Outcomes in adult critically Ill cancer patients with and without neutropenia: a systematic review and meta-analysis of the Groupe de Recherche en Réanimation Respiratoire du patient d'Onco-Hématologie (GRRR-OH)

PURPOSE: Whether neutropenia has an impact on the mortality of critically ill cancer patients remains controversial, yet it is widely used as an admission criterion and prognostic factor. METHODS: Systematic review and meta-analysis. Studies on adult cancer patients and intensive care units were searched on PubMed and Cochrane databases (2005-2015). Summary estimates of mortality risk differences were calculated using the random-effects model. RESULTS: Among the 1,528 citations identified, 38 studies reporting on 6,054 patients (2,097 neutropenic patients) were included. Median mortality across the studies was 54% [45–64], with unadjusted mortality in neutropenic and non-neutropenic critically ill patients of 60% [53–74] and 47% [41–68], respectively. Overall, neutropenia was associated with a 10% increased mortality risk (6%-14%; I(2) = 50%). The admission period was not associated with how neutropenia affected mortality. Mortality significantly dropped throughout the study decade [−11% (−13.5 to −8.4)]. This mortality drop was observed in non-neutropenic patients [−12.1% (−15.2 to −9.0)] but not in neutropenic patients [−3.8% (−8.1 to +5.6)]. Sensitivity analyses disclosed no differences in underlying malignancy, mechanical ventilation use, or Granulocyte-colony stimulating factor use. Seven studies allowed the adjustment of severity results (1,350 patients). Although pooled risk difference estimates were similar to non-adjusted results, there was no significant impact of neutropenia on mortality (risk difference of mortality, 9%; 95% CI, −15 to +33) CONCLUSION: Although the unadjusted mortality of neutropenic patients was 11% higher, this effect disappeared when adjusted for severity. Therefore, when cancer patients become critically ill, neutropenia cannot be considered as a decision-making criterion.