The Economic Costs of Alcohol and Drug Abuse in the United States - 1992
Alcohol and drug abuse cost society an estimated $176.4 billion during 1992 as a result of lost productivity resulting from premature death and illness among alcohol and drug abusers, associated crime-related costs of alcohol and drug abusers, time spent by alcohol and drug abusers in residential treatment, and developmental disabilities among fetal alcohol syndrome (FAS) survivors. An estimated $107 billion in overall productivity losses is attributable to alcohol abuse, and $69.4 billion is attributable to drug abuse (see table 5.1). The crime-related productivity losses include the value of lost potential productivity resulting from the following: (1) victimization by crime, (2) incarceration for a criminal offense, and (3) time spent by heroin and cocaine addicts in criminal activities (crime careers) rather than in legal employment.
The analysis of lost productivity in 1992 produced the following findings:
- More than 132,000 deaths were attributable to alcohol and drug abuse in 1992. The estimated loss of productivity resulting from these deaths was $45.9 billion (using a 6-percent rate of discount), including $31.3 billion for the 107,000 premature deaths attributable to alcohol abuse and $14.6 billion for the 25,000 deaths attributable to drug abuse.
- Alcohol and drug abuse can interfere with an individual's productivity and employment. Shortfalls in productivity and employment among individuals with alcohol and drug abuse disorders accounted for estimated losses of $80.9 billion in 1992, of which $66.7 billion is attributed to alcohol abuse and $14.2 billion is attributed to drug abuse.
- The estimated loss of potential productivity because of long-term residential treatment and short-term hospitalization of alcohol and drug abusers was $3 billion in 1992; drug problems cost $1.5 billion and alcohol problems, $1.5 billion. This estimate is somewhat conservative because it excludes time for hospitalization and recuperation associated with treatment for medical consequences of problem alcohol consumption.
- Adults who were born with FAS have mental impairments, which are associated with earnings lower than for adults who were not born with FAS. At a 1-in-1,000 incidence rate among adults, this results in an additional $990 million in lost earnings, all of which are alcohol caused.
- Victims of crime often experience loss of work time with the associated economic loss. An estimated 2.5 million alcohol-related victimizations (about 60 percent were assaults) occurred in 1992, with estimated costs of $1 billion. About 8.2 million drug-related victimizations occurred (about 75 percent were larcenies), with estimated victim work losses of about $2 billion.
- About 600,000 person-years of time spent in prisons and jails in 1992 (about 45 percent of the total time served in these facilities in 1992) is attributed to alcohol- (160,000 person-years, costing $5.4 billion) or drug-related crime (440,000 person-years, costing $17.9 billion).
The approximately 1.7 million "heavy" drug users often forsake the legitimate economy for the illicit drug trade or for careers in consensual (gambling and prostitution) or acquisitive (theft) crime, with losses of legitimate productivity of $19.2 billion.
Alcohol and drug abuse caused more than 132,000 deaths in 1992. The estimated loss of productivity resulting from these deaths was $45.9 billion using a 6-percent discount rate - 107,000 alcohol-attributable deaths cost an estimated $31.3 billion, and 25,000 deaths due to drug abuse cost an estimated $14.6 billion. This study, using 1992 mortality and cost data, found that the number of deaths attributed to alcohol had increased 13 percent since the previous study conducted by Rice et al. (1990) using 1985 data. Alcohol-attributable deaths increased from 94,768 in 1985 to 107,360 in 1992. The number of deaths attributed to drug abuse alone increased by 20 percent, and when other drug abuse-related conditions (tuberculosis, AIDS, and hepatitis) were added, the number of drug abuse deaths tripled from 7,454 to 25,493. The number of deaths during 1985 estimated by Rice et al. (1990) was 50 percent higher than the estimate of Harwood et al. (1984) for 1980. Table 5.2 summarizes the number of deaths and cost effects for alcohol and drug abuse.
Economic costs were estimated using the "human capital" approach to valuation of loss of expected years of life and productivity that would been undertaken over the course of life. These values are dependent on the age and gender of the decedent to reflect both the life expectancy as well as expected value of market and nonmarket/household productivity. This approach is in contrast to the "willingness-to-pay" approach, which attempts to reflect both expected loss of productivity (the essence of the human capital approach) as well as the "value" of pain and suffering. Studies performed for the U.S. Department of Health and Human Services and the Public Health Service have predominantly used the human capital approach, although cost-benefit studies (for example, studies of highway safety and pollution control) often use the willingness-to-pay approach (Miller et al. 1991). Human capital values are materially lower than "typical" willingness-to-pay values and, thus, will yield "conservative" estimates for economic costs relative to estimates based on the willingness-to-pay approach. Most of the prior studies on the economic costs of alcohol and drug abuse have used the human capital approach (an exception is Miller et al. 1997a,b). It has never been addressed whether special willingness-to-pay estimates would be needed for alcohol and drug abuse.
The following sections report the numbers of deaths attributed to alcohol and drugs and the estimated economic costs of those deaths using the human capital approach.
5.2.2 Methodology for Estimating Deaths
This study uses the same approach employed in the studies by Rice et al. (1990) and Cruze et al. (1981). The approach entails:
- Identifying the causes of death (ICD-9 diagnoses) that can be entirely or partially attributed to alcohol and drugs,
- Obtaining estimates of appropriate "attribution factors" by diagnosis,
- Accessing data on the number of deaths for each diagnosis by age and gender group, and
- Applying economic values by age and gender group for the present discounted value of expected lifetime productivity.
Mortality data are from 1992, the year for which detailed estimates have been developed. This study estimates values using discount rates of 3 percent, 4 percent, and 6 percent and primarily uses the 6-percent values in order to remain consistent with Rice et al. (1990) and previous studies. The concepts of expected lifetime productivity and discounting are discussed earlier, in section 3.3.3, as well as following, in section 5.2.3.
188.8.131.52 Deaths Attributable to Alcohol Abuse
This study first identified the ICD-9 codes - listed as the primary cause of death - that, according to the most recent research, are attributable entirely or partially to alcohol abuse. As a first step in this analysis, a list of applicable diagnoses, the proportion of deaths attributable to alcohol for each diagnosis (as designated by the attribution factor, or AF), and the corresponding age ranges were drawn from The Eighth Special Report to the U.S. Congress on Alcohol and Health (National Institute on Alcohol Abuse and Alcoholism [NIAAA] 1993). This list was supplemented with data from a report published by NIAAA in 1993 (Stinson et al. 1993). The final list of diagnoses, along with respective AF values, was reviewed and supplemented slightly by NIAAA staff.
The number of alcohol-related deaths for each ICD-9 code was calculated by multiplying the AF for each diagnosis by the total number of deaths in the relevant age range indicating the particular ICD-9 code as the primary cause of death. Mortality data with detailed age and gender distributions by diagnostic code for 1991 were initially obtained from the most recent public-use National Center for Health Statistics (NCHS) ICD-9 mortality data series (see table 5.3). Subsequently, this study gained access to NCHS mortality estimates for 1992, which have been used to adjust the 1991 values (NCHS 1996b).
The parameters for the alcohol analysis differ only modestly from the Rice et al. (1990) analysis. This study includes some of the deaths related to diseases in four categories not included in the Rice et al. (1990) study that have been recently studied for their relationship with alcohol abuse. The net effect of these additional categories is an extra 10,900 deaths attributed to alcohol abuse in the current study than would have been counted using the categories employed by Rice et al. (1990). These additional four diagnoses are the following:
- Cerebrovascular disease,
- Chronic hepatitis, and
- Other chronic liver diseases.
In addition, biliary cirrhosis has been dropped.
This study uses specified age breaks (e.g., age 15 and older) for 15 of the disease categories that were used by Rice et al. (1990) without age restrictions. This approach was adopted in accordance with current understanding of disease etiology (Stinson et al. 1993). Despite these restrictions, the number of deaths attributed to those categories for which there were age restrictions and for which there were no other methodological changes was slightly higher for this study than for the Rice et al. (1990) study.
Finally, there were 11 diagnostic codes for which the previous nonzero AF was reduced (e.g., automotive crashes and falls) or was increased (e.g., suicide, homicide, drowning, and fires). The net impact of such attribution changes applied to 1992 mortality in the indicated age groups is to increase attributed deaths for these 11 diagnoses from about 44,800 (using factors applied by Rice et al.) to 46,300 (using factors in this study).
184.108.40.206 Deaths Attributable to Drug Abuse
A similar process to that used for alcohol-attributable deaths was followed for identifying deaths attributable to drug abuse. The Rice et al. (1990) study (and previously Cruze et al. ) employed a list of drug abuse-related causes (diagnostic codes) originally designated by the National Institute on Drug Abuse for use by the Drug Abuse Warning Network (Gottshalk et al. 1977, 1979). This list includes diagnoses representing abuse of and dependence on psychoactive drugs as well as accidental and intentional (i.e., suicide) poisoning by a broad range of drugs and medicaments - psychoactive and otherwise.
The rationale for using the broad list of drug types had its origin in the difficulty of determining the motivation for drug ingestion. Inappropriate use of prescription or nonprescription drugs constitutes drug abuse, whether a person takes a prescription medication without a prescription or takes a drug for an objective other than that for which it is appropriate.
This study has augmented the basic list of codes by incorporating estimates of deaths from several health problems associated with drug abuse and dependence. Chapter 4 presents the reviews and analyses performed to obtain estimates of attribution factors for tuberculosis, hepatitis B and C, and AIDS for health expenditures. The primary data and references used to calculate the attributable deaths for each of these additional diagnoses are briefly reviewed below. Table 5.4 summarizes data on the number of deaths attributed in this study to drug abuse.
In 1992, 33,566 persons died of AIDS (NCHS 1996b). These data are obtained through the national death certificate data. Attribution of AIDS deaths was based on mortality data from national case reports for 1991 submitted to the Centers for Disease Control and Prevention (CDC) surveillance system. This source was chosen because the mortality data provide comprehensive information about AIDS exposure (CDC unpublished data, November 1994). The estimate of 33,566 AIDS deaths incorporates a downward adjustment for the fact that among deaths of AIDS patients, 88 percent were due to AIDS or AIDS-related causes, and the other 12 percent were unrelated (e.g., trauma) (Chu et al. 1993). The most recent year with complete data at the time that these calculations were developed was 1991.
This analysis applied to the reported 33,566 deaths the proportion of deaths within the following exposure categories from the CDC mortality data set (these values were adjusted upward to account for the 5 percent of deaths with no identified exposure category):
- Adults/adolescents who are injecting drug users (about 24 percent of deaths with identified exposure category);
- Adults/adolescents who have heterosexual sex with injecting drug users (about two-thirds of the 6 percent exposed through heterosexual contact);
- Adult/adolescent men who have sex with men and inject drugs (one-half of the 6 percent with dual exposure); and
- Children (under age 13) with a mother who has an infection or is at risk for infection because she is an injecting drug user or has sex with an injecting drug user (more than one-half of the 1 percent who were pediatric AIDS cases).
Out of the 33,566 AIDS deaths in 1992, this study attributes 10,737, or about 32 percent, to intravenous drug abuse. This rate is slightly below the 36-percent AF used for health care expenditures. The difference in these values lies in that the proportion of injecting-drug-user HIV cases has steadily increased since the beginning of the epidemic. The proportion of deaths (this section) actually represents AIDS cases reported several years previously.
220.127.116.11.2 Hepatitis B and C
The ICD-9 death certificate data undercount the number of deaths due to hepatitis B and C because they do not make the link between hepatitis and chronic liver disease. As an alternative to the ICD-9 data, the CDC Hepatitis Branch has developed annual death estimates, which take into account the toll of chronic liver disease brought on by hepatitis B and C (CDC 1993c, personal communication). The Branch estimates that there are 4,000 to 5,000 deaths a year from hepatitis B and 8,000 to 10,000 from hepatitis C. For the purposes of this analysis, the midpoint of these CDC-produced ranges is used to estimate that there were 4,500 deaths annually from hepatitis B and 9,000 annually from hepatitis C.
To estimate the numbers of these deaths attributable to intravenous drug use, this study relies on a sentinel risk factor study conducted by the CDC in four representative U.S. counties (Alter et al. 1990a). These studies reported the risk factors for acquiring hepatitis B and C, including parenteral (intravenous) drug use as a factor. Although these data do not link case development to death reports, it is thought that these are the most complete data presently available to estimate the number of hepatitis deaths associated with drug use. This study reported that 20 percent of the hepatitis C cases and 30 percent of the hepatitis B cases were linked to drug use for the range of years studied. This analysis uses these percentages as AF's to calculate the number of deaths attributable to drug use.
As described in chapter 4 of this report, it is estimated that drug abuse contributes to 4.5 percent of the tuberculosis cases that occur each year. This figure is applied to the deaths identified through the public-use data set from ICD-9 codes for tuberculosis (010 to 018). Co-infection with HIV does not lead to double counting of dually exposed individuals because both the death certificate ICD-9 system and the AIDS surveillance data from the CDC that this study uses impose a hierarchy of classification that assigns dually infected individuals to HIV/AIDS instead of tuberculosis.
5.2.3 Mortality Costs
The death of a person represents a loss of productive potential, whether in the marketplace (employment) or in the household. This productive potential has a clear economic value. For example, many families buy life insurance to protect against the potential loss of future productive contributions by those contributing to the upkeep of the household through jobs, homemaking responsibilities, or both. In instances of wrongful death, court cases make awards to family members of decedents based at least partially on their expected future earnings as well as the cost of replacing their expected contributions to maintaining the household. The "value of life" is recognized to be more than the replacement of future productive contributions. There are two major methods generally used to estimate the costs of premature mortality: the human capital and the willingness-to-pay approaches, both of which were briefly discussed in chapter 3.
This study applies the human capital approach to the estimation of mortality costs. This is the approach that has been used in most previous cost-of-illness studies (Hodgson and Meiners 1982). The approach to estimation of the cost for an individual is to sum the present discounted value of estimated productivity (market as well as nonmarket) over the person's expected remaining lifetime. Using this approach, a dollar value is assigned to production lost because of premature death.
The table of expected lifetime productivity (market plus nonmarket/household) by age and sex for 1992 was obtained in a personal communication from Dorothy Rice (1997). The formula for these calculations appears in Rice et al. (1990). Because Rice estimated lifetime productivity values for discount rates of 2, 4, and 6 percent, but not 3 percent, this value was interpolated from the values for 2 percent and 4 percent, which were provided by Rice. Analysis of the tables provided by Rice indicated that the interpolation should use the geometric mean, which actually provides a lower estimate of the value than use of the arithmetic mean or "average" provides. These estimates appear in appendix B of this report. Recent research suggests that a 3-percent discount value (as well as other, higher values) should be used in performing cost-benefit studies of health services (Gold et al. 1996).
The calculation of "present discounted value of expected lifetime earnings" for a person of a given age and gender begins with Bureau of Census "life tables." These tables provide average expected additional years of life (which depends on gender and current age) and use data on the average expected value of market (wages plus the value of fringe benefits) plus nonmarket productivity by age and gender. Expected productivity values are generally assumed to grow by 1 percent annually, reflecting expected growth in real labor productivity over time. Summing these values across the expected lifetime yields total expected lifetime productivity. For economic purposes, values in future years are adjusted down (discounted) by an amount related to the number of years in the future and the selected discount rate (equivalent to an interest rate). This adjustment reflects that a dollar today is worth more next year because it can accrue interest in the interim. Accordingly, a future value is reduced when considered in present discounted value terms. The higher the discount rate and the further in the future, the lower the present discounted value of future productivity.
Discussion is warranted on at least two particular issues in application of human capital estimates to alcohol and drug abusers: the discount rate and the expected future course of productivity. It is possible that a person with an alcohol or drug abuse problem may have lower productivity than a member of the general population - indeed, this is examined and supported in the next section of this report. The question is whether such a factor would justify using values lower than those for the general population in estimating future productivity. The rationale for not making such an adjustment or reduction is that if alcohol or drug abuse has caused the initial/current deficit relative to expected productivity, then future deficits will probably also be caused by those problems. Although it is likely that alcohol and drug abusers would have had lower future productivity, this would be caused by and attributable to alcohol and drug abuse, and making the adjustment would omit a material effect and cost that should be counted.
The appropriate discount rate for human capital estimates is somewhat controversial (Burkhead and Miner 1971). A high discount rate gives a lower present value of expected earnings, and a low discount rate gives a higher present value of expected earnings. There is extensive debate about what rate is appropriate for what kinds of studies. Higher rates (such as 8 to 10 percent) are used when there is more risk or uncertainty associated with a stream of values, while lower discount rates (2 to 4 percent) are used when there is less risk or uncertainty around events. Higher rates give preference to current benefits and consumption, whereas lower discount rates give more emphasis to future consumption and benefits. The major prior alcohol and drug studies have used a value in between - 6 percent - and the primary justification for using this value is to allow comparability with previous estimates. However, lower values, such as 3 percent, are apparently being used in much cost-benefit work for the Office of Management and Budget. This study reports costs using rates of 3, 4, and 6 percent. The more conservative, 6-percent figure (in that it gives lower cost estimates) is used in constructing and reporting the total estimates.
Using a 6-percent discount rate, the costs of alcohol and drug abuse-related deaths in 1992 are estimated to be $45.2 billion. About $30.7 billion of this is attributed to alcohol abuse (see table 5.5), and $14.6 billion is attributed to drug abuse (see table 5.6), or about $290,000 per alcohol-related death and $570,000 per drug-related death. The value per drug death is much higher because much of the alcohol-related mortality occurs at older ages as a result of chronic effects (cirrhosis and neoplasms), whereas virtually all the drug deaths are among young and prime-age adults (drug overdoses, homicide, and AIDS). Estimates are also calculated and reported using 3-percent and 4-percent discount rates. At a 3-percent discount rate, the estimated costs are $45.7 and $21.3 billion, respectively, or $425,000 and $835,000 per death for alcohol and drug abuse.
Much higher costs would result from application of the willingness-to-pay approach. A review of the willingness-to-pay literature found an average value per death across 47 studies of $2.3 million in 1989 dollars (cited in Miller et al. 1995). Adjusted for 3 years of inflation (about 12 percent to $2.55 million per death), this approach would indicate losses of $274 billion for alcohol-related mortality and $65 billion for drug-related deaths. These values are almost 9 and 4.5 times greater, respectively, than the estimates obtained using the human capital approach with a 6-percent discount rate. The application of the willingness-to-pay technique should be given further consideration. It was found that for smoking behaviors, willingness-to-pay values were substantially lower than for other populations and risks. It is worth considering whether the willingness-to-pay to save the lives of alcohol and drug abusers would be different from values for other types of risk and how such values would compare with other willingness-to-pay values.