This article originally appeared in the Harvard Business Review on October 24, 2017.
By: Anupam B. Jena, Michael Barnett, and Dana Goldman
By any metric, opioid-related overdoses in the United States have reached epidemic proportions. Many intertwined causes have led to this crisis, from reduced access to substance-abuse treatment, to increased unemployment spurring use of prescription opioids, to online pharmacies that illegally supply prescription opioids to patients.
But health care providers are also widely held responsible for overprescribing prescription opioids. While research testing this hypothesis is mixed, it’s clear that efforts to curb the epidemic need to involve physicians and hospitals. The adoption of prescription drug monitoring programs (PDMPs) is one such effort that holds promise, though it must be made more effective.
Are Doctors to Blame?
It is often argued that growing advocacy for the systematic measurement of patient pain in the late 1990s and early 2000s, as well as pay-for-performance measurement of provider quality based on their ability to relieve patients’ pain (among other things), spurred overprescribing.
But whether opioids are truly overprescribed is difficult to tease out, because pain is hard to objectively quantify, as is the amount of pain relief that patients may receive from opioids. Patients and doctors have recently raised concerns over pain being undertreated due to greater scrutiny causing a decline in opioid prescriptions.
Nevertheless, rates of prescribing are still very high, and opioid-related overdose deaths continue to rise. By some estimates, more than 50% of opioid pills are unused by the patients who are prescribed them after surgery, which suggests significant overprescribing exists.
Enormous variation exists across providers’ opioid-prescribing habits. In our own research, for example, we have found that U.S. hospitals vary by nearly twofold in the prescribing of opioids after hospital discharge. In particular, rates of prescribing opioids upon hospital discharge range from 10% to 20% of discharged patients, depending on the hospital, with factors such as a high ratio of nurses per bed, rural location, government ownership, and high performance on inpatient pain-assessment scores all modestly associated with higher rates of opioid prescribing.
Patients may also seek pain relief from multiple providers, with each provider being potentially unaware of opioid prescriptions written by others. We found that among a sample of 1.2 million Medicare beneficiaries who filled at least two opioid prescriptions in 2011, 34% received prescriptions from two providers, 14% from three providers, and 12% from four or more providers. We found that the more providers a patient received prescriptions from, the greater the likelihood the patient would suffer an adverse event related to an opioid such as respiratory arrest.
Our research also found that among similar patients treated in the same emergency department, rates of opioid prescribing varied by nearly threefold, ranging from 7% of an emergency physician’s patients being prescribed an opioid upon discharge to over 20% of an emergency physician’s patients. But patients treated by emergency physicians who prescribed fewer opioids were no more likely to return to the emergency department with pain-related conditions.
Can Prescription Drug Monitoring Programs Help?
Efforts to get physicians and hospitals to reduce opioid prescribing have seen limited success. The most notable effort to date has been the widespread development of PDMPs, which are now in place in 49 U.S. states. These state-level programs collect data on prescriptions for opioids and other controlled substances that any provider can access online, theoretically allowing providers to identify any suspicious patterns of opioid use. For example, the system would let a doctor see if a patient was obtaining prescriptions from multiple physicians within a short time span.
But the utilization of PDMPs by physicians has been low, largely because most states do not mandate that physicians check the PDMP before prescribing an opioid. In addition, use of PDMPs by physicians is time-intensive and poorly integrated into a physician’s daily workflow (for example, most PDMPs are not embedded within the electronic health records that physicians use in their daily practice).
Mandating the use of PDMPs and making them easier to use could make a big difference. The best available evidence by the economists Thomas Buchmueller and Colleen Carey suggest that PDMPs significantly reduce prescription opioid misuse, but only when use of the PDMP by physicians is mandated, which now occurs in approximately one-fifth of all states. For instance, implementation of a PDMP with a provider mandate for use is associated with substantive reductions in misuse — for example, an 8% relative reduction in the percentage of Medicare beneficiaries who obtain prescriptions from five or more providers and a 15% reduction in the percentage obtaining prescriptions from five or more pharmacies.
However, PDMPs are not a panacea. While they can be useful, their ability to stem opioid overprescribing is inherently limited by the specific form of misuse that the programs aim to address. They primarily focus on identifying aberrant use of prescription opioids. Although this is important to consider, it’s likely that the vast majority of prescriptions reflecting clinical overuse do not meet the criteria that would flag identification by a PDMP. For example, much routine opioid prescribing can be clinically inappropriate and lead to long-term opioid dependence, but it wouldn’t be flagged by a PDMP.
Accurately identifying and targeting interventions toward physicians, physician groups, and hospitals that prescribe opioids inappropriately — either in terms of frequency, dose, or duration — is critical to reducing prescription opioid overuse. This requires investments in data infrastructure and health information technology to measure real-time opioid prescribing patterns. For example, PDMPs can and should be integrated into electronic health records so that physicians prescribing an opioid can immediately observe a patient’s opioid utilization history at the time of prescribing.
PDMP data could also be used to rapidly evaluate the effectiveness of new interventions or policies. For instance, it was recently demonstrated that “peer comparison” letters, which compare a physician’s prescribing rates with other physicians to try to increase uniformity in prescribing, that Medicare sends to potential overprescribers of controlled substances had no effect on prescribing rates of these harmful medications. This differed from prior studies that showed peer comparisons can help influence physician prescribing behavior when it comes to other medications and suggested that peer comparisons alone, without higher-powered incentives to modify physicians’ prescribing behavior, are unlikely to influence opioid-prescribing behavior.
Other interventions that target individual physicians or hospitals may be effective in reducing opioid overprescribing. They include public reporting of opioid prescribing rates, pay-for-performance incentives that take opioid prescribing into account, mandatory prescriber education with repeat requirements for continual high-frequency opioid prescribers, and a specific standardized curriculum on opioid prescribing provided to medical students and resident physicians. All of these would require rigorous design and evaluation.
The opioid crisis has myriad economic and societal impacts, including declining labor-force participation of working-age men and women, increasing costs in the criminal justice system, and growing health care spending on substance-abuse treatment. But the most profound societal impact stems from the staggering loss of life itself and its inestimable toll on victims’ families and loved ones.
Ultimately, battling this epidemic requires flexible, carefully designed, and rapidly evaluated policies — and a focus on provider-prescribing behavior, while important, is only one lever. Without productive collaborations across the public and private sectors, we may lose precious time and lives in our efforts to slow the opioid epidemic.