). Whether this decline represents an actual decrease in the need for treatment or failure to recognize the need for treatment in older adults is not known. Elsewhere in this report, Henderson points out that referrals by health care professionals proportionately increase with age and that admissions for more than one substance problem increase between ages 70 and 75. The numbers are small at these older ages, and Henderson cautions against overinterpretation of these data. Studies are needed that give accurate and detailed estimates of the number of persons with psychoactive substance use problems alone and in combination at these older ages. Whether older persons will appear in publicly funded treatment depends on a variety of factors discussed in the Henderson chapter.



Figure 9 Total Number of TEDS Admissions, 1997


Note: There were 1,479,203 admissions to treatment in 1997.

Source: OAS (1999).


TEDS routinely reports the degree of overlap between nonalcoholic psychoactive drugs (excluding antidepressants and antipsychotics) and alcohol. Figure 10 shows the increase in use of only alcohol up to age 70, followed by a rapid increase in the use of alcohol and another drug. Whether this pattern reflects increased reliance on prescription medication is not now known. Overall, the overlap factor appears to change significantly over the life span.



Figure 10 TEDS Alcohol and Drug Overlap, 1997


Note: There were 1,479,203 admissions to treatment in 1997.

Source: OAS (1999).



The conceptual model described here can project the number of users and abusers into future years, provided the numbers can be found to fill the formula. Unless death rates and permanent recovery rates are estimated at several points in the life span, projections will continue to misrepresent the size of the problem.

Rough approximations are available for all the elements in the model with the exception of "drug overlap" (the combined use of the full range of psychoactive substances). The weakness is primarily due to missing information about the use of prescription (licit) psychoactive drugs that are increasing in number and popularity. Elsewhere in this report, Henderson begins to document the degree of drug overlap seen in admissions to substance abuse treatment.

The accuracy of estimates of recovery rate depends on accurate identification of substance abuse problems. In this model, admission to substance abuse treatment is used as a proxy for drug use having reached the level of a problem. This proxy is useful only to the extent that persons needing treatment are admitted to treatment. To the extent that persons waiting for treatment are unable to obtain it, their potential recoveries will be missing from the denominator and the rate estimates will be flawed.

The model describes the elements necessary to attempt an estimate of the number of older adults with substance misuse problems in the out years. Data to fill in the model are not yet completely available, but interest in the risks of prescription drugs is growing along with a demand for relevant data. This conceptual model highlights the specific missing data necessary to accurately estimate the substance-related health problems of the Nation's future elderly population.



Aparasu, R. R., Mort, J. R., & Sitzman, S. (1998). Psychotropic prescribing for the elderly in office-based practice. Clinical Therapeutics, 20, 603-616.

Central Office on ICD-9-CM. (1997). International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM). Chicago, IL: American Hospital Association.

Chen, K., & Kandel, D. B. (1995). The natural history of drug use from adolescence to the mid-thirties in a general population sample. American Journal of Public Health, 85, 41-47.

Edgell, R. C., Kunik, M. E., Molinari, V. A., Hale, D., & Orengo, C. A. (2000). Nonalcohol-related use disorders in geropsychiatric patients. Journal of Geriatric Psychiatry and Neurology, 13(1), 33-37.

Employee Benefit Research Institute. (1999, April). Prescription drugs: Issues of cost, coverage, and quality (Report No. 208, EBRI Issue Brief, 21 pp.). Washington, DC: Employee Benefit Research Institute.

Fingerhut, L. A., & Cox, C. S. (1998). Poisoning mortality, 1985-1995. Public Health Reports, 113, 218-233.

Golden, A. G., Preston, R. A., Barnett, S. D., Llorente, M., Hamdan, K., & Silverman, M. A. (1999). Inappropriate medication prescribing in homebound older adults. Journal of the American Geriatrics Society, 47, 948-953.

Hanlon, J. T., Fillenbaum, G. G., Schmader, K. E., Kuchibhatla, M., & Horner, R. D. (2000). Inappropriate drug use among community-dwelling elderly. Pharmacotherapy, 20, 575-582.

Kohn, L. T., Corrigan, J. M., & Donaldson, M. S. (Eds.). (2000). To err is human: Building a safer health system. Washington, DC: National Academy Press, Committee on Quality of Health Care in America.

McCloskey, A., & Schondelmeyer S. W. (2000, July). Cost overdose: Growth in drug spending for the elderly, 1992-2010 (Report No. 00-107). Washington, DC: Families USA.

National Center for Health Statistics. (2001). National Vital Statistics System: Mortality data, multiple cause-of-death public-use data files. Retrieved December 16, 2001, from

Office of Applied Studies. (1999). Treatment Episode Data Set (TEDS): 1992-1997: National admissions to substance abuse treatment services (DHHS Publication No. SMA 99-3324, Drug and Alcohol Services Information System Series S-7; available at /dasis.htm#teds2). Rockville, MD: Substance Abuse and Mental Health Services Administration.

Office of Applied Studies. (2000). Summary of findings from the 1999 National Household Survey on Drug Abuse (DHHS Publication No. SMA 00-3466, NHSDA Series H-12; available at /nsduh.htm). Rockville, MD: Substance Abuse and Mental Health Services Administration.

Rice, C., & Duncan, D. F. (1995). Alcohol use and reported physician visits in older adults. Preventive Medicine, 24, 229-234.

Steinberg, E. P., Gutierrez, B., Momani, A., Boscarino, J. A., Neuman, P., & Deverka, P. (2000). Beyond survey data: A claims-based analysis of drug use and spending by the elderly. Health Affairs, 19, 198-211.

U.S. Department of Health and Human Services. (1999). Mental health: A report of the Surgeon General. Retrieved October 26, 2001, from

Zito, J. M., Safer, D. J., dosReis, S., Gardner, J. F., Boles, M., & Lynch, F. (2000). Trends in the prescribing of psychotropic medications to preschoolers. Journal of the American Medical Association, 283, 1025-1030.


* To whom correspondence should be sent at the Office of Applied Studies, SAMHSA, Parklawn Building, Room 16-105, 5600 Fishers Lane, Rockville, MD 20857. Telephone: 301-443-0747. E-mail:

Top Of PageTable Of Contents

This page was last updated on June 16, 2008.