Combined Versus Detailed Evaluation Components in Medical Student Global Rating Indexes

To determine if there is any correlation between any of the 10 individual components of a global rating index on an emergency medicine (EM) student clerkship evaluation form. If there is correlation, to determine if a weighted average of highly correlated components loses predictive value for the final clerkship grade.

This study reviewed medical student evaluations collected over two years of a required fourth-year rotation in EM. Evaluation cards, comprised of a detailed 10-part evaluation, were completed after each shift. We used a correlation matrix between evaluation category average scores, using Spearman's rho, to determine if there was any correlation of the grades between any of the 10 items on the evaluation form.

A total of 233 students completed the rotation over the two-year period of the study. There were strong correlations (>0.80) between assessment components of medical knowledge, history taking, physical exam, and differential diagnosis. There were also strong correlations between assessment components of team rapport, patient rapport, and motivation. When these highly correlated were combined to produce a four-component model, linear regression demonstrated similar predictive power in terms of final clerkship grade (R(2)=0.71, CI95=0.65-0.77 and R(2)=0.69, CI95=0.63-0.76 for the full and reduced models respectively).

This study revealed that several components of the evaluation card had a high degree of correlation. Combining the correlated items, a reduced model containing four items (clinical skills, interpersonal skills, procedural skills, and documentation) was as predictive of the student's clinical grade as the full 10-item evaluation. Clerkship directors should be aware of the performance of their individual global rating scales when assessing medical student performance, especially if attempting to measure greater than four components.