Potentially Inappropriate Medication Prescribing by Nurse Practitioners and Physicians

Discussion

We found that NPs provided 6.4% of outpatient visits in primary care setting. Of all primary care visits to NPs and physicians, 24.1 per 1000 visits were associated with a PIM prescription; 9.0 per 1000 were for an initial PIM, and 15.1 per 1000 were for a refill PIM. Overall, NPs prescribed less PIM than physicians. Older age, male sex, black race, and metropolitan residence were associated with lower odds of receiving PIM. The odds of receiving an initial PIM prescription from an NP was substantially lower among black enrollees, but weaker for those living in nonmetropolitan areas.

A previous study^[18] reported that 10.5% of visits had a PIM initiated or renewed, a rate which was higher than our finding (2.4%). The difference could be in part due to differences in PIM definition and the study populations. First, the previous study restricted itself to a subset of drugs without dosage and duration criteria from the 1997 Beers criteria,^[28] whereas our study applied the HEDIS 2017 criteria, which were based on the 2015 Beers criteria.^[7] Second, that study was based on a national survey in an ambulatory setting, which covered all types of health service, from office, outpatient department, and emergency department settings, and medication information was retrieved from the medical record. Also, it did not distinguish PIM between initial prescription and refills. In our study, about one-third of PIM prescriptions were defined as an initial prescription. An initial PIM is more likely to be prescribed after an office visit from the same provider. However, a refill PIM might be given after other outpatient or telehealth services, and it could be prescribed by another provider in the same practice. Our study focused on prescription from the same provider after an office visit, so the PIM rate was lower than that reported in previous studies.^[18,29,30]

Female gender was associated with increased odds of PIM use in our study, a finding consistent with previous studies.^[19,31] However, findings from past studies on the association of age and PIM use were mixed, with some reporting a negative^[19] and others a positive association.^[19,31] The opposite findings could be a reflection of the divergent methods used and the differences in the populations studied. The positive association of age and PIM use was mainly reported from studies using medical records.^[19,31] We used administrative data and limited the prescriptions to those from the outpatient visit provider within 7 days. A refill prescription could be requested from a telehealth service or a web-based service platform. Older patients might tend to request a refill prescription through a telehealth service, not an in-person outpatient visit. Our study did not include prescriptions from telehealth services to accurately estimate the prescription associated with an outpatient visit. Otherwise, we reported increasing odds of PIM prescription in urban areas, which was opposite to the findings of previous studies.^[32,33] In previous studies, only central nervous system PIM was included, but not other more commonly prescribed PIM, like cardiovascular medication.^[34]

The increasing growth of the NP workforce has helped address the ongoing shortage of primary care physicians, particularly in medically underserved areas.^[11] This growth allows NPs to help narrow health disparities by increasing primary care access among racial and ethnic minorities.^[35] In our study, black enrollees were especially less likely to receive an initial PIM prescription from an NP visit. Older enrollees and those with more comorbidities also had substantially lower odds of receiving a refill PIM during an NP visit. Racial and ethnic minorities had lower healthcare utilization and were more likely to live in medically underserved areas, where primary care accessibility is limited, a situation accentuated in a state with restrictive NP scope of practice and lower accessibility to primary care.^[36] Texas—on whose Medicare data the current study is based—is a state with restricted prescriptive authority for NPs.^[37] Patients with multiple conditions or racial and ethnic minorities in NP-restricted states would tend to have physicians as PCPs.^[38,39] Thus, the reason for a physician visit and an NP visit might be dissimilar, with different care needs in terms of medication. This difference might also explain the lower odds of PIM prescribing by NPs compared with physicians.

Our analysis has some limitations. First, we focused on prescriptions made after an outpatient visit from physicians and NPs. Medications prescribed via telehealth or by other providers and a visit from physician assistants (PAs) or other types of providers were not counted. Because the prescriptions given from other providers were not counted, the burden of PIM prescription might have been underestimated. Second, the indications for the prescriptions were not evaluated. The intention of reducing PIM exposure is to minimize the unfavorable balance of benefit and risk in medication choice. Under certain conditions, a patient would receive benefit from a PIM. Third, our results involved Texans aged 65 and older on Medicare FFS and with Part D coverage. For individuals living outside of Texas, those in Medicare Advantage plans and those without Medicare Part D coverage, the results may be different. Fourth, provider type could be defined as a physician for part of a visit to an NP, under incident to billing rule. Thus, the difference between NP and physician prescription might be overestimated or underestimated due to misclassification of provider type. Fifth, beneficiaries who had outpatient visits with NPs only were more likely being younger, female, original entitlement of end-stage renal disease/disabled, dual eligibility, have >5 comorbidities, be hospitalized in the prior year, and reside in a rural community. The impact of these confounding factors on the association between provider type and PIM use was adjusted for in the multivariable analyses. However, the residual and unmeasured confounding factors might impact the association between provider type and PIM use. Finally, refill prescription in this study did not include the continuing medications prescribed by providers during other types of visit (e.g., tele visit) and those prescribed by other providers (e.g., PA). Therefore, the PIM prescription rate reported in this study should not be interpreted as prevalence, to compare it with other studies.

Despite some limitations, this study extends the knowledge about the PIM prescribing and correlates of PIM prescribing by NPs and physicians in outpatient settings. Previous studies rarely distinguished between initial and refill prescriptions, and few studies of prescribing have looked at how these might differ by type of provider. We observed that NPs prescribed fewer initial PIMs and were less likely to refill a PIM after an outpatient visit than physicians. The lower odds of receiving PIMs during an NP visit varied by age, race/ethnicity, rurality, and comorbidities. Future studies are needed to evaluate how the quality/outcome of NP prescribing may vary by training, degree (NP vs Doctor of Nursing Practice [DNP]), and the setting of NP practice. To improve medication management in primary care settings, further study is also needed to understand the geographic variation across states and the interaction of geographic factors with NP prescribing regulations in terms of quality and outcomes of prescribing practice.

J Am Geriatr Soc. 2021;69(7):1916-1924. © 2021 Blackwell Publishing