Originally posted by optimistique
I've heard 'talk' that MCAT verbal scores or O chem grades can predict how well you will do in med school. Wouldnt this just predict 1st & 2nd year performance in basic sciences?
Residency depends on clinical grades. What would you say is a good indication of clinical grades?* Leadership exp? Volunteering? Learning to be EMT or Nurse aide? Personality?
*Discliamer: As premeds, comments are mostly speculation. This thread is for entertainment purposes only.
Mainly LOR's are the best indicator of clincial performance. BUt there ahave been rticles written about this, check"Academic Medicine" Fields et al. 2000
Early Identification of Students at Risk for Poor Academic Performance in Clinical Clerkships
SCOTT A. FIELDS, CYNTHIA MORRIS, WILLIAM L. TOFFLER and EDWARD J. KEENAN
Correspondence: Dr. Scott A. Fields, Department of Family Medicine, Oregon Health Sciences University, 3181 SW Sam Jackson Park Road, Portland, OR, 97201; e-mail: <
[email protected]>.
Introduction
_
Many medical schools have revised, or are in the process of revising, their curricula.1 The impetus for this curricular change has been dependent on many factors. These factors include grant initiatives emphasizing the development of curricula to promote generalism and the Association of American Medical Colleges' Medical School Objectives Project (MSOP), as well as significant shifts in the health care system, such as the growing influence of managed care. The more innovative curricular revisions to date have included multidisciplinary, integrated courses with longitudinal curricula and early clinical experiences throughout the first two years (the preclinical curriculum).
Oregon Health Sciences University (OHSU) School of Medicine implemented its curriculum revision in 1992.2 The result of this effort was the restructuring of the first two years of the curriculum from 24 specific discipline-based courses to ten interdisciplinary units. One of the units, the Principles of Clinical Medicine (PCM), is a longitudinal two-year course composed of small-group activities half a day each week and a weekly half-day clinical preceptorship. In addition, there are nine integrated basic science courses in the first two years, and a one-week course, Transition to Clerkship, occurs at the end of the second year. The core clerkships, constituting the entire third year, include medicine, surgery, obstetrics-gynecology, family medicine, psychiatry, pediatrics, and rural primary care. Each of these clerkships is six weeks in duration with the exception of medicine, which occurs in two six-week blocks.
The premise for this study was that early identification of medical students who are at academic risk provides a basis for intervention with individualized remedial programs. Previously, studies have investigated predictors of performance for years one and two of medical school.3,4 Little has been done to address early identification of students at risk for academic difficulty in the third year of medical school. The hypothesis was that performance in PCM during the predominantly pre-clinical curriculum of the first two years predicts students at risk for academic difficulty in the clinical clerkships. Accordingly, this study analyzed the relationship between parameters of student assessment, including a number of admission, curriculum, and standardized testing criteria, and an accepted standard of graded performance in the third-year core clerkships.
__ METHOD TOP
Introduction
Method
Results
Discussion
References
_
The sample studied was a cohort of students beginning with those who matriculated at OHSU from 1992 to 1995 and who graduated between 1996 and 1999. Student data were available from OHSU databases; no major change in curriculum, grading policy, or calculation of student grade-point averages occurred during these years. In the study, all individual student performance data were treated as confidential.
The primary outcome of this analysis was performance in the core clinical clerkships of the third year curriculum, which serves as a critical component of the residency application process. All courses at OHSU, including clerkships, are graded as honors, near honors, satisfactory, marginal, or fail. Grade-point average (GPA) in year three was used as the outcome, with a GPA of 3.0 representing honors; 2.0 near honors; 1.0 satisfactory, and 0 marginal/failure. After initial analysis as a continuous variable, we identified the lowest quintile of performance in year three (GPA < 2.0).
A number of potential indicators were considered to predict performance in year three. These indicators included cumulative college GPA, separate MCAT scores (Verbal Reasoning, Biological Science, Physical Science, and Writing Sample), year one and year two basic science course performance as a mean percentage examination score, performance in the PCM course, and USMLE Step 1 score. The total MCAT score combined the Verbal Reasoning, Biological Science, and Physical Science scores. For the Writing Sample, the alphabetic score was coded from 4 to 15, with M = 8. Total points for the PCM course were used. In PCM, there are 80 points available for each of six quarters: 20 points for the clinical preceptorship, 10 points for small-group discussion activities, 10 points for patient examination activities, 10 points for an essay, 10 points for written exam, and 20 points for a group objective structured clinical examination (GOSCE).3
The first series of analyses were univariate, with all continuous predictor variables correlated with the primary outcome, year three GPA. Subsequently, a parsimonious logistic regression model was fit to predict this outcome using forward selection procedures. The odds of low performance (year three GPA < 2.0) were estimated. Cutoff points for categorizing each continuous predictor variable were based on the lowest quintile of each score or percentage, with latitude for ties. The significance of each predictor variable was assessed using a likelihood-ratio test statistic obtained from a logistic regression model fit to the outcome status.
__ RESULTS TOP
Introduction
Method
Results
Discussion
References
_
In total, data for 306 students were available. All data were complete except for one student who had attended a college without grades, seven students who had taken the earlier version of the MCAT, and two students whose USMLE scores were unavailable.
Correlation coefficients were obtained for each performance indicator as compared with the year-three GPA. Of all variables, this outcome was most significantly related to the score in the PCM course (r =.61, p <.001); year two percentage performance (r =.54, p <.001); year one percentage performance (r =.52, p <.001); and USMLE 1 score (r =.47, p <.001). Year-three GPA was only modestly related to undergraduate GPA (r =.19, p <.05) and MCAT Writing Sample score (r =.16, p <.05), and was not related significantly to the total MCAT score, or to the Biological Science, Physical Science, and Verbal Science Subscores. Figure 1 shows the relationship between the year-three GPA and the PCM score.
View larger version (31K):
[in this window]
[in a new window]
_ Figure 1. Scatterplot of the relationship between Principles of Clinical Medicine (PCAA) scores and year three grade-point averages at Oregon Health Sciences University School of Medicine, 1992-1999.
_
Prior to logistic regression analysis, the relationship of each variable to performance in the lowest quintile of year-three GPA was analyzed in order to determine the accuracy of prediction. Each was dichotomized by the lowest quintile and compared in a 2 x 2 table with low year-three GPA. A score in the lowest quintile of PCM ( 380) correctly predicted low year three performances of 38 of 68 students (positive predictive value = 56%). Of 238 students who had score above the lowest quintile, 212 (negative predictive value = 89%) had year-three GPAs above the lowest quintile. These values were similar considering performance in the lowest quintile in year two (positive predictive value = 53%, predicting 36 of 68; negative predictive value = 89%, 211 of 238). A USMLE Step 1 score in the lowest quintile ( 190) correctly predicted 28 of 68 students who scored in the lowest quintile of year-three GPA (positive predictive value = 41%), whereas a score above the lowest quintile predicted 206 of 238 (negative predictive value = 87%). No other variable performed similarly in univariate analysis.
A multivariate logistic regression model significantly predicted low year-three GPA (p <.001). (See Table 1.) Overall, performance in the lowest quintile in PCM was associated with a 9.45 times increased risk of performance in the lowest quintile of year-three GPA (95% CI, 4.71-18.98). Similarly, performance in the lowest quintile of year two conferred a 6.39 times risk of low year-three GPA (95% CI, 2.96-13.80). This model also included performance in the lowest quintile of the USMLE Step 1, although this was not significant (relative risk 1.83; 95% CI, 0.84-3.99). Last, performance by quintile of PCM score, after adjustment for USMLE Step 1 score and year-two percentage score, was linearly rela