Operative mortality for male versus female surgeons: a systematic review and meta-analysis

Abstract

Background:

This systematic review synthesizes published data concerning whether there is a difference in mortality rate for patients operated on by female surgeons compared to patients operated on by male surgeons. Several studies have attempted to compare surgical outcomes, including operative mortality, of male surgeons versus female surgeons. These studies include many different surgical subspecialties and surgery types. Despite current research on this topic, there has yet to be a published systematic review and meta-analysis quantitatively synthesizing these studies’ findings on operative mortality. This systematic review objectively synthesizes the current published data regarding mortality rate for patients operated on by female surgeons as compared to male surgeons.

Materials and Methods:

The study included patients of any age and sex undergoing any kind of surgical procedure in North America. This review considered all studies that evaluated the performance of any type of surgical procedure by female versus male surgeons and included intraoperative or postoperative mortality as outcomes. Eligible studies were appraised for risk of bias using the Newcastle-Ottawa Scale and for quality using standardized instruments from JBI SUMARI. Studies were pooled for raw outcomes for statistical meta-analysis including operative mortality counts and rates. Heterogeneity was assessed using Cochran’s Q test. Statistical analyses were performed using a random effects model. A forest plot was created to compare study associations and significance.

Results:

The search yielded five retrospective cohort studies published between 2000–2018 including various surgical subspecialties. In total, 1,055,122 operations were included, performed by 6,139 female surgeons and 47,666 male surgeons. There were 4,176 deaths among patients operated on by female surgeons and 55,666 among patients operated on by male surgeons, respectively. Forest plot analysis found no evidence of a difference between male and female surgeon mortality rate among the studies (pooled OR 0.96 [0.88–1.05]).

Conclusions:

All five studies provided odds ratios for mortality and four of five provided counts for mortality. There was no statistical difference in mortality between male and female surgeons among the studies.

Introduction

In surgical subspecialties, women are least represented in orthopedic surgery (15.4% female), neurological surgery (17.5% female), and thoracic surgery (21.8% female).¹ Within general surgery, women make up less than a quarter of the field (22% according to the 2021 physician specialty data report).² There is evidence that these disparities affect patient outcomes. A 2022 study found that sex discordance between surgeons and their patients was associated with a higher chance of adverse postoperative outcomes, complications, and death.³ Worse outcomes occurred among female patients treated by male surgeons, but not for male patients treated by female surgeons.³ However, this disparity is not consistent across populations or procedure type and requires further investigation.^4,5 The paucity of female representation within surgery has the potential to create, at best, an absence of knowledge regarding female surgeon outcomes, and, at worst, a patient and surgeon sex bias with a possible preference toward male surgeons.

Several studies have attempted to compare surgical outcomes, including operative mortality, of male surgeons versus female surgeons.^6–12 Previous studies that examined operative mortality included procedures performed by general surgeons, as well as many subspecialist surgeons including obstetricians/gynecologists, neurosurgeons, and otolaryngologists. Sharoky et al. looked at inpatient mortality for general surgery patients undergoing common procedures including appendectomy, cholecystectomy, and hernia repairs.⁶ In 2000, O’Neill et al. analyzed in-hospital mortality after carotid endarterectomy;⁷ in 2016, Xu et al. looked at inpatient deaths during index hospitalization for elective colectomy.⁸ While Wallis et. al. used a composite primary outcome encompassing death, readmission, and complications, they also reported 30 day mortality for 25 common surgical procedures ranging from cardiothoracic and vascular surgery to general surgery and obstetrics and gynecology.⁹ In 2018, Tsugawa et al. similarly looked at operative mortality rates for patients operated on by male versus female surgeons, including death during hospital admission or within 30 days of surgery.¹⁰ Chapman et. al., 2020 compared patient outcomes amongst male and female surgeons performing total joint arthroplasty by qualitative surgeon characteristics.¹¹ Lastly, Etherington et. al., 2020 identified the research and study knowledge gap related to physician’s sex as it relates to anesthesiologists’ practice.¹²

Despite current research on this topic, there has yet to be a published systematic review and meta-analysis quantitatively synthesizing these studies’ findings on operative mortality. The purpose of this systematic review and meta-analysis was to quantitatively synthesize published data comparing the operative mortality of patients operated on by male surgeons versus female surgeons.

Methods

Eligibility criteria:

Inclusion Criteria

Patients of any age and sex undergoing any kind of surgical procedure in North America. This review considered studies which evaluated the performance of any type of surgical procedure by female surgeons.

Comparator

This review considered studies which compared the performance of any type of surgical procedure by male surgeons.

Outcomes

This review considered studies which include the following outcomes: intraoperative or postoperative mortality. These outcomes were measured by the abstraction of comparative data from original manuscripts and statistical analyses.

Types of studies

This review considered prospective cohort, retrospective cohort, and observational studies. Randomized control trials were considered, but they were not expected to be found in this search or meet inclusion criteria because patients cannot be randomized to surgeons based on gender.

Information sources:

The databases searched include PubMed, MEDLINE, Embase, and Cochrane on June 3, 2021. Sources of gray literature were not included. We performed an additional brief literature search on PubMed June 15, 2023 revealing no new publications that fit this study’s criteria of inclusion in the analysis.

Search strategy:

An initial search of PubMed with the keywords and index terms: “surgeon” AND “gender” AND “patient outcomes” was undertaken to identify articles on the topic. The primary search text words contained in the titles and abstracts of relevant articles displayed a focus on mortality and death specific statistics as a primary outcome. Therefore, the index terms used to describe the articles were used to develop a full search strategy for all databases PubMed, MEDLINE, Embase and Cochrane search completed on June 3, 2021 with the keywords and index terms: “surgeon” AND “gender” OR “sex” AND “mortality” OR “death” (Appendix I). The search strategy, including all identified keywords and index terms was adapted for each database and/or information source. This review includes studies published in the English language and conducted in North America. Only studies published after 1999 were included. Studies were restricted to the English language and those conducted in the United States and Canada.

Selection process:

Following the search, all identified citations were collated and uploaded into EndNote 20 (Clarivate Analytics, PA, USA). Following a pilot test, all identified titles and abstracts were equally distributed and screened by five independent reviewers for assessment against the inclusion criteria for the review. Two supervisors also conducted secondary reviews of the primary reviewers’ assessments, ensuring that every abstract and paper was evaluated by at least two reviewers. Potentially relevant studies were retrieved in full, and their citation details were imported into the JBI System for the Unified Management, Assessment and Review of Information also known as JBI SUMARI (JBI, Adelaide, Australia). The full text of selected citations was assessed in detail against the inclusion criteria collectively by five independent reviewers. Reasons for exclusion of papers at full text that do not meet the inclusion criteria were recorded and reported in the systematic review. Any disagreements that arose between the reviewers at each stage of the selection process were resolved through discussion. The results of the search and the study inclusion process are reported in full in the final systematic review and presented in a Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) flow diagram.

Data collection process:

Data was extracted from studies included in the review by five independent reviewers supervised by two secondary reviewers using the standardized data extraction tool on JBI SUMARI. The data extracted includes specific details about the populations, study methods, interventions, and outcomes of significance to the review objective. Any disagreements that arose between the reviewers were resolved through discussion.

Data items:

This review considered studies which include the following outcomes: intraoperative or postoperative mortality and the comparison of outcomes among female and male surgeons. These outcomes are measured by the abstraction of comparative data from original manuscripts and statistical analyses. Missing data was not a complication in the systematic review. For the meta-analysis component, only studies with complete data were included, thus the Xu et al. study was excluded because the raw number of deaths from male and female surgeons was not reported. The contingency plan for missing data management included using statistical models that allow missing data to make assumptions of data relationships based on what data is available in the independent studies.

Study risk of bias assessment:

Eligible studies were critically appraised by two independent reviewers at the study level for risk of bias in the review using the Newcastle-Ottawa Scale. Any disagreements that arose were resolved through discussion.

Effect measures:

Effects measures include absolute risks for the treatment and control, estimates of relative risk, and a ranking of the quality of the evidence based on the risk of bias, directness, heterogeneity, precision, and risk of publication bias of the review results. The outcomes are reported in the Summary of Findings (SoF) created with GRADEPro GDT 3.0¹⁴ which include intraoperative or postoperative mortality.

Synthesis methods:

Five large studies met the inclusion criteria and were included in the systematic analysis shown in the PRISMA diagram, Figure 1. The systematic review and meta-analysis were conducted by the Rutgers University Biostatistics & Epidemiology Services (RUBIES) center, an advanced biostatistics unit within Rutgers Biomedical and Health Sciences and the Rutgers School of Public Health. Studies were, where possible, pooled for raw outcomes for statistical meta-analysis using the Meta package in R version 4.2.1 including operative mortality rates. The effect sizes are expressed as odds ratios (for dichotomous data), and their 95% confidence intervals were calculated for analysis. Heterogeneity was assessed statistically using Cochran’s Q test. Statistical analyses were performed using a random effects model. Where statistical pooling was not possible, the findings are presented in narrative form. A forest plot was created to compare study associations and significance for odds ratio only since five of five studies provided odds ratios. The handling of missing summary statistics or data includes only using the data available to provide answers that inform the study question.

Reporting bias and certainty assessment:

Management of risk of bias in missing data was determined per study independently based on all characteristics of the paper and available information. The authors of the paper were to be contacted to obtain additional information to assess risk of bias and certainty as needed based on the Newcastle-Ottawa Scale assessment.

Results

Study selection:

Five large studies met the inclusion criteria and were included in the systematic analysis shown in the PRISMA diagram, Figure 1. Studies excluded include those that did not evaluate the outcomes of focus, had an ineligible study design such as a review or comment paper, case report and case report series, wrong control, and wrong or unspecified design, for example, Chapman et al., 2020.

Study characteristics:

Of the five studies that met inclusion criteria for the systematic review, each compared operative outcomes, including operative mortality, for male and female surgeons (Table 1).^6–10 In total, 1,055,122 operations were included, which were performed by 6,139 female surgeons and 47,666 male surgeons. The studies ranged in size from 12,467 to 892,187 operations, 18 to 4,634 female surgeons, and 226 to 41,192 male surgeons. Five of five studies provided odds ratios for mortality and four of five studies provided counts of mortality. All five studies were published from 2000–2018, were conducted in the United States and Canada using a retrospective cohort study design, and were found to have a low risk of bias and designated as good quality. They included a variety of procedures performed by general surgeons as well as subspecialty surgeons such as vascular, neuro, cardiovascular, thoracic, orthopedic, plastic, otolaryngology, urology, and obstetrics/gynecology.^6–10

Risk of bias and certainty assessment:

The overall risk of bias for each of the five studies included in this systematic review is low as each study received a good quality rating. The results are attached in Table 3 using the Newcastle-Ottawa Scale.

Results of individual studies:

The studies included data from 126,846 patients operated on by female surgeons and 928,276 patients operated on by male surgeons (Table 2). There were 4,176 deaths among patients operated on by female surgeons and 55,666 among patients operated on by male surgeons, respectively.^6,7,9,10

Results of syntheses:

Forest plot analysis found no evidence of a difference between male and female surgeon mortality rate among the studies (pooled OR 0.96 [0.88–1.05]) (Figure 2).

Discussion

This is the first published systematic review and meta-analysis comparing surgical mortality for female versus male surgeons. Our findings add to a growing body of evidence that there is no relationship between whether surgeons are perceived as male versus female and their surgical ability. Our goal was to provide evidence about the mortality rate among patients operated on by female surgeons as compared to patients operated on by male surgeons.

In our meta-analyses, the studies combined present large sample sizes of more than 126,000 patients operated on by female surgeons and more than 920,000 patients operated on by male surgeons.^6–10 Four of five studies^6,7,9,10 provided counts for operative mortality, and all five studies^6–10 provided odds ratios for operative mortality. We can conclude that for these studies, there was no statistical difference between the odds of mortality for operations done by female surgeons when compared to operations done by male surgeons.

In the literature, there is a dearth of studies that compare clinical outcomes for patients operated on by male versus female surgeons. Systematic reviews synthesizing data only on clinical outcomes of this topic are even more scarce. A systematic review with a focus on cardiac operative care by Etherington et al. identified two papers with data comparing clinical outcomes between male and female surgeons.¹² Both papers included in the review are included in this study.^9,10 Thus, our study builds on their findings by providing additional data on post-operative mortality from three other published studies that is inclusive of other surgical operations. Furthermore, a scoping review by Champagne-Langabeer et al. showed that despite a search of 2,420 records, they only accepted 15 records, two of which report clinical outcomes for male versus female surgeons, one of which was included in this review;¹³ the other was excluded for not meeting inclusion requirements.¹⁵ The remaining categories in this review include a focus on clinical qualities, diagnosis of disease, and treatment comparisons between male and female surgeons.¹³ The authors’ concluding remarks advocate for further studies that look into differences of outcomes among male versus female surgeons. The findings of this systematic review reinforce studies that focus on comparing female versus male surgeon clinical mortality.

The limitations of this study, however, must not be overlooked. All the studies included are retrospective cohort studies, making it challenging to eliminate confounding variables.^6–10 For example, comorbidities could not be controlled for. This systematic review and meta-analysis also relied on studies that directly compared female and male surgeons’ mortality counts and rates. Three of the five studies used matching for patients and surgeons,^6,7,9 often to address the smaller sample size of female surgeons compared to male surgeons. Matching was also used to address possible confounders like differences in surgeon age, years in practice, specialty, and patient demographics.⁹ For the meta-analysis done based on raw counts, there is not sufficient information to control for confounding variables. Therefore, this presents a limitation in analysis. The meta-analysis based on odds ratios was adjusted for key confounding variables, including patient, doctor, and hospital characteristics, reported in the original studies. Each study included in this meta-analysis provided odds ratios derived from multivariable regression models that adjusted for these confounding factors (Table 4). For example, Xu et al., 2016 adjusted for patient age, race, elixhauser comorbidity index.⁸ Tsugawa et al., 2018 adjusted for patient age, sex, race, procedure type, elixhauser, median household income, Medicaid, year/day of surgery, surgeon age, sex, specialty, medical school, credentials, annual operative volume, and hospital.¹⁰ Wallis et al., 2017 adjusted for patient age at surgery, geographic location (local health integration networks), sex, socioeconomic status (based on geographic location), rurality, and general comorbidity according to the Johns Hopkins aggregate disease group score (ADG).⁹ Sharoky et al., 2018 adjusted for patient, hospital and surgeon characteristics using cardinality matching with refined covariate balance.⁶ O’Neill et al., 2000 adjusted by excluding cases in analysis with other procedures than carotid endarterectomy, those that could increase risk of adverse outcomes, presurgical mortality risk of 6% or greater, and adjusted for presurgical risk among remaining patients using proprietary Medisgroups severity index.⁷ Furthermore, our study focused solely on comparing mortality for male versus female outcomes, and as a result we did not investigate other factors that may influence outcomes including surgeon’s age, years of experience, training, workplace environment, discrimination based on sex or gender identity, or perception by others, among many factors.

Future studies should include more extensive investigation of other clinical outcomes such as reported common postoperative complications in addition to mortality. Additionally, they should consider the intersectionality of sexual and gender discrimination and sociological, psychosocial and contextual factors that introduce biases and influence outcomes in the clinical environment between surgeons and patients such as discordant patient-surgeon dyads. However, these are outside the scope of this study, which was designed to compare mortality outcomes among male versus female surgeons.

Conclusions

The systematic review and meta-analysis show that there is no evidence for a difference between mortality for patients operated on by male versus female surgeons. These results bring increased attention to the fact that female surgeons have similar rates of operative mortality as male surgeons. Overall, this study may assist in spearheading necessary discussions regarding biases of female versus male surgeon’s abilities. Additionally, this work should prompt further research regarding male versus female surgeon differences in clinical outcomes.

Figure 1. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Diagram.

Flowchart for the systematic review following PRISMA guidelines. A total of 2,591 titles were identified through our search. Five articles were identified as appropriate for inclusion. The studies included are all retrospective cohort studies.

Figure 2. Forest Plot Analysis of Operations Mortality Using Odds Ratio Only.

Forest plot comparing female to male surgeons for each included study with study measures. (I=percentage of variation across studies due to heterogeneity and not chance; Q Test=test for between-study heterogeneity). Data reporting number of operations resulting in death from the Xu, 2016 study are unavailable, thus the proportion of male and female surgeon operations that result in death is not displayed.

Table 1. Characteristics of Included Studies.

Summary data for included studies.

First Author, Year, Type of Study	Geographic Area(s)	No. of Female surgeons (N=5,929)	No. of Male surgeons (N=46,456)	No. of operations (N=1,348,507)	Surgical Specialties Included	Outcomes Analyzed
Wallis, 2017, retrospective cohort	Ontario, Canada	774	2,540	104,630	Cardiothoracic; general; neurosurgery; obstetrics/gynecology; orthopedic; otolaryngology; plastic; thoracic; urology; vascular	Death within 30 days; Readmission within 30 days; Complication within 30 days; Hospital length of stay
Sharoky, 2018, retrospective cohort	New York, Florida, Pennsylvania, USA	461	2,001	336,440	General surgery	Inpatient mortality; Any postoperative complication; Hospital and procedure-specific length of stay
Xu, 2016, retrospective cohort	Maryland, USA	42	234	2,525	General surgery, colorectal surgery, surgical oncology	Postoperative complications mortality during hospital stay
O’Neill, 2000, retrospective cohort	Pennsylvania, USA	18	489	12,725	Vascular surgery	In-hospital mortality; Nonfatal morbidity
Tsugawa, 2018, retrospective cohort	USA	4,634	41,192	892,187	General surgery, orthopedic surgery, colorectal surgery, obstetrics and gynecology, bariatric surgery, cardiothoracic surgery, vascular surgery	Operative mortality rate

Table 2. Analysis of Operative Mortality.

Summary of operative mortality by surgeon sex. The Xu, 2016 study is not included in this table because it does not provide counts for operative mortality.

Study Author	No. of patients operated on by female surgeons (N=126,846) (100%)*	No. of deaths among patients operated on by female surgeons (N=4,176) (100%)*	No. of patients operated on by male surgeons (N=928,276) (100%)*	No. of deaths among patients operated on by male surgeons (N=55,666) (100%)*	Lower limit of the 95% confidence interval (CI) of Relative Risk (RR)	Upper limit of the 95% confidence interval (CI) of Relative Risk (RR)	Relative Risk (RR)
Wallis, 2017	52,315 (41.2%)	480 (11.5%)	52,315 (5.64%)	543 (0.98%)	Mortality 0.78	Mortality 1.00	Mortality 0.88
Sharoky, 2018	22,444 (17.7%)	433 (10.4%)	23,394 (2.52%)	412 (0.74%)	Mortality 0.96	Mortality 1.25	Mortality 1.10
O’Neill, 2000	302 (0.24%)	1 (0.02%)	12,165 (1.31%)	85 (15.3%)	Mortality 0.07	Mortality 3.44	Mortality 0.48
Tsugawa, 2018	51,785 (40.8%)	3,262 (78.1%)	840,402 (90.5%)	54,626 (98.1%)	Mortality 0.94	Mortality 1.00	Mortality 0.97

^*Percentages do not add up to 100% in each column due to rounding.

Table 3. Risk of Bias Assessment.

Risk of bias was assessed using the Newcastle-Ottawa Scale. Good quality can be determined as: selection domain 3 or 4 stars; comparability domain 1 or 2 stars and 2 or 3 stars in outcome domain. Fair quality: selection domain 2 stars; comparability domain 1 or 2 stars and 2 or 3 stars in outcome domain. Poor quality: selection domain 0 or 1 stars, comparability domain 0 stars and 0 or 1 stars in outcome domain.

Study	Selection: Representativeness of sample	Selection: Selection of non-intervention cohort	Selection: Ascertainment of exposure	Selection: Demonstration that outcome of interest was not present at the start of the study	Comparability based on design and analysis	Outcome: Assessment of outcome	Outcome: Was follow-up long enough for outcomes to occur	Outcome: Adequacy of follow-up of cohorts	Total	Assessment
Wallis, 2017	1	1	1	1	2	1	1	1	9	Good
Sharoky, 2018	1	1	1	1	2	1	1	1	9	Good
Xu, 2016	1	1	1	1	2	1	1	0	8	Good
O’Neill, 2000	1	1	1	1	2	1	1	0	8	Good
Tsugawa, 2018	1	1	1	1	2	1	1	1	9	Good

Table 4. Factors Controlled for in Included Studies.

Summary of confounding factors controlled for in each study and method(s) used.

Study Author	Surgical Specialty and Procedures	Statistical Models or Approach Used to Control Factors	Factors Matched or Adjusted for
Wallis, 2017	Various surgical specialties and procedures	Johns Hopkins aggregate disease group score (ADG)	Matched: Patient factors including patient age, patient sex, comorbidity; surgeon factors including surgeon volume, surgeon age, specialty, income, years in practice; other factors including hospital, rurality, procedural fee code
Sharoky, 2018	Various surgical procedures within general surgery	Cardinality matching with refined covariate balance	Matched: primary/secondary specialties, years of experience Adjusted: patient, hospital, and surgeon age, surgeon sex, procedure volume, operation type, clinical factors
Xu, 2016	Various surgical specialties; elective laparoscopic/open colectomy procedures	Elixhauser comorbidity index	Adjusted: Patient factors including age, those of elixhauser comorbidity score; surgeon factors including surgeon experience, surgeon medical school ranking
O’Neill, 2000	Vascular surgery; carotid endarterectomy procedures	Medisgroups severity index	Adjusted: patient presurgical mortality risk, years since licensure, specialty-predicted surgical mortality rate
Tsugawa, 2018	Various surgical specialties and procedures	Elixhauser comorbidity index	Adjusted: Patient characteristics including age, sex, race/ethnicity, procedure type, comorbidity, income, dual Medicaid coverage, surgery timing; surgeon factors including specialty, medical school, credentials, operative volume; other hospital fixed effects

Highlights.

• Operative mortality by surgeon gender was compared for more than one million cases

• Five studies were pooled for meta-analysis including forest plot analysis

• Analysis found no difference in mortality rates for male and female surgeons

Data Statement

All data obtained for this study is from publicly available published articles included in this study.