What is the Geographic Distribution of Women Orthopaedic Surgeons Throughout the United States?

Abstract

Background

Orthopaedic surgery has a shortage of women surgeons. An even geographic distribution of women orthopaedic surgeons may provide more uniform care to patients. However, little is known about the geographical distribution of women orthopaedic surgeons.

Questions/purposes

(1) Is there substantial geographic variation in the distribution of orthopaedic surgeons who are women? (2) How does the geographic distribution of women orthopaedic surgeons compare with that of other physicians? (3) What are the variables associated with increased region-based proportions of orthopaedic surgeons who are women?

Methods

To obtain a national snapshot of orthopaedic providers, two Medicare databases were used (Medicare Provider Utilization and Payment Data and Medicare’s current and archived Physician Compare Data). These databases were used to identify physicians with self-reported specialties of “Orthopedic Surgeon,” “Hand Surgeon,” or “Sports Medicine” with at least 11 Medicare claims in 1 year for a single procedure type between 2012 and 2014. These databases are the only databases known to specifically report surgeon gender on a national scale and include physician demographics and education. The Dartmouth Atlas’s hospital referral regions and United States Census Bureau divisions were used to group physicians by geographic region. The Gini coefficient, a measure of statistical dispersion, was used to quantify the regional distribution of orthopaedic surgeons. This was compared with the dispersion of non-orthopaedic physicians within the same Medicare databases. Surgeon and regional characteristics were correlated with the proportion of women orthopaedic surgeons in the region.

Results

There is substantial geographic variation in the distribution of orthopaedic surgeons who are women, ranging from 0% to 15%. There was a greater prevalence of women orthopaedic surgeons in New England (7.3%, 107 of 1469 surgeons) and the Pacific region (6.5%, 208 of 3196 surgeons) than in the South Atlantic (4.5%, 210 of 4618 surgeons) and East South Central regions (3.5%, 50 of 1442 surgeons). This represents a greater level of variation (Gini coefficient = 0.37) compared with other specialties (0.30 and 0.37) and compared with men orthopaedic surgeons (0.16). Variables independently associated with an increased prevalence of women orthopaedic surgeons based on hospital referral region were an increased proportion of currently practicing women physicians who graduated from medical schools in that region (beta = 0.03; p = 0.01), increased proportion of Medicaid-eligible patients (beta = 0.12; p = 0.002), increased proportion of regional population is black (beta = -0.06; p = 0.03), and increased regional supply of women physicians (beta = 0.26; p < 0.0001).

Conclusions

Despite the recent increase in women orthopaedic surgeons nationally, gains have not been equally distributed throughout the United States.

Clinical Relevance

In other medical fields, gender diversity has been proven to be beneficial for patients. If this holds true in the field of orthopaedic surgery, we should be mindful of the geographic distribution of women orthopaedic surgeons as the percentage of these surgeons increases.

Introduction

Background

The number of women pursuing careers in medicine has been increasing steadily. According to the Association of American Medical Colleges, nearly half of medical students in 2014 were women.[1] A similar increase, however, has not been observed in the number of women entering orthopaedic surgery, nor has the percentage of women in orthopaedics kept pace with the number of women entering other medical and surgical specialties [14]. In 2014, the percentages of women who were full-time academic faculty in medical or surgical specialties were 51% in pediatrics, 37% in internal medicine, 48% in family practice, 22% in surgery, and only 16% in orthopaedic surgery [1]. Women only comprised 5.3% of all active orthopaedic surgeons in 2017 [2]. Additionally, the increase in the percentage of women in orthopaedics during the past 30 years has been much lower than those in every other surgical specialty including general surgery, neurosurgery, ophthalmology, obstetrics and gynecology, otolaryngology, and urology [6].

Despite this increase, we know very little about where those women have gone to practice. Additionally, there are data to suggest that women and men practice medicine differently [3, 4, 5, 16–18, 25, 26, 30], and that some patients express preference for a physician of their own gender in the fields of primary care, gynecology, urology, and breast surgery [15, 20–22].

Rationale

Our recent analysis showed that there were important differences in the practice characteristics of women orthopaedic surgeons providing care to Medicare patients [10]. Considering some patients’ demonstrated preference for a provider of a specific gender when choosing a physician, potential differences in practice patterns, and the fact that women continue to make up an increased percentage of the active orthopaedic surgery workforce, access to women orthopaedic surgeons seems to be a pertinent issue and may have implications for patient care. The concentrated distribution of women orthopaedic residents and surgeons in the United States warrants further investigation. It is important to understand the current geographic location of women orthopaedic surgeon practices to identify potential shortcomings.

Study Questions

Therefore, we asked: (1) Is there substantial geographic variation in the distribution of orthopaedic surgeons who are women? (2) How does the geographic distribution of women orthopaedic surgeons compare with that of other physicians? (3) What are the variables associated with increased region-based proportions of orthopaedic surgeons who are women?

Materials and Methods

Study Design and Setting

To evaluate the geographic distribution of women orthopaedic surgeons, we queried publicly available Medicare datasets to identify practicing physicians. To obtain a national snapshot of orthopaedic providers, two Medicare databases were used (Medicare Provider Utilization and Payment Data and Medicare’s current and archived Physician Compare Data). These databases were used to identify physicians with self-reported specialties of “Orthopedic Surgeon,” “Hand Surgeon,” or “Sports Medicine” with at least 11 Medicare claims in 1 year for a single procedure type between 2012 and 2014. These databases are the only databases known to specifically report surgeon gender on a national scale and included physician demographics and education.

Participants

To evaluate the geographic distribution of women orthopaedic surgeons, we queried publicly available Medicare datasets to identify practicing physicians. To formulate a national cohort of orthopaedic surgeons, we gathered Medicare Provider Utilization and Payment Data for available calendar years (2012, 2013, and 2014) [8]. This dataset includes any provider billing Medicare for greater than 10 claims of at least one Current Procedural Terminology code in a calendar year. A provider’s unique national provider identifier, gender, name, self-reported credentials, self-reported specialty, and billing ZIP code are provided in this dataset. Surgeons with self-reported specialties of orthopaedic surgery, hand surgery, or sports medicine were selected using specialized Medicare tags. Separately, Medicare’s current and archived (2014-2016) Physician Compare datasets were queried to provide more detailed information on the orthopaedic cohort and to identify physicians of all specialties [11]. Physicians were included in this dataset if they had submitted a Medicare claim within the previous 12 months, were enrolled in the Medicare Provider Enrollment, Chain, and Ownership System (Medicare’s enrollment system), and listed a specialty in the Provider Enrollment, Chain, and Ownership System. The Physician Compare database contains demographic data (gender, practice location, name of medical school, and year of medical school graduation), self-reported credentials, self-reported specialty, group practice identifier, and number of practice members. The previously described orthopaedic billing dataset was linked by the national provider identifier to the Physician Compare dataset with 95.7% (22,961 of 23,992 entries) concordance. To limit the non-orthopaedic cohort to physicians of interest, all entries were queried for primary specialty, and all non-physician specialties (such as certified registered nurse anesthetist) were eliminated. Women orthopaedic surgeons accounted for 5.3% (1269 of 23,994) of practicing orthopaedic surgeons in this cohort. Conversely, 30% (193,271 of 441,652) of the practicing physicians in all specialties were women (p < 0.001).

To identify the characteristics of medical schools graduating today’s practicing physicians, we aggregated all medical schools listed in the Physician Compare database. All medical schools with at least 250 graduates still practicing were included in this analysis. An internet-based search was performed for each of these schools to identify their ZIP code. The ZIP code was then used to determine the hospital referral region and geographic coordinates of each school.

Variables, Outcome Measures, Data Sources, and Bias

Hospital referral regions, as defined by the Dartmouth Atlas, were used to group physicians by location [12]. ZIP codes of the practice locations of all available physicians were used to identify their current hospital referral region from the ZIP code crosswalk files [13]. For a more general assessment of geographic variance, we grouped physicians by region and divisions as defined by the United States Census Bureau [31].

Variation was first assessed by comparing the proportion of orthopaedic surgeons who were women across hospital referral regions. To address potentially skewed results because of low orthopaedic surgeon supply in small-population hospital referral regions, we performed a second assessment comparing the gender distributions for only hospital referral regions with greater than 50 orthopaedic surgeons. Then, we compared the gender distributions of practicing orthopaedic surgeons throughout the United States Census Bureau regions and divisions for a more global assessment of variation.

To quantify variations in the supply of physicians, we calculated a Gini coefficient for women orthopaedic surgeons, men orthopaedic surgeons, all women physicians, and all men physicians. A Gini coefficient represents the uniformity of the distribution of a limited resource in a limited population—a coefficient of 0 represents perfect equality, and a coefficient closer to 1 suggests that a high proportion of the resource is distributed to a low proportion of the population.[23] For example, in a large group where only one person has all the income, and all others have none, the Gini coefficient will be very nearly one. This coefficient is calculated based on the Lorenz curve [24], which is a plot of the cumulative proportion of the specific physician groups (for example, women orthopaedic surgeons) against the cumulative proportion of the total number of Medicare beneficiaries in 2014 as segmented by hospital referral region in this instance.

To assess a potential source of physician gender-dependent geographic variation, we investigated the distribution of medical school graduates by gender and their chosen specialty. We assessed the variation of medical school graduates in a similar manner as the above-described assessment of practicing orthopaedic surgeons using the Lorenz curve and Gini coefficient segmented by medical school.

To identify factors associated with an increased proportion of women orthopaedic surgeons, we investigated the relationship between the regional proportion of women orthopaedic surgeons and hospital referral region-specific variables. We gathered the number of Medicare patients within the region (fee-for-service patients) in addition to average demographics (age and race), overall health and the proportion of the population with arthritis, and socioeconomic variables for each hospital referral region using the Centers for Medicare & Medicaid Services public use file [7]. This database reports the mean hierarchical condition category for each hospital referral region as a representative of the overall health of Medicare patients. The economic status of Medicare patients in each hospital referral region was represented by the proportion of the Medicare population eligible for Medicaid. To represent access to and use of general medical care, we gathered the proportion of regional Medicare patients visiting their primary care physician annually from the Dartmouth Atlas [13]. We used QGIS (version 2.8.9, www.qgis.org; Open Source Geospatial Foundation [OSGeo, Beaverton, OR, USA]), a mapping software, to identify neighboring hospital referral regions and the area of each hospital referral region. The following variables were assessed: total medical schools, the proportion of medical school graduates who were women; orthopaedic specialty hospitals, the state-wide proportion of Democratic voters for President in the 2016 election, total physicians per 1000 Medicare patients, and proportion of all physicians who were women.

Statistical Analysis and Study Size

To assess the gender distribution of orthopaedic surgeons nationally, we first used descriptive statistics to report differences in gender proportions by region. As described above, we used the Gini coefficient to quantify unequal geographic distribution of men and women orthopaedic surgeons, all men and women physicians, and men and women medical school graduates. To assess for an association between potential predictive variables and the regional proportion of orthopaedic surgeons who were women, we performed an initial analysis with Pearson’s correlation coefficient. We confirmed the data normality of potential predictors, with skewness of less than 2 and kurtosis of less than 12. A single variable (orthopaedic discharge prevalence) was log-transformed for non-normal data. To assess the independence of predictive variables, we performed a multivariate linear regression analysis including those variables that trended toward statistical significance (p < 0.1) in univariate analysis. For multivariate analysis, coefficients were standardized based upon the variable’s SD. The statistical analysis was performed with RStudio (Version 1.0.136; RStudio; Boston, MA, USA).

Results

Geographic Variation in Distribution of Orthopaedic Surgeons Who Are Women

There was regional variation in the proportion of orthopaedic surgeons who were women. Higher percentages of women orthopaedic surgeons and all physicians were present in New England, West North Central, and Pacific regions (p < 0.001) (Fig. 1). Each hospital referral region had at least five orthopaedic surgeons (mean 78; range 5 to 555) with an average of 2037 physicians (range 171 to 18,590). The average numbers of women orthopaedic surgeons and physicians per hospital referral region were 4.1 (range 0 to 42) and 618.3 (range 24 to 6313), respectively, and 80% (244 of 306) of hospital referral regions had at least one woman orthopaedic surgeon. A total of 28,092,841 of 31,014,226 Medicare patients (91%) had access to a woman orthopaedic surgeon in their hospital referral region. Of the 151 hospital referral regions with at least 50 surgeons, the proportion of orthopaedic surgeons who were women ranged from 0% to 15.1% (mean 5.3%) (Fig. 2). Of the 72 hospital referral regions with at least 100 orthopaedic surgeons, women comprised 5.4% of them (range 1.8% to 15.1%).

Fig. 1.

This map represents the geographic divisions reported by the United States Census Bureau. Within each division, the proportion of orthopaedic surgeons and all physicians are female is reported.

Fig. 2.

This map represents the proportion of currently practicing orthopaedic surgeons who are women based on the hospital referral region of practice location.

We identified 152 medical schools with at least 250 currently practicing physicians distributed over 101 hospital referral regions. These 152 medical schools were identified as the school of graduation for 19,387 currently practicing orthopaedic surgeons and 422,116 physicians. Again, we found geographic variation in the proportion of graduates who were women, ranging from 0% to 19% and 19% to 43% for currently practicing orthopaedic surgeons and all physicians, respectively (Fig. 3).

Fig. 3.

This map represents the proportion of medical school graduates who are women who now practice orthopaedic surgery based on the hospital referral region of graduation location.

Comparing Geographic Distribution of Women Orthopaedic Surgeons With Other Physicians

Orthopaedic surgeons who are women were distributed more unequally (Gini coefficient = 0.37) than orthopaedic surgeons who were men (0.16), than all women physicians (0.30), or than all physicians who were men (0.19) (Fig. 4A). However, this is an improvement from the geographic disparities of women who graduated in 2000 or earlier (Gini coefficient = 0.45), with no change for men orthopaedic surgeons (0.16). Another way to state this disparity: the 20% of hospital referral regions with the highest proportion of women orthopaedic surgeons contained 45% of all women orthopaedic surgeons. This was compared with 29%, 34%, and 39% for men orthopaedic surgeons, all men physicians, and all women physicians, respectively (p < 0.001). Among currently practicing physicians in all specialties, the medical school from which they graduated was nearly equally distributed for men and women (Gini coefficient, 0.04 and 0.10, respectively; Fig. 4B). However, graduates ultimately entering orthopaedics and currently in practice had more unequal distributions amongst medical schools, with greater inequality among women graduates (Gini = 0.28) than among men (Gini = 0.15).

Fig. 4 A-B.

This Lorenz curve represents the equality of resource distribution among (A) men and women orthopaedic surgeons and all physicians who are men or women by hospital referral region and (B) all physicians who are men and women and orthopaedic surgeons by medical school.

Factors Associated with Geographic Variation

After controlling for potential confounders, we found that social and economic regional characteristics explained 43% of the regional variation practice location for women orthopaedic surgeons (r² = 0.43; p < 0.0001). The specific variables independently associated with an increased proportion of women orthopaedic surgeons based on hospital referral region were an increased proportion of Medicaid-eligible patients (beta = 0.25; p = 0.005), an increased proportion of currently practicing women physicians who graduated from medical schools in that region (beta = 0.16; p = 0.03), an increased proportion of women orthopaedic surgeons in neighboring hospital referral regions (beta = 0.16; p = 0.03), and an increased regional supply of women physicians (beta = 0.37; p = 0.0004) (Table 1).

Discussion

A large gender gap in the orthopaedic surgeon workforce persists, despite the increase in the percentage of orthopaedic surgeons who are women over the past 40 years. Even with this increase, we know very little about where those women have gone to practice. There is data to suggest that women and men practice medicine differently [3, 4, 5, 16–18, 25, 26, 30], and some patients express preference for a physician of their own gender [15, 20–22]. Therefore, it was important to define the geographic distribution of women orthopaedic surgeons, as there may be implications for access to patient care.

An increased regional supply of women orthopaedists may provide care to patients who prefer a physician of a specific gender, promote diversity within the specialty, and ensure role models for women trainees in all geographic areas. However, we found that there was substantial geographic variation in the distribution of orthopaedic surgeons who were women, which was greater than the variation seen in orthopaedic surgeons who were men and physicians of all specialties of both genders. Additionally, we found that regions with the highest proportion of orthopaedic surgeons who were women were associated with an increased proportion of currently practicing women physicians who graduated from medical schools in that region, increased proportion of Medicaid-eligible patients, increased proportion of regional population who was black, and increased regional supply of women physicians.

Limitations

In interpreting these findings, it is important to be aware of this analysis’s limitations. First, the data were limited by the parameters of the Medicare database, which required physicians to have at least 11 Medicare billing encounters of the same procedure code (including office visits) between 2012 and 2014 before they were recorded in the database. This excluded surgeons who do not bill to Medicare, as well as physicians who did not bill enough of a single procedure. Relying on data from Medicare may also skew the type of patients and procedures recorded, potentially underreporting pediatric and sports medicine specialists. It is particularly important to account for the orthopaedic pediatric specialty, as it is the third most common subspecialty for women orthopaedic surgeons [28]. As of 2017, of the 1401 total registered members of the Pediatric Orthopaedic Society of North America, 296 were women [9]. Because of this shortcoming, we cannot reach any conclusions regarding the distribution of pediatric orthopaedic surgeons who are women.

To identify local demand for orthopaedic care, we collected the percentage of Medicare patients with arthritis from the Medicare dataset. However, these data do not specify arthritis location, quality, or type, and therefore may not accurately represent surgical demand. However, arthritis prevalence does provide an accurate representation of demand for orthopaedic surgeons, as orthopaedic surgeons are often involved in the longitudinal care of arthritic conditions with or without surgical intervention. Additionally, because of the nature of the data in the database, we were unable to subdivide surgeons to determine specific subspecialties within orthopaedic surgery other than hand surgery and sports medicine. Within hand surgery, we were unable to differentiate whether these surgeons were orthopaedic surgery or plastic surgery trained. However, despite the incomplete granularity of the data, it is still important to consider the findings as whole. All surgeons included in the analysis played a role in musculoskeletal care in the Medicare population and evaluating the distribution of women orthopaedic surgeons within that subset provides valuable insight.

Geographic Variation in the Distribution of Orthopaedic Surgeons who are Women

There was a higher proportion of women orthopaedic surgeons in New England and the Pacific regions than in the South Atlantic and East South Central regions. A previous study evaluated the discrepancies between urban and rural orthopaedic surgeons and found that from 1995 to 2010, there were fewer orthopaedic surgeons in rural areas than in urban areas (6.52 versus 8.73 per 100,000 surgeons in 2010) [19]. They also found that the urban cohort was younger than the rural cohort. However, the authors did not examine gender- and region-based distribution, but because women orthopaedic surgeons form a younger cohort of surgeons, it is reasonable to assume that they are more likely to be distributed in urban areas. Another study evaluated the general surgery workforce throughout the United States. In urban areas, the lowest number of surgeons per population of 100,000 was found in the Pacific (5.41), West South Central (6.11), Mountain (6.17), and East North Central (6.22) census regions. In large, rural areas, the West South Central (6.20) and Pacific (6.82) regions again had the lowest number of general surgeons relative to the population. Additionally, the authors found that women general surgeons were more likely to practice in urban areas (11.7%) than in large rural areas (6.1%) or small rural areas (7.3%) [27].

We found similar discrepancies in regional distribution for medical schools from which women orthopaedic surgeons had graduated. One prior study published a similar finding regarding which medical schools produce physician who practice in rural areas [29]. Medical schools varied greatly in the percentage of graduates who entered rural practice, ranging from 2.3% to 41.2% of the graduating classes studied. Additionally, 12 medical schools accounted for greater than one-fourth of the physicians entering rural practice. Variables associated with graduation of rural graduates included location in a rural state, public ownership, graduation of family physicians, and less funding from the National Institutes of Health. Similarly, we found that women orthopaedic surgeons who attended medical school in a given region were more likely to remain in practice in that hospital referral region. The same regional characteristics that attracted women orthopaedic surgeons to their medical school may retain them in the same region for their practice.

Comparing Geographic Distribution of Women Orthopaedic Surgeons With Other Physicians

This study revealed that there was greater geographic disparity for women orthopaedic surgeons than for men orthopaedic surgeons and women physicians in other specialties. Although only 80% of hospital referral regions had at least one woman orthopaedic surgeon, 90% of Medicare patients had access to a woman orthopaedic surgeon, a marker of disparate distribution. Previous studies have used the Gini index to examine the distribution of surgical subspecialties throughout the country compared with primary care providers [27]. They found that among surgical subspecialists, the Gini index was much higher than that in orthopaedics, indicating a more uneven geographic distribution. In 2008, the Gini index was 0.49 for neurosurgery and 0.37 for ophthalmology; orthopaedic surgery was close behind at 0.34 and primary care providers had relatively even distribution at about 0.24. However, to our knowledge, no prior studies have evaluated gender-specific discrepancy in the distribution of orthopaedic surgeons throughout the United States. It is important to examine these distribution trends longitudinally in orthopaedic surgery and other subspecialties to identify shortcomings that may inform beneficial change in physician geographic distribution to ensure equal access to health care.

Factors Associated with Geographic Variation

We identified multiple factors that were associated with an increased proportion of women orthopaedic surgeons in a given region, including percent of patients eligible for Medicaid, proportion of medical school graduates who were women, mean proportion of women orthopaedic surgeons in neighboring hospital referral regions, total physicians per 1000 Medicare patients, and proportion of all physicians who were women. The draw of a given geographic region did not seem limited to women practicing orthopaedic surgery, rather, it was generally more attractive for women physicians of all specialties. Because regions with a higher proportion of medical school graduates who were women also had higher proportions of women physicians and orthopaedic surgeons, it seems that something innate to the region attracts women to train and ultimately practice there.

This uneven distribution has also been shown at the level of residency training programs. A prior study found that the proportion of women orthopaedic surgery residents at different training programs is not uniform throughout the country. Between 5% and 20% of the residency programs in the United States have trained most women orthopaedic surgeons recently, whereas approximately one-third of the programs have trained few or no women [32]. The regions with residency programs with a higher proportion of women residents had, on average, a greater proportion of practicing women orthopaedic surgeons (5.2%) than regions with residency programs with lower proportions of women residents (3.9%). It is possible that regional culture contributes to the gender-based variation in distribution throughout medical school, residency, and practicing physicians.

Conclusions

Despite the limitations of this analysis, this study shows that there is increased geographic variation among women orthopaedic surgeons compared with men orthopaedic surgeons and compared with other female physicians. Higher percentages of women orthopaedic surgeons and all physicians were present in New England, West North Central, and Pacific regions, and factors associated with the increased proportion of women orthopaedic surgeons in given regions included percent of patients eligible for Medicaid, proportion of medical school graduates who were women, mean proportion of women orthopaedic surgeons in neighboring hospital referral regions, total physicians per 1000 Medicare patients, and proportion of all physicians who were women. Additional studies are needed to help define characteristics of specific regions that may be more attractive for women medical students, orthopaedic residents, and orthopaedic surgeons. From this analysis, it appears that regions successfully recruiting women into medical school have been more successful in retaining them for orthopaedic practice. As the proportion of women orthopaedic surgeons continues to grow, it will be important to understand the factors that drive a woman orthopaedic surgeon’s decision with regards to practice location to ensure continued growth and access to care for patients.