Serial platelet count as a dynamic prediction marker of hospital mortality among septic patients

Abstract

Background: Platelets play a critical role in hemostasis and inflammatory diseases. Low platelet count and activity have been reported to be associated with unfavorable prognosis. This study aims to explore the relationship between dynamics in platelet count and in-hospital morality among septic patients and to provide real-time updates on mortality risk to achieve dynamic prediction.

Methods: We conducted a multi-cohort, retrospective, observational study that encompasses data on septic patients in the eICU Collaborative Research Database (eICU-CRD) and the Medical Information Mart for Intensive Care IV (MIMIC-IV) database. The joint latent class model (JLCM) was utilized to identify heterogenous platelet count trajectories over time among septic patients. We assessed the association between different trajectory patterns and 28-day in-hospital mortality using a piecewise Cox hazard model within each trajectory. We evaluated the performance of our dynamic prediction model through area under the receiver operating characteristic curve, concordance index (C-index), accuracy, sensitivity, and specificity calculated at predefined time points.

Results: Four subgroups of platelet count trajectories were identified that correspond to distinct in-hospital mortality risk. Including platelet count did not significantly enhance prediction accuracy at early stages (day 1 C-index_Dynamic vs C-index_Weibull: 0.713 vs 0.714). However, our model showed superior performance to the static survival model over time (day 14 C-index_Dynamic vs C-index_Weibull: 0.644 vs 0.617).

Conclusions: For septic patients in an intensive care unit, the rapid decline in platelet counts is a critical prognostic factor, and serial platelet measures are associated with prognosis.

Keywords: Dynamic prediction; Intensive care medicine; Joint latent class model; Platelet count trajectory; Sepsis.

Figure 1

Flowchart illustrating the process of patient selection. eICU-CRD eICU Collaborative Research Database, MIMIC-IV Medical Information Mart for Intensive Care IV

Figure 2

Trajectory plots and Kaplan–Meier survival curves of patients with four dynamic platelet count trajectory patterns. (a, b) Trajectory plots of platelet count changes within the first 28 days after ICU admission in the eICU-CRD database, along with their corresponding survival curves. (c, d) Trajectory plots of platelet count changes within the first 28 days after ICU admission in the MIMIC-IV database, along with their corresponding survival curves. eICU-CRD eICU Collaborative Research Database

Figure 3

Mean platelet count and predicted survival probability over time for four classes in eICU-CRD. (a) Platelet measurements in the first 1 day and predicted survival probability in the next 27 days; (b) platelet measurements in the first 2 days and predicted survival probability in the next 26 days; (c) platelet measurements in the first 3 days and predicted survival probability in the next 25 days; (d) platelet measurements in the first 4 days and predicted survival probability in the next 24 days; (e) platelet measurements in the first 7 days and predicted survival probability in the next 21 days; and (f) platelet measurements in the first 10 days and predicted survival probability in the next 18 days

Figure 4

Individual prediction of two selected patients from the MIMIC-IV dataset. On the left is case 1, who stayed alive within 28 days after ICU admission, while on the right is case 2, who experienced death on day 18 of ICU stay. Individual predictions were updated every 3 days from 48 h after ICU admission to day 11. The x-axis represents the time since ICU admission, with platelet measurements on the left and model-predicted survival probabilities on the right. The vertical black dashed line indicates the prediction time, and the vertical black solid line represents the time of death