Report
An Ounce of Prevention, a Pound of Uncertainty: The Cost-Effectiveness of School-Based Drug Prevention Programs
Jan 1, 1999
Clarifying a Cloudy Issue
Research SummaryPublished 1999
The United States has for some time now been spending tens of billions of dollars a year in an attempt to control the trafficking and use of illicit drugs; most of those dollars have been used to support stricter enforcement. Few people, however, would claim that America's drug problem has been solved. There is growing doubt that the United States will be able to incarcerate its way out of its drug problem, and polls have shown a corresponding increase in public support for preventive approaches to reducing drug consumption.[1] But how effective is prevention—and at what price? Would a national program be feasible—and justifiable? These questions were addressed by a team of researchers from RAND's Drug Policy Research Center. The new research supported the following conclusions:
This brief shows how these results were obtained and expands on their implications. First, however, several aspects of the conclusions are worth emphasizing:
The new DPRC results are based on evaluations by other researchers[2] of two proven programs: Project ALERT (developed at RAND) and the Life Skills Training Program. Both use a "social-influence" strategy—one in which adolescents are motivated to resist pro-drug messages from peers and the media and taught methods for doing so. As evaluated, Project ALERT consisted of 11 sessions offered to seventh- and eighth-graders, while Life Skills Training consisted of 30 sessions during seventh, eight, and ninth grades.
These evaluations did not measure effectiveness at reducing cocaine consumption; typically, those who use cocaine do not begin doing so until well after their early adolescent years—and after the evaluations have stopped collecting data. Data were available, though, on use of a drug more commonly consumed by teens—marijuana. And it is known from the National Household Survey of Drug Abuse (NHSDA) that those who initiate marijuana use later are less likely to use cocaine. Furthermore, according to the survey, those who do use cocaine are apt to consume smaller quantities over the course of their lives. Finally, those who never initiate marijuana use are very unlikely ever to use cocaine, and the few who do use it consume relatively little.
The researchers assumed that, of the reductions in marijuana initiation reported over the course of the program evaluations, half were permanent (i.e., these students would never use marijuana) and half represented delays to the late teen years. The researchers then used the marijuana-cocaine relationships from the household survey to "translate" these preventions and delays into percentage reductions in lifetime cocaine use. Say, for example, that a 14-year-old is about to start using marijuana and a prevention program gets him or her to delay to age 18. It would then be inferred that his or her eventual cocaine consumption will decrease by the same percentage as the household survey has shown cocaine consumption to fall between those who initiate marijuana at 14 and those who do so at 18. But the estimate of aggregate cocaine use reduction so derived is only an initial one. To come up with their final one, the researchers took into account two other factors that would tend to decrease the estimate and two that would tend to increase it:
It was by taking all these factors into account that the researchers concluded it would be "reasonable" or "plausible" to infer that model school-based prevention programs could reduce cocaine consumption by between 2 and 11 percent. The range allows for the possibility that the best estimates the authors could make for the various factors might well be too high or too low, given the uncertainty involved.
The percentage reduction was converted into a cost- effectiveness value that could be compared with those of other cocaine control programs. That value was expressed as kilograms of cocaine consumption reduced per million dollars spent. The conversion required two steps:
The researchers estimated that model school-based prevention could reduce cocaine consumption by 7 to 60 kilograms per million dollars spent, with a "best estimate" of 26 kilograms per million dollars. These numbers are compared in Figure 1 with corresponding values for a range of enforcement approaches and for treatment, all of which had been estimated in previous DPRC studies. The mid-range estimate is more cost-effective than some enforcement approaches and less so than others. But uncertainty prevents confident comparisons with specific enforcement strategies. In fact, there is a small chance that prevention is either less cost-effective than all the enforcement approaches or more cost-effective than almost all of them. At this point in the epidemic, however, treatment is more cost-effective than even the most optimistic estimate for prevention.
It is worth emphasizing that the principal mathematical tool combining all the preceding factors is nothing more complicated than simple arithmetic (multiplication or division). This is important because, for a topic involving so much uncertainty, people should be able to substitute their own estimates for the estimates preferred by the authors, and new data should be easy to insert into the model. Someone might think, for example, that the authors have been too optimistic in inferring cocaine effects from a program's effects on marijuana or that they have been too pessimistic in their estimate of effects mediated through the market. The estimation paradigm makes it very easy for that person to compute his or her own overall cost-effectiveness estimate using a different program effectiveness number and a different market multiplier. This paradigm is thus a major contribution of the study.
Another way to evaluate the worth of prevention is to ask whether a dollar invested yields at least a dollar in benefits. Researchers outside DPRC had previously estimated the social costs of cocaine use in terms of health, crime, and other costs. A benefit-cost ratio for prevention could thus be straightforwardly derived from the use reduction estimates cited above. These worked out to be between 64 cents and about $5.60 for every dollar spent, with the mid-range preferred estimate at approximately $2.40.
These ratios, however, omit the non-cocaine-related benefits of school-based prevention. Such programs have also been shown to reduce cigarette and alcohol consumption. Although the effects for these more socially accepted drugs seem to be very small in percentage terms, the total social costs of these drugs are much larger than those for cocaine. Rough calculations suggest that, when reductions in those costs are taken into account, even the low benefit estimate is brought up to nearly a dollar. Furthermore, prevention programs generate benefits unrelated to reduced drug use that are not estimated in this study. It is thus highly likely that model school-based prevention programs are a good investment in terms of generating at least a dollar's worth of benefits for every dollar spent.
At $150 per student, a national school-based drug prevention program would clearly be affordable. It would cost about $550 million to offer it to all 3.75 million children reaching seventh grade in any one year. That's only a small fraction of the $40 billion the nation annually spends on drug control.
Such a program would not, however, dramatically alter the course of the current cocaine epidemic. The percentage effects cited above are modest, and it would take decades for them to accumulate, because the current population of cocaine users is too old to be affected directly by school-based prevention (see Figure 2).
It is important to realize, however, that the analysis discussed so far applies to the cocaine epidemic in the 1990s. Drug epidemics vary greatly in character over the course of their duration. Had prevention programs been in place in the late 1960s, about a decade before the peak in cocaine initiation, they might have been much more effective—by a multiple, perhaps a large multiple, of what is estimated here. There are two reasons for this: First, in the early years of an epidemic, initiation rates are higher, so there is more potential for reduction. Second, in those years, enthusiasm for the "new" drug has not yet been tempered by bad experiences, so the spillover effect on nonparticipants should be greater.
To the conclusions mentioned at the outset of this brief, the DPRC research team added the following insights:
This publication is part of the RAND research brief series. Research briefs present policy-oriented summaries of individual published, peer-reviewed documents or of a body of published work.
This document and trademark(s) contained herein are protected by law. This representation of RAND intellectual property is provided for noncommercial use only. Unauthorized posting of this publication online is prohibited; linking directly to this product page is encouraged. Permission is required from RAND to reproduce, or reuse in another form, any of its research documents for commercial purposes. For information on reprint and reuse permissions, please visit www.rand.org/pubs/permissions.
RAND is a nonprofit institution that helps improve policy and decisionmaking through research and analysis. RAND's publications do not necessarily reflect the opinions of its research clients and sponsors.