Lw. Greenberg et Pr. Getson, ASSESSING STUDENT PERFORMANCE ON A PEDIATRIC CLERKSHIP, Archives of pediatrics & adolescent medicine, 150(11), 1996, pp. 1209-1212
Objective: To determine whether any of 4 parameters used as evaluation
methods in the pediatric clerkship at Children's National Medical Cen
ter, Washington, DC, could predict a student's performance, as measure
d by the final grade. Design: A retrospective study in which the follo
wing data were collected: (1) a record of the diagnoses and total numb
er of patients seen during the rotation, (2) clinical performance grad
e, (3) the National Board of Medical Examiners (NBME) pediatric shelf
test score, (4) case presentation grade, and (5) the final clerkship g
rade for overall performance. Setting: Third-year pediatric clerkship
in the 1987-1988 academic year at Children's National Medical Center,
the George Washington University School of Medicine. Results: A total
of 128 students had complete data. Correlations among the clinical par
ameters, ie, number of patients seen, clinical grade, case grade, and
the NBME shelf test score were all statistically nonsignificant. In ad
dition, multiple regression modeling of NBME test scores, using measur
es such as clinical grade, average case grade, and number of patients
seen, was unsuccessful, with only clinical performance entering the re
gression model. In contrast, modeling of the final clerkship grade res
ulted in explaining almost 80% of the variation on a student's final g
rade (R(2)=0.79). Variables submitted to the model were number of pati
ents seen, clinical grade, average case presentation grade, and ordere
d examination score. Last, a highly significant relationship (chi(2)=1
5.98, P<.001) was noted between students receiving a final grade of ho
nors and an ''A'' on their case grade. Conclusions: Results confirmed
that there is no single best predictor of performance on a pediatric c
lerkship. While there was only a weak correlation between each single
measure and the final grade, together these accounted for 80% of the v
ariation in students' scores. Based on the study data, the use of vari
ed evaluative methods to determine a student's final grade is recommen
ded.