Using Objective Fluid Balance Data to Identify Pulmonary Edema in Subjects With Ventilator-Associated Events

A ventilator-associated events (VAEs) algorithm was developed to detect events in mechanically ventilated subjects using objective parameters, and we aimed to use objective data of fluid balance to identify pulmonary edema-associated VAEs.

This single-center retrospective cohort study was conducted in a medical ICU and enrolled all mechanically ventilated patients between July 2016 and June 2017. Electronic medical records were reviewed to obtain data regarding ventilator-associated conditions (VACs), infection-related ventilator-associated complications (IVACs), possible ventilator-associated pneumonia (VAP), and traditionally defined VAP.

Of the 1,158 mechanically ventilated subjects, 85 (7.3%) subjects developed VAEs with a corresponding incidence rate of 7.7 events per 1,000 ventilator days. Among the 85 subjects with VAEs, 52 (61.2%) were classified as IVACs, while 23 (27.1%) had possible VAP. Notably, pulmonary edema was the main etiology (29.0%) for VAEs in the 62 subjects with non-possible VAP VAEs. Compared with those without pulmonary edema, subjects with pulmonary edema had a higher positive fluid balance 2 d before (+1,228 vs +173.5 mL, P = .005) and 1 d before (+1,622 vs +313 mL, P = .002) the diagnosis of VAE. In the multivariate logistic regression analysis (adjusted odds ratio [OR]) adjusted for potential confounders, an older age (adjusted OR 1.072, 95% CI 1.001-1.147), receiving renal replacement therapy (adjusted OR 8.906, 95% CI 1.454-54.558), and a positive cumulative difference between fluid balance 2 d and 1 d before VAE indexing (adjusted OR 1.527 per L positive, 95% CI 1.153-2.023) were independently associated with pulmonary edema in subjects with VAEs.

These findings provide epidemiological evidence of VAEs in a medical ICU and showed that fluid balance may be used to identify pulmonary edema-associated VAEs. Further studies are warranted to validate and translate these findings into an automated surveillance system for VAEs.