Background Strongyloidiasis is a neglected tropical infection that can progress to life-threatening disease in immunocompromised hosts. However, predictors of severe disease and mortality remain incompletely defined, particularly in endemic regions. We aimed to identify clinical predictors of severe strongyloidiasis and 30-day mortality among hospitalized patients in southern Thailand. Methods We conducted a 16-year retrospective cohort study of hospitalized patients aged ≥13 years with laboratory-confirmed Strongyloides stercoralis infection at a tertiary-care hospital between 2009 and 2025. Severe strongyloidiasis was defined by larval detection in extraintestinal or respiratory specimens or compatible clinical manifestations involving multiple organ systems with infectious complications. Multivariable logistic regression was used to identify predictors of severe disease and mortality. Results Among 735 patients, 47 (6.4%) had severe strongyloidiasis. Corticosteroid exposure within 60 days was independently associated with severe disease (adjusted odds ratio aOR 3.25, 95% CI 1.65–6.73), whereas eosinophilia (absolute eosinophil count ≥500 cells/µL) was associated with lower odds (aOR 0.41, 95% CI 0.19–0.82). Bacteremia was more frequent in severe cases (38.3% vs 4.1%, p < 0.001), predominantly due to gram-negative organisms. Severe strongyloidiasis was associated with higher 30-day mortality (38.3% vs 14.0%, p < 0.001). In multivariable analysis, increasing age (aOR 1.22 per 10-year increase), higher National Early Warning Score (aOR 1.41 per point), and bacteremia (aOR 3.38) were independently associated with mortality, whereas eosinophilia was associated with lower odds of death and severe strongyloidiasis itself was not. Conclusions Corticosteroid exposure was the strongest predictor of severe strongyloidiasis, while eosinophilia was protective. Mortality was driven primarily by bacteremia and physiologic instability. Early recognition and treatment are essential to improve outcomes in endemic settings.
Chittrakarn et al. (Mon,) studied this question.