INTRODUCTION

Prescription drug monitoring programs (PDMP) are state-level electronic databases that collect and make available information regarding controlled substances (e.g., opioids) prescriptions.1,2 PDMPs have proven effective tools for addressing the prescription drug abuse epidemic.1,3 However, clinicians’ PDMP use is still suboptimal—the registration rate of PDMPs among prescribers was 35% in 2014.2 There is limited research on the up-to-date prevalence and determinants of PDMP use among physicians. This study examines physicians’ PDMP use and factors associated with use in a recent nationally representative sample.

METHODS

We used the 2018–2019 National Electronic Health Records Survey (NEHRS), a nationally representative survey of all US office-based practicing physicians that assesses the use of health information exchange and prescribing practices.4 We excluded physicians who answered “never” or had missing data to the question related to the frequency of controlled substances prescriptions (n = 187) due to skipping logic. The final analytic sample included 1730 physicians (representing 306,472 US physicians).

PDMP use was assessed using a question about the frequency of checking PDMP before prescribing a controlled substance, and we categorized those who responded “often” and “sometimes” as PDMP users and “rarely” and “never” as non-users. We dichotomized PDMP use status for ease of interpretability. We selected potential associated predictors for PDMP use in physician- and practice-level characteristics.1,2 Detailed questions and response options are described elsewhere.4

Survey design–adjusted bivariate analysis was used to analyze differences in physician- and practice-level characteristics by PDMP use status. Multivariable logistic regression was then estimated to examine the factors associated with PDMP use when controlling for other factors. For sensitivity analyses, we conducted (1) ordered categories of PDMP use (from “never” to “often”) using ordered logistic regression and multinomial logistic regression models by including the same set of predictors. This study was deemed exempt by Yale University School of Medicine.

RESULTS

Of 1730 physicians (72.2% aged ≥ 50 years old; 67.5% male), 78.3% reported PDMP use (Table 1). Frequency of PDMP use was (1) 63.8% “often,” (2) 14.6% “sometimes,” (3) 12.1% “rarely,” and (4) 9.6% “never.” In the multivariable logistic regression model, we found greater odds of PDMP use among (1) primary care physicians (vs. other specialties) (adjusted odds ratio [aOR] = 2.75, 95% CI = 1.55, 4.89), (2) those who more frequently (vs. “never or rarely”) sent e-prescriptions for controlled substances to the pharmacy (aOR = 1.76, 95% CI = 1.03, 3.00), and (3) having more midlevel providers (vs. none)—3 or more midlevel providers (aOR = 2.63, 95% CI = 1.35, 5.14) (Table 2). Findings from the sensitivity analyses were similar (results are not shown).

Table 1 Characteristics of Respondents and Bivariate Associations with PDMP Use Status
Full size table
Table 2 Results of Multivariable Logistic Regression Model on PDMP Use Status
Full size table

DISCUSSION

Most (78.3%) US physicians reported PDMP use in 2018–2019. This finding suggests that PDMP use has doubled since 2014, a trend that may be driven by state-level mandates of PDMP use.1

Consistent with previous studies,1 we found that surgeons or other medical specialties were less likely to use the PDMP compared to primary care physicians (PCPs). PCPs are more likely to prescribe opioids; they are more aware of the PDMP programs and acknowledge their importance.1 Additional effort may be needed to emphasize the benefits of PDMP use in specialty care settings. We also found a positive association between frequent use of e-prescribing and PDMP, which may be due to synergies created by using both tools (e.g., cross-referencing PDMP database).5 Having midlevel providers (e.g., nurse practitioners, physician assistants) was identified as a facilitator, which suggests that working with other health professionals might increase awareness and use of PDMPs.6 This finding is reasonable since nurse practitioners and physician assistants, in many states, require physician oversight to prescribe controlled substances.1

Study limitations include self-report biases, limited generalizability in non-ambulatory care settings, and limited information on other relevant variables (e.g., states, demographic). Particularly, the current study did not directly examine the PDMP use by specific specialties—for example, oncologists’ PDMP use might be higher than that of other specialties due to higher prescribing of opioids.1 Future studies should examine the patterns of PDMP use by specialties and states. Nevertheless, this study highlights the potential factors that increase the likelihood of PDMP use. Further efforts are still needed to optimize the PDMP systems to promote appropriate prescribing of controlled substances among US physicians. Additional studies are warranted to further examine the multi-level contexts of PDMP use and identify implementation barriers in practices.