Editor’s  Note: PPP sponsors educational programming for its members and invites all interested parties to take part in our periodic Journal Club.  For those new to the format, a journal club is a gathering of medical professionals to discuss a scientific paper.  The paper under discussion is distributed prior to the meeting date, so that all participants can read it.  The club meeting begins with a member presenting a summary of the chosen paper.  Then the discussion begins, including questions about the study design, methods, results, and conclusions.  Finally, a consensus is rendered regarding the overall value of the paper. 

This blog entry introduces and summarizes our next Journal Club paper, so you’ll be ready for the discussion!

Join us on Twitter to participate in this Journal Club
May 14, 2020
9p EST
@pppforpatients #PPPJC

Opioid Prescribing by Primary Care Providers:
A Cross-Sectional Analysis of Nurse Practitioner, Physician Assistant and Physician Prescribing Patterns
James Lozada DO, Mukaila A Raji MD, James S Goodwin MD, and Yong-Fang Kuo, PhD


What are the opioid prescribing patterns of primary care physicians, nurse practitioners, and physician assistants? 


The opioid crisis is critically important and impactful to the medical community, to lawmakers, and to the general public. As the authors point out, nearly 1/3 of Medicare Part D beneficiaries received at least one opioid prescription in 2018, and Medicare is the most common insurance carrier among hospitalized patients who are admitted for prescription opioid overdose. The CDC has issued guidelines for prescribing opioids to non-cancer patients stating that “clinicians should avoid increasing dosage to ≥90 morphine milligram equivalents (MME)/day…for acute pain, three days or less of opioids will often be sufficient.”

Nurse practitioners (NPs), and now physician assistants (PAs), have been given independent/unsupervised practice in several states by state legislators (and in some cases, via emergency order by state governors during this COVID pandemic crisis). Even when not given full autonomy, many non-physician practitioners are able to prescribe opioids without physician oversight. In fact, in 2015, only 8 states restricted NP and PA prescription authority, although the individual states did not always overlap. For example, NPs were restricted in Oklahoma and South Carolina, but not in Iowa or Kentucky. The opposite was true for PAs.

The Elixhauser co-morbidity index includes thirty conditions and identifies comorbidities based on ICD-9-CM codes.

The authors’ hypothesis was that NPs and PAs would prescribe fewer opioids with a lower MME/day than physicians, and that those NPs and PAs in independent states would prescribe more opioids than their counterparts in supervised states.


This was a retrospective, cross-sectional analysis of 2015 Medicare claims data with a 20% national sample. Any clinician who prescribed any medication and at least 50 prescriptions from Medicare Part D file was included, and clinician specialty was determined by Part B claims. Clinicians were divided into 3 groups: primary care physicians, primary care NPs, and primary care PAs. Clinicians were excluded if they billed ≥50% of emergency department visits or if they were listed as an NP/PA in psychiatric/mental health, surgery, critical care, acute care, or obstetrics/gynecology. Hospice and palliative care claims were not requested.

While prescriptions for methadone were included, prescriptions for buprenorphine, naltrexone, and injection opioids were not. The authors defined a high dose-dose opioid as containing ≥ 100 MME/day and a long-term opioid as being prescribed for >90 days. Overprescribers were defined as clinicians who prescribed at a high frequency (>50% of patients), high dose (100 MME/day to >10% of patients) or long-term (>90 days to >20% of patients). The definition of overprescriber was derived from two standard deviations above the mean for all primary care providers. Patients concurrently enrolled in Medicaid were qualified as “low-income.”

Comorbidities were identified for a subset of primary care clinicians and a percentage of patients who had ≥ 3 comorbidities (based on the Elixhauser comorbidity index) was added as a provider characteristic. 


In total, 156,161 primary care physicians (PCP), 42,655 primary care nurse practitioners, and 23,873 PAs were included for analysis.  Of these, 1.33% of PCPs, 6.34% of NPs and 8.84% of PAs were high-frequency prescribers. In addition, 0.95% of PCPs, 2.89% of NPs and 3.41% of PAs prescribed high-dose opioids to more than 10% of their patients. Lastly, 2.93% of PCPs, 3.9% of NPs and 3.65% of PAs prescribed long-term opioids to >20% of their patients.

Overall, NPs and PAs were more likely to be high frequency and high dose prescribers (OR NP 1.66, PA 5.73 for frequency, OR NP 1.66, PA 2.16).

Physicians were more likely to prescribe long-term opioids, but to a lower percentage of their patient population than NPPs.

In states where NPs and PAs had the authority to prescribe opioids independently, there were mores likely to prescribe at a high frequency (NP 7.5% vs. 0.2%, PA 10.0% vs. 0.7%).

The odds of over-prescribing in an independent prescribing state was 28 times that of over-prescribing in a supervised state. This held true for high-dose prescribing (OR 70.1) and long-term prescribing (OR 46.9).

PPP Initial Discussion/Critique

The authors focus on the population of Americans who are covered by Medicare insurance, which includes those age 65 and older and those who quality for Medicare due to disability. While this is a small subset of the population, it is a very relevant one given the impact opioids can have on the elderly. Opioids should be more restricted in this adult population compared to other adults and are, in fact, listed on the 2015 American Geriatrics Society Beers Criteria for Potentially Inappropriate Medication Use in Older Adults in those who have had a history of falls or fractures.

In reviewing the data, we note, along with the authors, that the dataset used was from 2015 which is near the peak of opioid prescribing in the US. This may impact the current relevance of the results, as practice habits may have changed since then. Further, the authors were not able to elucidate the indication for prescribing an opioid. Having this data may be helpful in identifying specific areas for improvement in prescribing patterns.

When examining the results, the most impactful data are the odds ratios that demonstrate that NPs and PAs were more likely to prescribe high frequency and high dose opioids and that this disparity worsened in independent practice states. While in the past, some may have attributed higher levels of opioid prescribing to NP/PAs to refilling opioids initially prescribed by their supervising physician or in oncology practices, the fact that the disparity is so significantly amplified in independent states argues that this prior conclusion was incorrect, as there would be no supervising physician, and NP/PAs should not be independently practicing in such a subspecialized field as oncology.

We believe this data should be disseminated to legislators who have legislated the practice of medicine through legislation rather than education. PAs and NPs should not have autonomous opioid prescribing privileges.

However, in the states where this is already occurring and may be resistant to change, educational requirements, including increased continuing education requirements surrounding opioid prescribing, should be instituted for non-physician practitioners. These requirements are already in place for physicians and may contribute to the lower opioid prescribing rates of physicians compare.

Discussion Questions

1) How does the population the authors studied impact the generalizability of results?

2) Were the authors’ definitions of high dose and long-term opioids and “overprescriber” correct? If you disagree, which definitions would you suggest?

3) The authors identified 156,161 primary care physicians, 42,655 primary care NPs and 23,873 PAs in a random sample. Is this a good sampling?  Any concerns?

4) What are some possible explanations for the extremely high odds ratios for overprescribing in states where NPs/PAs have unrestricted prescription authority?

5) The data for this study was in 2015. How may this impact the data and results?

6) The authors conclude that physicians and non-physician clinicians “have similar prescribing patterns.” Do you agree?

7) What are some ways we can lower opioid prescribing among physicians and non-physician clinicians alike?