P < .001 for both the inpatient and outpatient settings.
Positive values (red) reflect men having higher CMS payments, while negative values (blue) reflect women having higher CMS payments in a given state. The actual differences in mean CMS payments between men and women (per $1000) are displayed, ranging from −$31 483 in Vermont to $101 953 in Louisiana. Hawaii and Alaska are not included.
eTable 1. Centers for Medicare Services for Payments, number of charges, and unique billing codes in 2016 by gender excluding upper and lower 2.5th percentile
eTable 2. Provider and Medicare Beneficiary Characteristics stratified by provider gender
eTable 3. Median Payments in 2016 by Number of Charges, Quintile Groups
eTable 4. Centers for Medicare Services Payments per Charge in 2016 by gender for Top 3 Most Common Unique Billing Codes in the Inpatient Setting
eTable 5. Centers for Medicare Services Payments per Charge in 2016 by gender for Top 3 Most Common Unique Billing Codes in the Outpatient Setting
eTable 6. Payments, number of charges, and unique billing codes in the inpatient setting for cardiologists receiving Medicare payments in 2016 by subspecialty
eTable 7. Payments, number of charges, and unique billing codes in the outpatient setting for cardiologists receiving Medicare payments in 2016 by subspecialty
Customize your JAMA Network experience by selecting one or more topics from the list below.
Raber I, Al Rifai M, McCarthy CP, et al. Gender Differences in Medicare Payments Among Cardiologists. JAMA Cardiol. 2021;6(12):1432–1439. doi:10.1001/jamacardio.2021.3385
Are there gender differences in Medicare reimbursements to cardiologists?
In this cross-sectional analysis of 20 273 cardiologists who received reimbursements from the US Centers for Medicare & Medicaid Services in 2016, men received higher median reimbursements compared with women in both inpatient ($62 897 vs $45 288) and outpatient ($91 053 vs $51 975) settings. The gender difference in payments persisted after adjustment for physician experience and subspecialty, number of charges submitted, and Medicare beneficiary characteristics.
There may be significant gender differences in Centers for Medicare & Medicaid Services payments to and charges submitted by cardiologists, motivating more research into the mechanisms underlying this observation.
Women cardiologists receive lower salaries than men; however, it is unknown whether US Centers for Medicare & Medicaid Services (CMS) reimbursement also differs by gender and contributes to the lower salaries.
To determine whether gender differences exist in the reimbursements, charges, and reimbursement per charge from CMS.
Design, Setting, and Participants
This cross-sectional analysis used the CMS database to obtain 2016 reimbursement data for US cardiologists. These included reimbursements to cardiologists, charges submitted, and unique billing codes. Gender differences in reimbursement for evaluation and management and procedural charges from both inpatient and outpatient settings were also assessed. Analysis took place between April 2019 and December 2020.
Main Outcomes and Measures
Outcomes included median CMS payments received and median charges submitted in the inpatient and outpatient settings in 2016.
In 2016, 17 524 cardiologists (2312 women [13%] and 15 212 men [87%]) received CMS payments in the inpatient setting, and 16 929 cardiologists (2151 women [13%] and 14 778 men [87%]) received CMS payments in the outpatient setting. Men received higher median payments in the inpatient (median [interquartile range], $62 897 [$30 904-$104 267] vs $45 288 [$21 371-$73 191]; P < .001) and outpatient (median [interquartile range], $91 053 [$34 820-$196 165] vs $51 975 [$15 622-$120 175]; P < .001) practice settings. Men submitted more median charges in the inpatient (median [interquartile range], 1190 [569-2093] charges vs 959 [569-2093] charges; P < .001) and outpatient settings (median [interquartile range], 1685 [644-3328] charges vs 870 [273-1988] charges; P < .001). In a multivariable-adjusted linear regression analysis, women received less CMS payments compared with men (log-scale β = –0.06; 95% CI, –0.11 to –0.02) after adjustment for number of charges, number of unique billing codes, complexity of patient panel, years since graduation of physicians, and physician subspecialty. Payment by billing codes, both inpatient and outpatient, did not differ by gender.
Conclusions and Relevance
There may be potential differences in CMS payments between men and women cardiologists, which appear to stem from gender differences in the number and types of charges submitted. The mechanisms behind these differences merit further research, both to understand why such gender differences exist and also to facilitate reductions in pay disparities.
Gender differences in salaries are prevalent among physicians across medical specialties, and cardiology is no exception.1,2 Men cardiologists earn higher salaries compared with women, even after adjustment for physician and practice characteristics.2 Seemingly small differences in salaries can accumulate over time, evidenced by the lifetime gender wage gap in cardiology that is estimated at $2.5 million.3
As a growing proportion of individuals in the US are insured by Medicare and/or Medicaid (24% in 2008 and 34% in 2018),4 reimbursement for services provided by these programs represents an increasingly important source of revenue for clinicians. Prior studies have evaluated gender differences in payments from US Centers for Medicare & Medicaid Services (CMS).5-7 Mahr et al7 found that CMS payments and charges were higher for men than women across most specialties, including cardiology in 2013. We sought to determine whether gender differences exist in the charges submitted to and reimbursements received from CMS among men and women in cardiology and to further examine if these differences are influenced by practice location, geography, physician experience, physician subspecialty, and patient panel complexity in a more contemporary data set.
This cross-sectional analysis used publicly available data from the CMS Physician and Other Supplier Public Use File in 2016.8 This database contains the national provider identifier (NPI) of each physician submitting charges to Medicare and is linked to the physicians’ codes submitted through the Healthcare Common Procedure Coding System. The Physician and Other Supplier Public Use File only includes claims submitted under the Medicare fee-for-service program and does not contain information for beneficiaries in the Medicare Advantage plan. In 2016, 31% of Medicare patients were enrolled in a Medicare Advantage plan.9 Geographic location of practice and self-reported physician gender is attached to each NPI number. The CMS database does not report other sources of income for physicians. Individuals with 10 or fewer charges submitted are excluded from the database. Years since graduation for each physician was obtained from the National Downloadable File and merged with the Physician and Other Supplier Public Use File by linkage with physician NPI number.10 Medicare beneficiary characteristics were collected from the Medicare Physician and Other Supplier NPI Aggregate Report, 2016.11 As all data are publicly available and because no individual patient data were accessed, institutional review board approval and informed patient consent were not required. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
The mean (SD), median (interquartile range [IQR]), and maximum CMS payments, total charges, and number of unique billing codes were reported for men and women in the inpatient and outpatient settings in 2016. Differences in payments, charges, and unique billing codes between men and women were compared using Kruskal-Wallis testing when presenting median (IQR) values (with effect estimates determined by median regression) and t test when comparing mean (SD) values. To remove outliers, differences were again compared in a sensitivity analysis excluding the highest and lowest 2.5% earning physicians based on total CMS payments. The proportion of Medicare beneficiaries with heart failure, diabetes, hyperlipidemia, hypertension, and ischemic heart disease and the mean age of Medicare beneficiaries were compared using χ2 test for categorical variables and t test for continuous variables stratified by physician gender. Physician characteristics including years since graduation and cardiology subspecialty were compared. A multivariable-adjusted linear regression analysis was performed to determine the association of gender with payments, number of charges, and unique billing codes. Payments and number of charges were log transformed to ensure normality of the data; therefore, β coefficient values reported in our results section are on a log scale. Model covariates for linear regression included percent of patients with heart failure, diabetes, hyperlipidemia, hypertension, ischemic heart disease; years since graduation of physicians; and physician subspecialty. To assess the influence of coding volume on payment differences among men and women, cardiologists were divided into quintiles based on number of codes submitted to CMS. Median and mean payments were compared among men and women in each quintile. In addition, the 3 most common unique billing codes generated by cardiologists in the inpatient and outpatient settings were determined. For each of these 3 billing codes, the median payments per charge were compared for men and women in the inpatient and outpatient settings. Furthermore, a sensitivity analysis was performed to understand the influence of subspecialties with high procedure volume and with greater gender imbalance than general cardiology (electrophysiology and interventional cardiology) on gender differences in CMS payments. A state-by-state analysis of gender differences in mean CMS payments was performed and a US heat map was generated to show the geographic distribution of payment differences. A 2-sided P value less than .05 was considered statistically significant. All analysis was carried out using Stata/IC version 13.1 (StataCorp). Analysis took place between April 2019 and December 2020.
There were 20 273 cardiologists who received CMS payments in 2016 and met the study criteria. There were 17 524 cardiologists who received CMS payments in the inpatient setting, of whom 15 212 (87%) were men and 2312 (13%) were women. There were 16 929 cardiologists who received CMS payments in the outpatient setting, of whom 14 778 (87%) were men and 2151 (13%) were women.
In the inpatient setting, men received a mean of 45% ($24 229 [95% CI, $21 349-$27 109]; P < .001) more annual CMS payments compared with women in 2016 (mean [SD] payments, $77 421 [$68 461] vs $53 192 [$44 743]; P < .001) (Table 1). Similarly, the median payment was 39% higher ($17 609; [95% CI, $14 711-$20 508]; P < .001) for men compared with women (median [IQR], $62 897 [$30 904-$104 267] vs $45 288 [$21 371-$73 191]) (Figure 1). In the outpatient setting, men received a mean of 62% ($62 306 [95% CI, $52 255- $72 357]; P < .001) higher payments compared with women in 2016 (mean [SD] payments, $157 425 [$232 552] vs $95 119 [$130 411]). The median outpatient payment was also 75% ($39 078 [95% CI, $33 221-$44 934]; P < .001) more for men than women (median [IQR], $91 053 [$34 820-$196 165] vs $51 975 [$15 622-$120 175]). When the upper and lower 2.5% earning physicians were excluded from the analysis, there remained a statistically significant difference in the mean payments received by men and women in the inpatient setting (mean [SD], $71 627 [$50 527] vs $53 005 [$38 987]; mean difference, $18 622 [95% CI, $16 433-$20 811]; P < .001) and outpatient setting (mean [SD], $133 767 [$131 648] vs $91 800 [$107 142]; mean difference, $41 967 [95% CI, $36 010-$47 925]; P < .001) (eTable 1 in the Supplement).
Men submitted 24% (231 [95% CI, 171-289]; P < .001) more median charges in the inpatient (median [IQR], 1190 [569-2093] vs 959 [440-1696]) and 94% (815 [95% CI, 715-915]; P < .001) more median charges in the outpatient settings (median [IQR], 1685 [644-3328] vs 870 [273-1988]) (Figure 2). Men also submitted slightly more unique billing codes compared with women in the inpatient setting (median [IQR], 10 [6-15] vs 9 [5-13]; median difference, 1.00 [95% CI, 0.71-1.29]; P < .001) and outpatient setting (11 [6-17] vs 8 [4-13]; median difference, 3.00 [95% CI, 2.44-3.56]; P < .001).
Men cardiologists had more years in practice than women cardiologists; the mean years since graduation was higher for men than women in the inpatient setting (mean [SD], 22.6 [10.3] years vs 17.7 [8.1] years; mean difference, 4.9 [95% CI, 3.5-6.4]; P < .001) and in the outpatient setting (mean [SD], 23.1 [10.1] years vs 18.7 [9.5] years; mean difference, 4.4 [95% CI, 3.4-5.4]; P < .001). The proportion of physicians listed within each cardiology subspecialty is listed in eTable 2 in the Supplement.
The mean age of Medicare beneficiaries was marginally higher for men cardiologists than women cardiologists in the inpatient setting (mean [SD], 74.0 [3.0] years vs 73.2 [3.6] years; mean difference, 0.8 [95% CI, 0.7-1.0]; P < .001) and in the outpatient setting (mean [SD], 74.3 [2.6] years vs 73.4 [3.1] years; median difference, 0.9 [95% CI, 0.8-1.0]; P < .001). The proportion of Medicare beneficiaries with heart failure, diabetes, hyperlipidemia, hypertension, and ischemic heart disease stratified by physician gender is listed in eTable 2 in the Supplement. Patients cared for by men cardiologists had marginally higher rates of diabetes, hyperlipidemia, hypertension, and ischemic heart disease but lower rates of heart failure.
In a multivariable-adjusted linear regression analysis, women cardiologists received less mean CMS payments compared with men (β = −0.06; 95% CI, –0.11 to –0.02) after adjustment for number of charges, number of unique billing codes, complexity of patient panel, years since graduation of physicians, and physician subspecialty (Table 2). The number of charges and number of unique billing codes were also less for women than men (β = −0.35; 95% CI, –0.45 to –0.24 and β = −1.63; 95% CI, –2.25 to –1.02, respectively) again after adjustment for complexity of patient panel, years since graduation of physicians, and physician subspecialty. To ease interpretation of these β coefficients, Table 2 also reports the geometric mean for each point estimate among women, expressed as a percentage of the geometric mean among men. For example, the adjusted β coefficient for CMS payments of −0.06 translates to a ratio of geometric means of 0.94, which can be interpreted as women receiving a mean of 94% of the CMS payments received by men after adjustment.
Physicians were grouped into quintiles based on number of charges submitted in the inpatient setting with group 1 being the group with the lowest charges submitted (11-400 charges) and group 5 being the group with the highest number of charges (≥2282 charges) (eTable 3 in the Supplement). In the inpatient setting, there were 3216 men (91%) and 324 women (9%) in the highest charge group. The median (IQR) payment for the highest charge group was $129 732 ($94 742-$181 432) for men and $100 463 ($75 657-$138 041; median difference, $29 150 [95% CI, $19 689-$38 611]) for women (P < .001). Across all charge groups, the median payment was higher for men compared with women (charge group 1: $1898 [95% CI, $410-$3386]; P < .001; charge group 2: $7272 [95% CI, $4663-$9882]; P < .001; charge group 3: $6371 [95% CI, $2808-$9933]; P < .001; charge group 4: $15 574 [95% CI, $10 957-$20 190]; P < .001).
In the outpatient setting, the quintile with the lowest charges submitted had submitted 11 to 389 charges, and the highest quintile submitted 3687 or more charges. There were 3182 men (94%) and 201 women (6%) in the group with the highest charges. The median payments were not significantly different for men and women within each charge group in the outpatient setting (eTable 3 in the Supplement).
The 3 most common billing codes in the inpatient setting were code 99232 (subsequent hospital inpatient care, typically 25 minutes per day), code 99223 (initial hospital inpatient care, typically 70 minutes per day), and code 93306 (ultrasonographic examination of heart including color doppler blood flow rate, direction, and valve function) (eTable 4 in the Supplement). The median payments per charge for each of the 3 most common inpatient charges were numerically equivalent among men and women (eTable 4 in the Supplement).
The most common outpatient billing codes were code 93000 (routine electrocardiogram), code 99214 (established patient office or other outpatient visit, typically 25 minutes), and code 99213 (established patient office or other outpatient visit, typically 15 minutes) (eTable 5 in the Supplement). Once again, the median payments per charge for each of the 3 most common outpatient charges were numerically equivalent among men and women (eTable 5 in the Supplement).
A sensitivity analysis was performed, which found that gender differences in CMS payments persisted after excluding electrophysiology and interventional cardiology, subspecialties with high procedure volume and with greater gender imbalance than general cardiology. When categorized according to general cardiologists, electrophysiologists, and interventional cardiologists practicing in the inpatient (eTable 6 in the Supplement) and outpatient settings (eTable 7 in the Supplement), men received higher mean payments compared with women.
A state-by-state analysis was performed to evaluate the geographic distribution of payment differences between genders (Figure 3). The largest gender difference in mean CMS payments was observed in Louisiana where men cardiologists earn 235% ($101 953) more than women (mean [SD], $145 323 [$207 022] vs $43 370 [$39 024]). In Vermont, women cardiologists earned a higher mean CMS payment by 38% ($31 483) compared with men cardiologists (mean [SD], $81 995 [$72 619] vs $49 559 [$62 770]).
In this study of CMS reimbursements to cardiologists in 2016, we found a number of differences in CMS payments between men and women. First, we found that men receive higher payments from and submit more charges to CMS compared with women in both the inpatient and outpatient settings. This difference persisted after removal of outliers and importantly after adjustment for the characteristics of Medicare beneficiaries, number of charges submitted, number of unique billing codes used, and physician experience and subspecialty. Second, when divided into quintiles based on charges submitted, men received higher median payments across all charge groups in inpatient settings, although this pattern was not observed in the outpatient practice setting. Third, differences in CMS payments persisted in a sensitivity analysis for cardiology subspecialties with high procedural volumes (electrophysiology and interventional cardiology) that tend to have greater gender imbalances than general cardiology. Finally, when directly comparing payments per charge for the same unique billing codes (specifically the 3 most common charges applied in 2016 in the inpatient and outpatient settings), there is no difference in payments between men and women.
Why do gender differences in CMS payments to cardiologists exist? We believe there are several potential contributing factors. Perhaps most importantly, the differences in payments appears to be due to a difference in coding patterns or clinical practices and not a CMS reimbursement gender bias. Our findings suggest that women are submitting fewer charges compared with men. Accordingly, this will affect the total value of CMS payments to men and women. The lower number of charges submitted by women may reflect different practice styles between men and women. It could also reflect differences in the volume of patients seen and/or differences in assigned service responsibilities. A 2020 study found that female primary care physicians generated less revenue owing to a lower number of patient visits but spent more time with direct patient contact than their male counterparts.12 Prior studies have similarly demonstrated that women have longer clinic visits and spend more time communicating with patients and interacting with the electronic health record.13,14 The differences in charges may also reflect differences in testing and treatments performed by men vs women cardiologists. Indeed, the types of charges (ie, billable activity type) appear to contribute independently to the gender payment difference, demonstrated by the persistence of a difference in CMS payments after adjustment for the number of charges (ie, volume of billing) and number of unique billing codes (ie, diversity of billing) of each physician. While these billing differences may generate revenue differences, they do not necessarily translate into quality-of-care variability. For example, a prior study of Medicare patients found that inpatients with female physicians had marginally better outcomes than those with male physicians.15 Thus, it would be a mistake to assign employment priority or advancement based on billed charges only.
There are additional variables that could hypothetically cause the gender differences observed in our analysis. Women cardiologists may care for different patient populations than men in metrics not captured by the database, such as young patients (who are not represented in the Medicare database and often have a lower acuity of disease). There may be organizational barriers that affect women’s clinical volumes including less access to support from health care personnel. Women may have increased obligations in nonremunerated work than men, including mentoring, teaching, and serving on committees. Female physicians report spending more time on these citizenship tasks than male physicians.16 Some of the CMS payment differences may be a reflection of differences in part-time vs full-time work between men and women. A prior analysis found a 10% difference in the proportion of men and women working full time in cardiology.2 An analysis adjusted for work-relative value units by gender may represent the best test of gender differences in reimbursement for the same work and same volume. It is possible that the differences observed in CMS reimbursement reflect, at least in part, gender differences in the approach to care, with male cardiologists more likely to pursue care options that generate higher billing charges (eg, invasive angiography and stenting for stable angina) but female cardiologists are more likely to pursue care that generates less reimbursement (eg, optimal medical therapy for stable angina). While our data are hypothesis generating in this regard, requiring validation and motivating further study, we did find that the types of charge contributed to gender differences in CMS reimbursement, independently of other factors.
In the inpatient setting, we found differences in CMS payments persisted among men and women even among physicians who submitted the highest quintile number of charges. However, in the outpatient setting, there was no gender difference in CMS payments among physicians who submitted a similar number of charges (ie, in analyses comparing payment differences by gender among categories of charge quintiles). This suggests that the payment difference by gender may be driven primarily by number of charges in the outpatient setting and by both types of charge and number of charges in the inpatient setting. Interestingly, these patterns appear to be specialty dependent. In orthopedic surgery, among physicians who submitted a high number of charges, there was no gender difference in payments received.17 In radiation oncology, there was a gender difference in payments among the highest-charging physicians in both the outpatient and inpatient setting.5
This study has important limitations. The CMS database does not report work characteristics (eg, full time vs part time, clinical assignments vs administrative assignments) and quality of care (eg, adherence to guideline-directed therapies), and thus it was not possible to assess the association of any of these potential confounders with CMS charges and payments. While certain Medicare beneficiary characteristics were included in the database, the patient-level data are not all inclusive, and thus important differences in patient panels between men and women not accounted for in our analysis may play a role in the apparent gender pay differences. Time spent per patient encounter is not available in the CMS database, and this very important driver of billed charges is also not available for testing. Furthermore, we could not ascertain the appropriateness of billing practices. The CMS database includes 1 subspecialty listing per physician, and physicians who practice within multiple subspecialties (eg, general cardiology and critical care) may not have the full scope of their practice represented in the analysis. Lastly, Medicare payments account for only a portion of physicians’ salaries; we did not assess additional sources of income nor have access to salary adjustments to account for nonremunerative work. We acknowledge the possibility that, with the incorporation of additional data not available to us for the present analysis, a more fully adjusted analysis of CMS payment by gender may fail to substantiate our findings. Thus, our work establishes an important hypothesis and provides a direction for future study.
Women practicing cardiology may receive less CMS payments and submit fewer CMS charges than men in inpatient and outpatient practice settings. The apparent gender difference in payments persisted after adjustment for physician experience and subspecialty, number of charges submitted, number of unique billing codes used, and Medicare beneficiary characteristics. However, other important mitigating circumstances likely affect total CMS payment by gender. Further research is required to definitively identify gender differences in reimbursement and to understand why potential gender-associated differences in payment exist.
Corresponding Author: Inbar Raber, MD, Cardiovascular Institute, Beth Israel Deaconess Medical Center, Harvard Medical School, 330 Brookline Ave, Boston, MA 02215 (firstname.lastname@example.org).
Accepted for Publication: July 10, 2021.
Published Online: September 8, 2021. doi:10.1001/jamacardio.2021.3385
Author Contributions: Drs Raber and McEvoy had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Raber, Al Rifai, McCarthy, Wood, Smyth, McEvoy.
Acquisition, analysis, or interpretation of data: Raber, Al Rifai, Vaduganathan, Michos, Ibrahim, Asnani, Mehran, McEvoy.
Drafting of the manuscript: Raber, Al Rifai, Smyth.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Raber, Al Rifai.
Administrative, technical, or material support: McEvoy.
Supervision: Wood, Smyth, Ibrahim, Asnani, Mehran, McEvoy.
Conflict of Interest Disclosures: Dr Vaduganathan has received research grant support or served on advisory boards for American Regent, Amgen, AstraZeneca, Bayer AG, Baxter Healthcare, Boehringer Ingelheim, Cytokinetics, Lexicon Pharmaceuticals, Relypsa, and Roche Diagnostics; speaker engagements with Novartis and Roche Diagnostics; and participates on clinical end point committees for studies sponsored by Galmed and Novartis. Dr Michos has served on advisory boards for Novartis, Esperion, Amarin, and AstraZeneca. Dr Smyth has received presentation honoraria from Amgen, Bayer, Daiichi Sankyo, Menarini, and Novartis outside the submitted work. Dr Ibrahim has received presentation honoraria from Novartis and Roche Diagnostics. Dr Asnani reported grants from the National Institutes of Health and consults for Cytokinetics and AstraZeneca outside the submitted work. Dr Mehran has received institutional research funding from AstraZeneca, Bayer, Beth Israel Deaconess Medical Center, Bristol Myers Squibb/Sanofi, CSL Behring, DSI, Eli Lilly and Company/Daiichi Sankyo, Medtronic, Novartis, and OrbusNeich; is a consultant to Boston Scientific, Abbott Vascular, Medscape, Siemens Medical Solutions, Regeneron Pharmaceuticals (no fees), Roivant Sciences, and Sanofi; is an institution consultant (payment to institution) with Abbott Vascular and Spectranetics/Philips/Volcano Corporation; is on the executive committee for Janssen Pharmaceuticals and Bristol Myers Squibb; receives institutional (payment to institution) advisory board funding from Bristol Myers Squibb and Novartis; data and safety monitoring board membership funding to institution from Watermark Research; has less than 1% equity with Claret Medical (part of Boston Scientific, a company that makes medical devices) and less than 1% equity with Elixir Medical (a company that makes drug-eluting stents). No other disclosures were reported.