Sex-Specific Temporal Patterns of Haemoglobin and Haematocrit as Mortality Predictors in Polytrauma Patients: A...

Introduction: In polytrauma patients, effective triaging and resource allocation are critical in managing haemorrhage. This study aimed to analyse the association of haematocrit and haemoglobin with early death within 72 h in polytrauma patients stratified by sex. A sample of 3059 polytrauma patients was analysed using SPSS and the Watson Trauma Pathway Explorer up to 48 h after admission. Methods: A retrospective cohort of 3059 polytrauma patients (ISS ≥ 16) admitted 1996-2022 was analysed. Haemoglobin and haematocrit were measured at admission and serially through 48 h. Receiver operating characteristic (ROC) analysis assessed discriminatory performance for early death within 72 h. Multivariable logistic regression adjusted for injury severity, age, physiological status, and resuscitation volumes. Results: Significant sex-based differences in haemoglobin and haematocrit were observed at all time points (p < 0.05) in univariable analysis; haemoglobin and haematocrit showed significant associations with mortality in males at multiple time points (admission, 1 h, 3 h, 12 h, 24 h) and in females at admission and 1 h. After multivariable adjustment, haematocrit and haemoglobin for males at 2 h remained significant. Discriminatory performance was poor across all time points: in males, haemoglobin AUC ranged from 0.51 (95% CI 0.33-0.68) at admission to 0.59 (95% CI 0.43-0.75) at 48 h; in females, from 0.44 (95% CI 0.16-0.72) at 1 h to 0.68 (95% CI 0.44-0.93) at 3 h. Females exhibited a higher mortality in the early hours and at 8 h after admission. Mortality thresholds were lower in females than males at 1 h,3 h, 6 h, 24 h and 48 h for haematocrit and at 1 h, 2 h, 3 h and 8 h for haemoglobin, with the lowest value at 2 h for haemoglobin (females: 7.55 g/dL, males: 8.5 g/dL) and 1 h for haematocrit in females. Conclusions: While crude comparisons demonstrated sex-specific temporal patterns in haemoglobin and haematocrit between survivors and non-survivors, these associations did not remain statistically significant after multivariable adjustment and Benjamini-Hochberg correction for multiple testing (all FDR-adjusted p ≥ 0.200). Associations were substantially confounded by resuscitation practices. Discriminatory performance was poor across all time points (AUC 0.44-0.68). These findings should be considered hypothesis-generating and require prospective validation in cohorts with standardised resuscitation protocols before clinical implementation.