ABSTRACT
OBJECTIVE: Hyperinflammation (HI) that develops in week 2 of COVID-19 contributes to a worse outcome. Because week 2 laboratory findings can be relatively mild, the available criteria for classification of hemophagocytic lymphohistiocytosis or macrophage activation syndrome are not helpful. METHODS: Our study included a discovery cohort of patients from Turkey with symptomatic COVID-19 who were followed up while hospitalized during the initial wave and a replication cohort of hospitalized patients from a later period, all of whom required oxygen support and received glucocorticoids. Diagnosis of HI was made by an expert panel; most patients with COVID-19-associated HI (HIC) received tocilizumab or anakinra. Clinical and laboratory data from start day of treatment with tocilizumab or anakinra in HIC patients were compared with the data from day 5-6 in patients without HIC. Values maximizing the sensitivity and specificity of each parameter were calculated to determine criteria items. RESULTS: The discovery cohort included 685 patients, and the replication cohort included 156 patients, with 150 and 61 patients receiving treatment for HI, respectively. Mortality rate in HI patients in the discovery cohort (23.3%) was higher than the rate in patients without HI (3.7%) and the rate in patients in the overall replication cohort (10.3%). The 12-item criteria that we developed for HIC showed that a score of 35 provided 85.3% sensitivity and 81.7% specificity for identification of HIC. In the replication cohort, the same criteria resulted in 90.0% sensitivity for HIC; however, lower specificity values were observed because of the inclusion of milder cases of HIC responding only to glucocorticoids. CONCLUSION: The use of the 12-item criteria for HIC can better define patients with HIC with reasonable sensitivity and specificity and enables an earlier treatment start.