Purpose: To assess the prevalence of decisions to forgo life-sustaining treatment (DFLST), the patients characteristics, and to estimate the impact of DFLST stages on mortality.
Methods: Observational study of a prospective database between 2005 and 2012 from 13 ICUs. DFLST were defined as follows: no escalation of treatment (stage 1), not to start or escalate treatment even if such treatment is considered in the future; withholding (stage 2), not to start or escalate necessary treatment; withdrawal (stage 3), to stop necessary treatment. The impact of daily DFLST stage on day-30 hospital mortality was tested with a discrete-time Cox's model and adjusted for admission severity and daily SOFA score.
Results: Of 10,080 patients, 1290 (13%) made DFLST. The highest DFLST stage during the ICU stay was no escalation of treatment in 339 (26%) patients, withholding in 502 (39%) patients, and withdrawal in 449 (35%) patients. Older patients, patients with at least one chronic disease, and patients with greater ICU severity were significantly more numerous in the DFLST group. Day-30 mortality was 13% for non-DFLST patients, 35% for no escalation of treatment, 75% for withholding, 93% for withdrawal. After adjustment, an increase in day-30 mortality was associated with withholding and withdrawal (hazard ratio 95% CI 5.93 [4.95-7.12] and 20.05 [15.58-25.79], P < 0.0001), but not with no escalation of treatment (HR 1.14 [0.91-1.44], P = 0.25).
Conclusions: DFLST were made in 13% of ICU patients. Withholding, withdrawal, older age, more comorbidities, and higher severity of illness were associated with higher mortality. No escalation of treatment was not associated with increased mortality.
Keywords: Critical care; Decision-making; Do-not-resuscitate; End-of-life; Prognosis.