Continuation of Gender Disparities in Pay Among Academic Emergency Medicine Physicians
Abstract
Objectives
The objective was to identify the effects of gender and other predictors of change in the salary of academic emergency physicians over a four sequential time period of survey administration, across a sample of physicians within different emergency departments (EDs) and within states representing the four main geographical regions of the United States.
Methods
This was a successive cross‐sectional observational study of EDs in the United States using an annual salary survey distributed to all Association of Academic Chairs in Emergency Medicine (AACEM) and Academy of Administrators in Academic Emergency Medicine (AAAEM) members in 2013, 2015, 2016, and 2017 with a sample size of 7,102 respondents over all time periods. The primary variable of interest was the adjusted base salary, calculated to be the full‐time effort of the physician without any enhancements (e.g., without stipend, release time, extra hours). Institutional predictive variables included U.S. region that ED was in and if the site was an academic or community academic hybrid (“community”) ED. Individual level variables included gender, academic rank, years at academic rank, years at rank within the ED, and primary duty (clinical or other). A series of Wilcoxon tests were conducted to determine if the unadjusted difference in salaries by gender for each year of the survey were significantly different. The effects of relative change in adjusted base salary over time were assessed using a mixed‐effects regression model, with institutional‐ and individual‐level predictors included in the model.
Results
Data were provided by 81 departments across the four geographic regions of the United States (Northeast, South, West, and Midwest). Most of the survey respondents across the four time periods of administration were male (65%) and reported primary clinical appointments at an academic ED (94%). Overall salaries increased across the four time points of the data with an overall relative 10.8% (95% confidence interval [CI] = 9.6%–12%) change in median salary between 2013 and 2017; the relative percentage change for female respondents was 10.6% (95% CI = 9.4%–11.85%) and 11.1% (95% CI = 10.2%–12%) for males. Within survey years, not adjusting for academic rank, the median salary increase for males was higher ($226,746 in 2013 to $252,000 in 2017) than females ($217,000 in 2013 to $240,000 in 2017), with significance at all four time points (Z = 6.33, p < 0.001), with a median average salary gap of $12,000 in 2017. In the predictive model that adjusted for covariates, gender significantly predicted median adjusted salary, with males earning significantly more than females (F(1) = 22.5, p < 0.001).
Conclusions
Despite previously published data showing an inappropriate gender salary gap in emergency medicine, this gap has remained essentially unchanged over the past 4 years.
Salary parity has long been an unrealized goal of gender equality.1 Although the Equal Pay Act of 1963 prohibited unequal pay for “substantially equal” work,2 women working full time in the United States typically are paid just 80% of what men are paid—a gap of 20%.3 This gap in pay by gender has been demonstrated in a number of professional industries including law, marketing, and administrative services.4 For physicians, the earnings gap between men and women has been documented since the mid‐1970s5 with unexplained difference in earnings for women ranging from 52% to 57%. Although women now comprise half of medical school graduates and represent 38% of faculty members in U.S. medical schools, significant differences exist between male and female physicians in both compensation and job advancement.6 This disparity is one of the highest for any professional industry, trailing behind only dentists, according to one 2010 analysis.7
Unexplained gender salary disparities have been noted in internal medicine8 and surgery.9 After specialty, age, faculty rank, and metrics of clinical and research productivity had been taken into consideration, male physicians earned nearly $20,000 more annually than female physicians according to two analyses.10, 11 Previous studies of emergency medicine have shown on average, female faculty are paid 10% to 13% less than their male counterparts.12, 13
In this study we review compensation trends over a 5‐year period in academic emergency medicine departments for both clinical and academic faculty by gender to determine if there have been any changes in the compensation gender disparity trends. This is the first study we are aware of that evaluates this trend over such a long time period and that considers both traditional academic and academic‐community salaries by gender.
Methods
This was a cross‐sectional observational study of academic departments in the United States. An annual salary survey was distributed by email to a listserv of all Association of Academic Chairs in Emergency Medicine (AACEM) and Academy of Administrators in Academic Emergency Medicine (AAAEM) members in 2013, 2015, 2016, and 2017. In 2009, AACEM and AAAEM became responsible for periodically conducting this salary survey, which is an iteration of a survey that has been periodically conducted by SAEM since 1991.12, 14 The survey instrument has been revised multiple times since its development. Each iteration since 2009 have been approved by AACEM and AAAEM leadership. Gender items were added to the administrator's survey in 2013. Potential survey participants were identified by their membership in the AAAEM or the AACEM and were invited to participate in completion of the AAAEM salary survey via the AAAEM online community e‐mail system and the AACEM online community e‐mail system. Survey participants included academic emergency departments (EDs) or divisions (adult and pediatric) throughout the United States. Surveys are typically completed by department administrators. For participating centers, the department administrators were asked to log onto the Novi Survey portal to provide deidentified information regarding individual faculty members for the most previous academic year. A group of designated leads, consisting of selected administrators and/or chairs, followed up with all potential respondents during the collection period to answer questions and enhance the response rate. Survey data were collected and maintained by a central site. Each data point was reviewed. If it was a minimum or maximum value the accuracy of these potential outliers was confirmed with department administrator. This study was submitted to the institutional review board (IRB) of the institution that maintains the data set and was determined to be exempt from IRB approval.
Data Analysis
The data were imported into SAS, Version 9.4, for analyses. For descriptive analyses median salaries are reported with interquartile range, across all years of the survey, and by respondents’ characteristics.
The focus of the analyses was on changes in reported adjusted salary across the 4 years of survey administration. Predictors of these changes included institutional variables: U.S. region that ED was in, and if the ED was an academic or community academic hybrid (“community”) institution. Individual‐level variables included gender, academic rank, years at academic rank, years at rank within the ED, and primary clinical duty (clinical or other). The primary variable of interest was the adjusted base salary, calculated to be the full‐time effort of the physician without any enhancements (e.g., without stipend, release time, extra hours). A series of Wilcoxon tests were conducted to determine of the unadjusted difference in salaries by gender for each year of the survey were significantly different. The effects of relative change in adjusted base salary over time was assessed using a mixed‐effects regression model, with institutional and individual level predictors included in the model. Respondents were excluded from the analyses if the ED was only pediatric or primary role was indicated to be department chair, vice chair, or chief. Respondents were nested within the ED site, considered the random effect, with the interaction between gender and time also modeled as a random effect and the included predictor variables as fixed effects. The years of survey administration were not equally spaced (2013, 2015, 2016, 2017), and the year of survey variable included in the model was coded to reflect this unequal time distribution. As has previously been stated, the respondents across all the time points were not necessarily the same individuals.
Missing Data Approach
The data were examined for missingness in the outcome and predictor variables. The primary variable of interest was the adjusted base salary. In total 7,570 eligible respondents provided data across the four survey administration time points; prior to analyses and data univariate descriptive analyses were conducted on all included variables. For the adjusted salaries 192 respondents provided values that were below the bottom 1% of salaries reported (<$79,000 for full‐time clinical appointment) and these respondents were dropped from the analyses. Also 276 respondents reported their primary responsibility as either chief, chair or vice chair (less than 1% of all study respondents), and given the focus of this study, these responses were also dropped from the analyses. In total 7,102 responses were included for all analyses.
There were 2% missing data for adjusted salary and less than 1% missing data across the predictor variables (gender, region, and community versus academic ED) used to estimate the missing salary data. A multiple imputation approach using a SAS procedure was used to estimate the missing variables. This approach produces five regression estimates of the missing data from complete cases; the five models are combined to provide an estimate of the missing data with appropriate standard error of these estimates.
Results
Sample Characteristics
The sample size was 7,102 respondents over all time periods (2013, 2015, 2016, and 2017). Table 1 indicates the characteristics of the respondents across all time points. The data were provided by 81 EDs across the four geographic regions of the United States (Northeast, South, West, and Midwest). Most came from the Northeast (38%), were male (65%), and reported primary clinical appointments at a pure academic ED (94%). Across females 80% identified as white, 5.8% as black/African American, and 5.7% as Asian. For males, 86.5% identified as white, 3.2% as Asian, and 3.1% as black/African American.
| Characteristic | 2013 (n = 1,432) | 2015 (n = 1,417) | 2016 (n = 1,961) | 2017 (n = 2,292) |
|---|---|---|---|---|
| U.S. Region | Northeast: 605 (42.3) | Northeast: 509 (35.9) | Northeast: 691 (35.2) | Northeast: 900 (39.3) |
| South: 241 (16.8) | South: 280 (19.7) | South: 416 (21.2) | South: 559 (24.4) | |
| Midwest: 356 (24.9) | Midwest: 372 (26.3) | Midwest: 440 (22.4) | Midwest: 499 (21.8) | |
| West: 230 (16.1) | West: 256 (18.1) | West: 414 (21.1) | West: 334 (14.6) | |
| Gender | Females: 457 (32) | Females: 480 (33.9) | Females: 678 (35) | Females: 797 (35.8) |
| Males: 975 (68) | Males: 937 (66.1) | Males: 1,261 (65) | Males: 1,430 (64.2) | |
| ED setting | Community:55 (3.8) | Community: 63 (4.5) | Community: 107 (5.5) | Community: 210 (9.2) |
| Academic: 1,377 (96.2) | Academic: 1,354 (95.5) | Academic: 1,854 (94.5) | Academic: 2,082 (90.8) |
- Data are reported as n (%).
Salary Changes
Table 2 shows that overall salaries increased across the four time points of the data with an overall relative 10.8% (95% confidence interval [CI] = 9.6%–12%) change in median salary between 2013 and 2017; the relative percentage change for female respondents was 10.6% (95% CI = 9.4%–11.8%) and 11.1% (95% CI = 10.2%–12%) for males. During this time period by academic rank, the overall relative salary change (Table 3) was instructor (n = 532) increased by 11.1% (95% CI = 8.45%–13.8%), assistant professors (n = 4,087) by 10.1% (95% CI = 9.0%–11.2%), associate professors (n = 1,507) by 13.1% (95% CI = 11.4%–14.8%), and professors (n = 897) by 13.9% (95% CI = 11.5%–16.4%).
| Salary | ||||
|---|---|---|---|---|
| 2013 (n = 1,432) | 2015 (n = 1,417) | 2016 (n = 1,961) | 2017 (n = 2,292) | |
| All respondents | $223,000 ($204,556–$247,000) | $234,584 ($207,182–$260452) | $236,000 ($209570–$267169.72) | $247,000 ($222,803–$277,026) |
| Female | $217,000 ($193,000–$233,172) | $224,000 ($197,358–$248,981) | $225,222 ($200,889–$255,000) | $240,000 ($217,241–$266,012) |
| Male | $226,746 ($204,538–$247,000) | $240,000 ($210,143–$260,700) | $242,727 ($213,945–$273,276) | $252,000 ($228,286–$281,429) |
- Data are reported as median (IQR).
- IQR = interquartile range.
| Salary | ||||
|---|---|---|---|---|
| 2013 (n = 1,432) | 2015 (n = 1,417) | 2016 (n = 1,961) | 2017 (n = 2,292) | |
| Academic rank | Instructor: $197,532 ($180,444–$225,572) | Instructor: $225,000 ($190,000–$2582,100) | Instructor: $211,088 ($190,632–$247,531) | Instructor: $219,710 ($190,000–$262,032) |
| Assistant Prof: $218,000 ($200,000–$238,231) | Assistant Prof: $226,420 ($225,230–$273,900) | Assistant Prof: $229,750 ($205,500–$252,300) | Assistant Prof: $240,0000 ($220,000–$262,900) | |
| Associate Prof: $235,954 ($216,500–$260,000) | Associate Prof: $249,400 ($224,000–$273,600) | Associate Prof: $253,895 ($228,932–$280,000) | Associate Prof: $267,000 ($245,650–$287,290) | |
| Professor 264,875 ($228,660–$315,526) | Professor: $263,000 ($226,000–$303,347) | Professor: $282,000 ($241,000–$331,740) | Professor: $301,400 ($272,000–$367,575) | |
| Primary clinical responsibility | Clinical: $220,000 ($202,000–$242,515) | Clinical: $230,020 ($203,000–$256,416) | Clinical: $230,000 ($205,000–$260,858) | Clinical: $242,000 ($217,427–$268,450) |
| Director: $224,746 ($205,800–$247,000) | Director: $236,150 ($210,000–$260,700) | Director: $240,000 ($212,175–$269,711) | Director: $252,400 ($229,360–$277,027) | |
| ED setting | Academic: $232,275 ($204,528–$247,000) | Academic: $234,720 ($205,000–260,242) | Academic: $235,100 205,000–265,600) | Academic: $245,000 ($221,030–$274,000) |
| Community: $219,300 ($205,000–228,400) | Community: $229,525 ($215,550–$265,093) | Community: $255,000 ($215,250–$268,077) | Community: $262,955 239,000–283,710) | |
| U.S. region | Northeast: $222,222 ($207,500–250,000) | Northeast: $234,337 ($204,100–265,000) | Northeast: $230,3380 ($199,699–$267,903) | Northeast: $248,444 ($225,250–277,960) |
| South: $230,057 ($216,300–255,000) | South: $237,038 ($217,391–$258,100) | South: $247,822 ($221,800–$271,139) | South: $250,040 ($227,000–$277,800) | |
| Midwest: $223,160 199,300–247,000) | Midwest: $242,204 220,345–257,144) | Midwest: $239,322 ($220,768–269,638) | Midwest: $243,750 ($215,667–278,581) | |
| West: $211,740 ($190,000–$230,000) | West: $200,970 ($164,500–$241,000) | West: $225,000 ($193,125–$260,000) | West: $239,100 ($216,000–$273,500) | |
- Data are reported as median (IQR).
- IQR = interquartile range.
Within survey years, not adjusting for academic rank, the median salary of both men and women increased (Figure 1). However, the overall difference in salary for males was higher, and this difference was significant at all four time points (Z = 6.33, p < 0.001). Between 2016 and 2017, women's salaries increased at a rate of 6.56% compared to male salaries, which increased at a rate of 3.82% at academic departments. Across these time points the proportion of respondents at higher academic ranks (associate and full professor) at higher salaries (see Table 3) was always greater for males than females (2013 males = 39%, females = 22%, Δ = 17% [95% CI = 15.1%–18.9%]; 2015 males = 40%, females = 23.5%, Δ = 16.5% [95% CI = 14.5%–18.5%]; 2016 males = 40.5%, females = 22.5%, Δ = 17.5% [95% CI = 15.8%–19.2%]; 2017 males = 36.6%, females = 22.6%, Δ = 14% [95% CI = 12.6%–15.4%]). Males also reported longer median time as ED faculty than females (males median = 10 years, females = 7 years, Z = 14.1, p < 0.001).
There were also regional differences (Table 3) in reported salaries across survey years with EDs in the South consistently reporting significantly higher salaries than the other regions and EDs in the West reporting significantly lower median salaries than any other geographic region (χ2(3) = 180, p < 0.001). Across regions and survey years EDs that were identified as community consistently had significantly greater median salaries than those described as pure academic (Z = 5, p < 0.001).
Predictive Model
Table 4 shows the results of the mixed‐effects regression model. Main effects and interaction effects were entered in the model. As can be seen, and consistent with the univariate analyses, after covariates in the model were adjusted for there were significant differences in salary by gender, and the nonsignificant interaction effect of gender with time shows that this effect was consistent across all 4 years that the survey was administered, with males’ salaries being significantly higher than females. Time as a nonlinear effect was also modeled independently but this nonsignificant result suggests that there was no change in the linear effect of gender on salary across time.
| Predictor | Unstandardized Estimate | t‐value | p‐value |
|---|---|---|---|
| Intercept | $299,354 | ||
| Gender (female vs. male) | –$11,623 | 4.61 | <0.001 |
| Academic rank | |||
| Professor vs. associate | $45,654 | 19.3 | <0.001 |
| Professor vs. assistant | $70,151 | 29.6 | <0.001 |
| Professor vs. instructor | $83,081 | 25.1 | <0.001 |
| U.S. region | |||
| Midwest vs. West | $28,086 | 13.7 | <0.001 |
| Northeast vs. West | $28,025 | 14.8 | <0.001 |
| South vs. West | $30,592 | 14.7 | <0.001 |
| Community vs. academic ED | $10,925 | 3.95 | <0.001 |
| Years at faculty appointment | $338 | 3.22 | 0.001 |
| Year of survey | $18,311 | 8.4 | <0.001 |
| Gender × year of survey | $3,364 | 1.12 | 0.52 |
| Academic rank × gender | $10,680 | 1.54 | 0.09 |
Contrast effects were estimated between gender at each time point the survey was administered. These analyses supported the modeled effect of lack of significant interaction between gender and time, as at each time point the salary was significantly higher for males than females. It was of interest to see the nonsignificant interaction of gender and academic rank, which suggested that after other covariates were adjusted for (region, academic or community ED, years at academic rank), the gender difference in salary was not modified at any particular academic rank. As is shown, gender, adjusting for the other variables in the model, significantly predicted median adjusted salary, with males earning significantly more than females at all time points of survey administration.
Discussion
In this study, we found that despite previously published data showing an inappropriate gender salary gap in emergency medicine, this gap has remained essentially unchanged over the past 4 years. Gender disparities in academic medicine are pervasive. A 17‐year longitudinal random sample of faculty from 24 U.S. medical schools found that women earned a mean of $20,520 less than men (p = 0.03) and made 90 cents for every dollar earned by their male counterparts. This difference was still $16,982 (p = 0.04) after adjusting for covariates.11 Overall salaries for emergency physicians have increased over the past 4 years, but despite a call to end gender disparities in salary, men still make 18% more than women.13 And a $12,000 gender salary gap remains essentially unchanged. Compensation inequities have been demonstrated in other acute care provider settings including EMS16 and physician assistants.16 The reasons for salary disparities by gender are unclear and unexplained by covariates, but may include the presence of conscious and unconscious biases or initial recruitment negotiation skills despite the medical specialization.17
It is known that the gender pay gap has lifelong financial effects. In most academic institutions there are salary bands based on academic rank. Therefore, a theoretical opportunity to adjust salary disparities at each level of promotion exists. However, our data show that this is not occurring and that the salary disparity is greatest among more senior faculty, that is, associate and full professors. As more women enter the field and are promoted, conscientious chairs will need to prioritize pay parity to change this persistent trend.
As academic health centers start to consolidate and grow, more are developing community practice positions within the faculty practice. Consistent with other published reports, faculty in the community practice areas have high salaries overall compared to traditional academic faculty. This is not surprising as these faculty typically have limited to no academic responsibilities and have salaries that must compete with nonacademic clinical positions. Also, not surprising is the fact that clinical directors earn more than physicians with primary clinical responsibilities. These leadership roles in academic programs are becoming a more valued position as academic practices grow in scope and size.
Regional variation in pay has long been a noted trend in medicine and emergency medicine specifically.13 The West continues to have the lowest salaries. We are unaware of any published research that describes a justification for this variance including payment policy trends, affordability, or demand for resources.
Limitations
There are a several limitations of our study. Chairs and vice chairs were excluded from the analysis. This was because the salaries of these individuals are in many cases are substantially higher than those of the general faculty and therefore have a potential to skew the data. Additionally, the overall number of chairs, vice chairs, and chiefs are relatively small and this has the potential to make some of the information identifiable. Another limitation of the study is that the data have been blinded so that individual faculty members are not specifically identifiable to see how they are progressing in terms of rank or salary from year to year. We are able to examine the impact of the cohort as a whole, but faculty may have been included one year and not the next because of no‐response bias. This is also true for faculty who may have changed institutions during a year and been excluded from being counted in any institution survey, as the survey asked only for responses for people employed for the entire year. Additionally, the tool used to collect the data was refined over the years to add more individual specific data for each reported faculty member, so the earlier years have less data to analyze. The relatively low numbers of respondents from community ED versus academic ED settings also lends us to consider the possibility of bias in the sample and be cautious about interpreting differences between these ED settings.
Conclusion
Despite previously published data showing an inappropriate gender salary gap in emergency medicine, this gap has remained essentially unchanged over the past 4 years. Deliberate strategies should be developed to train academic emergency medicine employers how to prevent gender bias with regards to salary. There is a small positive trend in the larger percentage increase of salaries for women over the period than men, but this is a very small difference and has not made substantial progress in eliminating the overall gap. It may be that not enough time has passed for some departments to work on eliminating this gap. However, this is a long‐standing issue in emergency medicine, and in medicine in general, and without continued review and analysis the potential continues to exist that unconscious or implicit bias will continue.
