Alternatives to Opioids Prescribing Rates across States: Provider and Beneficiary Characteristics

Background. The path to opioid addiction often begins with doctors’ legal prescription of painkillers. To curb opioid over prescription, some health care providers are returning to alternatives to opiates (ALTOs) such as Tylenol, Neurontin and many non-steroidal anti-inflammatory drugs as safe alternatives for alleviating pain.

Objective. We consider the role that physician characteristics play when interacted with beneficiary characteristics (race sex, age, dual status, and average risk score) in shaping the prescription of ALTOs. We test if the strength of these associations is attenuated by the overall perceived quality of doctors based on online reviews (5 stars, 4 stars, less than 4 stars).

Methods. We combine Medicare Part D Prescriber Public Use File from years 2013 to 2016 using NPI numbers with data on physician characteristics obtained from ZocDoc, a website that provides online medical care appointment booking service. The physician's characteristics include gender, specialties, education and training, languages spoken, board certifications, office locations, and user-submitted reviews and ratings. We predict physicians’ race/ethnicity from their names using a machine learning algorithm trained from the US census data.

We define three rates of ALTOs prescription:

• Number of beneficiaries receiving an ALTOs prescription as a fraction of people being prescribed both ALTOs and opioids
• Number of claims prescribing ALTOs as a fraction of both ALTOs- and opioids- related claims.
• Cost of claims prescribing ALTOs as a fraction of both ALTOs- and opioids- related costs.

Preliminary results. Comparing 2016 to 2013, all states have experienced a systematic increase in the ALTOs claims prescription rates. ALTOs prescription rates are highest among specialties such as osteopathic manipulative medicine, psychology, sleep medicine, geriatric psychiatry, cardiology, and neuropsychiatry, and lowest among prosthetics and surgery. Much smaller gains in ALTOs are achieved when we look at the percentage of beneficiaries reached and at the amount spent on ALTOs compared to opioids. Most states experienced a decrease in the cost of ALTOs-related claims relative to opioids, even though the relative proportion of claims prescribing ALTOs increased. The possible explanation is the increase of the retail opioid prices during the same period.

Prescribers are less likely to provide ALTOs and more likely to favor opioids if they are in isolated small rural towns and if they are more likely to attend to older, white, healthier, and non-dual eligible patients. We find no gender match associations, but that physician-patient race-match leads to a 15% higher prescription rate of ALTOs. This association is smaller albeit significant as the overall rating of a doctor increases and becomes zero for physicians with 5-star ratings on waiting time.

ALTOs need to be considered as a safe strategy for pain management. Systematic disparities in pain management along prescriber and patient characteristics should not determine the quality of care. Benchmarking improvements by specialty can help inform achievable targets and help create guidelines and standards that reflect the characteristics of the population.