Jeffe DB, Andriole DA. Factors Associated With American Board of Medical Specialties Member Board Certification Among US Medical School Graduates. JAMA. 2011;306(9):961-970. doi:10.1001/jama.2011.1099
Author Affiliations: Washington University School of Medicine, St Louis, Missouri.
Context Certification by an American Board of Medical Specialties (ABMS) member board is emerging as a measure of physician quality.
Objective To identify demographic and educational factors associated with ABMS member board certification of US medical school graduates.
Design, Setting, and Participants Retrospective study of a national cohort of 1997-2000 US medical school graduates, grouped by specialty choice at graduation and followed up through March 2, 2009. In separate multivariable logistic regression models for each specialty category, factors associated with ABMS member board certification were identified.
Main Outcome Measure ABMS member board certification.
Results Of 42 440 graduates in the study sample, 37 054 (87.3%) were board certified. Graduates in all specialty categories with first-attempt passing scores in the highest tertile (vs first-attempt failing scores) on US Medical Licensing Examination Step 2 Clinical Knowledge were more likely to be board certified; adjusted odds ratios (AORs) varied by specialty category, with the lowest odds for emergency medicine (87.4% vs 73.6%; AOR, 1.82; 95% CI, 1.03-3.20) and highest odds for radiology (98.1% vs 74.9%; AOR, 13.19; 95% CI, 5.55-31.32). In each specialty category except family medicine, graduates self-identified as underrepresented racial/ethnic minorities (vs white) were less likely to be board certified, ranging from 83.5% vs 95.6% in the pediatrics category (AOR, 0.44; 95% CI, 0.33-0.58) to 71.5% vs 83.7% in the other nongeneralist specialties category (AOR, 0.79; 95% CI, 0.64-0.96). With each $50 000 unit increase in debt (vs no debt), graduates choosing obstetrics/gynecology were less likely to be board certified (AOR, 0.89; 95% CI, 0.83-0.96), and graduates choosing family medicine were more likely to be board certified (AOR, 1.13; 95% CI, 1.01-1.26).
Conclusion Demographic and educational factors were associated with board certification among US medical school graduates in every specialty category examined; findings varied among specialty categories.
Specialty-board certification by an American Board of Medical Specialties (ABMS) member board is an increasingly important credential for physicians engaged in clinical practice. Although lack of ABMS board certification does not necessarily mean that a physician is not well qualified,1,2 its presence is associated with the quality of medical care that physicians deliver to their patients.3- 5 Better patient outcomes have been observed for patients under the care of board-certified physicians compared with those under the care of non–board-certified physicians.5- 8 American Board of Medical Specialties member board certification9 and higher scores on certifying examinations among physicians certified by the American Board of Internal Medicine10 also have been associated with lower risk of physician disciplinary action, whereas lack of board certification has been associated with higher risk of such disciplinary actions as license revocation, practice suspension, probation, and public reprimand.11
American Board of Medical Specialties member board certification is currently among the criteria used by health maintenance organizations, hospitals, and health insurance plans in evaluating physicians who wish to obtain privileges or join provider organizations,7,12 by medical school promotion committees in evaluating physician faculty members for promotion and tenure,13,14 and by the Accreditation Council for Graduate Medical Education as criteria for selection of physicians to serve as graduate medical education (GME) program directors and residency review committee members.15,16 Thus, ABMS member board certification is emerging as a de facto requirement for the full participation of physicians in the US health care system, and non–board-certified physicians compose an increasingly marginalized group. We therefore sought to identify demographic, medical school, and GME variables associated with ABMS member board certification among a national cohort of US Liaison Committee on Medical Education–accredited medical school graduates.
After obtaining institutional review board approval at Washington University School of Medicine (nonhuman subjects research with waiver of consent), we constructed a database with individually linked, deidentified records for all 1993-2000 Liaison Committee on Medical Education–accredited US medical school matriculants who graduated from 1997 to 2000. Follow-up data through March 2, 2009, allowed more than 8 years of follow-up for all graduates in our database. The database included selected items from the Association of American Medical Colleges (AAMC) Student Record System; first-attempt US Medical Licensing Examination Step l and Step 2 Clinical Knowledge results, which were released with permission from the National Board of Medical Examiners; the AAMC Graduation Questionnaire; the AAMC GME Track; and the American Medical Association (AMA) Physician Masterfile.
The AAMC Graduation Questionnaire is administered voluntarily and confidentially to medical school graduates annually.17 Overall response rates among graduates in the 1997-2000 graduating classes ranged from 81% in 1999 to 91% in 2000.18- 22
The AAMC GME Track database contains the annual National GME Census data from all Accreditation Council for Graduate Medical Education–accredited programs; this census is conducted jointly by the AAMC and the AMA,23,24 with high completion rates. The training status was confirmed by program directors for 96% of all physicians in the GME Census database in the 2009-2010 academic year.24
American Medical Association Physician Masterfile data pertaining to active state medical licenses are provided by state licensing boards to the AMA and updated by these boards at least biannually.25 We used these licensing data to identify non–board-certified graduates in our study sample who were actively licensed as of March 2, 2009.
Demographic variables included graduation date and students' sex and self-identified race/ethnicity as reported on the American Medical College Application Service questionnaire. We categorized race/ethnicity as Asian/Pacific Islander, underrepresented minority in medicine (including Hispanic, black, American Indian, or Alaska Native), other/unknown (including graduates who self-identified as other or multiple races or who did not respond to this question), or white (reference group). We examined race/ethnicity because board certification rates were reportedly lower among nonwhite compared with white physicians.4,26- 29
We also included Graduation Questionnaire variables for graduates' age at graduation (<28 years vs ≥28 years), total debt, and planned specialty for board certification. Total debt at graduation was categorized as no debt, $1 to $49 999, $50 000 to $99 999, $100 000 to $149 999, and $150 000 or more. We included only graduates who answered yes to the Graduation Questionnaire item “Do you plan to become certified in a specialty?” and selected a planned specialty for board certification; questionnaire respondents who answered no or undecided to this item were not offered the opportunity on the Graduation Questionnaire to choose a specialty. Questionnaire respondents who planned to become certified in a chosen specialty/subspecialty were assigned to 1 of 8 specialty categories according to ABMS member board clinical and oral examination requirements for certification.30,31 Specialty categories were internal medicine, family medicine, and pediatrics (each 3 years of training); emergency medicine (3 years of training, oral examination); radiology (4 years of training, l year of clinical experience, oral examination); surgery/surgical specialties (each ≥5 years of training, oral examination), including surgery, urology (16 months of clinical experience), plastic surgery, orthopedic surgery (2 years of clinical experience), neurologic surgery (42 months of clinical experience), otolaryngology, colorectal surgery, thoracic surgery, and other surgical subspecialty; obstetrics/gynecology (4 years of training, 2 years of clinical experience, oral examination); and other nongeneralist specialties (each with ≥3 years of training and <2000 graduates who chose the specialty in the final study sample), including allergy and immunology, anesthesiology (oral examination), dermatology, medical genetics, neurology (oral examination), nuclear medicine, ophthalmology (oral examination), pathology, physical medicine and rehabilitation (oral examination), preventive medicine, psychiatry (oral examination), and choice of “other” specialty on the Graduation Questionnaire.
We included a dichotomous variable for first-attempt Step l results (pass vs fail) and a 4-category variable for first-attempt Step 2 Clinical Knowledge results (upper [range, 226-281], middle [range, 206-225], and lower [range, 170-205] tertiles of 3-digit passing scores vs all failing scores in the study sample) as predictor variables in the models. Using AAMC GME Track data, we created variables to distinguish between graduates who did or did not have a record of GME, complete specialty GME, transfer to a different specialty during GME, or take a leave of absence from GME and had or had not withdrawn or been dismissed from a GME program.
American Board of Medical Specialties records for member board certification activity for graduates in our database, including active and expired certification, were provided to the AAMC by Medical Marketing Services Inc, a licensed AMA Masterfile vendor, on behalf of the investigators through a data licensing agreement with the ABMS. In accordance with these ABMS records, we created a dichotomous variable for ABMS member board certification: having a record of certification by at least 1 of the 24 member boards vs having no record of certification by any board (reference group).30
We used χ2 tests to describe associations among categorical variables and analysis of variance to describe between-group differences in continuous variables. We report descriptive statistics for each independent variable and the dependent variable, ABMS member board certification, within each specialty category. We report crude and adjusted odds ratios and 95% CIs from separate multivariable logistic regression models for each specialty category to identify independent predictors of ABMS member board certification. All tests were performed with SPSS, version 17.0.3 (SPSS Inc, Chicago, Illinois); 2-sided P <.05 was considered significant.
Of all 57 437 graduates in the 1997-2000 graduating classes, 49 898 (86.9%) answered the Graduation Questionnaire item about plans to become board certified in a specialty, 47 035 of whom responded yes to this item. Of these 47 035 individuals, 46 757 (99.4%) chose a specialty on the Graduation Questionnaire, and 46 642 (99.2%) entered GME after graduation; 2098 of these 46 642 graduates (4.5%) changed specialties during GME and were excluded. We further excluded (because of small numbers) 956 graduates with multiple/unknown race/ethnicity reported, leaving 43 478 (75.7%) graduates eligible for inclusion in the analysis. Of those eligible, we included 42 440 (97.6%) graduates with data available for all items of interest on the Graduation Questionnaire, Step l, and Step 2 Clinical Knowledge results. Exclusions because of lack of all data of interest were greater for underrepresented minorities (220/5898 [3.7%]) than for Asian/Pacific Islanders (165/7441 [2.2%]) and whites (653/30 139 [2.2%]; P < .001). These exclusions were greater for graduates with Step l first-attempt failing scores (80/1698 [4.7%]) than for graduates with Step l first-attempt passing scores (949/41 771 [2.3%]; P < .001). They were also greater for graduates with Step 2 Clinical Knowledge first-attempt failing scores (57/1675 [3.4%]) than for graduates with Step 2 Clinical Knowledge first-attempt passing scores (964/41 786 [2.3%]; P = .004). The proportion of eligible graduates excluded did not differ significantly between board-certified graduates (902/37 956 [2.4%]) and non–board-certified graduates (136/5522 [2.5%]; P = .69) or between women (432/18 303 [2.4%]) and men (606/25 175 [2.4%]; P = .75).
Study sample characteristics grouped by board certification status are shown in Table 1, Table 2, Table 3, and Table 4 for each specialty category. Board certification rates and mean Step l and Step 2 Clinical Knowledge scores varied among specialty categories. Graduation year, race/ethnicity, age at graduation, US Medical Licensing Examination Step l and Step 2 Clinical Knowledge results, leave of absence during GME, and withdrawal/dismissal during GME were associated with board certification in all 8 specialty categories.
Tables 1-4 show results of the adjusted logistic regression models of variables associated with board certification for each specialty category. Results of both crude and adjusted models are shown in
the eTable In all 8 adjusted models, older graduates and graduates who had withdrawn/were dismissed from a GME program were less likely to become board certified. In 6 specialty categories (but not emergency medicine or radiology), graduates with first-attempt passing US Medical Licensing Examination Step l scores (vs first-attempt failing scores) were more likely to be board certified. In all specialty categories, graduates with first-attempt passing Step 2 Clinical Knowledge scores in the middle and upper tertiles were more likely to become board certified; in all but emergency medicine, graduates with first-attempt passing Step 2 Clinical Knowledge scores in the lowest tertile also were more likely to be board certified.
In the family medicine category, graduates with higher levels of debt were more likely to be board certified. However, in the obstetrics/gynecology category, graduates with higher levels of debt were less likely to be board certified. Compared with whites, underrepresented minorities in all specialty categories except family medicine were less likely to be board certified, as were Asians/Pacific Islanders in the surgery/surgical specialties category. Women in the obstetrics/gynecology, surgery/surgical specialties, and other nongeneralist specialties categories were less likely to be board certified.
Of the 5386 non–board-certified graduates, 3655 (67.9%) were actively licensed and had completed specialty GME, 678 (12.6%) were actively licensed but had not completed specialty GME, 628 (11.7%) were not actively licensed but had completed specialty GME, and 425 (7.9%) were not actively licensed and had not completed specialty GME.
Overall, 87.3% of our sample of 1997-2000 US medical school graduates were ABMS member board certified, similar to the 88% board certification rate in 2003 among of an earlier cohort of 1958-1994 graduates in selected specialties.32 Our study adds to this literature by identifying variables associated with board certification achievement among a national sample of US medical school graduates, with a composition reflective of the sex and racial/ethnic diversity of more contemporary US medical school graduates. Furthermore, to our knowledge the demographic and academic performance variables that we found to be associated with ABMS member board certification have not previously been examined among US medical school graduates in multivariable models. The study analyzed predictors of board certification separately for 8 specialty category groups. We identified differences across categories, which were expected, given differences in training duration, clinical experience, and oral examination requirements for board certification, as well as differences in written certifying examination first-attempt pass rates.31,33- 37
Each of 4 demographic variables was associated with board certification. Older graduates in each specialty category were less likely to be board certified, extending observations of 2 single-specialty studies.35,38 Older examinees were more likely to fail the internal medicine certifying examination,35 and residents who initially passed both qualifying and certifying American Board of Surgery examinations were younger than residents who initially failed.38 Our findings suggest that older graduates may experience greater difficulties, regardless of specialty choice, in timely advancement along the GME continuum toward board certification.
A 1997 study of US medical school graduates reported lower overall board-certification rates among women than men (67.0% vs 75.9%).39 We did not observe differences in the likelihood of board certification between the sexes in 5 of 8 specialty categories examined, suggesting that gender gaps in board certification may be narrowing among recent US medical school graduates in many specialties.39,40 We observed the gender gap in board certification rates among graduates choosing obstetrics/gynecology, which is currently the specialty with the largest proportion of physicians in training who are women (79%) among all specialties surveyed in the GME Census.24 A single-institutional study of 1964-1994 US medical school graduates practicing in obstetrics/gynecology in 2003 reported that sex was not a predictor of board certification.27 This finding suggests that, with longer follow-up of our cohort, the gender gap in board certification that we observed in the obstetrics/gynecology specialty category might diminish.
In every specialty category except family medicine, underrepresented minorities were less likely than whites to be board certified, as were Asian/Pacific Islander graduates in the surgery/surgical specialties category. These associations were observed in models that controlled for Step l and Step 2 Clinical Knowledge results and total debt, among other factors, raising concerns about ongoing efforts by US medical schools to increase the racial/ethnic diversity of the physician workforce, an issue of national concern.41,42 Because we observed these differences in a sample that included only graduates who reported specialty board certification intentions at graduation, there may be factors after graduation that disproportionately and negatively affect nonwhite—particularly underrepresented minority—medical school graduates' timely advancement along the postgraduation medical education continuum to board certification. Further research is warranted to identify factors after graduation that are associated with board certification and amenable to intervention so that these observed disparities in board certification can be eliminated.
Although there were differences in board certification rates on the basis of total debt among graduates in almost all specialty
categories (Tables 1-4), there was not a consistent relationship between higher debt and board certification among specialty categories. These mixed findings suggest that studies assessing possible relationships between debt and medical school graduates' career paths should control for specialty choices.
Both first-attempt Step 1 and Step 2 Clinical Knowledge passing scores were associated with greater likelihood of board certification, extending observations of earlier studies. Previous studies have been limited by inclusion of graduates in only 1 or a few specialties or graduates from a single institution, and some earlier studies pertained to performance on National Board of Medical Examiners Parts I and II rather than US Medical Licensing Examination Step l and Step 2 Clinical Knowledge.26,34 Recent studies have also been limited to examination of relationships between US Medical Licensing Examination Step scores and first-attempt performance on board-certifying examinations,38 not achievement of board certification itself. Such studies have reported differences in first-attempt results on the American Board of Surgery,38 the American Board of Orthopaedic Surgery,43 and the American Board of Pediatrics44 written examinations in relation to examinees' first-attempt Step l and Step 2 Clinical Knowledge scores.
Our study differs from these studies in that we analyzed Step l and Step 2 Clinical Knowledge results as categorical rather than continuous variables across several specialty categories, and we analyzed scores for associations with board certification, rather than certifying-examination scores. Nonetheless, we similarly demonstrated positive relationships between licensing examination results and board certification. Our findings provide support for program directors' use of first-attempt licensing examination results among criteria for evaluating applicants in a range of specialties.45
We identified 2 GME variables associated with a lower likelihood of board certification. Withdrawal/dismissal from a program during GME was associated with a markedly lower likelihood of board certification among graduates in all specialty categories, raising the possibility that, as a group, graduates who withdraw or are dismissed during GME may represent a particularly poorly performing group of graduates. We did not observe similar relationships across all specialty categories for graduates who took a leave of absence during GME, possibly because of the small number of graduates who took leave and their reasons for doing so (for which we lack information).
Most non–board-certified graduates in our study were actively licensed, indicating that they had satisfactorily completed at least 1 year of GME and ultimately passed US Medical Licensing Examination Step l, Step 2 Clinical Knowledge, and Step 3, all prerequisites for permanent state medical licensure,46 and many had completed specialty GME. However, we lacked information to determine which non–board-certified graduates in our sample might or might not have fulfilled all requirements to apply for ABMS member board certification by any member board. Indeed, this determination resides exclusively with member boards. Not every graduate who has completed a program of specialty GME of a specific duration has necessarily fulfilled requirements regarding the nature and scope of specific training experiences, and some ABMS member boards accept training credit for non–Accreditation Council for Graduate Medical Education–accredited residency training.30,31 Moreover, because ABMS member board certification is not required for physicians, lack of certification might reflect a physician's choice, depending on the physician's professional activities, not to proceed with the specialty board certification process even though the physician may have fulfilled all requirements to do so.
Strengths of the study included the use of both active and inactive ABMS records of board certification data rather than self-reported board certification data or data pertaining only to current board certification status. Another strength was the inclusion of a large national cohort of graduates with complete data for factors along the full extent of the medical education continuum.
Limitations include that, although the study used data about a nationally representative cohort of US medical school graduates, the observational design precludes making causal inferences from the findings. In addition, lack of board certification within the study's duration does not necessarily mean that a graduate will never become board certified; longer follow-up might show that some graduates become board certified, which may be especially true among graduates in those specialty categories with relatively lengthier GME requirements that also mandate clinical practice and oral examination requirements for board certification, such as obstetrics/gynecology. These results cannot be generalized to other groups of medical school graduates, such as graduates of osteopathic medical schools or international medical school graduates. Nevertheless, our findings can inform an understanding of factors contributing to US medical school graduates' advancement along the medical education continuum to board certification, an outcome of interest for medical school graduates, their patients, and the relevant professional organizations involved in undergraduate medical education, GME, and board certification.
Corresponding Author: Donna B. Jeffe, PhD, Division of Health Behavior Research, Washington University School of Medicine, 4444 Forest Park Ave, Ste 6700, St Louis, MO 63108 (email@example.com).
Author Contributions: Dr Jeffe had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Both authors contributed equally to this work.
Study concept and design; Acquisition of data; Analysis and interpretation of data ; Drafting of the manuscript ; and Critical revision of the manuscript for important intellectual content: Jeffe, Andriole.
Financial Disclosures: Both authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Drs Jeffe and Andriole report receiving travel funds from the National Institutes of Health (NIH) to attend various meetings, at which their research on medical education was presented. Dr Andriole reports receiving an honorarium and travel reimbursement from the University of Cincinnati for a lecture on MD-PhD programs and their graduates. Colleagues at the Association of American Medical Colleges (AAMC) and American Medical Association (AMA) did not receive compensation from the authors for their support, but data files provided by the AAMC and AMA were purchased with grant funds.
Funding/Support: Funding for the study was provided by the NIH National Institute of General Medical Sciences (grant R01 GM085350-03).
Role of the Sponsors: The National Institute of General Medical Sciences of the NIH was not involved in the design and conduct of the study; collection, management, analysis, and interpretation of the data; or preparation, review, or approval of the manuscript.
Previous Presentation: Presented in part at the 7th Annual AAMC Physician Workforce Research Conference, National Harbor, MD, May 5-6, 2011.
Disclaimer: The conclusions made by the authors are not necessarily those of the AAMC, the National Board of Medical Examiners, the NIH, the AMA, the American Board of Medical Specialties, or their respective staff members. The AMA is the source for the raw Physician Masterfile data; the statistics, tables, and tabulations of the data were prepared by the authors with the AMA Masterfile data. The board certification information presented herein is proprietary data maintained in a copyrighted database compilation owned by the American Board of Medical Specialties.
Additional Contributions: Data management and statistical services were provided by James Struthers, BA, and Yan Yan, MD, PhD (Washington University School of Medicine), who were supported in part by the NIH–Institute of General Medical Sciences. We thank Paul Jolly, PhD, Gwen Garrison, PhD, Jason Cantow, MS, MBA, Franc Slapar, MA (AAMC), and Robert M. Galbraith, MD, MBA (National Board of Medical Examiners), for their support of our research efforts through provision of data and assistance with coding, and Sarah Brotherton, PhD (AMA), for her support and assistance with coding, none of whom received financial compensation for his/her contributions.
This article was corrected for errors on September 23, 2011.