Time series analysis between platelet counts and 60-day mortality in sepsis patients with thrombocytopenia: a retrospective cohort study.

BACKGROUND

Thrombocytopenia is a prevalent complication and an adverse prognostic factor in sepsis; however, there is a paucity of studies examining the dynamic changes in platelet counts and their association with the prognosis of sepsis patients with related thrombocytopenia. This study aims to retrospectively to investigate the relationship between the dynamic trajectories of platelet counts and early adverse prognostic outcomes in such patients.

METHODS

A retrospective observational study was conducted using data from 946 sepsis patients with thrombocytopenia extracted from the Medical Information Mart for Intensive Care (MIMIC-IV) database. A joint latent class model (JLCM) was employed to analyze platelet trajectories over time as part of a time series analysis (TSA). Time-dependent piecewise Cox regression models and Kaplan-Meier survival curves were utilized to evaluate the association between trajectory patterns and 60-day mortality. Sensitivity analyses included the Schoenfeld residual test, subgroup analysis and reclassification of groups based on corresponding mortality risks.

RESULTS

The platelet trajectories during the first 28 days of hospitalization were classified into four classes, each associated with distinct 60-day mortality risks. Class 1 consistently remained at the lowest level, while Class 2 exhibited a gradual increase, yet still remained below normal levels. Class 3 demonstrated a rapid increase to normal levels, and Class 4 continued to increase closely toward normal level. The multivariable Cox proportional hazards model, with Class 1 as the reference group and adjusted for potential confounders, revealed significant associations between the trajectories and 60-day mortality for Class 3 (HR 0.27, 95% CI 0.08-0.85, P = 0.025) and Class 4 (HR 0.68, 95% CI 0.48-0.96, P = 0.029). Class 2 suggested a non-significant trend toward higher early mortality (HR 1.38, 95% CI 0.85-2.24, P = 0.198). Kaplan-Meier curves from the multivariate analysis indicated that Class 1 had the lowest 60-day survival probability, while Class 3 exhibited the highest, followed by Class 4. After merging Classes 1 and 2 into Group 1 for sensitivity analysis, with Group 1 as the reference, the groups remained significantly associated with 60-day mortality. The Schoenfeld residual test, subgroup analysis, regrouping were performed as sensitivity analyses to ensure the robustness of our study results.

CONCLUSIONS

This study demonstrates that trajectories of platelet in sepsis patients with thrombocytopenia is clinically significant. Sustained low platelet levels or low amplitude increases that remain below normal levels indicate a higher early mortality rate. Early rapid increase of platelet count is associated with better prognosis.