Communication Features Associated with Clinical Performance and Non-technical Skills in Healthcare Settings

Yuhao Peng, Purdue University

Abstract

Effective teamwork and communication are critical to patient outcomes, and subjective assessment tools have been developed for measuring team performance using both technical and non-technical skills. However, inherent biases remain with using subjective assessment tools. The objective of this thesis is to investigate the relationships between objective communication measures (e.g., speech duration, ratio, rate, etc.) and healthcare providers’ clinical performance and Non-technical Skills (NTS) performance in simulated trauma care team scenarios. In this study, 3rd-year medical students participated in the Acute Care Trauma Simulation (ACTS). The student performed the role of clinician in a team that included a nurse and a simulated patient. Participants conducted post-operative patient management, patient care diagnoses, and treatment. Audio from all team members was recorded, and speech variables (e.g., speech duration, frequency of interaction, etc.) from student’s audio were extracted. For Research Question I, correlation and regression models were used to explore the relationships between vocal features and clinical performance; for Research Question II, additional vocal features were extracted from audio recordings, and these features were used to developed multiple regression models relating vocal features with NTS overall scores and with the communication construct of the NTS score. Findings showed that a majority (67%) of the communications were initiated by the student. Speech ratio, intensity, and frequency of communications differed when students communicate with the nurse than with the patient (e.g., communication from student to the patient resulted in a higher intensity). The models for Research Question I showed that increasing frequency of checkbacks between student and nurse (p<0.05) and speech duration from student to patient (p=0.001) significantly increased student’s clinical performance score. In Research Question II, a positive association (ρ=0.456, p<0.001) between speech duration from student to patient and overall NTS scores was observed, and this correlation was the strongest amongst all other vocal features with overall NTS score. The forward stepwise regression model predicted overall NT skills scores with adjusted R-squared value of 0.537. Similarly, the forward stepwise regression model predicted communication construct with adjusted R-squared value of 0.54. Both studies showed significant positive relationships between key vocal features (e.g., speech duration), frequency of communication with respect to performance. Metrics and vocal features derived from audio recordings can be measured in predicting clinical performance and NTS, moreover, it can further contribute to the understanding of communication in the healthcare setting. Most importantly, the potential of providing an objective approach for simulation-based trauma care training.

Degree

M.Sc.

Advisors

Yu, Purdue University.

Subject Area

Artificial intelligence|Medical personnel|Surgery|Communication|Computer science|Electrical engineering|Health care management|Management|Medicine

COinS