2019 HSC Section 2 - Practice Management

Otolaryngology–Head and Neck Surgery 154(6)

technical skills if underemployed/unemployed, and being required to consider working in a less-desired city/province were most concerning to those surveyed. While the job market in Canada is not the same as in the United States, it is possible that some of these same forces may be driving resident deci- sions in America. In this study, we analyzed responses to the American Academy of Otolaryngology—Head and Neck Surgery (AAO-HNS) Section for Residents and Fellows (SRF) Annual Survey from 2008 to 2014. We sought to answer 2 primary questions: (1) what are the current trends with regard to pursuit of subspecialty fellowship training in oto- laryngology, and (2) what are the factors that affect interest in fellowship during the course of residency? Materials and Methods The AAO-HNS SRF Annual Survey has been conducted annually since 2002 using an online, electronic resource (sur- veymonkey.com). The survey link is distributed via email to all otolaryngology residents and fellows-in-training who are AAO-HNS members. In addition, program directors and SRF resident representatives (1-2 residents per program who volunteer as a liaison between the SRF and their coresidents) are contacted through email to encourage participation in the survey. Membership in the AAO-HNS is not required to access and complete the survey. A broad range of questions, including demographics, features of their current residency programs, and future plans, are asked. Results are anon- ymous. Each resident can complete the survey once per year. The number of questions varies each year, but to allow for analysis of longitudinal trends, the majority of the survey is unchanged. Although the survey was started in 2002, individ- ual responses have been available only since 2008. Data used for this study included all responses from 2008 to 2014. This study uses a database that is already established and has never contained any identifiers. Therefore, institutional review board approval was not needed. Statistical analysis was performed using SPSS (version 22; SPSS, Inc, an IBM Company, Chicago, Illinois). The x 2 trend tests were conducted to analyze differences in data. Univariate logistic regression analysis was used to identify statistically significant predictors of pursuing fellowship as the dependent variable (coded as either declared intent of pursuing fellowship or already matched). A multivariate logistic regression model using a forward stepwise regression was then constructed using the decision to pursue fellowship as the dependent vari- able. Due to age not being asked in 2008, survey data used to construct the regression model came from 2009 through 2014. Model explanatory power and fit was assessed using the c -statistic and Hosmer-Lemeshow goodness-of-fit test. An a of 0.05 level of significance was used for all tests. Results From 2008 to 2014, there were 2422 survey respondents, with a mean response rate of 26% ( Table 1 ). Demographic information is shown in Table 2 . Descriptive detail of the percentage of residents falling in each category for variables

Table 1. Number of Survey Respondents Each Year and Response Rate.

Respondents by Year (N = 2422)

No. (%) of Total

Response Rate, %

2008 2009 2010 2011 2012 2013 2014

412 (17) 139 (6) 396 (16) 347 (14) 336 (14) 331 (14) 462 (19)

31 11 30 26 26 25 35

of interest is shown in Figure 1 . Interest in fellowship training has fluctuated between 2008 and 2014 ( Figure 2 ), as have plans for future practice types ( Figure 3 ). The data in this study show a decrease in fellowship interest as residents progress through training. Of 1921 postgraduate year (PGY) 1 through 5 residents, 79% of those in PGY-1, 73% in PGY-2 and PGY-3, and 64% in PGY-4 and PGY-5 reported planning to pursue fellowship or had already matched ( x 2 = 28.84, P \ .0001). This was observed for both males and females. Educational debt was also found to significantly correlate negatively with a resident’s response regarding plans to pursue fellowship. Of 1344 resident respondents who reported their educational debt status and had not already matched into fellow- ship, 17% reported no debt, and the most frequently reported amount of educational debt was $150,000 to $200,000, with 20% reporting this amount. Fellowship plans differed between groups, with 77% of respondents with no educational debt reported pursuing a fellowship, while this percentage was statis- tically significantly lower for each category of educational debt ( x 2 = 28.7, P \ .0001). Older age, higher training level, any amount of educational debt, having dedicated research time . 6 months during resi- dency, and nonacademic practice setting all showed a statisti- cally significant negative correlation to fellowship interest on univariate logistic regression analysis ( Table 3 ). Sex was not significant. The multivariate logistic regression model con- structed using the variables with P \ .05 on univariate analy- sis was statistically significant ( x 2 = 507.69, P \ .0001), and the Hosmer-Lemeshow goodness-of-fit test indicated accepta- ble model fit to the data ( x 2 = 1.532, P = .992). Statistically significant predictors included in the final model were age category older than 29 years, training level at PGY-4 or PGY- 5 level, all amounts of educational debt, and academic practice setting. The area under the receiver operating characteristic (ROC) curve, or c -index, for the final model was 0.86, indicat- ing excellent discriminatory power ( Table 4 ). Discussion Using the results of a survey administered to residents and fellows in otolaryngology–head and neck surgery, we found

176