feedback (47% versus 27%) (Table). When there was disagreement between residentand examiner, learning goals favored self-assessment over feedback (4% versus 1%).

Conclusions: The findings highlight the perceived strengths and weaknesses inclinical competence and self-assessment skills and provide direction for programtraining needs. While self-assessments may not realistically indicate ability, it is stillcritical to determine how students perceive their ability because their opinions drivetheir learning goals. The interactive examination appeared to be a convenient tool forproviding deeper insight into students’ ability to prioritize, self-assess and steer theirown learning.

114 Emergency Medicine Residency Applicant InterviewScoring: Can a Global Visual Analog Scale ProduceReliable Results?

Roper J, Melanson S, Jeanmonod R/St. Luke’s University Hospital,Bethlehem, PA

Study Objective: The interview process is integral to choosing emergencymedicine residents. Historically, the interview score has accounted for 75% of thevariability in rank position at our institution, and this has been reported in anotherstudy. There is little data exploring the reliability of interview scoring in emergencymedicine . We sought to determine if a single global visual analog scale (VAS) toscore applicants would create a broad enough spread of values to be useful while beingreliable between faculty interviewers.

Methods: This is a retrospective observational study performed at a community-based university-affiliated institution with 8 allopathic emergency medicine positionseach year. Applicants in the 2011 pool were interviewed by 3 or 4 individuals drawnfrom a pool of 9 academic faculty and 1 chief resident. Interviewers scoredapplicants using a VAS which was 100mm long with the words “bottom of therank list” on one side and “top of the rank list” on the other. Interviewers wereinstructed to place a single hash mark on the VAS corresponding to where theybelieved the applicant should be ranked. The distance in millimeters from the“bottom of the rank list” side was measured and recorded as the applicant’s score.Interviewers had access to the applicants’ academic records and board scores for15 to 30 minutes prior to the interview. Scores by each of the 10 interviewerswere compared to determine if any interviewer scored applicants consistentlylower or higher than the global pool of interviewers. Each applicant’s scores wereanalyzed for reliability among the interviewers. Regression analysis was performedon interview scores as compared to board scores as well as rank position. Scoresfor the entire interview season were examined through descriptive statistics. Thisstudy was institutional review board exempt.

Results: Over 600 applications were received in the 2011 season, and about 100applicants were invited to interview. 91 applicants accepted the invitation and wereinterviewed during the 2011 interview season. The mean interview score for theseapplicants was 61.9 (median 61.7) with a standard of deviation of 16.5. The 10interviewers did not differ significantly in their scoring of applicants (ANOVAp�NS). For applicants interviewed by 4 interviewers, the intraclass correlation was0.85. For those interviewed by 3 interviewers, the intraclass correlation was 0.84. Inregression analysis of interview scores and USMLE Step 1 scores, r � 0.28 (rsquared � 0.08). In regression analysis of interview scores and final rankposition, r � 0.87 (r squared � 0.75).

Conclusion: A global VAS provides a wide distribution of scores and is a reliablemethod for assessing interview applicants with 3 or 4 interviewers. Interview scorewas responsible for 75% of the variability in position on the rank list. There was onlya weak relationship between interview score and USMLE Step 1 score. Future studieswill determine if applicant interview scores correlate with clinical success in residency.

115 Review of Emergency Medicine Publications:Methodology, Ethics and Setting

Ducasse J, Dehours E, Houze Cerfon V, Bounes V/Samu 31 - Chu Purpan,Toulouse, France

Study Objective: To describe emergency medicine publications in terms ofmethodology, approval by IRB, method of consent, external validity and setting (eg,out-of-hospital or emergency department).

Methods: The 12 top ranked emergency journals were selected. We manuallyreviewed the last 30 original articles in each emergency medicine journal, to representmore than 2 months of publications for all emergency medicine journals (range 2 to 6months). Only clinical original articles on human subjects were included. To ensureaccurate data transcription, each article was read at least twice by 2 different reviewersand graded by written criteria using an extraction standard chart.

Results: Over the articles reviewed, 330 were analyzed. 189 (57.3%) wereprospective studies, 29 (8.8%) were randomized studies. 226 studies (68.5%)mentioned an institutional review board approval or a waiver of authorization, and aninformed consent was not mentioned in 227 (68.8%) of studies. 59 (17.9%) wereconducted in a out-of-hospital setting. Two hundred and thirty 8 (72.1%) of thesestudies were at single center institutions, the USA contributed 158 (47.9%) of thetotal publications. Trauma, Management-Overcrowding-Policies and Cardiology(including cardiac arrest) represented more than 50% of all the publications.

Conclusion: This study describes publications in the field of emergency medicine.Randomized studies represent 9% of publications, most studies are cross sectional,and more than half have a retrospective design. We found that in one third of thestudies, an institutional review board review was not mentioned and informedconsent was not specified in 2/3 studies. Emergency medicine research volume,quality, and grants activity must increase in order for emergency medicine to progresswithin academic medicine.

116 Impact of Learners on Emergency MedicineAttending Physician Productivity

Bhat R, Maloy K, Dubin J/Georgetown University Hospital, Washington, DC;Washington Hospital Center, Washington, DC

Background: Several prior studies have examined the impact of learners (medicalstudents or residents) on overall emergency department flow as well as the impact ofresident training level on the number of patients seen by residents per hour. No studyto date has specifically examined the impact of learners on emergency medicineattending physician productivity, with regards to patients per hour (PPH). We soughtto evaluate whether learners increase, decrease, or have no effect on the productivityof emergency medicine attending physicians in a teaching program with one studentor resident per attending.

Methods: This was a retrospective database review of an urban, academic tertiarycare center with 3 separate teams on the acute care side of the emergency department.Each team was staffed with one attending physician paired with either one resident,one medical student or with no learners. All shifts from July 1, 2008 to June 30th2010 were reviewed using an electronic database. A shift was predefined as “Resident”if � 5 patients were seen by a resident, “Medical Student” if any patients were seenby a medical student, and “No Learners” if no patients were seen by a medicalstudent or resident. Shifts were removed from analysis if more than one learner sawpatients during the shift. Resident shifts were further stratified by emergencymedicine training level or off service rotator. For each type of shift, the total numberof patients seen by the attending physician was then divided by 8 hours (shiftduration) to arrive at number of patients per hour.

Results: A total of 7360 shifts were analyzed with 2778 removed due to multiplelearners on a team. For the 2199 shifts with attending physicians with no learners, theaverage number of PPH was 1.87(95% CI 1.86,1.89). For the 514 medical studentshifts, the average PPH was 1.87(95% CI 1.84,1.90), p � 0.99. For the 1935resident shifts, the average PPH was 1.99(95% CI 1.97,2.00), p �0.005, with nostatistically significant difference found between emergency medicine 1: 1.98,emergency medicine 2: 1.99, emergency medicine 3: 1.99, off-service: 1.99.

Conclusion: The addition of a resident to an attending team in a one-on-oneteaching model increased attending productivity while the addition of a medicalstudent had no impact. The results of this study may help guide emergencydepartments seeking to expand or establish a residency training program to assess theproductivity impact of this decision.

Research Forum Abstracts

S42 Annals of Emergency Medicine Volume , . : October