The Political Polarization of Physicians in the United States: An Analysis of Campaign Contributions to Federal Elections, 1991 Through 2012

Importance  Few current data are available regarding the political behavior of American physicians as the number of female physicians has increased and the number of solo practitioners has decreased.

Objective  To analyze campaign contributions that physicians made from the 1991 to 1992 through the 2011 to 2012 election cycles to Republican and Democratic candidates in presidential and congressional races and to partisan organizations, including party committees and super political action committees (Super PACs).

Design, Setting, and Participants  We explored partisan differences in physician contributions by sex, for-profit vs nonprofit practice setting, and specialty using multiple regression analysis. We studied the relation between the variation in the mean annual income across specialties and the mean percentage of physicians within each specialty contributing to Republicans.

Main Outcomes and Measures  Differences in contributions to Republicans and Democrats, for all physicians and for subgroups.

Results  Between the 1991 to 1992 and the 2011 to 2012 election cycles, physician campaign contributions increased from $20 million to $189 million, and the percentage of active physicians contributing increased from 2.6% to 9.4%. Of physicians who contributed during the study period, the mean percentage contributing to Republicans was 57% for men and 31% for women. Since 1996, the percentage of physicians contributing to Republicans has decreased, to less than 50% in the 2007 to 2008 election cycle and again in the 2011 to 2012 election cycle. Contributions to Republicans in 2011 to 2012 were more prevalent among men vs women (52.3% vs 23.6%), physicians practicing in for-profit vs nonprofit organizations (53.2% vs 25.6%), and surgeons vs pediatricians (70.2% vs 22.1%). In 1991 to 1992, these contribution gaps were smaller: for sex, 54.5% vs 30.9%; for organizations, 54.2% vs 40.0%; and for specialty, 65.5% vs 32.7%. The percentage of physicians contributing to Republicans across specialties correlated 0.84 with the mean log earnings of each specialty; specialties with higher mean earnings had higher percentages of physicians contributing to Republicans.

Conclusions and Relevance  Between 1991 and 2012, the political alignment of US physicians shifted from predominantly Republican toward the Democrats. The variables driving this change, including the increasing percentage of female physicians and the decreasing percentage of physicians in solo and small practices, are likely to drive further changes.

Although few systematic analyses have been conducted on the political behavior of physicians in the United States, it is often assumed that they sit to the right on the political spectrum. Generalizing from the American Medical Association’s strong opposition to the 1965 passage of Medicare, the belief is that physicians share the wariness of Republicans about government interventions, particularly in health care.^1,2 Ostensibly, this outlook persists today.³

Given the scarcity of data and alert to the many changes in the composition and organization of the physician workforce, we examined physician contributions to presidential and congressional political campaigns from 1991 to 2012. Information on campaign contributions to federal elections is publicly available. The data illuminate patterns of support of physicians for Democratic and Republican candidates and how these patterns compare to those for all donors.

Between 1991 and 2012, campaign contributions in the United States increased substantially. Inflation-adjusted to 2012 dollars, contributions from all individuals increased from $716 million in 1991 to 1992 to $4.64 billion in 2011 to 2012, a 6.5-fold increase. Contributions from physicians increased at a greater rate, from $20 million to $189 million, or by nearly 9.5 fold.

We grouped contributions by the 2-year congressional election cycles. There are important differences in voter participation between midterm election years and presidential election years; for example, people with low incomes are less likely to vote in midterm elections.^4(pp130-133) Contributions also varied between midterm and presidential years, with greater contributions in presidential years.

Over our 22-year study period, the composition of the medical profession changed—most notably, there were more female physicians and fewer solo practitioners^5-7—and politics in the United States became increasingly polarized.⁴ We hypothesized that the increased number of female physicians and the changes in medical practice altered the patterns of political partisanship within the profession.

Because all data came from public sources, this study was not human subject research and did not require IRB review or informed consent. We obtained data on contributions to individual candidates or party-connected organizations in primary and general elections from the Federal Election Commission (FEC); contributions to candidates or organizations, including super political action committees (Super PACs), that total $200 or more over a 2-year election cycle must be reported to the FEC. Contributions to political advocacy groups filing as 527 organizations and tax-exempt nonprofit 501(c)(3) organizations are reported to the Internal Revenue Service; we obtained these data from the Database on Ideology, Money in Politics, and Elections, which augments^8,9 the raw FEC data about donors with geocoding and identity resolution (see eAppendix in Supplement for explanations). The geocoding software permitted us to map the address of the physician to his or her congressional district. Identity resolution software groups contributor records that differ, for example, in spelling, omitted middle names, or the use of initials in place of middle names. We aggregated the contributions of each individual across all candidates and organizations, calculating the total given to Democrats and to Republicans in each election “cycle,” the 2-year period comprising a federal election year and the year preceding it.

Our basic unit of observation is a physician in an election cycle. Of the 140 423 physicians who contributed in at least 1 of the 11 election cycles spanning 1991 through 2012, 71 822 contributed in 1 election cycle, 28 914 in 2, and 39 687 in 3 or more.

Our dependent variable is a physician's contributions to Republicans as a percentage of his or her total contributions to candidates or affiliated organizations. We identified candidates’ political party through the required reports that they file with the FEC. For nominally nonpartisan organizations that are in practice affiliated with Republicans or Democrats (eg, Super PACs), we coded partisanship on the basis of estimates of the liberalism or conservatism of their donors. (See eAppendix in Supplement for fuller discussion.)

Almost all values of the dependent variable are either 0 or 100 because few physicians contribute to both Republicans and Democrats. Among physicians making 2 or more contributions in an election cycle, 2.7% split their contributions during the 2012 election cycle, compared with 5.9% who split their contributions during the 1992 election cycle. Percentage contributions to Republican candidates in a specific election cycle include donations from physicians who split their contributions. For example, the rare donor who split contributions 50-50 between parties would be a 50% Republican donor and a 50% Democratic donor.

We identified physicians through the National Provider Identifier (NPI) public use file from the National Plan and Provider Enumeration System. The file contains records on all individual and organizational health care providers covered by the Health Insurance Portability and Accountability Act. To access records prior to the creation of the NPI database in 2006, we used its precursor, the Unique Physician Identifier Number (UPIN) database, which extends back to 1992.

Our independent variables are physician sex, specialty, employment status, and the partisanship of the congressional district in which the physician practices. The measure of partisanship was the Republican share of a district's 2-party presidential vote. For midterm election cycles, we use the district’s vote share in the preceding presidential election cycle. Identification of physician sex is from the NPI or an automated coding scheme described in the eAppendix (in Supplement). Specialty and employment information are from the NPI or UPIN databases, which include detailed specialty codes for all physicians from 1992 to present. (See eAppendix in Supplement for the specialties excluded from our analysis.) We used both the size (small and large) and the for-profit or nonprofit status of the physician’s organization as variables related to employment status. We defined large practices as organizations employing 10 or more physicians; this group consists primarily of hospitals, health maintenance organizations, government agencies, and large group practices. We classified all physicians not employed by large employers, including those reporting themselves as self-employed, as in small practices. Thus, our employment status variable had 4 categories: large for-profit, large nonprofit, small for-profit, and small nonprofit.

We used contributors’ addresses as entered in the contribution database to locate their congressional districts. We then assigned our measure of district partisanship (the Republican 2-party vote share in the most recent presidential election) to each physician in each election cycle. We standardized the variable with a mean (SD) of 0 (1). Aggregating over all observations, the mean (SD) Republican vote share was 44.7% (14.9%). Thus, the partisan variable would have a value of 0 in a district that voted 44.7% Republican and 1 in a district that voted 59.6% Republican.

We used ordinary least-squares multivariate linear regression. The dependent variable was the percentage given to Republicans. Physician donors in multiple election cycles were treated as multiple observations. From election to election, short-term effects can change contribution patterns for many types of physicians. For example, it is possible that physician contributions in 2007 to 2008 may have shifted away from Republicans and toward the Democrats because of the unpopularity of the Iraq war and other short-term factors. In our regression, we adjusted for such effects by allowing the regression constant to assume a different value in each election cycle. “Fixed effect” is the technical term frequently used to describe such a varying constant.

Table 1 shows physicians’ contributions in each 2-year election cycle by the number of physicians and the percentages of all active physicians, measured as the number of physicians in the NPI or UPIN database for the even year in the cycle. The percentage of active physicians contributing to federal election campaigns increased from 2.6% in the 1991 to 1992 election cycle to 9.4% in the 2011 to 2012 election cycle. In general, percentages and means in the results, tables, and figures are based on the number of active physicians who contributed to federal election campaigns, not the total number of active physicians.

Table 2 shows aggregated physician contributions to Republican candidates from the 1991 to 1992 through 2011 to 2012 election cycles by sex and for 50 specialties. (Because the findings are very similar if they are weighted by dollar amount given, only the unweighted data are shown.) The mean percentage of physicians contributing to Republicans was 57% for men and 31% for women. Data for the 2012 election cycle do not include the $43 million given to Republicans by Dr Miriam Adelson, the spouse of casino owner Sheldon Adelson. Most of her contributions were to political advocacy groups filing as 527 organizations and Super PACS; such contributions have no limits. Dr Adelson was classified by NPI as a “miscellaneous provider,” and we excluded all miscellaneous providers from our analysis. However, if Dr Adelson’s contributions had been included, the percentage of money contributed by female physicians to Republican candidates would have increased from 27% to 56% because her contribution was so large.

Table 3 shows the aggregated physician contributions to Republican candidates by sex for the employers who had the largest number of physicians who made campaign contributions and who reported the employer as their primary employer, according to the NPI or UPIN database.

Between the 1991 to 1992 and 2011 to 2012 election cycles, substantially more physicians contributed to Republicans than to Democrats. Since 1996, however, the percentage of physicians contributing to Republicans has declined, decreasing to less than 50% in the 2007 to 2008 election cycle, when President Obama was first elected (Figure 1). In the 2009 to 2010 election cycle, as Congress debated and voted on the Affordable Care Act, the percentage of physicians contributing to Republicans increased. The percentage then decreased again in the 2011 to 2012 election cycle.

The change in the pattern of physician contributions is similar to that in the general population of donors, although the trend toward Democrats has been more pronounced among physicians than in the general population. Until 2000, the percentage of physician contributions to Republicans was consistently higher than the percentage for all contributors. In the last 3 election cycles, however, the percentage of physician contributions to Republicans has matched that of all contributors. Results are similar for the percentage of all dollar amounts given to Republicans (data not shown). The partisan affiliations of physicians who donate small amounts of money are similar to the affiliations of those who donate large amounts of money (data not shown).

The political alignments of physicians are increasingly shaped by sex, employment type, and medical specialty. When employment type and specialty are considered separately for men and women, female physicians are far more likely to contribute to Democrats than their male counterparts. When employment status and specialty are controlled for, the difference persists. In fact, women were more likely than their male colleagues to contribute to Democrats than to Republicans in all 50 specialties that we analyzed (Table 2). Also, with 4 exceptions among the specialties, the percentage of the total dollar amount that female physicians contributed to Republicans was 50% or less. By comparison, in 30 specialties, the percentage of the total dollar amount that male physicians contributed to Republicans was greater than 50%. For example, although women are well represented among both dermatologists and pediatric endocrinologists, 45% of female dermatologists making campaign donations contributed to Republicans in the 11 election cycles, compared with 23% of female pediatric endocrinologists. However, within both of these specialties, the mean percent contributions by women to Republicans were substantially less than the mean percent contributions by men, 45% vs 63% for dermatologists and 23% vs 30% for pediatric endocrinologists, respectively.

Figure 2 shows the percentage of physician campaign contributions to Republicans, by sex, employment status, and for surgeons and pediatricians, without controlling for other variables. For each of these categories, the trends observed were similar to those for all physicians (Figure 1). Since 1996, the percentage of campaign contributions to Republicans has declined and the percentage to Democrats has increased. In the 2009 to 2010 election cycle, as Congress acted on the Affordable Care Act, the percentage of contributions to Republicans increased in each category. Among surgeons, however, the trend toward donations to Democrats was minimal; approximately 75% of surgeons who made campaign contributions contributed to Republicans in the 2011 to 2012 election cycle. By comparison, approximately 25% of pediatricians contributed to Republicans in the same cycle. Between the 1991 to 1992 and 2011 to 2012 election cycles, the gap between the percentage of contributions to Republicans for surgeons and pediatricians widened considerably, as did the gaps between physicians working in for-profit and nonprofit organizations and, to a lesser extent, between male and female physicians. In sum, the political preferences of physicians are increasingly polarized when analyzed by sex, employment type, or specialty.

Multiple regression analysis also found that physician contributions differ sharply when the independent variables of sex, employment type, specialty, and the Republican presidential vote share in a congressional district are taken into account (Figure 3). For the regression, we assigned the reference physician a value of zero on all the independent variables. Such a reference physician is a female psychiatrist employed by a large nonprofit organization and located in a congressional district in which 46.3% of the votes in 2004 were cast for President George W. Bush. Because the regression coefficients are all positive, a physician of the reference type had the least likelihood of contributing to a Republican candidate. Thus, only 3% were estimated to have contributed to Republicans. (The 3% corresponds to the value of the fixed effect for 2004.) In contrast, 88% of male orthopedic spine surgeons working in a small for-profit organization and located in a congressional district that voted 61.1% for Bush in 2004 were estimated to have contributed to Republicans.

After controlling for the effects of the other independent variables, we found that male physicians were more likely to contribute to Republicans than female physicians. The gap between the sexes in the percentage of donations to Republicans was 16%; the gap in the percentage of donations to Republicans between physicians employed in small for-profit and large nonprofit organizations (less to Republicans in nonprofits) was 17%; and the gap between the percentage of donations to Republicans by orthopedic surgeons and pediatricians (less to Republicans by pediatricians) was 30% (Figure 3).

Partisanship in physician contributions also reflects the political preferences of the physicians’ congressional districts. Physicians located in congressional districts that vote Republican, after the other independent variables were controlled for, were more likely to contribute to Republicans. We estimated that a 1 standard deviation shift in the presidential vote (1 unit of our standardized variable, or approximately 15% of the actual vote) was accompanied by a 10% increase in the mean percentage of physicians donating to Republicans.

Figure 4 shows how the variation in mean income across specialties correlates with the relationship between physician specialty and partisanship. The percentage of physicians contributing to Republicans across the specialties varied almost exactly with the mean log earnings of each specialty (r = 0.84). Contributions to Republicans were aggregated for all 11 election cycles; the earnings data are a 6-year average as reported in a survey conducted in September 2011.¹⁰ The estimated tripling in the percentage of donations to Republicans associated with an increase in annual earnings from $200 000 to $600 000 is greater than the effects of income on partisanship among all voters. For example, in the 2012 presidential election, voters reporting incomes less than $30 000 were just 19% less likely to vote for Mitt Romney, the Republican candidate, than were those with incomes greater than $100 000.¹¹

To better understand the shift in physician contributions from Republican candidates and organizations to Democratic ones, we compared contributors in 2 midterm election cycles, the 1993 to 1994 and 2009 to 2010 election cycles. In both midterm elections, Republicans captured control of the House of Representatives. In both cycles, a short-term surge in Republican giving occurred when Democratic presidents—Bill Clinton in 1994 and Barack Obama in 2010—promoted health care reform. We found similar results when comparing the 1995 to 1996 and 2011 to 2012 election cycles or other pairs (data not shown).

To better understand why the percentage of physician contributions to Republicans decreased from the 1993 to 1994 to the 2009 to 2010 election cycle, we assigned each contributor to 1 of 8 groups {male, for-profit employment, high-pay specialty}, {male, for-profit, low pay}, {male, nonprofit, high pay}, {male, for-profit, low pay}, {female, for-profit, high-pay specialty}, {female, for-profit, low pay}, {female, nonprofit, high pay}, {female, for-profit, low pay}. We divided specialties into high-paying and low-paying categories on the basis of a break in the data, shown in Figure 4 at $350 000.

In all 8 groups, the percentage of physician contributions to Republicans declined between 1993 to 1994 and 2009 to 2010, from 68.6% to 53.4%. The decline, however, was much less pronounced for physicians in high-paying specialties working in the for-profit sector. Whereas male physicians in low-earning specialties working in the nonprofit sector shifted from 60.6% contributions to Republican to 33.4%, the shift for male physicians in high-earning specialties working in the for-profit sector was modest, from 76.6% Republican to 71.3%.

We also examined whether this decline in physician contributions to Republicans reflected a general shift to Democrats in the profession or changes in the relative size of the 8 groups among the physicians who made campaign contributions. Between 1993 to 1994 and 2009 to 2010, the number of donors substantially increased in all 8 groups, from 13 925 to 41 244. The patterns varied by group, however. In the nonprofit sector, female donors in lower-paying specialties increased from 204 to 1811, and male donors in higher-earning specialties increased from 2218 to 4158. If the 8 groups had kept their rates of contributing to Republicans in 1993 to 1994 but were reweighted by their relative size in 2009 to 2010, the decline in contributions to Republicans would have been from 68.6% to 66.1%, a smaller but substantial shift. Thus, most of the actual decline, from 68.6% in 1993 to 1994 to 53.4% in 2009 to 2010, resulted from the shift in physician donors toward Democrats (Figure 2).

Between the 1991 to 1992 and 2011 to 2012 election cycles, campaign contributions by differing groups of physicians in federal elections have been sharply divided between Democrats and Republicans, with regard to both the percentages of physicians who contribute to each party and the absolute amounts contributed. Exceptions do occur. For example, although a 2009 survey of physicians showed that more than 70% supported expanding health care coverage¹² as Congress considered the Affordable Care Act, physician contributions to Republicans surged in the 2009 to 2010 election cycle to the point that a majority of campaign donors donated to Republicans. Then in 2011 to 2012, as in 2007 to 2008, a majority of physician campaign donors contributed to Democrats.

Between 1991 and 2012, physician contributors shifted away from Republicans and toward Democrats, particularly in specialties that are dominated by women and that are lower paying (eg, pediatrics), not in specialties that are dominated by men and higher paying (eg, surgery). Most of the shift resulted from an influx of new donors who were more likely to support Democratic candidates than prior donors, as shown by our comparison of the 1993 to 1994 and 2009 to 2010 election cycles.

The results shown in Figure 4 suggest that the polarization of physician contributors relates to their economic status; physicians in specialties with higher earnings are more likely to contribute to Republicans than those with lower earnings. These effects of income are particularly striking because most physicians have high earnings compared with the population at large. (Mean income in the lowest-paying medical specialties is close to $200 000. In contrast, median household income in the United States in 2011 was $50 054.¹³) Moreover, physicians employed by for-profit organizations tend to contribute to Republicans, as do male physicians. As a group, male physicians are likely to be older than female physicians; the large increases in the enrollment of women in medical schools occurred only in recent decades. Male physicians are more likely to work longer hours than female physicians and to work in specialties with higher earnings.^14,15

The large and increasing differences in physicians’ partisanship by sex far exceed the differences among voters in general. Already more than 20% in 1992, the gap between the sexes in physician donations between Democrats and Republicans without adjustment for other independent variables neared 28% by 2012 (Figure 2). These gaps are substantially more than the 10% gap among all voters in the 2012 presidential election.¹⁶ To be sure, the gap between the sexes is smaller for midterm elections. Women who contribute to Democrats are less likely to donate in midterm election years than in presidential years. Overall, however, over the last 2 decades, the political polarization of male and female physicians has increased.

Increasing differences in physician contributions also appear by employment type and specialty. Surgeons contributed 32.7% more to Republicans than did pediatricians in 1992. In 2012, the difference increased to 48.1%, which exceeds the 39.4% difference in the 2012 presidential vote between “red” Wyoming and “blue” Vermont.

The full consequences of the changes in the patterns of physician campaign contributions should be further studied, in particular, the relation between partisanship and specialty income. Partisanship may follow economic interests, emerging, for example, only after physicians in training enter a specialty. But it is also possible that physicians in training have characteristics that result in their being both partisan Republicans and entering higher-paying specialties, whereas others are partisan Democrats and trade earnings for work characteristics other than high pay. One survey of entering medical school students reported that those with Republican preferences were more likely to pursue careers in surgical specialties.¹⁷

Between 1991 and 2012, the political alignment of physicians in the United States changed dramatically. A profession once firmly allied with Republicans is now shifting toward the Democrats. Indeed, the variables driving this change—sex, employment type, and specialty—are likely to continue to be active forces and to drive further changes. For example, the percentage of women in the physician work force is likely to continue to increase and the number of solo and small practices is likely to continue to decrease.

The implications of these changes should be carefully tracked. Two questions are key. Now that the medical profession sits on both sides of the political aisle, will the Republican and Democratic parties devote unprecedented energy and shape policy to attract physician loyalty? Or will the increasing polarization of the profession increase the political participation of physicians even as it reduces the ability of physicians to reach consensus on health care policy? The number of people with medical degrees who serve in Congress has doubled over the last decade; in 2014 there are 20 physicians in Congress, of whom 16 are Republicans.¹⁸ At present, all that is certain is that in political terms, the profession is in play.

Accepted for Publication: April 11, 2014.

Published Online: June 2, 2014. doi:10.1001/jamainternmed.2014.2105

Correction: This article was corrected on January 28, 2019, to correct transposed row headings for “Pediatric surgery” and “Pediatrics” in Table 2.

Corresponding Author: David J. Rothman, PhD, Center on Medicine, Columbia College of Physicians and Surgeons, 622 W 168th St, PH15-25, New York, NY 10032 (djr5@columbia.edu).

Author Contributions: Dr Bonica had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: All authors.

Acquisition, analysis, or interpretation of data: Rosenthal.

Drafting of the manuscript: All authors.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Bonica, Rosenthal.

Administrative, technical, or material support: Rothman.

Study supervision: Rosenthal, Rothman.

Conflict of Interest Disclosures: None reported.

Additional Contributions: Whitney Adair, BA, Center on Medicine as a Profession, provided exceptional research and administrative assistance for this article. She was not compensated for her contribution beyond her salary.