Objective: Describe modes of death and factors involved in decision-making together with life support limitation (LSL) procedures.
Design: Prospective, descriptive, longitudinal, and noninterventional study.
Setting: Sixteen pediatric intensive care units in Argentina.
Patients: Every patient who died during a 1-yr period was included.
Measurements and main results: Age, sex, length of stay (LOS), primary and admission diagnosis, underlying chronic disease (CD), postoperative condition (PO). Deaths were classified in four groups: a) failed cardiopulmonary resuscitation (CPR); b) do-not-resuscitate (DNR) status; c) withholding or withdrawing life-sustaining treatment (WH/WD); and d) brain death (BD). Justifications were classified as a) imminent death; b) poor long-term prognosis; c) poor quality of life; and d) family request. Data were collected from medical records and interviews with the attending physicians. Descriptive statistics were performed. Differences among groups were analyzed through contingency tables and analysis of variance when required. Relative risks and confidence intervals of variables potentially related to LSL were analyzed, and logistic regression was performed. There were 6358 admissions and 457 deaths. CPR was performed in 52%, DNR in 16%, WH/WD in 20%, and BD in 11% of dead patients. BD patients were older, LOS and CD prevalence were higher in the WH/WD group. Inotropic drugs were the most frequently limited treatment in 110 patients (55%), CPR in 72 (35.6%), and mechanical ventilation in 63 (31%). Imminent death was the most frequently reported justification for LSL. CD and more staff were associated with a higher probability of LSL.
Conclusions: Most of the patients in Argentina underwent CPR before their death. We have a high proportion of patients with CD (65%) and low BD diagnosis. PO condition decreased LSL probability in chronically ill patients. Do-not-resuscitate orders and withholding new treatments were the most common LSL. Active withdrawal was exceptional. The Ethics Committee was consulted in 5% of the LSL population.