Residents' Attitudes and Behaviors Regarding Care for Underserved Patients: A Multi-Institutional Survey

Caring for underserved patients is a fundamental obligation of physicians, and accrediting committees in undergraduate and graduate medical education have emphasized the importance of maintaining training environments that foster altruistic behaviors among residents and medical students.1,2 Previous studies have shown that students' altruistic attitudes towards underserved patients decreases as medical school training progresses,3 but why students' attitudes change and how they relate to subsequent behaviors during residency training and practice are unclear. Additionally, while resident physicians are often the front line of care for underserved patients, little is known about their attitudes and behaviors regarding these populations.

The Accreditation Council for Graduate Medical Education (ACGME) has identified systems-based practice (SBP) as a core competency which is “… manifested by an awareness of and responsiveness to the larger context and system of health care and the ability to effectively call on system resources to provide care that is of optimal value.”2 Consistent with this definition, curricula are emerging to enhance knowledge about the roles of health care and socioeconomic systems in caring for underserved patients so that physicians in training may advocate more effectively for these populations.4–21 Understanding residents' attitudes towards topics of underserved care and behaviors regarding underserved populations should enhance the content of future curricula and provide information that will improve our ability to engage future physicians in volunteer service. Therefore, we administered a cross-sectional, multi-institutional survey to better understand resident attitudes towards topics in underserved care and behaviors regarding underserved patients.

A total of 956 surveys were distributed to residents in 18 programs representing 7 different specialties at 10 institutions in the Midwest, West, Northeast, mid-Atlantic, and Southeast regions of the United States. A convenience sample of program directors was invited to participate. Surveys were distributed to all residents in each participating program. The institutional review boards at each of the participating institutions approved the study.

The validity paradigm for educational research states that validity evidence is collected from the following sources: content, response process, internal structure, criterion and consequences.22–25 For this study, we demonstrated content and internal structure (reliability) evidence for assessment of attitudes regarding topics relevant to medically underserved patients, which is consistent with the validity emphasis found in many education research reports.24,26 

The questionnaire utilized in this investigation was designed to assess residents' attitudes regarding education topics relevant to medically underserved patients. Content validity was determined by reviewing the literature and seeking expert input to determine the domains relevant to underserved patients. First, a panel of content experts on underserved populations and health equity was identified based on reputation, publications, and leadership in the field. Nine experts from seven institutions (Harvard; Johns Hopkins; Mayo Clinic; Rockefeller Foundation; US Public Health Service; University of California, San Francisco; and the University of North Carolina) were interviewed in June 2006 by telephone using an unstructured interview script with one main question and follow-up questions as appropriate. The primary question to the experts was: “what are the issues related to medically underserved populations that are most important for physicians-in-training to understand?” Probing questions were then used to elicit additional ideas from the experts. We defined “underserved populations” as groups that are traditionally less advantaged or have limited access to the health care system (eg, low income, low education, racial and ethnic minorities, uninsured).

Content validity was also determined by a comprehensive MEDLINE search that was conducted through January 2007 to obtain published literature relevant to the topics identified during the phone interviews; we also obtained references from the experts themselves. Keywords were taken directly from the initial list of topics identified by the experts. Manuscripts selected for review were those that corresponded to these topics. This process yielded a distilled list of topics relevant to underserved populations. This list was then resubmitted to the expert panel for further revision and assessment of importance. All this resulted in 3 core domains, each with 5 sub-domains, which were agreed upon by most of the experts. The 3 major domains were access to health care, socioeconomic position and health, and racial and ethnic health disparities.27 

Survey items were designed by the investigators and survey methodologists at the Mayo Clinic Survey Research Center. Resident physicians rated their attitudes on each of the 15 items with a 3-point scale (very important, somewhat important, or not important). An overall attitude score for each resident was then calculated on a scale of 0–1 by averaging their responses across all 15 survey items, where 1  =  very important, 0.5  =  somewhat important, and 0  =  not important.

Internal consistency reliability of scores for each of the 3 attitude domains was determined by calculating Cronbach alpha, with acceptable values being >0.7.28 

The behavior assessment evaluated perceived volunteerism with underserved patients in the past (medical school), present (residency) and future (anticipated practice) on a scale of average time spent volunteering per week (no time, 1 to 3 hours, 4 to 7 hours, 8 to 10 hours, more than 10 hours per week).

The survey was administered online by the Mayo Clinic Survey Research Center between October 18, 2007 and January 31, 2008. A multiple-contact data collection protocol was implemented consisting of the following steps: an initial survey with a cover letter/e-mail message explaining the study was sent to all physician residents listed in the rosters provided by the respective sites; a reminder either thanking them for their response if they completed the survey or an exhortation to respond was sent one week after the initial mailing; and a second survey 2 weeks after the reminder, again with cover letter/e-mail message, to nonrespondents to the previous surveys.

Survey results were reported using standard descriptive statistics, such as Mean ± Standard deviation (SD) or Frequency (Percentage). Attitude scores were compared between 2 groups using 2-sample t test, and among more than 2 groups using one-way ANOVA, where appropriate.

Volunteering percentages were compared between 2 groups by Fisher Exact test, or among more than 2 groups using Pearson's Chi-Square test, where appropriate, with levels of significance set at P <.05. All analyses were handled by SAS version 9.1.3 software (SAS Inc., Cary, NC).

A total of 498 (response rate  =  52%) surveys were completed by residents in all of the programs surveyed.

Demographic characteristics of the survey respondents are shown in table 1. The majority of residents surveyed were younger than 30 years old (74%), men (54%), and white (70%). Most respondents (93%) were in their first 3 years of training, and distribution across these training levels was relatively even. The majority of respondents were enrolled in internal medicine training programs (59%), and the remainder was represented by general surgery (11%), pediatrics (6%), medicine-pediatrics (5%), neurology (5%), and family medicine (9%). Most respondents were raised by parents who earned more than $50 000 per year (71%), and who had at least a 4-year college education (79%), if not postgraduate education (59%).

Regarding validity of “attitudes” survey items, overall Cronbach's alpha for all of the survey items was 0.93. Cronbach's alpha was >0.9 for all 15 survey items.

Attitude assessments revealed that overall, 58.4% of residents considered issues of medically underserved populations to be very important, 39.1% somewhat important, and 2.7% not important. Residents (percent) agreed that issues of access to health care (63.8%), racial and ethnic health disparities (59.6%), and socioeconomic position and health (51.8%) were very important (table 2). Women had more favorable attitudes than men across all 15 items (P value range of .04 to <.001), and the overall attitude score was significantly higher (more favorable) for women than for men (0.83 versus 0.74; P  =  <.001). There were no significant differences in attitude scores in terms of ethnicity, parent income, parent education level, age, or postgraduate year.

Behavioral assessments revealed that percentages of residents who regularly volunteered with medically underserved populations were 76% during medical school, 19% during residency, and 84% anticipated in future practice (figure 1). Nonwhite respondents had volunteered more often than whites in the past (82% versus 73%; P  =  .01). Likewise, more women than men anticipated volunteering in the future (90% versus 79%; P  =  .001). Finally, postgraduate year-1 (PGY-1) residents reported being more likely to volunteer in the future (91%) when compared to PGY-2 (80%), PGY-3 (81%), and PGY-4 (79%) residents; P  =  .02. There were no differences in volunteerism in terms of age, parent income, or parent education level.

When correlating attitudes with behaviors, respondents who volunteered on a weekly basis had more favorable attitudes regarding topics of underserved populations than those who did not volunteer (overall average attitude score of 0.81 versus 0.73; P  =  <.001). This was true for medical school (0.80 versus 0.71; P  =  <.001), residency (0.83 versus 0.76; P  =  .007) and anticipated practice (0.79 versus 0.70; P  =  <.001) (figure 2). Furthermore, the relationship between hours volunteered per week and favorable attitudes were significant across all 15 items during medical school (P value range of .029 to <.001) and anticipated practice (P value range of .035 to <.001). This relationship occurred in a linear, dose-dependent fashion (table 3).

Our survey revealed that US residents' attitudes towards issues relevant to medically underserved populations are generally favorable. This is reassuring given the perceived cynicism among practicing physicians regarding underserved patients.29 Our survey found no difference in attitudes between any of the first 3 years of residency. This finding contrasts with previous studies of medical students, which demonstrated that student attitudes towards underserved patients become more negative over the course of medical school.3,30 Furthermore, our survey found more favorable attitude scores among women compared with men. This finding supports previous studies showing gender discordance for attitudes towards underserved populations among medical students,3 yet it also provides fresh evidence that this gap between men and women may persist beyond undergraduate medical education.

In our survey, rates of volunteering to serve underserved patients were high in medical school and anticipated future practice, yet low during residency. The low volunteerism among residents likely reflects more rigorous time commitments and inflexible schedules during residency training. Moreover, residents may feel that they are already caring for underserved patients in their training programs and thus feel less inclined to provide additional, voluntary care. Yet another explanation for our findings is that residents may feel less integrated with their communities during the often transient career phase of residency. Further, residency curricula and culture may not offer the same community focus as those found in medical school or future practice settings. Finally, residents frequently start families during their training, which adds to their competing priorities. Overall, the finding of low volunteerism among residents suggests the need to identify barriers to volunteering during residency training.

A remarkable finding of our survey was that anticipated weekly volunteerism after completing training among residents was very high (84%), even higher than reported rates of volunteering during medical school. Although these findings are encouraging, they contradict actual rates of volunteerism among practicing physicians, only 39% of whom reported ever volunteering over the course of a year,31 and only 54% of whom reported ever volunteering over 3 years.32 However, if charity care is included as volunteerism, then rates of volunteerism among practicing physicians have been estimated at upwards of 68%.33 The reasons for the discrepancy between volunteerism among residents versus practicing physicians are unclear and likely multi-factorial. First, there may be generational differences in that younger physicians have been raised in an educational environment that places a greater emphasis on volunteerism. Second, residents' perceptions of practice may be overly optimistic as they may fail to accurately assess the amount of time available for volunteering within a busy medical practice.

We found that female residents have more favorable behaviors than male residents regarding the underserved, which is consistent with existing demographic data on overall volunteerism in the United States.34 Likewise, we found that nonwhite residents demonstrated more volunteerism in the past and anticipated future than white residents. These findings build upon a study by Weisman et al demonstrating that female and nonwhite residents were more likely than male and white residents to rate poor inner city communities as desirable practice locations.35 Furthermore, our findings are similar to those of practicing physicians, which found that underserved patients are more likely to be treated by nonwhite physicians.36,37 

This survey identified a strong correlation between residents' attitudes and behaviors regarding medically underserved populations. Similarly, O'Toole et al showed an association between medical students' attitudes towards underserved populations and volunteering at homeless clinics.38 Likewise, Chirayath found an overall association between practicing physicians' attitudes towards underserved populations and caring for the underserved.39 To explain these associations, existing theoretical frameworks have identified that factors associated with physicians' service to underserved patients may include practice setting, favorable changes in income, emotional connections to underserved populations, and commitments to underserved populations throughout the life course.39–42 Our findings add to the existing literature by showing the new finding of associations between attitudes and caring for underserved patients among resident physicians, and indicate that residents who express an interest in topics relevant to underserved patients may be more likely to provide volunteer service.

Our study has limitations. Although this was a multi-institutional survey of residents from numerous training programs, internal medicine residents were disproportionately represented, so the results may not generalize to physicians in all training programs. However, the attitude scores across the different specialty training programs in this study were similar, especially for the nonsurgical trainees (Table 1). Additionally, while the survey response rate was modest, it was representative of physician surveys, which generally have lower response rates (about 52% to 54% for larger physician surveys) than do surveys of the general population.43,44 Our study was a cross-sectional survey and thus failed to account for changing attitudes and behaviors within each PGY level over time. Further, participants were not asked whether they attended medical school in the US, so we are unaware of potential differences between US and foreign medical graduates. Another limitation is that residents interested in underserved patient care may have been more likely to complete the survey, thereby skewing the results towards more positive overall attitudes and reported behaviors. Likewise, recalling past volunteerism may be less accurate than reporting current volunteerism. Another limitation is that respondents were not instructed on the meaning of “volunteerism” and they were allowed to interpret the term as they saw fit; however, it is plausible that a commonsense interpretation of “volunteerism” would have broad agreement.

This study's main conceptual limitation is that volunteerism is only one surrogate for physician behavior regarding underserved populations. Comparing volunteer experiences across medical school, residency, and practice could be problematic as these experiences may be qualitatively different. Additionally, volunteer experiences are likely variable regarding their availability and impact on underserved populations. Furthermore, practitioners may find that their clinical practices, social advocacy and research benefit underserved patients more than existing volunteer opportunities in their communities. Despite these limitations, our study presents the unique opportunity to examine perceived “life cycles” of volunteerism among individual residents. Moreover, volunteerism is arguably the leading surrogate for behaviors regarding underserved populations in the literature. The finding in this study of a strong association between reported volunteerism and attitudes regarding underserved populations across every domain further supports the use of volunteerism as a measure of caring for the underserved.

In summary, we found that US resident physicians in this study had generally positive attitudes towards issues contributing to underserved patients. The association between residents' attitudes and behaviors (volunteerism) was strong, indicating that residents' expressed intentions may be a reliable indicator of their future service. Also, female and nonwhite residents reported doing more volunteer work than male and nonwhite residents respectively. All these findings, while basically consistent with previous research in medical students and practicing physicians, provide new knowledge regarding resident physicians' attitudes and behaviors towards underserved patients. Additionally, these findings suggest that residents should be provided with more formal opportunities to volunteer during residency training, and should inform future research on reasons for gender and racial differences in the desire to serve, and possible barriers to volunteering during residency training. Finally, previous studies have suggested the potential for medical education to change attitudes towards underserved populations.38,45,46 Therefore, our study's findings may provide insights that will enhance curricula and the care of underserved patients.

Additional Members of the Underserved Care Curriculum Collaborative*

Sandra L. Argenio, MD (Mayo Clinic, Florida), Christopher J. Boes, MD (Mayo Clinic, Rochester), Suzanne Brandenburg, MD (University of Colorado, Denver), Fred Edwards, MD (Mayo Clinic, Arizona), David R. Farley, MD (Mayo Clinic, Rochester), Robert T. Flinchbaugh, DO (Mayo Clinic, Rochester), Katherine A. Julian, MD (University of California, San Francisco), Cindy J. Lai, MD (University of California, San Francisco), Lia S. Logio, MD (Indiana University, Indianapolis), John D. Myers (Texas A&M, College Station), Michael P. Phy, MD (Texas Tech University, Lubbock), Robert G. Voigt, MD (Mayo Clinic, Rochester), Lawrence D. Ward (Temple University, Philadelphia), Renee D.E. Wright, MD (Johns Hopkins University, Baltimore).

Funding: The authors report no external funding source.

Acknowledgments: We would like to thank the content experts in underserved patient population topics who were interviewed for this project: Darren Dewalt, MD, MPH (University of North Carolina, Chapel Hill), David Himmelstein, MD, MPH (Harvard Medical School, Boston), Nancy Krieger, PhD (Harvard College, Cambridge), David Rutstein, MD, MPH (US Public Health Service), Barbara Starfield, MD, MPH (Johns Hopkins University, Baltimore), Robert Stroebel, MD (Mayo Clinic, Rochester), Sally Trippel, MD, MPH (Mayo Clinic, Rochester), Margaret Wheeler, MD, MPH (University of California, San Francisco), and Derek Yach, MBChB, MPH (Rockefeller Foundation, New York).

This project was supported by an institutional Educational Innovations grant through the Mayo Foundation and by the Mayo Clinic Internal Medicine Residency Office of Educational Innovations as part of the ACGME Educational Innovations Project.