Describing and Predicting Drug Epidemics
» Project Details
Drug use and related problems can evolve very rapidly over time, suggesting that they are driven by internal (endogenous) feedback, not just by exogenous factors. The most widely recognized example of such feedback is that initiation can be contagious, with new users introducing others to the drug, who in turn spread the drug to still others producing a chain reaction. There are many other such feedbacks.
It is increasingly clear that drug policy ought likewise to evolve dynamically over time along with a drug epidemic. For example, an argument can be made that enforcement is relatively more cost-effective in the early, explosive growth stage of an epidemic, whereas treatment should figure more prominently after the epidemic has peaked.
Analysis linking policy to the evolving nature of drug epidemics has to date rested on a very narrow empirical basis, heavily influenced by US national data series for cocaine from roughly 1975 to the present. The principal objective of this project is to study additional time series documenting rapid changes in drug use and problems. The focus will be on emergency department mentions from the DAWN system because they offer the best combination of length (at least 22 bi-annual observations), consistency in data collection, and specificity by drug, geographic unit (the city), and demographic group.
The methods will be simple and data-driven. Series will be characterized qualitatively (distinguishing, e.g., between steady increases and spikes) and quantitatively (e.g., the height of a spike relative to baseline). Then patterns will be sought across these series (e.g., if it happened to turn out that spikes were sharper for stimulants than for marijuana or for men than for women, that would be noted). The goal is to generate fundamental insights by looking at data in a novel way, not to use very sophisticated methods to tease out patterns that are otherwise invisible.
It is expected that the project will generate some policy relevant insights, e.g., identifying bellwether cities or demographic groups. It is not expected that this small project will exhaust the topic of empirical examination of drug epidemic time-series. However, it should help give focus and direction to further inquiries in this vein.