Background: We intended to create a new acute respiratory distress syndrome (ARDS) prediction score in high-risk critically ill patients.
Methods: We recruited 200 patients [63 (43-70) years, 120 (60%) males] admitted to the ICU with APACHE-II scores of ≥15 and at least one ARDS risk factor, after excluding patients with ARDS on admission, cardiac patients, and readmissions. The presence of risk factors together with the admission and 48-hour CRP (CRP-0 and CRP-48) were tested in univariate then multivariate regression models for identifying significant predictors whose weights were assigned according to the β-coefficient of the regression model. Our score was compared with the score previously proposed by Trillo-Alvarez et al. on 2011 (LIPS-T). The primary and secondary outcomes were the development of ARDS and in-hospital mortality, respectively.
Results: ARDS developed in 88 patients (44%). Logistic regression revealed that pneumonia, tachypnoea, increased heart rate, and increased CRP-48 are significant ARDS predictors. The weight of each predictor was estimated according to its β-coefficient. The new score was 35.5 (27-44) and 14 (9-24.3) in ARDS and non-ARDS patients, respectively (p=0.000). The AUC of the new score was 0.827 compared with 0.74 for the LIPS-T (p=0.014). A score of 20 had a sensitivity and specificity of 82% and 71%, respectively, in predicting ARDS. Our score was significantly lower in survivors compared with non-survivors (p=0.000) and its AUC in predicting in-hospital mortality was 0.761 compared with 0.657 for the LIPS-T (p=0.0045).
Conclusions: We have created a new simple LIPS score which could be better than the scores previously proposed in terms of ARDS and in-hospital mortality prediction in critically ill patients.
Back to issue - Download PDF