Publication Misrepresentation Among Radiation Oncology Residency Applicants

Volume 87  Number 2S  Supplement 2013 Author Disclosure: E. Holliday: None. R. Jagsi: None. C.R. Thomas: None. L.D. Wilson: None. C.D. Fuller: None.

2785 Gender Distributions in Radiation Oncology Residency Compared to Other Specialty Residencies: A 20-Year Perspective G. Dhami, S. Amarnath, W. Gao, A.D. Trister, and G.E. Laramore; University of Washington Affiliated Hospitals, Seattle, WA Purpose/Objective(s): Physician gender parity has been shown to play an important role in patient satisfaction. Over the last 20 years, there has been an increase in the number of women matriculating into US medical schools (29% in 1980 to 48% in 2012). However, a corresponding change in gender distribution among residents has not been noted in all medical specialties over this same time period. We assessed trends of gender distribution in various specialties over the last 20 years, with specific focus on radiation oncology. Materials/Methods: The Association of American Medical College FACTS database, Group on Women in Medicine and Science, and Journal of the American Medical Association Medical Education supplement, were queried for resident gender distribution data. Data was available for radiation oncology, ENT, plastic surgery, general surgery, urology, orthopedics, dermatology, internal medicine, family medicine, OB/Gyn, radiology, pediatrics, and ophthalmology. The percentage of women in US residencies per year was collected for the years 1989-2011. Granger causality testing was performed to compare trends in percentage of female residents, and ARIMA was used to develop forecasts of future distribution of women in these specialties. Results: In 1989, women represented the minority of residents in most fields, but the gap was most pronounced in procedural subspecialties (orthopedics 5%, urology 5.3%, radiation oncology 24.7%, and internal medicine 27.4%). Between 1989 and 2011, most specialties have had significant increases in number of female residents entering the field, notably urology (350% from 1989) and general surgery (180% from 1989). In contrast, there was minimal increase in female residents entering radiation oncology, with 25.2% females in 1989, and 32.3% in 2011 (27% increase), and no increase between 1994 and 2011 (32.3% in both years). Of the specialties included in our survey, orthopedic surgery (p Z 0.03) and diagnostic radiology (p Z 0.03) were the only other specialties that had as minimal increases in gender parity by Granger causality. Furthermore, ARIMA forecasts for radiation oncology predict no growth in the next five years. Even at the upper end of the 95% confidence interval, this would represent a very modest change in the percentage of women in radiation oncology residency. Conclusions: Despite significant increase in enrollment of women in United States medical schools over the last 20 years, this has not translated into an increase in female resident representation in radiation oncology, in contrast to other specialties. Projected trends show this will not change in the next five years. Further studies are needed to determine the potential barriers to entering the field and to improve the recruitment of female medical students into radiation oncology. Author Disclosure: G. Dhami: None. S. Amarnath: None. W. Gao: None. A.D. Trister: None. G.E. Laramore: None.

2786 The Radiation Oncology In-Training Exam: An Appeal for Better Testing A. Morris; University of Iowa, Iowa City, IA Purpose/Objective(s): Resident physician assessment plays a key role in graduate medical education. As of 1992, 20 medical specialties offered an in-training exam for the purposes of formative assessment. In 1984, the American College of Radiology developed an in-training exam (TXIT) for radiation oncology. Consisting of approximately 350 questions, this exam addresses the topics of radiation biology, radiation physics and clinical oncology. Appropriate use of the TXIT has not been determined, yet 33%

Poster Viewing Abstracts S503 of radiation oncology residents surveyed in 2004 felt the exam had a moderate to significant impact on their annual review. Given this perception, critical evaluation of the evidence supporting use of the TXIT is needed in order to optimize use of this exam in formative assessment. Materials/Methods: A review was conducted of all literature published on the topic of the American College of Radiology TXIT between 1984 and 2012. All relevant papers were analyzed for evidence of psychometric reliability and validity, using a framework of content validity, construct validity, and concurrent/predictive validity. Results: From 1988 to 2008, four papers were published on the topic of the ACR in-training exam (TXIT). Only one article addressed the issue of reliability, demonstrating a mean item difficulty of 60.7%  22.4% with mean item discrimination of 0.19 and an overall reliability coefficient of 0.92. While favorable, these values were generated from a single year of data (2007). A review of the 2004-2007 TXIT attempted to address content validity by showing that each exam contains a similar number of questions on each of 12 clinical subsites; no data was provided, however, to demonstrate the clinical relevance of the items themselves. Content analysis of the 2012 TXIT revealed 53 items (17%, n Z 321) in which all four foils were percentages and 43 items (13%) that focused on a clinical trial; these items may or may not be viewed as directly relevant to clinical practice. Regarding construct validity, articles from 1988, 1992 and 2008 showed that TXIT scores tend to improve with year in training. Residents who took the exam from 1984-1991 and 2004-2007, however, only improved their average clinical score by 8.5% over the course of training (average score: PGY2 54.6-71.5%, PGY5 65.3-77.8%). No inferential statistics have been published on this data. Conclusions: Very little evidence exists regarding the psychometric properties of the ACR TXIT. Reliability data exists for one administration of the test, and average scores show a modest trend towards improvement with increased years in training. Evidence for content, concurrent and predictive validity, however, remains lacking and represents a large gap in the literature, diminishing the exam’s utility. Future research should focus on defining the role of the TXIT as a formative assessment in resident education through rigorous psychometric analysis. Author Disclosure: A. Morris: None.

2787 Publication Misrepresentation Among Radiation Oncology Residency Applicants S.R. Amarnath, A.D. Trister, and R.D. Ermoian; University of Washington Affiliated Hospitals, Seattle, WA Purpose/Objective(s): Radiation oncology has become an increasingly competitive Match specialty. Applicants’ publication records are increasingly used as a metric of assessment; however, ensuring the accuracy of citations is increasingly difficult as the number of publications per applicant continues to increase each year (6.3 in 2007 to 8.3 in 2011). Publication misrepresentations in residency applications have been well described in multiple specialties, including radiation oncology (for the 2004 Match). In this study, we assessed changes in the rate and types of publication misrepresentation in radiation oncology applicants participating in the 2011 match compared to the 2004 Match. Materials/Methods: The Electronic Residency Application Service (ERAS) was queried for all complete applications submitted to our institution by United States graduates in the 2011 match. We identified 176 applicants out of 181 total United States applicants (97%) and excluded 3 applicants matched to our program. Applicant data that we collected included other graduate degrees and significant research experiences. We then identified all listed peer-reviewed manuscripts and checked each citation for accuracy using PubMed, Google Scholar, Web of Science, and journal websites. Each inaccuracy was counted and categorized by the type of misrepresentation. Results: Fifty-one percent of all applicants had significant research experiences: 33% of applicants had other graduate degrees (35 PhD, 26 master degrees) and 40 applicants had spent a median of one year in a non-

S504

International Journal of Radiation Oncology  Biology  Physics

degree related research position. One hundred sixty-two of 173 (94%) of all applicants listed publication citations on their applications compared to 42% in 2004. Sixty-four percent of 2011 applicants listed peer-reviewed publications or abstracts. In total, 1597 citations were listed, of which 523 (33%) were recorded as peer-reviewed manuscripts. We found a total of 125 (23.9%) incorrect manuscript citations submitted by 39 applicants (22.5% of all applicants) compared to 9% of all applicants listing incorrect citations in 2004. Types of misrepresentation included: change in authorship order (20.4%), incorrect title of publication (18.4%), not in peerreviewed journal (6.1%), incorrect journal listed in citation (12.2%), manuscript not published at time of application (8.2%), unable to find publication using search tools (16.3%) and other (18.4%). Even though applicants with more research experience were more likely to have a larger number of publications (p Z 0.001), they were not more likely to misrepresent (p Z 0.25). Conclusions: Publication misrepresentations are increasingly seen in applications to radiation oncology. Future study will focus on specific applicant factors that may predict for ethical breaches and ways to mitigate misrepresentation among applicants. Author Disclosure: S.R. Amarnath: None. A.D. Trister: None. R.D. Ermoian: None.

years with subtle increases for Asians and Females, and either decrease (Blacks) or no significant changes (AI/AN, NH/PI, and Hispanics) for traditionally underrepresented groups in medicine. Author Disclosure: C. Deville: None. W. Hwang: None. S. Both: None. C. Thomas: None. C.H. Chapman: None.

2788 United States Radiation Oncology Residency Diversity Over the Past 20 Years C. Deville, W. Hwang, S. Both, C. Thomas, and C.H. Chapman; University of Pennsylvania, Philadelphia, PA Purpose/Objective(s): Women and traditionally underrepresented groups in medicine (URM) e Blacks, American Indians and Alaska Natives (AI/ AN), Native Hawaiians and Pacific Islanders (NH/PI), and Hispanics e are currently underrepresented in the US Radiation Oncology physician workforce. The purpose of this study was to assess changes in Radiation Oncology resident diversity by race, Hispanic ethnicity, and sex, over the past 20 years. Materials/Methods: Racial groups included: White; Black or African American, referred to as Black; Asian or Asian American, referred to as Asian; AI/AN; NH/PI; and Other, here defined as any person with unknown racial information and/or not classifiable into previous categories. Ethnic groups included Hispanic and non-Hispanic. Resident data for academic years 1991-1992 to 2011-2012 were accumulated from annual Journal of the American Medical Association (AMA) supplements, reporting AMA/AAMC database information on Accreditation Council for Graduate Medical Education training programs. Data for Asian, Black, Other, and Hispanics were first reported for 1995-1996, and for AI/AN and NH/PI in 2001-2002. To assess the changes in the percentages of different race, ethnicity, and sex in residents over 20 academic years, the slope and the associated 95% confidence intervals for each group were estimated using a simple linear regression model where year was used as independent variable. Results: Minimum and maximum percentages (in the most recent year recorded) were for: males 65.2% (2007), 78.9% (1999); Whites 50.7% (1999), 63.8% (1995); Asians 21.3% (1995), 35.8% (2001); females 27.2% (1996), 34.8% (2007); Hispanics 1.6% (1997), 5.2% (2001); Blacks 2.6% (2008), 6.4% (1998); AI/AN 0.0% (2003), 0.4% (2005); and NH/PI 0.0% (2011), 0.9% (2007). When analyzing differences over time (change in percent/year; 95% CI; p value), representation was found to be increasing for Asians (0.6544; 0.2688, 1.0223; p Z 0.002) and females (0.2650; 0.1349, 0.395; p < 0.001), unchanged for Whites (0.1994; 0.2141, 0.6128; p Z 0.320), AI/AN (0.0183; 0.0113, 0.0478; p Z 0.196), NH/PI (0.0446; 0.0969, 0.0078; p Z 0.087), and Hispanics (0.0551; 0.0483, 0.1584; p Z 0.274), and decreasing for Blacks ( 0.1353; 0.2227, 0.0480; p Z 0.005) and other ( 0.5214; 0.9093, 0.1335; p Z 0.012). Conclusions: Despite the ongoing diversification of medical school graduates over the past 20 years, overall diversity by race, ethnicity, and sex, of Radiation Oncology residents has changed minimally over the past 20

2789 WITHDRAWN

2790 Correlation of Clinical and Dosimetric Parameters With Radiographic Lung Injury Following Stereotactic Body Radiation Therapy A. Kishan, J. Wang, V. Yu, D. Ruan, M. Cao, S. Tenn, D. Low, and P. Lee; University of California - Los Angeles, Los Angeles, CA Purpose/Objective(s): Radiographic lung density changes occur in over half of patients treated with stereotactic body radiation therapy (SBRT) and correlate histopathologically with injury. We sought to quantify radiographic lung density changes at 3, 6 and 12 months, and investigate the relationship between the volume of density change and various clinical and dosimetric parameters. Materials/Methods: The primary study population consisted of patients treated with SBRT to the lung for stage I primary lung cancers (n Z 39) or oligometastatic lesions (n Z 17). Fractionation schemes included three fractions of 10, 12, 14, or 18 Gy and four fractions of 12 or 12.5 Gy prescribed to cover 95% of the planning target volume (PTV). Initial planning computed tomography (CT) scans were rigidly registered to follow-up CT scans obtained at intervals of 3, 6, and 12 months. Regions of density change, as determined by two investigators (A.K. and P.L.), were contoured on the follow-up scans. A threshold dose (TD) was defined as the dose to the total lung corresponding to the volume of the density change normalized to the treatment dose. The percentage of overlap between density changes and isodose curves were compared at 6 and 12 months. Associations between the volume of density changes and clinical and dosimetric parameters were investigated using multivariable linear regression. Results: Follow-up data were available for 65 and 47 lesions at 6 and 12 months, respectively. Age, years since quitting smoking, and GOLD score (if applicable) were significantly associated with increasing volume of density change (p < 0.05). The total dose, PTV, and volumetric parameters (V0 to V55) were also significantly associated with increasing volumes of density change (p < 0.01). The strength of the association was strongest at V55. The average TD was 84.0% and 90.1% of the treatment dose at 6 and 12 months, respectively. The average percentage of overlap between density changes and the 90% isodose curves was 45.0% and 48.7% at 6 and 12 months, respectively. Conclusions: Age, significant smoking history, and GOLD score were significantly associated with increasing volumes of radiographic density change following SBRT. Further, there is a strong dose-response relationship between the volume of lung exposed to a certain dose and the volume of density change. The overlapping regions between the density changes and the 90% isodose curves correspond to nearly 50% of the density changes, volumetrically, while the TD that would theoretically create a volume equaling that of the density changes at 6 and 12 months is between 84% and 90% of the treatment dose. Author Disclosure: A. Kishan: None. J. Wang: None. V. Yu: None. D. Ruan: None. M. Cao: None. S. Tenn: None. D. Low: None. P. Lee: None.

2791 Radiation Pneumonitis Following Simplified Intensity Modulated Radiation Therapy for Lung Cancer L. Zhou,1 J. Liu,2 Y. Zhang,3 Y. Gong,1 S. Bai,3 and Y. Lu1; 1Department of Thoracic Cancer, Cancer Center and State Key Laboratory of Biotherapy, West China Hospital of Sichuan University, Chengdu, China,