Behavioral Coding for Evaluation of Medical Student Communication: Clarification or Obfuscation?

Purpose. To understand which specific student behaviors predict performance ratings from standardized patients and behavioral scientist preceptors. Method. In 1996-98, objective, real-time ratings of student verbal and nonverbal behaviors were conducted on 75 videotaped interviews between second-year medical students and standardized patients at University of North Carolina at Chapel Hill School of Medicine. The coding system used in these analyses was developed based on evidence-based literature reviews and used software that provides for real-time recording; 30 nonverbal and 33 behaviors were coded. The coded behaviors were then compared with four ratings: a global rating of the encounter by the standardized patient, a summary score derived from a checklist completed by the standardized patient, a global rating of the encounter by a behavioral science preceptor who observed the encounter, and a summary score derived from a checklist completed by the preceptor. Results. Analyses identified strong correlations between all four of the preceptor and patient ratings, a strong independent effect of case scenario, and significant between-rater variation. When multivariable analysis was used to predict these global ratings based on coding of specific behaviors, a relatively high proportion of observer variation was explained by a small group of coded behaviors. Conclusions. This study suggests that the coding of specific behaviors may hold promise as a student evaluation technique thereby improving medical training techniques and ultimately enhancing the communication skills in physician-patient encounters. It also illustrates the need to better clarify which specific behaviors are most critical in influencing patient satisfaction.