의과대학 입학 전형 요소들의 4년 후 레지던트 성적 예측성

의과대학 입학 전형 요소들의 4년 후 레지던트 성적 예측성

Effectiveness of medical school admissions criteria in predicting residency ranking four years later

Christopher Peskun, Allan Detsky & Maureen Shandling

배경

캐나다의 의과대학들은 우수한 지원자 풀에서 학생을 선발하기 위해서 많은 노력을 기울이고 있다. 의학을 행하는데(practice of medicine) 있어서 기본적으로 중요한 비인지적 평가 대부분의 학교에서 이뤄지고 있다. 우리는 University of Toronto의 학업적, 비학업적 입학 평가요소를 분석하여 내과와 가정의학과 레지던트 프로그램에서의 을 예측하고자 하였다.

방법

연구 대상은 University of Toronto에 1994년부터 1998년 사이에 입학한 학생들 중, 캐나다 레지던트 매칭 프로그램을 통해 졸업시 University of Toronto의 내과나 가정의학과에 매칭된 학생들이다. 입학 요소들이 의과대학 기간, 또는 레지던트 기간의 수행능력에 미치는 영향을 보았다.

결과

내과 레지던트의 랭킹은 학부의 학점평균, 그리고 입학시의 비인지적

평가요소와 유의한 상관관계를 보였다. 또한 2학년의 OSCE점수, 내과 임상실습 점수, 의과대학 최종 성적과도 연관성을 보였다. 

가정의학과 레지던트의 랭킹은 입학시 면접 점수, 2학년의 OSCE점수, 가정의학과 임상실습 점수, 내과병동 임상실습 점수, 그리고 의과대학 최종 성적과 연관성을 보였다.

결론

의과대학 입학시 평가하는 인지적 요소와 비인지적 요소 모두 졸업후의 성공수준을 예측하는데 중요하게 작용했다. 비인지적 평가요소들은 레지던트 프로그램에서의 랭킹을 예측함으로서 입학전형 요소로서의 유용성을 보여줬다.

KEYWORDS *school admission criteria; *schools, medical; Ontario; internship and residency ⁄ *standards; clinical competence ⁄ *standards; education, medical, undergraduate ⁄ *standards.

Overview

What is already known on this subject

Undergraduate GPA and MCAT scores are correlated with academic performance in medical school

What this study adds

Non-cognitive measures assessed at admission to medical school correlate with non-cognitive performance measures in medical school as well as overall residency ranking

Suggestions for further research

Methods for optimising cognitive and noncognitive evaluation of students at the time of medical school application to predict future success in Medicine

---

Statistical techniques

Using both univariate and multivariate techniques the analyses described above were completed. The numerical rankings of candidates applying to Internal Medicine were collapsed into 3 approximately equal categories: rank 1–40, 41–90 or >90. These categories were chosen based on the fact that those in the top category had a very high likelihood of acceptance, whereas those in the lowest category were certain not to be accepted. As a result of this categorical designation, a logistic regression model based on binary collapse was utilised during analysis.

Because there was less predictability in the matching process in Family Medicine, ranking was weighted equally from highest to lowest and was treated as a continuous variable. A linear regression model was therefore used in the analysis involving Family Medicine residency rank as the outcome variable.

The results are expressed in terms of slope of the regression line. As a result of these differences in data and statistical analysis, results for Internal Medicine are expressed throughout in terms of odds ratios, while those for Family Medicine are expressed in terms of slope. The units of the odds ratios for Internal Medicine results and the slopes for Family Medicine results were standardised for 1 standard deviation (SD) of the variable of interest. 

Through this standardisation, comparison of the results between both programmes is possible. A small proportion of candidates had applied for postgraduate training in both Internal Medicine and Family Medicine. Due to the blinded nature of the residency application process, these applicants had the ability to be ranked equally and independently by each residency programme and as such were considered to be independent members of both cohorts.

For all analyses involving the continuous outcome variables of OSCE score and final grade in medical school the results were expressed in terms of slope. The units of slopes in these cases were once again standardised for 1 SD of the variable of interest.

---

Med Educ. 2007 Jan;41(1):57-64.

Effectiveness of medical school admissions criteria in predicting residency ranking four years later.

Peskun C, Detsky A, Shandling M.

Source

University of Toronto, University of Toronto, University of Toronto, Canada. chris.peskun@utoronto.ca

Abstract

BACKGROUND:

Medical schools across Canada expend great effort in selecting students from a large pool of qualified applicants. Non-cognitive assessments are conducted by most schools in an effort to ensure that medical students have the personal characteristics of importance in the practice of Medicine. We reviewed the ability of University of Toronto academic and non-academic admission assessments to predict ranking by Internal Medicine and Family Medicine residency programmes.

METHODS:

The study sample consisted of students who had entered the University of Toronto between 1994 and 1998 inclusive, and had then applied through the Canadian resident matching programme to positions in Family or Internal Medicine at the University of Toronto in their graduating year. The value of admissions variables in predicting medical school performance and residency ranking was assessed.

RESULTS:

Ranking in Internal Medicine correlated significantly with undergraduate grade point average (GPA) and the admissions non-cognitive assessment. It also correlated with 2-year objective structured clinical examination (OSCE) score, clerkship grade in Internal Medicine, and final grade in medical school. Ranking in Family Medicine correlated with the admissions interview score. It also correlated with 2nd-year OSCE score, clerkship grade in Family Medicine, clerkship ward evaluation in Internal Medicine and final grade in medical school.

DISCUSSION:

The results of this study suggest that cognitive as well as non-cognitive factors evaluated during medical school admission are important in predicting future success in Medicine. The non-cognitive assessment provides additional value to standard academic criteria in predictingranking by 2 residency programmes, and justifies its use as part of the admissions process.

PMID:

 

17209893

 

[PubMed - indexed for MEDLINE]