LAUSR

In 2016, 11.5 million persons in the United States reported nonmedical use of prescription opioids; close to half a million reported nonmedical use of sedative agents, such as benzodiazepines; and 1.7 million reported nonmedical use of stimulant medications (1). The current prevalence of nonmedical use of controlled substances raises an essential question: How can we provide better stewardship of these medications? It is in this context that Carey and colleagues (2) conducted a study in 627391 Medicare beneficiaries who received opioid prescriptions between 2008 and 2012. The authors examined the relationship between various opioid prescription patterns, including quantity of opioids, overlapping prescriptions, and use of multiple pharmacies or prescribers (including those out of state), and the outcomes of opioid overdose and all-cause mortality. They demonstrated that patients who received a larger quantity of opioids or had a greater number of prescribers or pharmacies were more likely to have an opioid overdose, die within 30 days of an opioid overdose, or die of any cause. Further, the authors noted a doseresponse effect, with higher risk associated with a greater number of misuse patterns, and found an increased incidence of adverse outcomes even with patterns falling below the typical threshold that might trigger prescriber concern. This study adds to the growing literature demonstrating that risks for overdose and all-cause mortality are associated with specific prescribing patterns. It also shows that by analyzing prescribing data, one may assess risk and identify characteristics and patterns associated with adverse events. The strengths of this study include its use of a large national database; the longitudinal and cogent sensitivity analyses; and the provision of absolute, rather than relative, risk estimates. Limitations include the use of data from 6 to 10 years ago, given the recent decrease in opioid prescribing and the increasing role of heroin and fentanyl in the current epidemic (3); a lack of information on benzodiazepine prescriptions or patient-level data on nonmedical use of opioids or other substances; and the challenge of generalizing these findings beyond the Medicare population. Methodological challenges include known surveillance and detection biases in opioid overdose coding and the depletion of susceptibles that may occur when the cohort comprises prevalent rather than incident recipients of a medication (4). An important finding from the study is the demonstration of how prescription databases may be analyzed to provide important information to help guide clinicians' prescribing decisions. Theoretically, greater access to information about patients' prescriptions for controlled substances would help prescribers improve stewardship of these medications. The authors extrapolate their findings to state prescription drug monitoring programs (PDMPs) that contain information on controlled substances. The use of PDMPs has increased, and these programs are now available in 49 of the 50 states. Guidelines now endorse the use of PDMPs by prescribers to better inform their provision of controlled substances. The introduction of PDMPs has been associated with fewer opioid prescriptions and overdose deaths in some states (5, 6). The current use of PDMPs is not without challenges. These programs were established for a variety of uses, as reflected in their funding sources, which include such disparate stakeholders as the U.S. Department of Justice, states, and the U.S. Department of Health and Human Services (7). Further, states vary with regard to many of the basic PDMP elements, including but not limited to the schedules of controlled substances that are included, who may access that information (for example, health profession students, patients, pharmacists, or law enforcement), and the extent to which enrollment and use by prescribers are mandatory. Carey and colleagues' study highlights the need to create new PDMPs designed to efficiently and effectively make complex data, such as the type of patterns observed in large databases, available and useful to clinicians. Given the heightened recognition of the risk for overdose among patients prescribed opioids for pain, the high prevalence of nonmedical use of controlled substances, and the all-hands-on-deck attitude to address the opioid epidemic, the moment seems right for a new generation of PDMPs. Developers of next-generation PDMPs will have to address specific problems. One challenge is that standardization of PDMP data across states is limited (7). Various stakeholders and vendors control the content of PDMP data, and providers are not uniformly presented with risk patterns. If risk information is available, it may be presented through proprietary formats and risk calculators. In addition, as noted a decade ago (8), PDMPs are not uniformly integrated into the electronic medical record, requiring prescribers to log into a separate computer interface with a unique password, creating barriers to the use of these programs. Finally, PDMPs have an important limitation that must be acknowledged to avoid false reassurance: Most persons who use prescription medications for nonmedical purposes do not receive the drugs directly from prescribers; therefore, their information is not included in PDMPs (9, 10). Strategies to enhance PDMPs so that they provide more sophisticated, user-friendly data, as well as research on how to improve outcomes associated with prescriber use of PDMP data, are needed. As prescribers, we are privileged to control access to controlled substances. With that privilege comes the responsibility of stewardship. Stewardship programs provide coordinated efforts to promote the appropriate use of medications, improve patient outcomes, and reduce adverse public health effects. The most common and successful stewardship programs have been established in the infectious disease field, where they have increased guideline-concordant provision of antibiotics, decreased microbial resistance, and reduced the spread of multidrug-resistant organisms. Given the risks for overdose and death associated with opioid prescribing and the widespread diversion and nonmedical use of controlled substances in the United States, the current findings should spark processes designed to strengthen our stewardship of controlled substances. As a start, an acute need exists to improve and customize PDMPs to better serve the needs of prescribers and, in turn, the patients to whom they provide care.