Hospital length of stay, do not resuscitate orders, and survival for post-cardiac arrest patients in Michigan: A study for the CARES Surveillance Group.

Objective: Current guidelines recommend deferring prognostic decisions for at least 72 h following admission after Out of Hospital cardiac arrest (OHCA). Most non-survivors experience withdrawal of life sustaining therapy (WLST), and early WLST may adversely impact survival. We sought to characterize the hospital length of stay (LOS) and timing of Do Not Resuscitate (DNR) orders (as surrogates for WLST), to assess their relationship to survival following cardiac arrest.

Design: We performed a retrospective cohort study of probabilistically linked cardiac arrest registries (Cardiac Arrest Registry to Enhance Survival (CARES) and Michigan Inpatient Database (MIDB) from 2014 to 2017.

Patients: Adult (≥18 years) patients admitted following OHCA were included. We considered LOS ≤ 3 days (short LOS) and written DNR order with LOS ≤ 3 days (Early DNR) as indicators of early WLST. Our primary outcome was survival to hospital discharge. We utilized multilevel logistic regression clustered by hospital to examine associations of these variables, patient characteristics and survival to hospital discharge.

Measurement And Main Results: We included 3644 patients from 38 hospitals with >30 patients. Patients mean age was 62.4 years and were predominately male (59.3%). LOS ≤ 3 days (OR = 0.11) and early DNR (OR = 0.02) were inversely associated with survival to discharge. There was a non-significant inverse association between hospital rates of LOS ≤ 3 days and survival (p = 0.11), and Early DNR and survival (p = 0.83). In the multilevel model, using median odd ratios to assess variation in LOS ≤ 3 days and survival, patient characteristics contributed more to variability in surviival than between-hospital variation. However, between-hospital variation contributed more to variability than patient characteristics in the provision of early DNR orders.

Conclusions: We observed that LOS ≤ 3 days for post-arrest patients was negatively-associated with survival, with both patient characteristics and between-hospital variation associated with outcomes. However, between-hospital variation appears to be more highly-associated with provision of early DNR orders than patient characteristics. Further work is needed to assess variation in early DNR orders and their impact on patient survival.