American Board of Ophthalmology Certifying Examination Performance and Opioid Prescription Patterns

Board certification is viewed by many health care institutions, insurance companies, and the public as a necessary credential for practicing physicians.^1-3  Once certified, participation in continuing medical education (CME) and continuing certification activities (also known as Maintenance of Certification [MOC]) is associated with improved patient outcomes.^4,5 

An important aspect of the delivery of safe patient care is management of patients’ pain in the everevolving opioid epidemic. Ophthalmologists are not spared from the opioid epidemic and routinely prescribe opioids following surgery.^6-8  Using the current pain score algorithms is problematic in determining the best course of action for pain control. Additionally, as many specialties attempt to thwart over-prescription of opioids without compromising patient care, physicians across specialties are using different measures to limit opioid prescriptions.^9-13  The Department of Ophthalmology at Mayo Clinic has proposed guidelines for appropriate opioid prescriptions following ophthalmic surgery¹⁴  and other groups similarly have developed protocols for improved opioid prescribing.¹⁵  Even with appropriate prescription patterns, however, a short course of opioids has the potential to lead to opioid dependence following ophthalmic procedures.¹⁶ 

Ophthalmologists are responsible for only a fraction of all opioids that are prescribed following surgery.⁶  Still, in light of the current opioid epidemic and the risk for opioid dependence following any opioid prescription,¹⁶  ophthalmologists need to be cognizant of prudent and appropriate opioid prescription practices. Although many regard initial board certification as predictive of delivery of high quality and safe patient care, it is important to test this assumption through analysis of practice patterns amongst BC and non-BC physicians. Achieving American Board of Ophthalmology (ABO) Certification, thus, may serve as a means of identifying physicians who will appropriately prescribe opioids in clinical and surgical settings. This study aims to answer this question by identifying opioid prescription patterns amongst physicians within the Medicare Part D database based on ABO certification status.

This is a quality improvement project examining opioid prescribing patterns using the Medicare Part D Database from January 1, 2013, through December 31, 2018. The study adhered to the tenets of the Declaration of Helsinki and was deemed exempt from review by the Mayo Clinic Institutional Review Board.

The total number of opioids, mean cost of opioids, and mean days’ supply of opioids on an annual basis were analyzed for all ophthalmologists. Only short-acting opioids as defined by the Centers for Medicare and Medicaid Services (CMS) were analyzed for this investigation.¹⁷  The National Provider Identifier (NPI) number from the Medicare Part D database was then cross-referenced to the ABO database of diplomates. All ophthalmologists within the Medicare Part D database from 2013 – 2018 were included. Those ophthalmologists who sat for ABO certification exams prior to this period were included if they were included within the Medicare database. Each physician was assigned a unique identifier and their opioid prescription patterns were analyzed based on ABO certification status, as follows: (1) certified without failing any qualification examination; (2) certified after previously failing the WQE, the OE, or both, or; (3) never achieving ABO certification. Medicare Part D data is presented on an annual basis and thus the opioid prescribing data was aggregated annually based on ABO certification status for each diplomate during this study period. Individual physicians may have multiple entries during the study period as the data available are presented on an annual basis.

The Medicare data is provided for all physicians, however, the actual number of prescriptions for each physician is reported only if no prescriptions are written or if more than 10 prescriptions are written for that calendar year. Physicians writing between 1 and 10 opioid prescriptions in a calendar year were reported with the value left “blank” within the Medicare database as a measure to protect privacy. These physicians writing between 1 and 10 prescriptions were estimated to write 5 prescriptions annually, similar to the methodology of Patel and Sternberg.⁶  The opioid costs and duration of supply were not adjusted.

The Medicare Part D databases were downloaded and entered into Microsoft Excel 2010 (Microsoft Corporation, Redmond, Washington, USA). For statistical analysis, BlueSky Statistics (Version 7.4, Chicago, Illinois, USA) was used. A mixed linear regression model was performed to account for the multiple observations per physician within the database. The 3 main opioid metrics were analyzed using the certification status as the explanatory variable of interest and then an additional model with gender and physician rural-urban commuting area (RUCA) codes included as covariates was performed. A Wilcoxon-rank sum test was performed comparing metrics over time. A p-value less than 0.05 was considered to be statistically significant.

From 2013-2018 there were 22749 ophthalmologists with a unique NPI registered in the Medicare Part D prescriber database with a total of 110995 entries (annual records) during the study period. Of these physicians, 6822 (30.0%) also had records within the ABO database. Analyzing only the physicians with ABO records, the 6,822 ophthalmologists had 32258 records in the database during the study period. Analyzing all the available ophthalmologists, 5765 (84.5%) passed the WQE on the first attempt, with only 191 (2.8%) not passing the WQE on the most recent attempt within the dataset. As regards the OE, 5594 (82.0%) of ophthalmologists passed on the first attempt with only 48 (0.7%) not passing the OE on the most recent attempt. Within this database, 4929 (72.3%) of ophthalmologists achieved board certification on their first attempt; 1693 (24.8%) failed 1 or both qualifying examinations but ultimately became board certified (BC); and 199 (2.9%) ophthalmologists never became BC. Over the course of the study period, 1165 (17.1%) ophthalmologists never wrote an opioid prescription, 2124 (31.1%) wrote between 1-10 opioid prescriptions, and 3532 (51.8%) wrote more than 10 opioid prescriptions.

In this study, the mean number of opioid prescriptions written per year per ophthalmologist was 9.4 ± 31.6 and when extrapolated to all ophthalmologists, the mean number of opioid prescriptions per year per ophthalmologist (including those ophthalmologists with “blank fields”) was 7.4 ± 23.4 prescriptions annually during the study period. The mean total cost of the opioids was $68.0 ± $377.4 dollars per year and the mean total supply of opioids was 42.7 ± 436.9 days. Table 1 high-lights the trends in opioid prescription patterns over time from 2013 to 2018. There were significant reductions in all metrics from 2013 to 2018, except in the mean number of days supplied which dropped from 50.1 ± 216.3 in 2013 to 28.3 ± 231.1 in 2018, however this was not significant, p = 0.118.

Opioid prescription patterns were then compared based on ABO certification as outlined above. Analyzing all total opioid prescriptions extrapolated to all ophthalmologists, the point estimate for total prescriptions per year was 7.33, standard error (SE) 0.3 for ophthalmologists who were certified on the first attempt (n=4929). Using these ophthalmologists as a reference point, the estimate for those who failed one or both qualifying examinations, but then became BC (n=1693) was −1.7, SE 0.6 (p = 0.0031); and the estimate for those ophthalmologists who never became BC (n=199) was +0.8, SE 1.5 (p = 0.5318, Figure 1A).

Controlling for physician gender and RUCA code covariates, the point estimate for total opioid prescriptions per year was 5.5, SE 0.4 for ophthalmologists who were certified on the first attempt (n=4929) and then using this cohort as a reference point, the estimate for those who failed one or both qualifying examinations, but then became BC (n=1693) was −1.8, SE 0.6 (p = 0.0015); and the estimate for those ophthalmologists who never became BC (n=199) was +0.5, SE 1.5 (p = 0.7582).

Comparing BC versus non-BC ophthalmologists, the point estimate for total prescriptions per year was 8.1, SE 1.4 for non-BC ophthalmologists (n=199). Using non-BC ophthalmologists as a reference point, the estimate for BC ophthalmologists was −1.2, SE 1.5 (p = 0.4060).

The point estimate for cost of opioids prescribed per year was $55.17, SE 4.4 for ophthalmologists who were certified on the first attempt (n=4929). Using these ophthalmologists as a reference point, the estimate for those who failed one or both qualifying examinations, but then became BC (n=1693) was $−12.29, SE 8.7 (p = 0.1581); and for those ophthalmologists who never became BC (n=199) was +$69.54, SE 21.7 (p = 0.0014, Figure 1B).

Controlling for physician gender and RUCA code covariates, the point estimate for cost of opioids prescribed per year was $36.15, SE 6.5 for ophthalmologists who were certified on the first attempt (n=4929). Again, using these ophthalmologists as a reference point, the estimate for those who failed one or both qualifying examinations, but then became BC (n=1693) was −$13.30, SE 8.7 (p = 0.1272); and the estimate for those ophthalmologists who never became BC (n=199) was +$67.40, SE 21.8 (p = 0.0020).

Comparing BC versus non-BC ophthalmologists, the point estimate for cost of opioids prescribed per year was $124.70, SE 21.2 for non-BC ophthalmologists (n=199). Using non-certified ophthalmologists as a reference point, the estimate for BC ophthalmologists was −$72.79, SE 21.6 (p < 0.001).

The point estimate for total days’ supply of opioids prescribed per year was 30.67, SE 5.0 for ophthalmologists who were certified on the first attempt (n=4929) and using these ophthalmologists as a reference point, the estimate for those who failed one or both qualifying examinations, but then became BC (n=1693) was −6.7, SE 9.7 (p = 0.4895); and for those ophthalmologists who never became BC (n=199) the estimate was +130.0, SE 24.2 (p<0.001, Figure 1C).

Controlling for physician gender and RUCA code covariates, point estimate for total days’ supply of opioids prescribed per year was 18.0, SE 7.2 for ophthalmologists who were certified on the first attempt (n=4929). Using these ophthalmologists as a reference point, the estimate for those who failed 1 or both qualifying examinations, but then became BC (n=1693) was −7.34, SE 9.7 (p = 0.4490); and the estimate for those ophthalmologists who never became BC (n=199) was +129.0, SE 24.3 (p <0.001).

Comparing BC versus non-BC ophthalmologists, the point estimate for total days’ supply of opioids prescribed per year was 160.6, SE 23.7 for non-BC ophthalmologists (n=199). Lastly, using these ophthalmologists as a reference point, the estimate for BC ophthalmologists was −131.7, SE 24.1 (p < 0.001).

In this report analyzing performance on the ABO's certification examinations found that board certified ophthalmologists prescribed opioids for a shorter duration on annual basis and prescribed less costly opioids on an annual basis than those ophthalmologists who were not board certified. Although it appeared as though BC ophthalmologists wrote fewer opioid prescriptions per year, this value was not statistically significant on mixed model regression analysis. Interestingly, those ophthalmologists who initially failed the WQE, the OE, or both but then ultimately went on to become BC appeared to have the most prudent opioid prescription patterns with statistically fewer opioid prescriptions overall. Despite prescribing fewer days’ supply and less costs per year compared to the other 2 cohorts, those values were not statistically different than those ophthalmologists who became certified on the initial attempts. Overall, it appears that ABO-certified ophthalmologists have more appropriate opioid prescription patterns compared to those ophthalmologists who never become ABO certified as evidenced by the prescription of less costly opioids, fewer opioid prescriptions, and lower opioid supply.

The findings in this report support studies from other surgical subspecialties that board certification status may be positively correlated with improved patient outcomes.^18,19  Kopp and colleagues reported that board certification from the American Board of Surgery was associated with lower rates of disciplinary action by state medical boards.¹⁸  Xu and coauthors found that BC surgeons who had completed MOC requirements had lower complication rates.¹⁹  This was a detailed analysis of the Medicare claims database and analyzed over 1.9 million procedures by nearly 15000 surgeons and defined “outlier surgeons” as those in the highest decile of complication rates. Although BC surgeons had a lower odds ratio of being an outlier, completion of MOC requirements was not associated with decreased complication rates in orthopedic surgery or urology. ¹⁹  As seen in our study, BC ophthalmologists wrote less costly and fewer days’ supply of opioids, the correlation to patient safety events and or complications are not addressed in this manuscript. Writing fewer opioid prescriptions is not a defined patient safety metric within ophthalmology, but as mentioned above, the opioid epidemic continues to rage on and thus the use of the 3 metrics in this study may be appropriate surrogates for improved patient care outcomes.

Managing post-operative pain can be challenging and prudent prescription practices are warranted. Gray et al. found that physicians with higher clinical scores on the American Board of Internal Medicine's MOC exams were less likely to prescribe opioids than those physicians who performed less well on the exam.²⁰  This observation may reflect a paradigm shift in the management of patients’ pain as physicians and surgeons are more cognizant of the addictive nature of opioids and are transitioning to multi-disciplinary approaches to pain control.^20-22  Reducing the number of initial post-operative opioids does not lead to increased rates of refills of opioids.^13,14  Perhaps board certification and MOC performance are indicative of those physicians committed to improving opioid prescription patterns while attempting to appropriately address patient pain and limiting the use of opioids. Certainly, there are scenarios where prescribing more opioids is appropriate and thus still should be labeled a judicious opioid prescription pattern, however, within the field of ophthalmology, there are few surgeries which require opioid prescriptions more than 40 oral morphine equivalent (OME) and per the Mayo Guidelines, no ophthalmic surgery should require more than 80 OME in a patient that is opioid naïve. Thus, within ophthalmology, fewer opioid prescriptions are likely indicative of appropriate use of opioid prescriptions.

As our analysis has shown, failure to achieve board certification on the initial attempt does not preclude subsequent improvement and success. There are many factors, anxiety amongst them, that may negatively affect a candidate's performance on a licensure or certification examination.²³  This at first seems counter-intuitive, that those who failed one or both qualifying examinations, but then went on to achieve BC perhaps had improved patient safety metrics as defined by opioid prescription patterns in this study. Candidates who are ultimately successful may have benefitted on subsequent attempts from additional study or added clinical experience. There are many factors as mentioned that may affect a diplomate's performance during a qualifying examination, and thus initial failure does not translate to poor physician performance. The initial setback may be an impetus to always seek continued improvement as defined in this study by opioid prescription patterns. Diplomates that become BC without failure may become complacent and thus this may explain the differences found in our study. Furthermore, board certification itself does not guarantee improved outcomes.^24,25  For example, in a study from the vascular surgery literature, increased surgical experience was a better predictor of quality care (improved cardiovascular and morbidity outcomes) than initial performance on a board certification examination.²⁴  Interestingly, within ophthalmology, one of the factors associated with improved anatomic outcomes following rhegmatogenous retinal detachment surgery, was younger surgeon age.²⁶  Moreover, many metrics and patient factors affect surgical outcomes, making it difficult to determine specific factors associated with improved post-operative outcomes. Still, within our study, board certification overall, regardless of initial results, appeared to be associated with improved opioid prescription patterns compared to those ophthalmologists that were non-BC.

This study is limited by its retrospective design. It is difficult to draw conclusions regarding practice patterns using a large-scale Medicare database that does not capture all ophthalmology patients, namely it likely does not capture elective surgeries such as refractive surgeries which may often necessitate the use of opioids nor are we able to differentiate physicians based on their subspecialty within ophthalmology. This was partially accounted for on covariate analysis accounting for physician gender and geographic setting, but still this analysis could not account for specific subspecialty practice. Physicians may have different prescription patterns for privately insured patients which are not reflected in this dataset as well. The inability to know the exact number of prescriptions and supply of opioids in those physicians writing a small number of opioid prescriptions per year may affect the results. Comparing physicians across the time periods by aggregating the data, though, may not be the solution either given the incomplete nature of the dataset when physicians prescribed a small number of opioids, however, we attempted to account for this using a mixed model analysis. This dataset was also compiled during a time when many physicians were writing fewer opioid prescriptions as seen in Table 1. Certainly, the growing evidence around the harm of opioids may have triggered improved prescription patterns, independent of board certification status. Additionally, although unlikely, it is possible that the different cohorts of ophthalmologists managed different patient populations that required different opioid regimens. Regardless, this study is an initial step towards assessing ABO certification and physician practice patterns on a large-scale basis.

Many factors or metrics may promote improved patient care and safety such as the integration of new technology and innovation into clinical practice, the use and assessment of surgical and clinical patient logs, and the continued application of Plan-Do-Study-Act (PDSA) cycles. This study does not demonstrate causality, ABO certification likely does not directly lead to improved opioid prescription patterns, but as demonstrated by this study, board certification may be a quality metric associated with safer opioid prescription practice patterns. Advocating for all newly graduating physicians, not just ophthalmologists, to achieve board certification within their specialty may be warranted. Achieving certification and then continued maintenance of certification will allow physicians to continually improve upon their practice patterns, such as opioid prescriptions. As healthcare evolves, maintaining board certification will allow physicians to remain knowledgeable on current practice patterns and evolve their care to patients. As noted above, this study does not demonstrate directly that board certification leads to improved patient care within ophthalmology, but further studies examining ophthalmic practice patterns are warranted to better elucidate our initial findings.