Prognostic Factors for Long-Term Mortality in Critically Ill Patients Treated With Prolonged Mechanical Ventilation: A Systematic Review.

OBJECTIVES: Long-term survival for patients treated with prolonged mechanical ventilation is generally poor; however, patient-level factors associated with long-term mortality are unclear. Our objective was to systematically review the biomedical literature and synthesize data for prognostic factors that predict long-term mortality in prolonged mechanical ventilation patients. DATA SOURCES: We searched PubMed, CINAHL, and Cochrane Library from 1988 to 2015 for studies on prolonged mechanical ventilation utilizing a comprehensive strategy without language restriction. STUDY SELECTION: We included studies of adults 1) receiving mechanical ventilation for more than or equal to 14 days, 2) admitted to a ventilator weaning unit, or 3) received a tracheostomy for acute respiratory failure. We analyzed articles that used a multivariate analysis to identify patient-level factors associated with long-term mortality (≥ 6 mo from when the patient met criteria for receiving prolonged mechanical ventilation). DATA EXTRACTION: We used a standardized data collection tool and assessed study quality with a customized Newcastle-Ottawa Scale. We abstracted the strength of association between each prognostic factor and long-term mortality. Individual prognostic factors were then designated as strong, moderate, weak, or inconclusive based on an a priori previously published schema. DATA SYNTHESIS: A total of 7,411 articles underwent relevance screening; 419 underwent full article review. We identified 14 articles that contained a multivariate analysis. We abstracted 19 patient-level factors that showed association with long-term mortality. Six factors demonstrated strong strength of evidence for association with the primary outcome: age, vasopressor requirement, thrombocytopenia, preexisting kidney disease, failed ventilator liberation, and acute kidney injury ± hemodialysis requirement. All factors, except preexisting kidney disease and failed ventilator liberation, were measured at the time the patients met criteria for prolonged mechanical ventilation. CONCLUSIONS: Despite the magnitude of the public health challenge posed by the prolonged mechanical ventilation population, only 14 articles in the biomedical literature have tested patient-level factors associated with long-term mortality. Further research is needed to inform optimal patient selection for prolonged mechanical ventilation.