Objective
The purpose of this study was to examine retention rates of entry-level physician faculty members in obstetrics and gynecology.
Study Design
Ongoing data were collected by the Association of American Medical Colleges between 1981 and 2009 for full-time, entry-level assistant professors to determine whether they remained at their original departments, switched to another school, or left academia. Retention curves and 5- and 10-year retention rates at their original department and for academia were determined.
Results
The number of entry-level faculty members per year increased significantly for women and those faculty members in general obstetrics and gynecology. Retention rates at the original departments improved for all disciplines in recent years (2000-09), regardless of sex. Among those faculty members who left their original department, faculty members in general obstetrics/gynecology were more likely than subspecialists to leave academia.
Conclusion
Growth in the number of entry-level physician faculty members was accompanied by higher retention rates at their original departments only in recent years.
Development of faculty requires financial, personnel, and facility resources to meet departmental education, research, clinical care, and service objectives. The intent of academic leaders would be to retain promising faculty members for long-term department stability and productivity. An Analysis in Brief , which was published by the Association of American Medical Colleges (AAMC), indicated that, of clinical faculty members who began their appointments between 1981 and 1997, 52% remained at their original departments for 10 years; 10% switched to other medical schools, and 38% left academic medicine.
Autry et al examined the reasons that obstetrics/gynecology faculty members leave academic practice. From a sample of 5000 American College of Obstetricians and Gynecologists fellows, they surveyed 551 obstetricians/gynecologists in current faculty positions and 233 obstetricians/gynecologists who left academia. Faculty members rated the opportunities for teaching and administration to be important components of job satisfaction, although practitioners who left academics were more likely to be in general obstetrics/gynecology and were junior faculty members.
The primary objective of this investigation was to determine whether the retention of entry-level faculty members in obstetrics/gynecology declined, remained the same, or increased between academic years 1981 and 2009. We also examined whether long-term retention at their original department differed for faculty members with first-time appointments in obstetrics/gynecology and other clinical departments.
Materials and Methods
Selection of entry-level faculty members
This project was approved by the University of New Mexico Human Research and Review Committee (HRRC 08-442) and the American Institutes for Research (approval no. EX0008) of the AAMC Institutional Review Board. Data came from the AAMC’s Annual Faculty Roster, which is the only national database that contains demographic background and employment records of every full-time faculty member at all accredited US medical schools from academic year 1978-1979 onward.
Entry-level obstetrics/gynecology faculty members were those who were first-time assistant professors who worked at academic health centers, not community-based residencies, and who had had no previous appointments. We examined retention of female and male physician faculty members (MD, MD-PhD, DO) only, rather than PhD or other nonphysician faculty members. The faculty members were categorized into their respective discipline (general obstetrics/gynecology, gynecologic oncology, maternal-fetal medicine, reproductive endocrinology). Entry-level faculty members who had been trained in more recent years in urogynecology/pelvic support disorders were not recognized by the AAMC because of a lack of board certification and therefore were included in the general obstetrics/gynecology discipline group. Our analysis began with academic year 1980-1981, when the AAMC began to examine faculty member retention rates and ended with completion of the most recent academic year (2008-2009) of recorded information. Cohorts of entry-level faculty members were grouped by decade (1981-1989, 1990-1999, and 2000-2009), because there were sparse data in beginning years. The tenure for each faculty member was tracked from the beginning of his/her initial appointment until any departure date during their first 10 years.
Retention and attrition rates
We chose 5 and 10 years as primary endpoints to calculate proportions of entry-level faculty members who remained at their original medical school (remained), left for another school (switched), remained in academia (remained at original department or switched to another school), or left academia altogether (attrition). These periods are commonly used industry standards and are critical in a junior faculty member’s career development. An entry-level faculty member would be expected to become board certified in 5 years and usually to be promoted to the associate professor level with or without tenure by 10 years.
We also compared 10-year retention and attrition rates of entry-level faculty members in obstetrics and gynecology (all disciplines combined) with other clinical specialties. Data were provided from the AAMC for entry-level assistant professors in other clinical departments combined (eg, internal medicine, pediatrics, surgery, psychiatry, neurology, radiology, and anesthesiology). Appointments for these other clinical departments began during the same period.
Retention curves
Analysis of these retention data was done by survival analysis, because the data reflect the time until an event (either leaving original department or leaving academia) and there is considerable censoring (many faculty members do not experience 1 of these events). The events in this case were aggregated into relatively few years, hence discrete time survival analysis methods were required. We followed the method of Singer and Willett, in which hazard functions (probability of an event during a particular year) are estimated directly with the use of either logistic regression or complementary log-log regression. Results were nearly identical for the 2 methods, so we chose to report logistic regression results. Factors that fit in the analysis were sex, decade cohort, and discipline.
A quadratic form of the hazard function in the year of first appointment allowed for nonlinear hazard over time. Stepwise regression methods were used to reduce the large number of potential interactions from fitting 24 separate hazard functions. The fitted hazard functions were used to produce the estimated retention (survival) curves with the use of the same methods. Analysis was performed separately for retention in original department and for retention in academia.
Statistical analysis
All data were entered, and calculations were made, using SAS software (version 9.2; SAS Institute Inc, Cary, NC). χ 2 tests were used to compare group proportions for 5- and 10-year retention rates. Statistical significance was reached at a probability value level of < .05. Discrete time survival (or retention curve) analysis was a refinement of information in the 5- and 10-year retention rates, because it permitted a separate estimation of effects from disciplines, sex, and decade cohort. Because logistic regression was the basis of the analysis, odds ratios (ORs), rather than the familiar hazard ratio, served as the measure of effect. A 95% confidence limit was reported after each OR.
Results
Entry-level faculty
Responses were received from all US medical schools for each academic year from 1981-2009. A total of 5072 full-time physician faculty member became entry-level assistant professors in obstetrics/gynecology during this 29-year period. On average, newly recruited entry-level faculty members accounted for 6.9% of all obstetrics/gynecology physician faculty members annually.
As shown in Figure 1 , the number of entry-level assistant professors each year increased consistently, from 104 in 1981 to 228 in 2009. The number of entry-level male faculty members per year remained relatively constant (median, 86 per year). The number of entry-level female faculty members increased significantly ( P < .001) during this 29-year period (from 22 in 1981 to 152 in 2009).
