Journal Logo

EQUITY AND SOCIAL JUSTICE

Underemployment of Female Surgeons?

Chen, Ya-Wen MD, MPH; Westfal, Maggie L. MD, MPH; Chang, David C. PhD, MPH, MBA; Kelleher, Cassandra M. MD

Author Information
doi: 10.1097/SLA.0000000000004497
  • Free

Abstract

Most sex disparity research has focused on implicit bias,1 sexual harassment in the workplace,2 or sex differences in family obligation as drivers of professional inequity amongst physicians.3,4 However, what has not been determined is a potential relationship between these factors and underemployment among female surgeons. Underemployment is defined as being employed at less than full-time or working at jobs inadequate with respect to one's training.5 Underemployment is a significant and persistent economic problem in our society, as documented in regular reports from the Federal Reserve Bank.6 Despite the perception that physicians are highly valued professionals with virtually no unemployment, recent employment research has found that underemployment exists among physicians and that is more common among women in the general workforce.7,8

Our prior research suggests that underemployment may be significant among female surgeons relative to male surgeons.9 Specifically, female surgeons perform a fewer total number of operative cases and perform fewer cases that require their subspecialty training than their male peers. Anecdotal observations also suggest that an additional measurable difference may exist, in that female surgeons perform less complex operations (determined by work relative value unit, wRVU) than their male peers, despite having equivalent training and seniority. Such differences in practice patterns would further represent the underemployment of female surgeons and have negative repercussions on surgeons’ psychological health, career trajectories, job turnover, individual and team performance,10 and healthcare quality.11–13 The goal of this study was to validate these anecdotal observations by analyzing the complexity of operations across sexes to determine whether female surgeons perform less complex operations than their male peers.

METHODS

Data Source

The case records of general surgeons at a large academic medical center from 1997 to 2018 were obtained from institutional sources, including the wRVU, each surgeon's years in practice, and procedure codes. Furthermore, through unique surgeon identification numbers we could identify the sex, race, and clinical division of all surgeons included in the dataset. Case complexity was chosen as an outcome because it is defined by a universally accepted metric, wRVU, which encompasses the technical difficulty of the operative case, is applied equally to operative cases regardless of surgeon or patient characteristics, and is often used to determine remuneration. In addition, the average clinical effort for male and female surgeons, calculated as the percentage clinical full-time equivalent effort (%FTE), was obtained from hospital administrative records. These reported FTE values may not accurately capture the work effort that is spent in clinical versus nonclinical responsibilities; however, such inaccuracies are likely similar for both sexes and therefore are unlikely to change our findings.

Inclusion and Exclusion Criteria

All operative cases performed by attending surgeons with wRVU values greater than 2 were included. Records with wRVU values less than or equal to 2 were excluded as they overwhelmingly represented office-based procedures. Current procedural terminology codes <10004 and >61000 were excluded because they are Evaluation and Management (E and M) codes or codes not relevant to general surgery. The following subspecialties were included in the analysis but are not identified in our results to protect surgeon anonymity: surgical oncology, colorectal, vascular, cardiac, thoracic, pediatric, trauma/critical care/burns, community surgery, minimally invasive, transplant, and endocrine. Subspecialties that had no female surgeons were excluded from subspecialty analyses.

Primary Variables and Statistical Analysis

The primary end point was wRVU for each case (wRVU/case). The primary independent variable was sex of the surgeon. The coefficient of variance [(CV) ] was also calculated to express the variability of wRVU distribution for male surgeons and for female surgeons. A higher CV indicates a larger variation in distribution. The secondary analysis was performed at the surgeon-month level, with dependent variable of wRVU per month for each surgeon (wRVU/mo). Months during which a surgeon had zero wRVU were excluded. For example, if surgeons were only available for 10 months, then they would only contribute 10 surgeon-months.

The project was approved by the Partners Healthcare Institutional Review Board (2018P002141). All analyses were performed in Stata 15.1 (StataCorp, TX). Statistical significance was defined at P values <0.05.

RESULTS

Data Characteristics

A total of 551,047 cases from 131 surgeons performed across 13,666 surgeon-months were included. Among them, 104,424 (19.0%) of cases were performed by female surgeons, who make up 20.6% (n = 27) of the surgeon population, and 2879 (21.1%) of the surgeon months (Fig. 1). Furthermore, male surgeon's average clinical %FTE was 75% and female surgeon's was 85% (P = 0.07). The cases done by male surgeons were more often performed by mid-career or senior surgeons, whereas the opposite trend was found for female surgeons (P < 0.01). Over the study period, a higher percentage of cases done by male surgeons were prior to 2010 than that by female surgeons (41.5% vs 36.2%, P < 0.01) (Table 1).

FIGURE 1
FIGURE 1:
Clustered column graph of cohort characteristics by sex.
TABLE 1 - Operative Case Characteristics by Sex
Male Surgeons (n = 104) Female Surgeons (n = 27) P
Number of cases 446,623 (81.0%) 104,424 (19.0%)
wRVU per case
 Mean (SD) 10.8 (12.6) 8.3 (8.4) <0.01
 Median (IQR) 6.3 (2.8, 14.5) 6.3 (2.7, 10.1) <0.01
 CV 116.5% 101.5%
Surgeon practice year
 <10 90,350 (20.5%) 41,167 (39.9%) <0.01
 10–19 161,932 (36.7%) 41,974 (40.7%)
 ≥20 188,488 (42.8%) 20,112 (19.5%)
Procedure calendar year
 <2010 185,568 (41.5%) 37,832 (36.2%) <0.01
 ≥2010 261,055 (58.5%) 66,592 (63.8%)
The sex distributions of subspecialty were not identified to protect surgeon anonymity.IQR indicates interquartile range; SD, standard deviation.
n indicates the number of surgeons in our study population.

Unadjusted Analysis

The mean wRVU/case for male surgeons was significantly higher than that for female surgeons (10.8 vs 8.3, P < 0.01) (Table 1). In addition, male surgeons were more likely to have high outlier values in their wRVU/case (0.38% for wRVU ≥100 for male surgeons, vs 0.10% for female surgeons, P < 0.01) (Fig. 2) and the CV of wRVU/case distribution was higher for male surgeons (CV: 116.5% vs 101.5%).

FIGURE 2
FIGURE 2:
Box plot of wRVU/case stratified by surgeon sex. Yellow dots represent outliers of ≥100 wRVU.

Adjusted Analysis

On adjusted analysis, male surgeons earned an additional 1.65 wRVU/case, compared to female surgeons [95% confidence interval (CI) 1.57–1.74]. In addition, seniority of a surgeon was also associated with increasing wRVU/case (+1.41 for 10–20 years in practice, and +1.58 for years ≥20, compared to surgeons with <10 years in practice) (Table 2). On secondary analysis, a similar trend was observed when wRVU/mo was analyzed. Male surgeons earned an additional 110.6 wRVU/mo, compared to female surgeons, even after adjusting for surgeon subspecialty, seniority, calendar year, and race (Table 3).

TABLE 2 - Adjusted Analysis of Work Relative Value Unit per Case
Estimates (95% CI) P
Sex
 Female surgeon Ref.
 Male surgeon +1.65 (+1.57, +1.74) <0.01
Surgeon practice year
 <10 Ref.
 10–19 +1.41 (+1.33, +1.49) <0.01
 ≥20 +1.58 (+1.50, +1.67) <0.01
Procedure calendar year
 <2010 Ref.
 ≥2010 +0.38 (+0.32, +0.44) <0.01
The linear regression models were additionally adjusted for race and clinical subspecialty of surgeons (data not shown).

TABLE 3 - Adjusted Analysis of Monthly Work Relative Value Unit
Estimates (95% CI) P
Sex
 Female surgeon Ref.
 Male surgeon +110.6 (+96.7, +124.5) <0.01
Surgeon practice year
 <10 Ref.
 10–19 +58.1 (+45.1, +71.1) <0.01
 ≥20 +78.2 (+64.0, +92.3) <0.01
Procedure calendar year
 <2010 Ref.
 ≥2010 –19.7 (–30.5, –8.8) <0.01
The linear regression models were additionally adjusted for race and clinical subspecialty of surgeons (data not shown).

Subset Analysis

A subset analysis found that sex disparity increases with surgeon seniority (Fig. 3A). The adjusted difference in wRVU/case between male and female junior surgeons (<10 years in practice) was +0.24 (95% CI 0.04–0.45), for “mid-career” surgeons (between 10 and 20 years in practice) was +2.16 (95% CI 2.02–2.29), and for senior surgeons was +3.30 (95% CI 3.12–3.48). In addition, subset analysis by subspecialties demonstrated qualitatively similar findings of sex disparity in case complexity across most subspecialties, with only 1 subspecialty (11.1%) having a higher wRVU for female surgeons (Fig. 3B). Strikingly, sex disparity did not improve over the 20-year study period (Fig. 3C).

FIGURE 3
FIGURE 3:
Subgroup analyses on sex disparity in wRVU per case by (A) surgeon practice year (seniority), (B) subspecialty, and (C) procedure calendar. In (B), the red line indicates the reference group for each subspecialty, female surgeons. Values above the line show that male surgeons in that subspecialty have higher wRVU/case.

DISCUSSION

The goal of this study was to compare the complexity of operations between female and male surgeons at a population level, and determine whether female surgeons perform less complex operations than their male peers. We found that female surgeons are more likely to perform cases with lower wRVU value than their male peers, even when accounting for differences in clinical subspecialty and years in practice. This sex disparity increased with surgeon seniority, and there was no measurable improvement over the 20 years of this study. This work extends our previous research showing significant underemployment of female surgeons relative to male surgeons.9 This underemployment was evident across many facets of practice, with female pediatric surgeons performing fewer total numbers of operative cases, fewer cases that require their subspecialty training, and cases with lower complexity than their male peers.

Our study uniquely advances the literature on sex disparity in surgery with our analysis of wRVU data from a large institution. There is no rational reason to think that women would willingly take on “simpler” cases than their male peers with equal training, especially when such disparity persists across specialty, seniority, and decades. Common arguments such as differences in availability due to competing obligations, even if they were true (which may not be, as will be discussed below), should not affect the complexity of cases that a surgeon performs. Many surgeons have significant administrative responsibilities and yet have a complex surgical practice. Because wRVU directly relates to revenue, our finding at least partially highlights the mechanism behind the salary differential between male and female surgeons. There are similar findings in other professions. For example, in the legal profession, while females account for 38% of all attorneys in the United States,14 female attorneys account for only 25% of all attorneys appearing in courtrooms and are less likely to be lead counsel in complex civil litigation, comprising 31.6% in 1-party cases but less than 20% in cases involving 5 or more parties.14,15 Moreover, female attorneys appear in Supreme Court arguments between only 12% and 21% of the times for every year since 2011.16 Similarly, in the engineering profession, female engineers are less likely to be mentored to take on tasks that require “hard” engineering skills such as technical ability and problem solving.17

Several other reasons have been proposed to explain why female surgical underemployment may occur and include availability affected by surgeon factors—alternate obligations or subspecialty, and referral factors—patient preference, social network among physicians, and confidence of referring colleagues. This work addresses several of these arguments as well.

First, we countered the narrative that females are less available to take consults, perform operations, or see referrals because of competing obligations. Our data show that female surgeons are equally available as their male colleagues when representation in the workforce is factored in. In addition, using our internal human resources data, we showed that male surgeons tend to have more competing obligations than female surgeons as demonstrated by their lower clinical %FTE. Therefore, differences in income or promotion, both of which are frequently attributed to a lower female surgeon availability, are rendered unlikely by these data. We also found that regardless of subspecialty choice, female surgeons more often performed cases with lower wRVU/case than males within the same subspecialty. In only 1 of 9 subspecialties (11.1%) did women have higher wRVU/case than their male colleagues. Therefore, this is a phenomenon that crosses subspecialties and is not a repercussion of a female surgeon's subspecialty choice.

Surgery is primarily a referral-based practice. Therefore, patient bias for male physicians could lead patients to prefer male surgeons for more complex surgery, perhaps because of a societal perception that men are more competent than women across many industries. Although unfortunate, this bias also exists for medicine, but is less likely relevant for a surgeon's practice because physician colleagues are the primary source of patient referrals to surgeons, and the literature supports that most patients follow the referral recommendations of their primary care physician rather than choose a surgeon on their own.18 Therefore, patient bias is unlikely to have any significant effect on disparate case complexity between female and male surgeons.

The primary source of referrals to surgeons is from colleagues. Interprovider referral patterns have been shown to depend on social networks that exist within hospitals and hospital systems.19 These networks show how many connections there are between colleagues, and how strong that network connection is. When starting a practice in a new hospital, young physicians may not receive the most complex referrals because they are less well known, have a limited social network, and have not yet demonstrated their skill level. It would follow then that female and male junior surgeons would be expected to have fewer referrals than senior surgeons. However, it would not explain why the complexity of the referrals differs between sexes. Nevertheless, our data counter that social networks have any significant effect on the complexity of referrals made to surgeons as female surgeons were disadvantaged from the outset of their careers, performing lower wRVU/case than their male colleagues. Further discounting this narrative, the disparity between male and female surgeons wRVU/case actually worsens with surgical seniority and presumed visibility in the hospital social network, such that mid-career male surgeons earn 2.2 wRVU/case and senior male surgeons earn 3.3 wRVU/case more than their female peers at equivalent career stages. As the disparity between case complexity of female and male surgeons only widened with surgeon seniority, eminence within a hospital social network does not appear to correlate with surgeon's case complexity.

Having ruled out the other potential rational explanations, we are left with a “diagnosis of exclusion”; the possibility that physician colleagues refer less complex cases to female surgeons because of their own bias.20 This may be due to referring providers having less confidence in their female surgical colleagues, or other sex biases. This hypothesis is supported by studies that show referring physicians’ discrepant beliefs are driving sex inequality in surgical practice, rather than female surgeons’ own behaviors of refusing referrals for complex surgeries.20 Specifically, when a female surgeon has a patient complication, the referring physician is not only less likely to refer to that female surgeon, but also less likely to refer to all other female surgeons in the future. This spillover effect is not seen with male surgeons. In addition, despite changes in referring physician practice, data show that female surgeons continue to accept the same number of referrals after an adverse outcome.20 It has also been suggested that female surgeons receive referrals for less complex cases than male surgeons because of suspected differences in patient outcomes by surgeon sex, although there is no evidence that such outcome differences exist; in fact, most studies show equivalent outcomes between male and female surgeons, whereas a few studies suggest female physicians have better outcomes than male physicians.21–23 The extent to which these discrepant attitudes contribute the female surgeon underemployment requires separate investigation.

Our study has several strengths. Our primary endpoint of wRVU/case is novel and uniquely accounts for potential differences in surgeon availability due to other professional and familial obligations. We also examined data across subspecialty, years in practice, and over time to increase the generalizability of our work. Our study also has limitations. The single-institutional nature of our study, while it provided us with the necessary granular data elements to perform these analyses, remains to be validated in other institutions. Because this institution has historically had a larger majority of male surgeons, the sex differences in operational complexity may be a result of having few female surgeons. These differences could be attenuated in institutions where the sex distribution is more even. However, we doubt that sex distribution in surgeon numbers can independently explain our results. In addition, we were not able to account for a surgeon's subspecialty choice. Although it may be true that female surgeons are more concentrated in lower-complexity subspecialties, our result showed that in addition to differences across subspecialties, sex differences in case complexity exist within most subspecialties as well. In other words, whether female surgeons choose subspecialties with high or low average wRVU/case, the disparity within each subspecialty still favors male surgeons.

Our study has important implications. The goal of “equal pay for equal work” may ultimately not be a pay equality issue, but rather a work equality issue—the inability of female surgeons to attain equal work. Our work thus sheds important light on the issue of work equality, and provides a novel way to measure a previously unmeasured component of sex inequality in surgery, namely, surgeon's clinical practice portfolios. With no change over 20 years, our data suggest that increased attention to sex inequality alone will not be a solution for the surgical workplace. This problem requires systems-level interventions. Potential interventions include centralized referral systems, improved sponsorship of female surgeons, and cognitive error training.

In conclusion, female surgeons perform less complex cases than their male peers, even after accounting for differences in clinical subspecialty, availability, and years in practice. These differences are evidence of underemployment of female surgeons and are not due to competing professional or familial obligations or the surgeon's choice of subspecialty but may instead be due to bias from referring physicians.

REFERENCES

1. Phillips NA, Tannan SC, Kalliainen LK. Understanding and overcoming implicit gender bias in plastic surgery. Plast Reconstr Surg 2016; 138:1111–1116.
2. Ceppa DP, Dolejs SC, Boden N, et al. Sexual harassment and cardiothoracic surgery: #UsToo? Ann Thorac Surg 2020; 109:1283–1288.
3. Rogers AC, Wren SM, McNamara DA. Gender and specialty influences on personal and professional life among trainees. Ann Surg 2019; 269:383–387.
4. Acai A, Steyn C, Reid SE, et al. A solution to gender inequity in surgery? Better caregiving policies. Can J Surg J Can Chir 2018; 61:6–7.
5. Are You Underemployed? Infographics. PayScale. Available at: https://www.payscale.com/data-packages/underemployment. Accessed April 23, 2020.
6. Abel JR, Deitz R, Su Y. Are Recent College Graduates Finding Good Jobs? Curr Issues Econ Finance Fed Reserve Bank N Y. Available at: https://ssrn.com/abstract=2378472. 2014. Accessed May 20, 2020.
7. Underemployed by Education Level? Infographics. PayScale. Available at: https://www.payscale.com/data-packages/underemployment/education-level. Accessed April 24, 2020.
8. Luckwaldt JH. Women Are More Likely to Be Underemployed Than Men. PayScale. Available at: https://www.payscale.com/career-news/2016/06/underemployed-women. 2016. Accessed April 23, 2020.
9. Zhang B, Westfal ML, Griggs CL, et al. Practice patterns and work environments that influence gender inequality among academic surgeons. Am J Surg 2020; 220:69–75.
10. Maynard DC, Feldman DC. Underemployment: Psychological, Economic, and Social Challenges. New York, NY: Springer Science + Business Media; 2011.
11. Jones CM, Gautier L, Kadio K, et al. Equity in the gender equality movement in global health. Lancet 2018; 392:e2–e3.
12. Gomez LE, Bernet P. Diversity improves performance and outcomes. J Natl Med Assoc 2019; 111:383–392.
13. Rotenstein LS, Jena AB. Lost Taussigs—the consequences of gender discrimination in medicine. N Engl J Med 2018; 378:2255–2257.
14. Bureau UC. Number of Women Lawyers At Record High But Men Still Highest Earners. The United States Census Bureau. Available at: https://www.census.gov/library/stories/2018/05/women-lawyers.html. Accessed May 12, 2020.
15. Bazelon L. What It Takes to Be a Trial Lawyer If You’re Not a Man. The Atlantic. Available at: https://www.theatlantic.com/magazine/archive/2018/09/female-lawyers-sexism-courtroom/565778/. Accessed May 5, 2020.
16. Women Argue Only a Fraction of Supreme Court Cases Available at: https://news.bloomberglaw.com/us-law-week/women-argue-only-a-fraction-of-supreme-court-cases-3. Accessed May 5, 2020.
17. Cardador MT, Caza BB. The Subtle Stressors Making Women Want to Leave Engineering. Harvard Business Review, 2018 Available at: https://hbr.org/2018/11/the-subtle-stressors-making-women-want-to-leave-engineering. Accessed May 5, 2020.
18. Forrest CB, Shadmi E, Nutting PA, et al. Specialty referral completion among primary care patients: results from the ASPN referral study. Ann Fam Med 2007; 5:361–367.
19. Landon BE, Keating NL, Barnett ML, et al. Variation in patient-sharing networks of physicians across the United States. JAMA J Am Med Assoc 2012; 308:265–273.
20. Sarsons H. Interpreting Signals in the Labor Market: Evidence from Medical Referrals. Job Mark Pap. Available at: https://scholar.harvard.edu/sarsons/publications/interpreting-signals-evidence-medical-referrals. Accessed May 20, 2020.
21. Sharoky CE, Sellers MM, Keele LJ, et al. Does surgeon sex matter?: practice patterns and outcomes of female and male surgeons. Ann Surg 2018; 267:1069–1076.
22. Tsugawa Y, Jena AB, Orav EJ, et al. Age and sex of surgeons and mortality of older surgical patients: observational study. BMJ 2018; 361:k1343.
23. Wallis CJ, Ravi B, Coburn N, et al. Comparison of postoperative outcomes among patients treated by male and female surgeons: a population based matched cohort study. BMJ 2017; 359:j4366.
Keywords:

complexity of surgery; sex inequality; work relative value unit

Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved.