Quantifying Nurse and Physician Clinical Performance for Mechanically Ventilated Patients

Expanding on previous research applying value-added modeling to clinician performance measurement, we quantified simultaneous nurse and physician performance among mechanically ventilated patients.

In a retrospective cohort spanning five academic hospitals, we used multiple linear regression to determine relationships between nurse and physician assignments and changes in patient illness severity (Laboratory-based Acute Physiology Score) during intensive care unit (ICU) admission. Individual clinician performance was derived from regression coefficients. Collective clinician contribution to disease severity variance was assessed by comparing the coefficient of determination across clinician-only and joint-clinician models. Consistency of clinician performance estimates was cross-validated using random 1:1 partitions of the patient population using Pearson's correlation.

Electronic health record data were extracted from mechanically ventilated patients at study ICUs between 2018 and 2022.

Among 17,082 clinical encounters, 215 physicians, and 1719 nurses, nurse and physician assignments improved model fit. Together, they attributed 7% of variability in disease severity trajectory; separately, nurse assignments accounted for greater variance than physician assignments (6% vs. 4%, respectively). Cross-partition clinician performance was moderately correlated (r = 0.31 [nurses], r = 0.4 [physicians], p < 0.001 for both comparisons).

Individual nurse and physician care independently contribute to disease severity trajectory among mechanically ventilated patients.