ObjectivesTo compare the incidence of withdrawal or withholding of life support (WD/WHLS), and to identify similarities and differences in the process of the withdrawal of life support (WDLS) between teaching and community hospitals' intensive care units (ICUs).DesignProspective cohort study, with some data obtained by retrospective chart review.SettingThe ICUs of three teaching hospitals and six community hospitals.PatientsAll patients who died in these nine ICUs over a 6-mo period.InterventionsNone.Measurements and Main ResultsData on admitting diagnosis, cause of death, mode of death (death despite active treatment, withdrawal or withholding of life support), those initiating and involved in WDLS, and modalities of life support withdrawn were gathered for patients dying in the ICU over a 6-mo period. One hundred sixty patients in community hospitals and 292 in teaching hospitals died in their respective ICUs over the 6-mo period. We found a difference in the distribution of mode of death between community hospitals and teaching hospitals, resulting from a greater proportion of patients dying as a result of withholding life support in community hospitals (11.9% vs. 3.8% withheld, respectively, p = .004). Among the six community hospitals and three teaching hospitals, we found a difference in the proportion of patients dying despite active treatment compared with those dying as a result of WD/WHLS (p = .042 and p = .044, respectively). Initiation of WDLS by physicians was more frequent at teaching hospitals (81% vs. 61%, p = .0005), while families more commonly initiated WDLS at community hospitals (34% vs. 19%, p = .005). A greater proportion of patients in teaching hospitals were receiving mechanical ventilation (99% vs. 89%) and vasopressors (76% vs. 65%) before WDLS. Similar proportions had mechanical ventilation withdrawn (68% and 74%, community hospitals and teaching hospitals, respectively), while there was a trend for fewer patients in community hospitals to have vasopressors withdrawn (56% vs. 70%, p = .082). The time to death after WDLS had begun was longer in community hospitals compared with teaching hospitals (0.74 +/- 1.38 days vs. 0.27 +/- 0.79 [SD] days, p = .0028).ConclusionsThe incidence of WD/WHLS was similar in community hospitals and teaching hospitals; however, withholding of life support was more common in community hospitals. The process of WDLS appears to differ between community hospitals and teaching hospitals. (Crit Care Med 1998; 26:245-251)