LAUSR

Dear Editor, Do-not-resuscitate (DNR) orders have been widely used for critically ill patients [1]. However, their impact on survival outcome is largely unknown for patients with severe sepsis and/or septic shock. The aim of the study was to investigate the impact of DNR on mortality in patients with severe sepsis and/or septic shock. The Medical Information Mart for Intensive Care (MIMIC) III database was employed for analysis [2]. Data were collected from patients admitted to the Beth Israel Deaconess Medical Center in Boston. Severe sepsis and/or septic shock was defined as presence of infection and acute organ dysfunction [3, 4]. DNR status was further categorized as early (<12 h after ICU entry), middle (12–72 h), and late (>72 h) DNR. The primary endpoint was the 100-day survival. A Cox proportional hazard regression model incorporating covariates such as sequential organ failure assessment (SOFA), metastatic cancer, and simplified acute physiology score II (SAPSII) was employed to control for confounding factors [5]. Interaction between DNR status and SOFA score was also included because we hypothesized that the effect of DNR on survival could vary depending on the severity of illness. A total of 17,168 subjects were included for analysis, including 14,096 subjects without DNR and 3072 with DNR order (Fig. E1, ESM). Patients with DNR were more likely to die during hospital stay than patients without DNR (42.0 vs. 11.0%; p < 0.001). Non-survivors had more comorbidities such as congestive heart failure (33.0 vs. 27.3%; p < 0.001), renal failure (22.6 vs. 21.0%; p = 0.038), and metastatic cancer (9.5 vs. 4.0%; p < 0.001) than survivors (Table E1, ESM). The Cox regression model showed that early (HR 2.31; 95% CI 1.85–2.89), middle (HR 4.07; 95% CI 3.21–5.14), and late (HR 11.56; 95% CI 9.64– 13.87) DNR were significantly associated with 100-day survival (Table E2, ESM). The interaction between early DNR and SOFA was not statistically significant. However, there were interactions between SOFA and middle (HR 0.96; 95% CI 0.94–0.99; p = 0.003) or late (HR 0.87; 95% CI 0.85–0.89; p < 0.001) DNR. For subjects with high SOFA score (SOFA = 20), the DNR order did not show a significant impact on survival probability, as compared with non-DNR patients. However, for less severely ill patients, the presence of DNR had a negative impact on survival outcome (Fig. 1). SOFA score had a limited impact on death for patients with late DNR (e.g., given a certain change in SOFA, the magnitude of changes in probability of survival was much smaller in patients with late DNR than that in other DNR status, Fig. E2, ESM). Patients had consistently high mortality rate in the presence of late DNR irrespective of the initial SOFA score (Fig. E3, ESM). The study confirmed previous findings that DNR was associated with higher risk of death. A novelty of the study was that there was an interaction between the presence of DNR and severity of acute organ dysfunction. While DNR was associated with increased risk of death