Risk model of short-term prognosis for patients with acute respiratory distress syndrome

  Objective  A 28-day death risk prediction model for acute respiratory distress syndrome (ARDS) was explored to provide early intervention strategies for stratification of patient risk and improvement of short-term prognosis.

  Methods  A retrospective analysis of the clinical data of 628 ARDS patients diagnosed and treated in the respiratory medicine of department of comprehensive internal medicine, Qionghai City Hospital of Traditional Chinese Medicine and department of respiratory medicine, Qionghai City People's Hospital from March 2010 to March 2020 was done. The patients were divided into the death group(267 cases)and survival group(361 cases)based on the 28-day survival status. Multivariate logistic regression analysis was used to show the factors influencing fate of ARDS patients at day 28. A risk prediction model was constructed, and the ROC curve was used to evaluate the risk model.

  Results  Multivariate logistic regression analysis showed that compared with body mass index (BMI) ≤ 25.00 kg/m², or oxygenation index (PaO₂/FiO₂) ≤ 122.50, the factors BMI > 25.00 kg/m² (OR = 0.37, 95% CI: 0.15-0.72, P = 0.002) and PaO₂/FiO₂ > 122.50 (OR = 0.59, 95%CI: 0.21-0.83, P = 0.007)were protective factors. Compared with the neutrophils lymphocyte ratio(NLR)≤ 13.85, Clara cell protein-16(CC16)≤ 53.50 μg/L, or acute physiology and chronic health evaluation (APACHE) II score ≤ 23.80 points, the factors NLR > 13.85 (OR = 1.58, 95%CI: 1.26-3.39, P < 0.001), CC16 > 53.50 ng/L (OR = 1.66, 95%CI: 1.13-3.90, P = 0.013), and APACHEⅡ score > 23.80 (OR = 2.39, 95%CI: 1.55-4.12, P = 0.030) were the risk factors. Then the risk prediction model was constructed by using new derived variable "Predicting" based on the regression equation. ROC curve analysis showed that the AUC of Predicting to predict the fate of ARDS patients at day 28 was 0.893(95%CI: 0.847-0.939, P < 0.001), and the accuracy rate was 83.28%, the sensitivity was 83.90%, the specificity was 82.83%, and the critical value was 0.768.

  Conclusions  The risk model constructed based on factors NLR, CC16, APACHEⅡ score, BMI, and PaO₂/FiO₂ can help early identification of critically ill patients with ARDS and guide clinical work.