Farrar H, Green J, O'Sullivan P. University of Arkansas for Medical Sciences and Arkansas Children's Hospital, Little Rock, AR.
Poor math problem solving skills of physicians can contribute to medication errors. Pediatric residents have highly variable math problem solving skills, with average test scores of 42 to 87%. There is currently no data on the math skills of medical students.
This retrospective study evaluated the math problem solving skills of a single class of 125 JMS's using a written, clinical problem solving exam. This was an analysis of 4 math questions addressing common pediatric problems, with a total possible score of 0-10 (internal consistency reliability=0.38; inter-rater reliability coefficient=0.91). The overall performance was calculated, and differences in mean scores between quartiles in class rank in pediatrics were evaluated using ANOVA. A one-tailed Pearson correlation coefficient was used to correlate performance in math skills with other measures of student performance during the pre-clinical and clinical (3rd and 4th) years (significance set at p≤0.01 to correct for multiple statistical tests).
The mean score for math problem solving was 67±19%, with 24% of students scoring ≤50% on the test. There were significant differences in performance between the quartiles in pediatric class rank (p=0.005). Math skills significantly correlated with MCAT scores (p=0.01) but not the USMLE Step 1 scores or the GPA after 2 years. During the 3rd year, the NBME subject exam (p=0.005) and the 3rd year GPA (p<0.001) correlated with math skills but not the subjective evaluations while on the inpatient service. USMLE Step 2 scores (p=0.002) and GPA through 3 years (p=0.002) also correlated with math skills scores.
Performance of JMS's on this test was similar to that reported for pediatric residents on similar tests, and was also highly variable. Class rank and performance on standardized tests during the 3rd year may identify students with poor math skills prior to beginning residency.