Benedetti, Thomas J. MD, MHA; Baldwin, Laura-Mae MD, MPH; Andrilla, C Holly A. MS; Hart, L Gary PhD
The gender composition of the obstetrics and gynecology workforce in the United States has undergone rapid transformation in the last 20 years. The percentage of women in the workforce has increased from 12% in 1980 to 32% in 2000. Because more than 70% of entering obstetrics and gynecology residents are women, this percentage is projected to increase to 50% by 2014.1
Concerns have been voiced regarding the effect of a predominately female obstetrician–gynecologist workforce on practice productivity. These opinions are partly based on the belief that female obstetrician–gynecologists practice fewer hours than male obstetrician–gynecologists, are more likely to interrupt their careers for childbearing and childrearing activities, and are more likely to prematurely terminate full-time practice (The obstetrician–gynecologist workforce in the 21st century. ACOG Today 2000;March:12–3).1,2 The few studies that have examined gender differences in practice productivity tend to support these anecdotal concerns.3–5
Current obstetrics and gynecology training program enrollment levels were developed by using productivity estimates based on the practices of a predominately male workforce. Substantial differences in the productivity of male and female obstetrician–gynecologists could have important implications for the size and number of obstetrics and gynecology graduate medical education training programs nationally. This study expands the literature on gender-based practice differences by comparing practice patterns and productivity of all identifiable male and female obstetrician–gynecologists in Washington State by using data from a 1998–1999 licensure survey and the state's vital records.
MATERIALS AND METHODS
This descriptive study was approved by the University of Washington Human Subjects Division. The Washington State licensure database on health care providers licensed between mid-May 1998 and mid-May 1999 served as the core database for this study. A supplementary survey with questions on practice specialty, locations, and volume was sent with each provider's license renewal and then linked with the licensure data. We linked the 1998 American Medical Association (AMA) Masterfile to the licensure data to supply additional information on physician specialty. Last, we linked the 1999 Washington State birth-certificate data with licensure data to identify those physicians practicing obstetrics and their annual number of deliveries. If no 1999 birth certificate linked with the licensure database, we linked the 1998 birth certificate data to ensure ascertainment of obstetrical practice and volume status.
Washington State's obstetrician–gynecologists were defined as those doctors of medicine and doctors of obstetrics who 1) listed a practice location ZIP code in Washington State and 2) listed gynecology or obstetrics and gynecology as their specialty area on the survey, or educational, administrative, or public health as their primary specialty area and gynecology or obstetrics and gynecology as their secondary specialty area. If these survey data were unavailable, any physicians who reported maternal-fetal medicine, reproductive endocrinology, obstetrics, gynecology, gynecologic oncology, or obstetrics and gynecology as their primary specialty on the AMA Masterfile were also included. We identified 587 obstetrician–gynecologists in Washington State. Forty-six of these obstetrician–gynecologists did not complete the licensure-renewal survey, which contained critical information on practice productivity, and thus were excluded from the study. The final study population included 92% (541) of all obstetrician–gynecologists identified in Washington State.
Gender was the primary variable of interest in this study. Demographic and practice characteristics available from the study databases included age (reported as mean age and by decades, eg, 30–39 years, 40–49 years), race, Latino/Hispanic ethnicity, primary work location (eg, office, hospital), and geographic practice location (ie, rural, urban). The Washington State Department of Health has created 124 Health Service Areas throughout Washington State. We defined rural practices as those with practice ZIP codes located in 1 of 52 rural Health Service Areas of Washington State.
The licensure survey provided 5 practice productivity measures: weeks worked in the past year, total professional hours excluding on-call time (eg, direct patient care, meetings, continuing medical education, other professional activities) per week, direct patient care hours per week at the two main work locations, outpatient visits per week, and inpatient visits per week. We developed 3 additional variables from these measures: nondirect patient care hours (total professional hours minus direct patient care hours), less than 32 total professional hours per week, and 60 or more total professional hours per week. We used birth certificate data to develop variables identifying whether each obstetrician–gynecologist was practicing obstetrics during the study period and the number of each obstetrically active provider's deliveries in either 1998 or 1999. An obstetrician–gynecologist was considered obstetrically active if he or she attended more than 5 deliveries in the study year. Practice productivity was evaluated by using these 10 variables.
We compared the demographic and practice characteristics and the practice productivity measures between the male and female obstetrician–gynecologists using standard t tests for continuous variables and χ2 tests for categorical variables. We calculated female-to-male productivity ratios for each of our study variables. Because of significant differences in the age distribution between male and female obstetrician–gynecologists, we used analysis of covariance with the Scheffé test to adjust for gender differences in the continuous practice productivity measures, adjusting for age and examining whether the effect of gender on practice productivity varied by age. Multiple logistic regression was used to test for gender differences in the categorical practice productivity variables (proportion practicing obstetrics, working less than 32 hours per week or working 60 or more hours per week).
Of the 541 obstetrician–gynecologists in this study, two thirds (366) were men and one third (175) were women (Table 1). Women were significantly younger than men. There were no significant differences between these men and women in race, ethnicity, primary work location, and urban-rural practice location.
Overall, the crude practice productivity rates and the distribution of obstetric deliveries were similar between male and female obstetrician–gynecologists (Table 2, Figure 1). The only significant differences in practice productivity between women and men were lower mean inpatient visits per week for women (10.1 versus 12.8, P ≤ .01) and a lower proportion of women working 60 or more hours per week (22.1% versus 31.5%, P ≤ .05). After controlling for age, the analysis of covariance (not shown) confirmed these findings and, in addition, showed that women worked an average of 4.1 fewer total professional hours than men (P < .01). The multiple logistic regression analysis demonstrated that the odds of women working 60 or more hours per week was 0.48 that of men after controlling for age (P < .01). There were no statistically significant differences in direct or nondirect patient care hours or any other productivity measures between men and women in the adjusted analyses.
Because of the dramatic age differences between female and male obstetrician–gynecologists, we performed a stratified analysis by 10-year age groups to determine whether there were differences within age groups that could not be observed in the overall results (Figure 2). In the 40–49-year age group, we observed the lowest productivity ratio in women compared with men for several of the study variables: total professional hours, direct patient care hours, nondirect patient care hours, percentage practicing obstetrics, average number of deliveries per year, working more than 60 hours per week, and working less than 32 hours per week (Table 3). In general, these differences were on the order of 10–20% lower productivity for the women aged 40–49 years compared with men. The analysis of covariance, however, did not reach statistical significance for these measures, as our study's sample was large enough to detect differences of 20–40%, but not smaller.
Not shown in Table 2 are the data for obstetrician–gynecologists age 60 years and older. Although there were 75 men in this category, there were only 2 women. There were not enough women for a meaningful comparison. It should be noted that 48% of men in the 60-and-older category worked less than 32 hours per week.
In this study, we found differences in clinical productivity between female and male obstetrician–gynecologists in only 3 of the 10 practice productivity variables we examined: total professional hours worked per week, inpatient visits per week, and proportion of providers working 60 or more hours per week. We did not find gender differences in the number of outpatient visits per week, direct or nondirect patient care hours worked per week, the number of births attended per year, or the proportion of providers working less than 32 hours per week. In addition, we found no significant gender difference in the percentage of licensed practitioners who attended births in the 3 physician age groups studied.
Our study's finding of minimal differences in practice productivity between female and male obstetrician–gynecologists in practice productivity differs from that of Pearse et al,3 who estimated a 15% discount in productivity for female compared with male obstetrician–gynecologists. Pearse et al3 used data from 2 national survey sources: 1998 AMA Socioeconomic Survey (331 respondents) and a 1998 survey of The American College of Obstetricians and Gynecologists (ACOG) fellows (1,230 respondents). In that study, female-to-male productivity ratios in the provision of obstetric care from the ACOG fellows and the AMA Socioeconomic Survey suggest that female obstetrician–gynecologists attended fewer deliveries in all 3 age groups, but the AMA survey's ratios were much lower than those from ACOG fellow's survey. Our study's stratified results were closest to those from the ACOG fellows’ survey, yet our adjusted analysis demonstrated no statistically significant difference in the number of births attended.
The most striking difference in practice behavior between male and female obstetrician–gynecologists was found in their proportions working 60 or more hours per week. Men were more likely than women to work 60 or more hours per week in all age 3 age groups. The difference was especially noteworthy in the 40–49-year age group. In this age group, 46% of male obstetrician–gynecologists worked more than 60 hours compared with 22% of women. Whether this is representative of the particular cohort of 40–49-year-olds in this study or indicative of a career pattern for male and female obstetrician–gynecologists cannot be distinguished in this cross-sectional study. Nonetheless, the observation that a lower proportion of women choose to work more than 60 hours per week, especially during the prime years of a medical career, may explain the questions that have been raised about women's productivity in the field of obstetrics and gynecology.
There were several differences in the practice patterns of male and female obstetrician–gynecologists that did not reach statistical significance but that may be of clinical significance. Our sample size was insufficient to detect a difference of 10–20% in practice productivity between men and women among the individual 10-year age cohorts. Pearse et al3 estimated the overall gender difference to be of this order of magnitude but did not perform any statistical test to confirm this finding. We noticed that for several of our study measures men, but not women, in the 40–49 year age group appeared to increase their practice productivity when compared with the 30–39-year age group. This was apparent for the categories total professional hours, inpatient visits per week, percentage practicing obstetrics, percentage working less than 30 hours per week, and percentage working 60 hours or more per week. These differences were narrowed or reversed when compared with the 50–59-year age group for all but the inpatient visits measure.
A closer look at the total professional hours worked per week in the 40–49-year age group shows the difference between men and women to be due primarily to the decrease in nondirect patient care hours for the women. For women, the direct patient care hours ratio was reduced by only 4%, but the nondirect patient care hours ratio was reduced by 16%. Nondirect patient care hours include these professional activities such as administration, continuing medical education, and professional leadership activities (hospital committee work, regional and national professional committee work, and speaking/lecturing).
Our data show that men in the 40–49-year age group increase their practice activities compared with the decade earlier. Women do not appear to follow this pattern. They maintain their prior levels of direct patient care but do not engage in as many other professional activities as men. One hypothesis to explain this observation is that childcare needs may reach their peak among families in which women have delayed child bearing until the fourth decade of life. In these families, female obstetrician–gynecologists may be caring for their families rather than taking on extra professional activities. If a significant portion of nondirect patient care activities involves leadership activities, the field of obstetrics and gynecology may be missing the important leadership contributions of its female members. A more specific look at this issue of the content of male and female obstetrician–gynecologists’ careers over time is an important future task for the field of obstetrics and gynecology.
This study has both limitations and strengths compared with others in the literature. First, because this study's findings are from Washington State only, we cannot assert that they are generalizable to the remainder of the United States. Second, most of the variables in our database were self-reported and therefore subject to unverifiable underreporting and overreporting. However, there is no reason to suspect that men and women would have different self-reporting biases. In addition, the self-reported practice survey data were consistent with the objective obstetric delivery data with regard to practice patterns and ratios. Third, this study was able to provide information on the productivity at a given point in time only. Whether the study practitioners in their 30s will behave in a manner similar to those in their 40s, 50s, or 60s when they reach these ages is an unanswered question. Whether age or gender is the more important influence on practitioners’ behavior has never been evaluated. Increasing numbers of women in obstetrics and gynecology will provide an opportunity to answer this question. The answer may be different for different specialties.
This study's strength is its nearly complete population of obstetrician–gynecologists for an entire state. More than 92% of practicing obstetrician–gynecologists in Washington State completed the licensing survey during the period of interest. In addition, we were able to count the actual number of births attended by individual practitioners by linking the licensure-survey data with state birth-certificate data. This likely produced a more accurate volume of deliveries than by self-reporting, as in the ACOG and AMA data.
Many variables influence whether the size of the obstetrician–gynecologist workforce will meet the population's needs. Predicting demand for childbirth and gynecologic services is beyond the scope of this article. Factors influencing the supply of obstetrics and gynecology providers include the number graduating from residency, the productivity of those practitioners, and the total duration of practice life of the practitioners. In 1990, Kletke et al6 observed that even though female physicians had higher early retirement rates than male physicians, as a group their expected work lives were just as long because of their lower early-mortality rates. In that study they also predicted the work life of obstetrician–gynecologists to be 23 years after age 50. Currently, this projection of working into the eighth decade seems unrealistic in light of external influences in obstetrician–gynecologist practice, such as increasing administrative demands from managed care companies, professional liability concerns, and changing practice structures. The satisfaction of obstetric and gynecologic practice in the United States is one of the lowest of all medical or surgical specialties.7 These external influences and other both positive and negative factors could affect the workforce projections more substantially than the small gender-based differences we observed in our study.
This study suggests that overall clinical productivity of female obstetrician–gynecologists in the late 1990s in Washington State was not dramatically different from their male counterparts. Although we have data on more than 90% of obstetrician–gynecologists in our state, apparent differences between age cohorts that may be clinically significant could not be substantiated because of our sample size. Projecting these findings into the future is even more difficult as our cross-sectional data represent a single state during a 1-year period. An ongoing longitudinal study of age- and gender-specific practice productivity patterns is critical to provide accurate data regarding the behavior and capacity of the obstetrician–gynecologist workforce in the United States.