The goal of service learning is to provide a meaningful, hands-on learning experience while serving the needs of the community. It is both different from and more encompassing than either a volunteer experience in a community setting or a didactic course on community health taught in the classroom. There are three fundamental tenets of service learning. First, there must be direct, hands-on experience in a community setting using skills commensurate with the students’ level of education and training. Second, the service must meet a community-identified and directed need. Finally, there must be a formal learning component to the experience that meets educational objectives and allows the student to reflect on the service with the assistance of faculty and community mentors.

Since 1993 the University of Pittsburgh’s Program for Health Care to Underserved Populations has coordinated a service-learning program with the Salvation Army for health professional students and internal medicine residents. As part of this program, primary care interns in the first postgraduate year (PGY1) are required to spend 15 hours at a health care clinic for homeless adults during their ambulatory care rotation. At the same clinics, the university’s medical students volunteer in a variety of capacities throughout their 4-year curriculum. First-year and second-year medical students conduct structured patient admission and history-taking sessions and give health education talks. Third-year students participate in the physical examination and treatment, while fourth-year students work as “acting interns” at the clinics and on a public health project. Full-time academic medical center physicians volunteer at the clinics as preceptors on evenings and weekends along with internal medicine residents in the second and third postgraduate years (PGY2 and PGY3). Aside from the mandatory rotations, all staffing is voluntary, with no scheduling allowances or financial compensation provided. In this article we report outcome data on the impact of a required homeless clinic rotation for primary care interns on their future behavior as residents, and on correlates of volunteerism among preclinical medical students.

The survey contained questions about demographics, past and current volunteering, future career plans, and ratings of 26 attitude statements about caring for indigent patients and 15 factors that could encourage or discourage students from volunteering. Attitude questions and potential encouraging/discouraging factors were derived from focus group discussions with previous participants and from the literature.^6,7 All attitudes and encouraging/discouraging factors were rated on a 5-point Likert scale (1 = strongly disagree, 5 = strongly agree; or 1 = strongly discouraged, 5 = strongly encouraged). Responses to the attitudinal questions were analyzed individually and aggregated into summary scores based on principal components extraction. Two scores (encouraging and discouraging) were created for each student for each influencing item. Encouraging scores were created by assigning a “1” if the student responded strongly or somewhat encouraged and a “0” otherwise. Discouraging scores were created by assigning a “1” if the student responded strongly or somewhat discouraged and a “0” otherwise. Then, the sum of the number of encouragers and number of discouragers was counted for each respondent.

Pearson χ²and student’s t tests were used to identify significant correlates of volunteering. Multivariate logistic regression analysis was used to identify significant independent correlates of volunteering. Independent variables included in the modeling were race, year in school, gender, previous volunteering experience, projected debt burden, the summary attitude measures, and the sum of encourager and discourager influencing factors. The dependent variable was current volunteering.

Table 1 Volunteer Status of PGY2 and PGY3 Residents in the General Medicine Residency Training Program atthe University of Pittsburgh Before and After Mandatory Rotations for Primary Care PGY1 Residents

Of the 210 students who completed the survey, second-year students were somewhat less likely to volunteer (46.9% vs 53.1%, χ²= 2.94, p = .09). Race, age, marital status, projected debt burden, and plans to be a primary care physician were not associated with volunteering. Women were more likely to volunteer than men (52.2% vs 47.8%, χ²=5.49, p = .02). Previous volunteer experience was not related to current participation unless the specific type and place of past volunteering was considered. As displayed in Table 2, only previous experience with an urban Indian Health Service or with providing direct medical care to an indigent population was associated with current volunteering.

Table 2 Description of Medical Student Population (Survey Respondents)

The principal components analysis of the 26 attitude items (Table 2) extracted 4 components that accounted for 43% of the variance. Overall, there were seven individual items that differentiated volunteers from nonvolunteers with ( p < .05). Of these, six fell on the first principal component, and one fell in the final component. Only the factor score from the first component, which we named positive attitudes for caring for indigent patients, was significantly associated with volunteering. For ease of interpretation a summary measure of the score on this first factor was generated by summing the responses to the 12 attitude ratings on this component (coefficient α =0.84) and subtracting 36 (to make a neutral score = 0). The potential range of scores was −24 through +24 with −24 being the least positive attitude. The observed range was −13 through +23 with mean = 9.3, SD = 6.8. The mean of the volunteering students was higher (11.1) than that of nonvolunteering students (7.2, t = 4.35, p < .001).

Only one encouraging factor (role models in the health profession) differentiated volunteers (52%) from nonvolunteers (37.1%). Also, the number of encouraging factors did not differentiate the volunteers from the nonvolunteers. Six discouraging factors differentiated the two groups (Table 3). Univariate analyses indicated that students with five or more discouraging factors (n = 75) were less likely to volunteer than students with no discouraging factors (n = 39) or one to four discouraging factors (n  = 96) (33% vs 59% vs 68%, χ²= 20.5, p < .001).

Table 3 Influential Factors in the Medical Student’s Decision to Volunteer

Multivariate logistic regression analysis of current volunteering indicated that previous direct service with an urban Indian Health Service (odds ratio [OR] 3.1; 95% confidence interval [CI] 1.6, 6.2), first year in medical school (OR 2.5; 95% CI 1.3, 4.9), and the summary measure of “positive attitudes for caring for indigent patients” (1.05 [per scale point]; 95% CI 1.00, 1.11) were associated with more volunteering. Having five or more discouraging factors was associated with less volunteering (OR 0.26; 95% CI 0.1, 0.7). Gender was not significantly related to volunteering when the attitude measure was included in the model.

Volunteer service with a homeless population is influenced by the education and training we provide. Among residents, participation in a structured, direct care experience with homeless patients was associated with increased subsequent volunteerism. Among medical students, voluntary participation was associated with year in school, the type of previous volunteer activity, fewer discouraging factors, and positive attitudes toward caring for the indigent.

The findings from the medical residency intervention demonstrate that a required, structured rotation with a homeless population can result in sustained volunteerism in this capacity. We assume that a positive experience and readily available opportunities for volunteering are critical in subsequent volunteering. The socialization process that this rotation provides may help dismantle preconceived stereotypes and fears that may have existed about the patients being served. We do not have data on whether there is a carryover to subsequent volunteerism with other special-need communities or in other settings, or if the experience has an impact on postresidency practice characteristics. Previous research has shown the benefits of such curricular efforts in enhancing attitudes and perceptions.^8,9 However, this is the first report to our knowledge of an impact on volunteerism. Given the challenges compounded by the maldistribution of physicians away from underserved areas,¹⁰ and the difficulties of recruiting physicians to McKinney Act Clinics,⁵ these findings suggest a useful strategy for training physicians who will be receptive to providing care to underserved and disadvantaged populations.

Consistent with previous research reporting that being in the first year of medical school was associated with more altruistic attitudes,⁷ our data also associate year in school with volunteerism (notably, the relation for gender did not hold up in the multivariate analyses.) How much the increased volunteerism among first-year students is a reflection of a desire to find a balance to the academic course work or more free time available for volunteering is unclear from our results. Our finding that the type and quality of previous volunteer experiences are associated with volunteering is noteworthy. As more schools place a service requirement on graduation and medical schools prioritize volunteerism and community service as entry criteria, it may be important to distinguish the type and relative quality of that experience if we are to expect it to influence future behavior.

As Flexner noted at the turn of the century, “medical education is not just a program for building knowledge and skills in its recipients … it is also an experience which creates attitudes and expectations.”¹¹ Attitudes, perceptions, and future plans among medical students are fluid and pliable and directly influenced by the academic experience.^12–14 The use of academic faculty as preceptors in our clinics reinforces the value and viability of professional volunteerism in the context of other demanding professional constraints. The preceptors are all full-time faculty members at the University of Pittsburgh and most are on a tenure track and have significant research and clinical responsibilities. Most of them balance their volunteering with families. The students who actively volunteer are exposed to these role models who are able to balance successful careers and a family life with volunteer activities within the community.

There are several limitations to our findings. Participants in the intern service-learning experience were not randomly assigned. It is possible that intrinsic differences between primary care and categorical residents account for the different PGY2 and PGY3 volunteer rates. However, we are encouraged that this potential bias is minimal because volunteerism among primary care residents after the PGY1 rotation exceeded volunteerism among primary care residents before the service-learning experience was instituted. Among medical residents it is unclear whether this willingness to volunteer extends beyond residency training, when financial and time constraints may be less, but personal commitments greater. Longitudinal tracking of this cohort is needed to assess the nature and sustainment of this association. For medical students, the survey responses are self-reports of current beliefs and behaviors. We need to follow this cohort of students to see if future volunteering is also associated with these responses. Despite these limitations, our findings demonstrate that early experiences during medical school and residency present a significant opportunity to instill social responsibility and sustained volunteerism in future physicians.

The authors acknowledge Maria Warhenberger, MD, MPH, for her assistance in developing the survey instrument and Joan Harvey, MD, for her assistance in administering the survey.

This study was funded in part by a grant from the Health Professions in Service to the Nation Program and the Corporation for National Service’s Learn and Serve America: Higher Education Program.