VPB0116 - Prediction of severe covid, proposition of a severe index

Background: In a pandemic context with limited intensive care beds, it’s important to prioritize which patients will need to be hospitalized in intensive care.

Aims: We aimed to develop and evaluate a simple but accurate biological based index which is associated with intensive care hospitalization.

Methods: It’s a retrospective study including sex, age, CRP, Natriemia, D-Dimers, Leukocytes, Polymorphonuclear Neutrophils (PMN), Lymphocytes (Ly), Platelets (Plt) from June to September 2021, in the covid unit and in the intensive care unit in Taher Maamouri Hospital, in Tunisia. Index was calculated as below: (Plt*PMN)/(Ly*age). ROC curves were obtained using SPSS software (IBM®, v22.0).

Results: Four hundreds and four patients were included in our study. Sex ratio was 1.38 and median age was 62 (from 3 to 98). Covid severity was significantly correlated with age, high leukocytes, PMN, Plt counts.
Single parameters didn’t show a good area under the curve (AUC). However, for a 56.0 value, our index, which combined the better single parameters, was associated with an AUC of 0.71 (CI95: [0.66-0.76], OR:4.7, p< 0.001) sensitivity of 65.5% and a specificity of 72.8%. The negative predictive value was 74.1% and the positive predictive value was 64.0%. Youden index was 0.38. Yule’s Q was 0.67 which shows a strong link between severity and our index.

Conclusion(s): Simple parameters may lack of signification, but combining them may improve their performance and may constitute a decision tool to help clinicians identify severe cases based on biological datas.