Modification of the prolonged mechanical ventilation prognostic model score to predict short-term and 1-year mortalities

Abstract

Background and objective: We aimed to validate the use of the Prolonged Mechanical Ventilation Prognostic Model (ProVent) score in medically ill patients with co-morbidities and to modify the score to improve the prediction power of 1-year mortality. Methods: We conducted a retrospective study of all patients who required at least 14 days of mechanical ventilation (MV) and established two groups (14–20 and ≥21 days of MV) based on the MV duration. We performed external validation of the present ProVent Model in our patients on Day 14 (or Day 21 for the ≥21-day MV group) of MV, and established the extended ProVent model, while considering the albumin and bilirubin levels and co-morbidities (chronic obstructive pulmonary disease and cancer). Results: A total of 1288 patients (666 and 622 with 14–20 and ≥21 days of MV, respectively) with at least 14 days of MV were enrolled. The 1-year mortality was 79.9% and 78.7% in the ≥21- and 14–20-day groups, respectively. Most of the observed mortality rates in all groups were within the 95% CI of predicted mortality as per the ProVent Model, except for the ProVent scores of 0 and 5. In the ProVent model, the area under the curve for the prediction of 1-year mortality was 0.69 in all patients with ≥14 days of MV, whereas in the extended ProVent model, the area under the curve was 0.89. Conclusion: The extended ProVent model, which considers co-morbidities and laboratory data, increases the prediction power of 1-year mortality in patients who require prolonged MV.