To investigate the factors influencing the 28-day prognosis in patients with sepsis-associated severe thrombocytopenia and to develop and validate a clinical prognostic model.

Patients from the Medical Information Mart for Intensive Care (MIMIC-Ⅳ) v2.2 database were selected as the training cohort, and patients from the emergency intensive care unit (eICU) Collaborative Research Database served as the validation cohort. Patients diagnosed with sepsis-associated severe thrombocytopenia were included. Demographic data, vital signs, laboratory parameters, chronic comorbidities, treatment measures, and 28-day survival outcomes were collected. The least absolute shrinkage and selection operator (LASSO) and Bayesian information criteria were used to select variables for the prognostic model, and the model was externally validated using the eICU database.

1489 patients were included in the training cohort. Eight variables were selected: age, acute kidney injury stage, urine output, minimum blood urea nitrogen level, maximum heart rate, maximum international normalized ratio, minimum partial thromboplastin time, and the presence of liver disease. The C-index was 0.714 for the training cohort and 0.64 for the validation cohort. In the training cohort, the area under the curve for the 7-, 14-, and 28-day survival probabilities were 0.723, 0.733, and 0.736, respectively.

Key factors affecting the prognosis of patients with sepsis-associated severe thrombocytopenia were identified through multivariate analysis. A prognostic model incorporating eight variables was developed, which demonstrated good predictive performance for the prognosis of these patients.