consumed per incident; possible examples include overdoses, domestic violence, and motor vehicle accidents. Other risks may vary mostly as a function of the frequency of consumption; possible examples include dependency, income-generating crime, lost productivity, and poor parenting.
Are there lagged or delayed effects of drug use? Examples of this would include AIDS deaths as a lagged consequence of injection drug use with contaminated needles, and liver cirrhosis as a lagged effect of alcohol consumption.
How do these dose-response parameters vary across types of drugs, types of users (age, gender, socioeconomic status, etc.), and geographic, temporal, and cultural variations in the purity and potency of the substance and the ways in which it is consumed (e.g., snorting versus smoking)?
How can we determine the causal direction of dose-response relationships? One approach is through laboratory experiments; for example, many studies have examined the effects of various drugs on cognitive or psychomotor functioning or on aggressive behavior. But these studies provide limited evidence on drug use in realistic social environments or on the aggregate contribution of drugs to various categories of harm. Correlational field studies are vulnerable to the possibility that individuals with a higher propensity for danger self-select higher consumption levels (Zuckerman, 1994). This will spuriously inflate the quantity-risk relationship.
A better understanding of this full range of dose-response relationships would be valuable for many reasons. First, this information might provide an important deterrent to initiation for nonusers and to escalation for casual users. Second, such information would provide a firmer foundation for estimates of the aggregate costs of drug use (Harwood et al., 1998; Rice, 1999). Third, it would facilitate longitudinal inferences regarding trends in drug use and its outcomes. For example, to the extent that morbidity and mortality are sometimes lagged consequences of drug use or consequences of cumulative rather than incidental use, data on emergency room visits and drug-related deaths are potentially misleading as proxies for otherwise underestimated hard drug prevalence. Fourth, a better understanding of dose-response relationships might support more effective decision making about the allocation and targeting of drug policy instruments and resources (e.g., arrests, prison space, treatment slots, prevention efforts) across types of users, drugs, and settings. Finally, this kind of information might facilitate the development of more sophisticated and credible analytical models of the drug problem, its