eMethods. Principal Component Analysis
eTable 1. Survey Questions Constituting the 9 Factors and Variance Explained by Each Factor
eTable 2. Factor Loading Pattern After Monotonic Spline Transformation
eTable 3. Correlations Between Factors Constructed by Principal Component Analysis
eTable 4. Cronbach α Test of Factor Analysis Measuring Internal Consistency
eTable 5. Summary of Key Prior Literature on Risk Factors for Attrition
Customize your JAMA Network experience by selecting one or more topics from the list below.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.
Abelson JS, Sosa JA, Symer MM, et al. Association of Expectations of Training With Attrition in General Surgery Residents. JAMA Surg. 2018;153(8):712–717. doi:10.1001/jamasurg.2018.0611
What attitudes and expectations of training are associated with attrition in general surgery residents?
In a national cohort of 828 categorical general surgery interns, those with realistic expectations of residency were less likely to undergo attrition. Those who chose their training program based on reputation were more likely to undergo attrition.
Attrition from general surgery training programs among incoming interns might be reduced by increased awareness of the rigors of surgery residency.
Attrition from general surgery training is highest during internship. Whether the expectations and attitudes of new trainees affect their subsequent risk of attrition is unknown.
To identify the expectations of general surgery residency associated with attrition from training.
Design, Setting, and Participants
This prospective observational cohort study included categorical general surgery interns entering training in the 2007-2008 academic year. Residents were surveyed regarding their expectations of training and of life as an attending at the start of their intern year (June 1 to August 31, 2007). Expectations were grouped into factors by principal component analysis, and a multivariable model was created using these factors in addition to known demographic and program characteristics associated with attrition. Follow-up was completed on December 31, 2016.
Main Outcomes and Measures
Attrition from training was determined by linkage to American Board of Surgery resident files through 2016, allowing 8 additional years of follow-up.
Of 1048 categorical surgery interns in the study period, 870 took the survey (83.0% response rate), and 828 had complete information available for analysis (524 men [63.3%], 303 women [36.6%], and 1 missing information [0.1%]). Most were white (569 [69.1%]) and at academic programs (500 [60.4%]). Six hundred sixty-six residents (80.4%) completed training. Principal component analysis generated 6 factors. On adjusted analysis, 2 factors were associated with attrition. Interns who choose their residency based on program reputation (factor 2) were more likely to drop out (odds ratio, 1.08; 95% CI, 1.01-1.15). Interns who expected as an attending to work more than 80 hours per week, to have a stressful life, and to be the subject of malpractice litigation (career life expectation [factor 6]) were less likely to drop out (odds ratio, 0.90; 95% CI, 0.82-0.98).
Conclusions and Relevance
Interns with realistic expectations of the demands of residency and life as an attending may be more likely to complete training. Medical students and residents entering training should be given clear guidance in what to expect as a surgery resident.
Despite changes in resident duty hours and programmatic efforts aimed at reducing attrition, withdrawal from general surgery training remains a national problem with important workforce and training implications.1,2 Several prior studies3-12 have investigated individual and programmatic variables that affect attrition, but none have evaluated whether expectations of trainees are associated with successful completion of general surgery residency.
Previous reports have demonstrated that the highest rate of attrition occurs among interns.13 A mismatch between expectations of surgical training and the reality of internship may in part explain these higher attrition rates. Understanding expectations of internship would help to inform the selection of candidates, better prepare medical students for the rigors of surgical training and the stress of a surgical career, and facilitate creation of appropriate support programs in residency training. In the future, such interventions could in turn lead to lower attrition rates, especially early in residency.
In this prospective national study of categorical general surgery interns, a principal component analysis (PCA) was used to categorize resident expectations of training into factors. These factors were subsequently used to measure the association between expectations of surgical training and careers and noncompletion of residency. We hypothesized that interns who expected adversity, longer hours, and a more challenging environment during residency would have higher rates of completion of training.
Study participants included categorical general surgery interns from the class of 2007-2008 who participated in a nationwide survey from June 1 through August 31, 2007, as a prelude to the National Study of Expectations and Attitudes of Residents in Surgery administered during the American Board of Surgery In-Service Training Examination in January 2008.14,15 This study was approved by the institutional review board at Yale School of Medicine and subsequently transferred and approved by the institutional review board at Weill Cornell Medicine. Completion and return of the survey was considered implied consent.
The development of an intern survey was based on qualitative interviews with surgery residents who had left their training programs, with focus on reasons why surgery residents drop out.16 Based on these interviews, prior literature on attrition, and collaboration with the American Board of Surgery, the intern survey was crafted to test hypotheses for how mismatches in expectations of residency and reasons for pursuing a particular program may be associated with a higher risk of noncompletion of training.
Survey data included intern demographics, expectations for residency and a career in surgery, and reasons for selecting a residency program. Questions related to demographics included intern gender, race (white, black or African American, Asian, and other), ethnicity (non-Hispanic or Hispanic), and marital status (married, never married with steady partner, never married with no partner, or divorced or widowed). Questions targeting interns’ expectations of residency training assessed a variety of topics, including program supportiveness, free time outside work, perceptions of gender differences in training, and expectations for operating room exposure. Questions assessing the reasons why interns selected specific residency programs included program reputation, location, adherence to work-hour limitations, and collegiality. Interns also were asked about their expectations for life as an attending surgeon. Data on the program where each intern started residency were collected; characteristics of programs included program type (academic, community, or military), program size determined a priori based on the number of chief residents in the graduating class of 2007-2008 (1-3, 4-5, or ≥6), and program location (Northeast, South, Midwest, or West).
For each question assessing intern expectations, participants were asked to select their level of agreement with survey statements using a Likert scale. Options included strongly agree, agree, neutral, disagree, and strongly disagree. Survey responses were analyzed as ordinal variables (1 indicates strongly disagree; 5, strongly agree).
Quiz Ref IDInterns who participated in the intern survey were followed up for 8 years (until December 31, 2016) using linkable identifiers provided by the American Board of Surgery. The primary end point was completion of a general surgery residency within this period. Of note, the American Board of Surgery never had access to the interns’ demographics or responses to the survey.
We used PCA to test the primary end point. Principal component analysis is a validated statistical method used to examine interrelated survey results by grouping many variables into a smaller number of correlated variables. With that smaller number of correlated variables ultimately grouped into multiple factors, analysis of the association between these factors and the primary outcome of noncompletion of residency training was undertaken.17-19 Univariable and multivariable analyses were performed testing the association of expectations of surgical training with resident attrition. The adjusted analysis accounted for intern gender, race, ethnicity, marital status, and program size, location, and type. To test for within-program clustering, we tested the use of a random-effect variable to account for correlations within programs, and the coefficient was not significant. An odds ratio (OR) of attrition for 1-unit increases in the factor-based score and 95% CI were obtained. Additional detail about our methodologic approach using PCA is available in the eMethods in the Supplement).
Subset analysis was performed for all variables in this study; however, only statistically significant findings from logistic regression analysis assessing the association between expectations and attrition within subgroups (P < .05) are reported. All statistical analysis was conducted with SAS software (version 9.3; SAS Institute Inc).
Of 1048 categorical general surgery interns in the 2007-2008 academic year, 870 took the survey, yielding a response rate of 83.0%. Among the respondents, 836 had sufficient linkage data to permit confirmation by the American Board of Surgery Resident Roster that they were interns in the class of 2007-2008. Eight of these interns had incomplete survey data and were excluded from subsequent analysis, resulting in a final study cohort of 828 interns. Quiz Ref IDOf these interns, 666 (80.4%) completed their residency training. Five hundred twenty-four (63.3%) were male, and 303 (36.6%) were female (missing for 1 [0.1%]); 569 (69.1%) were white, 500 (60.4%) trained at an academic program, and 247 (29.8%) trained in the Northeast (Table 1).
With use of PCA, the 31 survey questions were grouped into 9 factors based on their concordance. The total variance explained by these 9 factors was 58.1% (eTables 1 and 2 in the Supplement). eTable 3 in the Supplement shows the intercorrelation between summarized factors.
A Cronbach α coefficient calculation revealed that factors 7 (expectation of operating room exposure) and 8 (80-hour work week and career benefits) generated a coefficient of less than 0.5 and thus had unacceptable internal consistency; this finding meant that these factors could not reliably be grouped together (eTable 4 in the Supplement). Therefore, these 2 factors were excluded from the final multivariable analysis along with factor 9 (collegiality), which mapped to only 1 variable and therefore could not be reliably compared with other variables. Interns who did not complete residency had higher mean agreement with factor 2, meaning that they were more likely to select a residency program based on its reputation and advice from mentors (Table 2). Quiz Ref IDAfter adjustment for resident demographics and program characteristics, residents who selected their residency program based on its reputation and advice from mentors (factor 2) had an increased likelihood of attrition compared with residents who did not choose their residency based on similar variables (OR, 1.08; 95% CI, 1.01-1.15). Interns who expected to (1) work more than 80 h/wk as an attending, (2) have a very stressful life as an attending, and (3) be the subject of malpractice litigation at some point during their career (factor 6) were less likely to experience attrition (OR, 0.90; 95% CI, 0.82-0.98) (Table 3).
When stratifying interns based on whether they attended academic or community programs, only interns at academic programs who chose their training program based on its reputation and advice from mentors (factor 2) had a higher likelihood of noncompletion (OR, 1.09; 95% CI, 0.99-1.19). For interns at community programs, choosing a program based on reputation and advice had no influence on attrition (Table 4).
Quiz Ref IDUnmarried interns who chose a residency program based on its reputation and advice from mentors (factor 2) had a higher likelihood of attrition (OR, 1.11; 95% CI, 1.02-1.21), but choice of program based on location (factor 4) and expectation of a stressful life as an attending (factor 6) had no difference in likelihood of attrition (Table 5).
Among married interns, the choice of a program based on its reputation and advice from mentors (factor 2) did not influence the likelihood of attrition, whereas selection of a training program based on geographic location (factor 4) increased the likelihood of attrition (OR, 1.20; 95% CI, 1.03-1.40), and the expectation of a stressful life as an attending (factor 6) decreased it (OR, 0.82; 95% CI, 0.69-0.98).
In this longitudinal, nationwide analysis of categorical general surgery interns from the 2007-2008 academic year, we provide, to our knowledge, the first large-scale assessment of how interns’ expectations of residency and reasons for choosing a training program affect their likelihood of completing training. Although other studies have assessed risk factors for attrition (eTable 5 in the Supplement), none have evaluated expectations of residency and a career as a surgeon as a risk factor for attrition1,3-5,8-13; furthermore, none have used PCA based on learners’ survey responses to estimate future attrition. The results of this study may inform interventions among medical students interested in surgery and interns at the start of residency with the purpose to reduce attrition.
Those interns who were married and expected to have a stressful career (ie, to work >80 hours/week as an attending, to have a stressful life as an attending, and to be the subject of malpractice litigation) were less likely to experience attrition. Quiz Ref IDThis finding suggests that an intervention to reduce attrition could reasonably focus on better preparing medical students and interns for the stress of a surgical career. Another possible intervention could be the development of wellness programs to teach strategies to cope with stress and prevent burnout.20 In addition to preparing for the rigor of residency, interns may need to be given a better understanding of the stress an attending surgeon might experience during a lifetime. This process may include training for how to balance a need to be productive at work while also enjoying personal relationships outside work. It may also include specific training on how to handle a malpractice lawsuit.21
Attrition also was associated with some factors that motivated an intern’s program choice. For example, residents who chose their training programs based on reputation were more likely to experience attrition. On subset analysis, this finding was only true for interns who were unmarried. In addition, interns who were married and chose their training programs based on location were more likely to experience attrition. Having close family nearby has previously been shown to be associated with attrition.15 Married interns may be more focused on family, and thus having family nearby may actually influence them to leave residency, although more data are needed to confirm this possibility. Together these findings point to the importance of deciphering the exact reasons why applicants express interest in particular programs. Future surgery residents may consider de-emphasizing program reputation and location and instead focus on individual fit between residents and attendings. In addition, future research should focus on developing and validating the association of expectations among applicants to surgery residency with attrition from residency to potentially assist program directors with candidate selection.
The results of this study should not be interpreted as de-emphasizing the importance of creating a collegial, respectful, and diverse training environment. Our study assesses expectations of interns and not their actual experiences in residency. Thus, our results should not be interpreted to mean that the experience of training at a program with better work environment, collegiality, and respect for diversity has no influence on resident attrition. Instead, our main conclusions are that interns who are more prepared for the arduous journey are more likely to complete residency. Therefore, 1 implication from this study is that investing in efforts to make programs more collegial and diverse may reduce attrition among interns who are less prepared for the rigors of surgery residency.
This study is unique in its use of PCA to use survey results to estimate outcomes. The advantage of using PCA in this context is the ability to summarize patterns of data and reduce the number of variables for incorporation into multivariable analysis. A few examples in the surgical literature use PCA to analyze survey responses for a variety of research topics, including evaluating medical student performance on surgical clerkships,22 guiding psychological evaluation before bariatric surgery,23,24 and understanding perceptions of surgical competence.25
This study has a few limitations. The data used in this analysis were only collected at a single point (the start of residency), and they were not reassessed as they progressed through training. Impending career or life events unaccounted for at the time of our survey may have prompted interns to leave surgery residency, in which case attrition was not related to interns’ expectations of training. The results of this study apply to the intern class of 2007-2008 and may not be generalizable to other intern classes, especially given changes in duty hour restrictions. More contemporary data are needed to identify whether such changes have affected attrition. Although several subset analyses were undertaken, a more robust analysis of all subgroups (eg, female Hispanic residents at a program in the Northeast) could not be performed owing to the sample size. Although our response rate was excellent, we do not have individual or programmatic data on nonrespondents of our survey to test for nonresponse bias for our final cohort.
Categorical general surgery interns with more realistic expectations of surgery residency and life as an attending surgeon are more likely to complete training. These findings in a large national cohort using PCA highlight possible points of intervention to reduce attrition from residency. Future work should focus on better informing the expectations of medical students interested in surgery. Program directors may also consider these results to help better support residents and prepare medical students for surgical training.
Accepted for Publication: January 24, 2018.
Corresponding Author: Heather L. Yeo, MD, MHS, Department of Surgery, New York–Presbyterian Hospital, Weill Cornell Medicine, 525 E 68th St, PO Box 172, New York, NY 10065 (firstname.lastname@example.org).
Published Online: April 18, 2018. doi:10.1001/jamasurg.2018.0611
Author Contributions: Dr Yeo had full access to all 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: Abelson, Sosa, Mao, Yeo.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Abelson, Symer, Mao, Yeo.
Critical revision of the manuscript for important intellectual content: Sosa, Symer, Mao, Michelassi, Bell, Sedrakyan, Yeo.
Statistical analysis: Symer, Mao, Sedrakyan, Yeo.
Obtained funding: Sedrakyan, Yeo.
Administrative, technical, or material support: Abelson, Bell, Sedrakyan, Yeo.
Study supervision: Sosa, Yeo.
Conflict of Interest Disclosures: Dr Sosa reported being a member of the data monitoring committee of the Medullary Thyroid Cancer Consortium Registry supported by GlaxoSmithKline, Novo Nordisk, AstraZeneca, and Eli Lilly and Company. Dr Bell reported being the owner and president of Discourse LLC, a company that creates educational software for health care. Dr Yeo reported that her spouse received salary from Bioscrip, Inc. No other disclosures were reported.
Funding/Support: This study was initially supported by the Robert Wood Johnson Foundation.
Role of the Funder/Sponsor: The sponsor had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: Frank Lewis, MD, and the American Board of Surgery provided support throughout the project and assisted with data collection, for which they were not compensated.
Create a personal account or sign in to: