Measures to quantify the abuse of prescription opioids: a review of data sources and metrics

Abstract

Purpose: The abuse and nonmedical use of prescription opioids and its subsequent consequences are an important public health concern. This phenomenon has paralleled the increase in the therapeutic use of opioids for pain management. There is thus a need to measure prescription opioid abuse to understand trends over time and to compare abuse of one product to another. The purpose of this review is to provide an overview of the strengths and weaknesses of frequently used numerators and denominators in "abuse ratios" (ARs).

Methods: For this review, we critically evaluated the various measures to quantify drug availability and the available data sources to measure prescription opioid abuse.

Results: There are currently no commonly adopted metrics for measuring either the prevalence of opioid abuse, or abuse relative to drug availability. Because the settings, manifestations, and severity of abuse can vary from one person to the next, no one measure of abuse, abuse-related outcome, or drug exposure is ideal. Each measure of abuse captures a specific facet of abuse, but not the whole spectrum. Reliable estimation of population-adjusted or utilization-adjusted rates of abuse can be accomplished with a prescription opioid AR. This metric estimates the prevalence of abuse in a given population or abuse relative to how much drug is available, and, in certain cases, can be used to compare abuse among various opioid drugs. AR measurements in the literature vary in the inclusion of specific measures of abuse and availability, and there is little consensus in the field regarding which measures allow for the most appropriate approximation of the extent of abuse, and for comparisons among opioids. Crude numbers of outcomes related to abuse (e.g., emergency department visits, treatment admissions, and overdoses) cannot be properly understood without context as these may overestimate or underestimate the true scope and severity of prescription opioid abuse. They can, however, serve as numerators in properly constructed ARs. The denominator of the AR provides the necessary context by accounting for populations at risk or drug availability (e.g., prescriptions or tablets dispensed, unique recipients of dispensed drug, total patient days of therapy, or kilograms sold), and each comes with its own set of assumptions to consider.

Conclusions: Moving forward, it is important that there be a common understanding in the scientific community regarding how to select appropriate measures to serve as numerators and denominators in AR calculations, and how to interpret the resultant findings. There is no single best measure of abuse for use as a numerator in an AR, and each must be chosen and interpreted in the context of what it measures. For public health considerations, one must always look at both absolute numbers and adjusted numbers. When conducting multiple analyses using different measures of exposure as denominators, differences in ARs are not unexpected, but one should explore why there are differences and assess the appropriateness of each of the denominators.

Keywords: abuse; data sources; diversion; measurement; metrics; pharmacoepidemiology; prescription opioid.