Do incentives nudge physicians to prescribe opioids for pain?

Humans have used opioids to treat pain, both physical and mental, for more than 3,000 years. Ancient Egyptian documents like the Ebers Papyrus recommend using an extract of poppy seeds to calm crying children. In the “Odyssey,” Homer describes a drug “to lull all pain and anger and bring forgetfulness of every sorrow.”

Both the analgesic and addictive properties of these drugs have long been recognized, and the tension between relief and dependence now lies at the heart of our country’s opioid epidemic. Opioids are effective at easing pain, but the number of opioid overdose deaths has quadrupled since 1999, and a new report from the Centers for Disease Control and Prevention shows that an estimated 24,000 babies a year are born dependent on opioids.

There’s plenty of blame to go around, but one central concern is whether doctors are overprescribing opioids — and if we might inadvertently be offering them incentives to do so.

The Centers for Medicare and Medicaid Services currently makes incentive payments to hospitals based partially on how well they do on patient satisfaction surveys. These surveys ask patients how they feel their pain was controlled and whether providers did everything they could to help with pain.

Critics have argued that these incentive payments are making the problem worse — that questions like, “Did hospital staff do everything they could to help with your pain?” set unrealistic expectations and foster more aggressive opioid prescribing by clinicians to avoid financial penalties.

Last month, CMS announced a proposal to divorce incentive payments from the answers to the pain surveys while at the same time arguing that these payments have a “very limited connection” to survey responses. The proposal comes on the heels of the Promoting Responsible Opioid Prescribing Act, a bipartisan bill introduced in February, which, if passed, would have a similar effect.

Our work as physicians and as health policy researchers suggests that the move might help — a bit. But to truly curb the nation’s opioid epidemic while still ensuring that patients’ pain is adequately controlled, we also need a clearer framework for how and when to prescribe opioids, a definitive way to measure how appropriately physicians are prescribing them, and accountability for these measures.

On the surface, payments linked to pain scores are relatively small. Pain management accounts for less than 5 percent of Medicare incentive payments. (About 30 percent of a hospital’s total performance score is determined by patient satisfaction scores, only one-eighth of which is based on pain management.) But quantifying the dollar amount allocated to specific survey questions may ignore the outsized influence that these scores can have on prescribers’ psyches. Some polls have found that between half and three-quarters of physicians feel increased pressure to prescribe opioids because of pain-related questions on patient surveys.

Hard evidence on whether patient pain surveys actually lead to more opioid prescriptions — or whether withholding opioids results in lower patient satisfaction scores — is limited and mixed. But a recent study by one of us (A.J.) offers some new insights and a possible path forward.

This study looked at more than a half-million hospitalized Medicare beneficiaries, none of whom had recently been prescribed opioids. Nearly 15 percent of them left the hospital with a new opioid prescription. More than 40 percent of these prescriptions were still being filled 90 days later, suggesting that hospitalization may be an important risk factor for opioid dependence. Interestingly, even after accounting for patient diagnoses and severity of illness, the rate of opioid prescribing varied dramatically among hospitals — some were twice as likely as others to send patients home with a prescription for these potentially addictive painkillers.

Hospitals with better pain-control scores on patient satisfaction surveys were more likely to have prescribed opioids, but the strength of the association was small. That tells us that breaking the link between pain scores and incentive payments may be worthwhile, but it isn’t likely to make a big dent in opioid prescribing.

Patient satisfaction scores didn’t create the opioid epidemic, nor will removing them solve it. Nonetheless, removing payments for pain control may be an important step because even the perception that financial incentives are tied to pain scores may contribute to opioid prescribing.

Creating a gold standard

The central problem this research highlighted is that we lack a gold standard for defining the “appropriate use” of opioids. Without one, there’s no way to systematically measure opioid underuse and overuse.

In health care, variation in how often a treatment is used — like some hospitals prescribing opioids twice as often as others — often indicates overuse by some providers and underuse by others. But identifying which is which can be challenging.

If we want to make sure that people who need opioids get them while at the same time reducing the risk of dependence for others, we need validated tools that measure opioid underuse and overuse. This would make it possible for opioid prescribing to be used as a hospital quality measure — much like door-to-balloon time for angioplasty and cardiac stenting is for people having heart attacks — and for CMS to offer incentives for smarter opioid prescribing instead of more opioid prescribing.

A deeper understanding of appropriate opioid prescribing would also pave the way for creating validated public reports of opioid prescribing patterns by hospitals and health systems along with relative rankings. Such reports are already available at www.hospitalcompare.hhs.gov and other sites for things like hospital-acquired infections and 30-day readmission or death after being treated for a heart attack.

We support the CMS proposal to delink incentive payments from patient surveys of pain control. But a deeper problem remains: how to measure appropriate versus inappropriate opioid prescribing.

Opioids are necessary in some situations, dangerous in many, and both necessary and dangerous in others. Only by clearly identifying where on the spectrum each patient falls can we effectively and safely treat pain during hospital stays and prevent addiction-related tragedies afterward.

Dhruv Khullar, MD, is a resident physician at Massachusetts General Hospital and Harvard Medical School. Anupam B. Jena, MD, is an associate professor of health care policy at Harvard Medical School, an internist at Massachusetts General Hospital, and a scientific advisor at Precision Health Economics. He has received consulting fees from Pfizer, Novartis, Bristol-Myers Squibb, Vertex Pharmaceuticals, and Hill Rom Inc.

This article was updated to include Dr. Jena’s financial disclosures.