Sources of Satisfaction: A Second Administration of the Program Director Satisfaction Survey
Article Outline
In the past decade, the short half-life of leaders in academic medicine has been an important issue,1, 2, 3, 4 but residency program directors have been of particular interest because of their roles in developing and maintaining a high-quality academic environment, ensuring continuous cycles of program and process improvement, and monitoring the development and well-being of physicians-in-training from the time of recruitment to graduation.5 Program directors have short job durations and high burnout associated with administrative “hassles,” long work hours, dissatisfaction with promotion opportunities, and concerns about resources.6, 7, 8
In 1996, 34% of internal medicine program directors reported some degree of “burnout,” defined as “the loss of enjoyment or enthusiasm for a job, so that an individual is no longer able to devote emotional energy to its accomplishment.”9 Among this group of 262 program directors, a 3-year cohort study demonstrated a 29% turnover and mean job duration of 2.4 years. Turnover was highly associated with overall job satisfaction. Program director characteristics that were independently predictive of job turnover included low satisfaction with colleague relationships, high percentages of administrative work time, perceptions of the job as a stepping stone, and the availability of training to deal with problem residents.8
The Association of Program Directors of Internal Medicine (APDIM) Survey Task Force sought to measure the job satisfaction of program directors at routine intervals to monitor issues affecting turnover during years of rapid change in internal medicine education. This knowledge is important for strategic planning among national academic medicine organizations to meet the continuing education needs of academic administrators. The survey also serves to identify potentially remediable variables that affect job turnover. To that end, APDIM repeated the job satisfaction measures used previously to determine changes in satisfaction since the study in 1996 and the characteristics that are most strongly associated with satisfaction.
Materials and Methods
Questionnaire Development
The APDIM Survey Task Force developed the Program Director Satisfaction Survey (PD-Sat) to gather information on certain characteristics of the 391 internal medicine residencies in the United States and to address current issues facing residencies and residency directors. In particular, the questionnaire asked about housestaff financing (eg, salaries, benefits), program director job support and satisfaction, and the Accreditation Council for Graduate Medical Education 6 general competencies.10
The PD-Sat is a reliable and valid facet-specific job satisfaction measure for residency program directors.9 A 3-year cohort study demonstrated its predictive validity through its independent association with program director job duration.8 Program directors of other disciplines, including obstetrics and gynecology residencies, used the survey as well.7 There are 6 facets of the PD-Sat: satisfaction with work with residents, colleague relationships, resources, patient care, pay, and promotion. The total PD-Sat score correlates positively with a standard general job satisfaction measure.9
The task force distributed the questionnaire in November 2004 to all residency programs registered with APDIM. Results were collected by the end of March 2005. A program administrator or an associate program director could complete the first section about baseline program characteristics, but the program director was requested to complete the remaining questions containing the job satisfaction. The survey was confidential with respondents tracked by numeric codes.
Data Analysis
Survey responses were entered into a Microsoft Access (Microsoft Corp, Redmond, Wash) database with intentional double-checks to limit entry errors. Multiple questions were used to measure each specific facet of job satisfaction. If a respondent failed to complete one of the questions of a facet, we assigned an average value as calculated using the average value of the other questions of the specific facet questions (mean substitution).
Program directors provided salary data as an ordinal variable, and we used this to calculate a weighted average to compare data between 1996 and 2005 after adjusting for inflation.11 For the top and bottom ordinals, for which there was no range, we used the weighted limit value in the calculation. When provided, exact salary data were used to validate the weighted average of salaries for the larger group.
Continuous variables were examined for evidence of skewing, outliers, and non-normality, and were described using distributions, means, medians, standard deviations, and ranges. We compared data from 2005 with the initial 1996 survey to assess changes in total and facet-specific satisfaction scores. For the sake of presentation, many continuous variables (eg, percent work time) were assessed in tertiles after reviewing frequency distributions. Mean PD-Sat scores were compared using analysis of variance test and Tukey's honestly significant difference test. To account for multiple comparison, we report as significant only those variables with a P value less than .01 (Bonferroni correction). Linear regression analysis with a series of stepwise regressions was used to determine the strongest independent predictors of program director job satisfaction, using the F value (entry = .05 and removal = .10) to determine inclusion in the model. All tests are 2 sided.
Results
A total of 272 programs responded to the survey, and 5 programs closed during the survey collection period (response rate = 272/(391-5) = 70%). The mix of university (30%) and community programs (57%) was representative of programs across the nation (personal communication with Nicole V. Baptista, APDIM Policy Coordinator, APDIM office, April 2, 2007). Nonrespondent programs were not significantly different than respondents in their rolling 3-year (2004-2006) American Board of Internal Medicine certifying examination pass rate (91.4% vs 92.0%, P = .44)
Table 1 describes respondents, including the number of years in the position, the academic rank, and salary, along with comparative demographic data from the 1996 survey. The 2005 respondents' mean job tenure was 6.8 years, with more than one third holding this title for 3 years or fewer. Program directors held the rank of associate professor most frequently (41%), working an average of 58 hours per week. The mean yearly salary calculated from the ordinal data was $169,000, whereas the average of the 118 respondents who provided their exact salary was $171,439, demonstrating the validity of using a weighted average on the ordinal salary variable.
Table 1. Characteristics of Program Director Respondents to Program Director Satisfaction Survey in 2005 and 1996
| Demographic Characteristics | 1996 (n = 301) | 2005 (n = 272) | P Value |
|---|---|---|---|
| No. of years in position, Mean (SD, range) | 5.3 (5.8,1-39) | 6.8(5.9,1-32) | .002 |
| <3 y in position, % | 48.1% | 36.1% | .004 |
| Academic Rank: | |||
| Professor, % | 30 | 32 | .59 |
| Associate Professor, % | 32 | 41 | <.001 |
| Assistant Professor, % | 29 | 26 | .40 |
| Other, % | 9 | 2 | <.001 |
| Salary,a Mean | $194K | $169K | <.001 |
a1996 salary figures adjusted for inflation. Source: US Dept of Labor, Bureau of Labor Statistics (http://www.bls.gov/cpi); accessed October 18, 2007. |
Table 2 compares the facet-specific job satisfaction measures of the 2005 respondents with the 1996 respondents. Overall, there is a trend toward higher overall job satisfaction (PD-Sat, P = .007) and in the individual facets of satisfaction with resources (P = .003), colleague relations (P = .03), and promotion (P = .011). Satisfaction with patient care was significantly higher (11.4 vs 12.0, P = .0013) in 2005. Satisfaction with pay trended lower.
Table 2. Mean Scores by Factor: Program Director Satisfaction Survey 2005 and 1996
| Facets of Satisfaction | 1996 Mean (SD) | 2005 Mean (SD) | P Value |
|---|---|---|---|
| Work with residents | 24.1(4.5) | 24.5(2.5) | .20 |
| Colleague relations | 26.7(4.4) | 27.6(5.6) | .03 |
| Resources | 17.5(3.2) | 18.5(4.8) | .003 |
| Patient care | 11.4(2.6) | 12.0(1.6) | .0013 |
| Pay | 10.4(3.2) | 9.8(3.2) | .03 |
| Promotion | 14.5(3.5) | 15.4(4.6) | .011 |
| Total Satisfaction (PD-Sat) | 104.0(16.7) | 107.6(13.8) | .007 |
Overall program director satisfaction scores are listed in Table 3 for characteristics of both the residency programs and program directors. Although the program director salary is the only variable significantly associated with overall job satisfaction using the most conservative definition of statistical significance, several variables demonstrate trends that suggest a graded influence on job satisfaction, including the academic rank (P = .05), the number of hospital ward or consult months per year (P = .01), the program director's travel budget (P = .02), and the number of full-time equivalent administrative assistants for the program director and associate program directors (P = .03).
Table 3. Total Job Satisfaction Measure for Different Characteristics of the Residency and the Program Director, n = 272
| Characteristics | N | Mean Total PD-Sat Scorea |
|---|---|---|
| Residency Type | P = .05 | |
| Hospital-based | 152(57) | 109.6(13.1) |
| University-based | 79(30) | 104.9(13.5) |
| Consortium | 19(7) | 109.4(11.6) |
| VA/military/other | 17(6) | 104.6(13.8) |
| Region | P = .11 | |
| Northeast | 99(38) | 107.6(14.6) |
| Midwest | 60(23) | 110.4(11.0) |
| South | 67(26) | 104.9(14.5) |
| West | 33(13) | 109.7(9.8) |
| RRC Cycle Length | P = .22 | |
| <1 y | 14(5) | 101.5(18.6) |
| 2 y | 21(8) | 109.3(12.0) |
| 3 y | 55(21) | 106.4(13.8) |
| 4 y | 64(24) | 108.3(12.8) |
| 5 y | 89(33) | 109.7(12.9) |
| Program Director Tenure | P = .31 | |
| <3 y | 68(26) | 108.0(11.8) |
| 3-6 y | 77(29) | 105.2(13.8) |
| 6-9 y | 38(14) | 108.2(14.4) |
| 9-12 y | 31(12) | 109.6(12.2) |
| 12-15 y | 22(8) | 109.6(12.3) |
| >15 y | 29(11) | 111.4(15.5) |
| Program Director Rank | P = .05 | |
| Assistant Professor | 67(26) | 105.6(11.9) |
| Associate Professor | 105(41) | 107.1(13.3) |
| Professor | 84(32) | 110.8(14.7) |
| % Program Director Administrative Work | P = .83 | |
| <45% | 84(33) | 107.4(12.8) |
| 45%-60% | 106(41) | 107.7(13.9) |
| >60% | 68(26) | 108.7(13.5) |
| % Program Director Work in Clinics | P = .32 | |
| <10.3% | 85(33) | 109.0(11.2) |
| 10.3%-24.7% | 87(33) | 108.1(15.2) |
| >24.7% | 88(33) | 106.0(13.3) |
| % Program Director Teaching Work | P = .86 | |
| <15% | 71(28) | 106.9(13.5) |
| 15%-25% | 112(43) | 108.1(13.4) |
| >25% | 74(29) | 107.6(13.1) |
| % Program Director Research Work | P = .19 | |
| 0% | 110(42) | 106.2(13.2) |
| 1-5% | 95(36) | 108.3(12.5) |
| >5% | 58(22) | 109.9(14.9) |
| No. of Hospital Months/Consult Months | P = .01 | |
| <2 | 58(23) | 110.7(12.6) |
| 2-4 | 140(52) | 108.6(13.8) |
| >4 | 65(25) | 104.1(11.3) |
| Program Director's Salary | P <.001 | |
| <$125,000 | 33(13) | 105.5(11.4) |
| $125-150,000 | 42(16) | 102.9(13.1) |
| $150-175,000 | 53(20) | 105.8(13.3) |
| $175-200,000 | 68(26) | 107.4(14.1) |
| $200-225,000 | 27(10) | 111.1(10.2) |
| >$225,000 | 37(14) | 115.6(12.4) |
| Program Director's Travel Budget | P = .02 | |
| <$2000 | 57(24) | 104.9(15.4) |
| $2000-$3500 | 108(43) | 106.9(12.4) |
| >$3500 | 80(33) | 111.3(13.5) |
| Program's No. of FTE Administrative Help | P = .03 | |
| <2 | 67(26) | 104.3(13.2) |
| 2-3 | 153(57) | 108.4(13.6) |
| >3 | 47(18) | 110.7(12.0) |
| No. of Associate Program Directors | P = .14 | |
| 0-1 | 96(36) | 105.2(13.1) |
| 2 | 76(28) | 108.2(11.6) |
| 3 | 47(17) | 109.3(16.4) |
| 4 | 27(10) | 110.8(11.3) |
| ≥5 or more | 21(8) | 111.0(14.4) |
| Total No. of Residency Teaching Faculty | P = .10 | |
| <35 | 83(33) | 105.6(12.7) |
| 35-100 | 107(40) | 107.2(13.5) |
| >100 | 79(27) | 110.3(13.7) |
aAnalysis of variance for differences between categories within each characteristic. |
Regression modeling on these and other variables of interest found 4 variables that explained 14% of the variability in the job satisfaction scores of internal medicine residency program directors: the annual take-home pay, the total number of full-time equivalent administrative support available for the program director and associate program directors, the number of months per year on hospital wards or consulting services (a negative predictor), and the program director's percent research time.
Discussion
Between 1996 and 1999, internal medicine program directors in the bottom two thirds of scores on the PD-Sat job satisfaction measure were 2.5 times more likely to change jobs than individuals in the highest one third.8 In the present study, we used the same valid and reliable job satisfaction measure to gauge changes in job satisfaction during the last 9 years and to identify characteristics that are associated with higher job satisfaction. If departments of internal medicine can modify characteristics of their residency programs and the program directors' work environments, they may be able to improve job satisfaction and increase job stability of their program directors.
Our study shows signs that the situation of program directors may be improving. Respondents were more likely to be associate or full professors in 2005 (73% vs 62% in 1996, P = .005), and data trended toward an increase in satisfaction with promotion (15.4 vs 14.5 in 1996, P = .011). Moreover, the 2005 cohort has a longer tenure (6.8 vs 5.3 in 1996, P = .002). The data suggest program directors are remaining in their positions longer and their likelihood of academic promotion has improved.
Some trends may not be so hopeful. The inflation-adjusted salary is lower. Although 22% of the 1996 sample were concomitantly department chairs, we know from internal APDIM data that this number has decreased over the last decade to approximately 18% (personal communication with Nicole V. Baptista, APDIM Policy Coordinator, APDIM office, April 2, 2007). We suspect this decrease in program directors who are concomitantly department chairs may account for a small decrease over this period in the average salary, but not all. Given the concomitant trend toward a decrease in satisfaction with pay in 2005 compared with 1996 (P = .03), program director salaries may not be keeping up with inflation.
Not surprisingly, salary is the strongest independent predictor of total job satisfaction among program directors, adding nearly 2 points to the satisfaction measure with every $25,000 increment (Table 4). However, higher salary and satisfaction with pay do not necessarily predict job longevity. The longitudinal follow-up on the 1996 cohort demonstrated no difference in job changes between those making the highest salary range versus the lowest (hazard ratio = 1.0).8 In addition, when it came to satisfaction with pay, respondents with the lowest satisfaction had only a small, nonsignificant increase in job change rate (hazard ratio = 1.3).8
Table 4. Linear Regression to Determine Independent Predictors of Program Director Job Satisfactionabc
| Independent Predictors | Unstandardized Coefficient βb | Standardized Coefficient β'c | P Value |
|---|---|---|---|
| Annual take-home pay (in increments of $25K) | 1.78 | 0.232 | .001 |
| No. of FTE administrative support staff for program director and associate program directors | 3.56 | 0.170 | .011 |
| No. of months per year the program director is on inpatient or hospital consulting services | −0.75 | −0.167 | .013 |
| Program director's total percent work time dedicated to research | 2.50 | 0.138 | .039 |
aThe model's constant has coefficient = 89.2, P <.001; F = 7.9, P <.001; df = 201; R = 0.371, R2 = 0.138. |
bThe model's constant (89.2) gives the baseline PD-Sat job satisfaction a residency program director would be predicted to have. The unstandardized β coefficient gives the number of PD-Sat units to add to the constant for each interval of the variable. For instance, job satisfaction = 89.2 + (No. $25K increments of pay over $125K × 1.78) + (No. FTE administrative support × 3.56) + (No. months hospital × −0.75) + (% research time × 2.5). |
cStandardized coefficients (β′) compare the magnitudes of the regression coefficients and indicate the relative importance of each individual variable. |
What was strongly associated with time to job change between 1996 and 1999 was the satisfaction with colleague relationships, with a hazard ratio for individuals in the lowest tertile more than 3 times the number in the highest.10 In the current study, we found that administrative assistants, the number of teaching faculty, and the number of associate program directors all had trends indicating their impact on job satisfaction. Program directors might not appreciate feeling isolated with the burden of all of the duties of the residency. When administrative staff, including program administrators, associate program directors, chief residents, and key clinical faculty, contribute to the work of the residency, the program director can enjoy working with a high-functioning team and likely feels more affinity for the job.
The number of months of hospital service was the only negative predictor of total job satisfaction that remained in the multivariable regression model. Since 2003, the Accreditation Council for Graduate Medical Education has required sponsoring institutions to “provide at least 50% salary support” for internal medicine program directors and that program directors should “dedicate no less than 50% of his or her professional effort” to administering the program.12 Hospital ward service might interrupt the focus of attention on the administration and management of ongoing processes within the residency because of the high demands of inpatient care. One quarter of the sample performed more than 4 hospital service months per year, and their job satisfaction seems to have suffered from the increased workload. Similarly, the opportunity for a program director to engage in meaningful research requires support from the institution and is likely a good marker of whether a program director is able to find time away from clinical and administrative work.
There are several limitations to our study. This survey was by self-report and only captured the respondents' job satisfaction at one point in time. However, our previous study demonstrated that PD-Sat had predictive validity in determining job change over a 3-year period. In addition, the demographics of program directors likely have shifted over the last 9 years, given that fewer are in combined roles of chair and program director. Yet, the PD-Sat has been found valuable for program directors in other specialties7 and measures facets that at face value would apply to academic administrators at several levels. Finally, the associations of the varied facets of job satisfaction with job turnover are intricate tapestries and not always intuitive. Fortunately, we have our previous cohort study to use as a guide. Also, the study was powerful enough to produce a regression model that is highly significant (P <.001).
Conclusions
Internal medicine program directors who responded to our survey in 2005 trended toward higher total job satisfaction scores than did their peers from 1996. However, satisfaction scores assigned to pay trended lower in 2005 than 1996, perhaps reflecting inflation-adjusted salaries that were significantly lower for the 2005 cohort. This realization is important given that the program director's salary level was the strongest independent predictor of the total job satisfaction. Other potentially modifiable variables that are associated with program director job satisfaction include the amount of administrative support available, the number of hospital patient care months assigned to the program director, and the amount of available research time. Departments of internal medicine and national organizations should ensure that these job characteristics are addressed in the program directors' job descriptions and support provided to enhance both program director job satisfaction and residency program leadership stability.
References
- Tenure of academic chairs in obstetrics and gynecology: a 25-year perspective. Obstet Gynecol. 2006;108:1217–1221
- . Leadership trends in academic pediatric departments. Pediatrics. 2005;116:342–344
- . Educational programs in US medical schools, 2003-2004. JAMA. 2004;292:1025–1031
- . Turnover among APM members since 1971. Am J Med. 2002;113:706–710
- . The program director—a competency-based job description. ACGME Bulletin. 2007;28–30
- . Directing an emergency medicine residency: the problems and their potential solutions. Acad Emerg Med. 1995;2:914–918
- Job satisfaction among program directors in obstetrics and gynecology: a national portrait. Am J Obstet Gynecol. 2003;189:628–630
- . Job turnover and its correlates among residency program directors in internal medicine—a three year cohort study. Acad Med. 2001;76:1127–1135
- . A job-satisfaction measure for internal medicine program directors. Acad Med. 1999;74:263–270
- Chaudhry S, Beasley BW, Holmboe E. The state of evaluation in internal medicine residency. Submitted.
- . Inflation Calculator. www.bls.gov/cpi/home.htmAccessed November 5, 2007
- . Program Requirements for Residency Education in Internal Medicine. http://www.acgme.org/acWebsite/downloads/RRC_progReq/140_im_07012007.pdfAccessed December 13, 2007
Funding: None.
Conflict of Interest: None.
Authorship: All authors had access to the data and a role in writing the article.
PII: S0002-9343(08)01065-6
doi:10.1016/j.amjmed.2008.10.022
© 2009 The Association of Professors of Medicine. Published by Elsevier Inc. All rights reserved.