Performance on the American Board of Family Medicine Certification Examination by Country of Medical Training

Discussion

In this study, we have demonstrated that USMGs and CMGs have higher rates of passing the ABFM CE for initial board certification and recertification than NC-FMGs. This finding is consistent with our study hypothesis and with conclusions in the published Internal Medicine literature.^{3⇓⇓⇓⇓–8} CMGs have higher overall pass rates than USMGs. This difference is possibly because of the lower numbers of CMGs compared with USMGs. Statistically significant differences occur in the overall comparison but not for each year of the study period. The reasons behind the differences in examinee performance are ultimately unknown. There are likely differences in the international medical school curricula, and these differences should be studied in the future.

We also found statistically significant decreasing pass rate trends for USMGs and CMGs. This finding is very concerning. A similar decreasing pass rate trend has been seen recently on the American Board of Surgery Certifying Examination.¹⁰ The explanation for these decreasing trends on the ABFM CE is unclear. The examination blueprint changed in 2006 from the classic test theory psychometric model to one that relies on item response theory. This significant change was felt to be necessary to direct relevant field content while focusing on the knowledge required to practice a higher quality of family medicine.¹¹ However, this chang in examination blueprint did not go into effect until a few years into the study period. This study would suggest that the decreasing pass rate trend had been significant even before the blueprint changed. Further examination of these trends and efforts to determine the root cause is warranted and recommended.

There are several notable limitations to this study. First, because our study is retrospective, a cause and effect for the results is not able to be determined. Confounding variables such as the applicant quality, medical school and residency program curricula, and ability to attract effective teaching staff or clinical material are likely some of the reasons behind our findings. Second, initial certification candidates could not be differentiated from recertification candidates. Even though both groups of candidates took the same examination, there are inherent differences between these groups. Third, it is unclear where examinees completed residency training. This may have an impact on examination outcome. Finally, there were no data available for the age or length of time since medical school and residency training for recertification examinees.

These limitations are a source of future directions with regard to research related to the ABFM CE. Separating initial certification candidates from recertification candidates would make for more specific and homogenous study groups. Further research looking at residency training would be worthwhile to describe other potential demographic predictors of ABFM CE performance for initial certification and recertification. In addition, evaluating time from medical school and residency training would give insight into how timing factors are important for recertification candidates.

Strengths of this study include its relatively straightforward design. Because the examination results used here are available publically, this study could easily be repeated. Our study is the first to show that the location of medical school training can affect board examination outcomes in Family Medicine for initial certification and recertification. This study also incorporates the entire population of examinees as opposed to a cohort of local or regional examinees.