Association of Intensive Care Unit Patient-to-Clinician Ratios with Mortality across Two U.S. Health Systems.

Rationale: The association of interprofessional team member workload with intensive care unit (ICU) outcomes is understudied. Objectives: To evaluate the association of patient-to-intensivist ratio (PIR), patient-to-respiratory therapist ratio (PRTR), and patient-to-clinical pharmacist ratio (PpharmR) with hospital mortality. Methods: We conducted a retrospective study of adults admitted from the emergency department to an ICU with acute respiratory failure or sepsis within two U.S. healthcare systems (2013-2018). Our primary exposures were patient-to-clinician ratios (PIR, PRTR, and PpharmR) averaged over the ICU stay; our primary outcome was hospital mortality. We used multivariable mixed-effects regression, with patient-to-clinician ratios modeled as restricted cubic splines (four knots). We primarily considered each exposure separately, then included all ratios together. Results: Our cohort included 45,036 patients (mean age, 66.0 [standard deviation, 16.6] years; 23,420 [52.0%] men) across 27 ICUs within 24 hospitals. Of these, 29,326 (65.1%) had acute respiratory failure, 32,434 (72.0%) had sepsis, and 9,675 (21.5%) died in the hospital. The average PIR was 9.3 (standard deviation, 3.6), and the average PRTR was 7.9 (standard deviation, 3.2); the average PpharmR was 15.0 (standard deviation, 5.5) among patients (n = 8,950 of 45,036) in ICUs with clinical pharmacists (n = 8 of 27). We found no significant association between average daily PIR (Wald test for all spline terms: P = 0.24) or PRTR (P = 0.18) and hospital mortality in the full cohort; similarly, among patients in ICUs with pharmacists, no significant association of PpharmR with mortality was observed (P = 0.08). Models including ratios together yielded similar null results. Conclusions: We did not identify an association of any average daily patient-to-clinician ratio with hospital mortality for U.S. ICU patients with sepsis or respiratory failure.