Period-Specific Determinants of Mortality After Geriatric Hip Fracture: Insights into Inflammation, Comorbidity and...

Mortality remains high after geriatric hip fractures. Evaluating postoperative mortality across distinct time windows may allow a more accurate clinical interpretation of phase-specific risk determinants.

To assess clinical and laboratory factors associated with mortality at 0-30 days, 30-90 days, and 90-365 days after hemiarthroplasty for geriatric hip fractures.

In this single-center retrospective cohort study, 433 consecutive patients aged ≥65 years who underwent hemiarthroplasty between January 2019 and November 2024 were analyzed. Phase-specific determinants of mortality were evaluated using a three-interval landmark approach with multivariable logistic regression models (0-30, 30-90, and 90-365 days). Discrimination was assessed with ROC analysis; Kaplan-Meier analyses were reported as supportive findings.

Mortality was 17.1% at 0-30 days, 6.2% at 30-90 days, and 12.9% at 90-365 days. In multivariable analyses, 0-30-day mortality increased with age (OR 1.17) and Charlson Comorbidity Index (CCI) (OR 1.72), while female sex was protective (OR 0.28). Shorter time-to-surgery showed an inverse association with early mortality (OR 0.97 per hour). For 30-90 days, mortality was associated with log(CRP+1) (OR 1.62). For 90-365 days, mortality was associated with age (OR 1.06), higher ASA score (OR 2.60), and lower albumin (OR 0.92 per g/L). In ROC analyses, age yielded an AUC of approximately 0.70; CRP and CCI showed time-dependent AUC values ranging from 0.55 to 0.66, indicating limited discrimination for use as standalone clinical decision-support tools.

Determinants of mortality after geriatric hip fracture vary across postoperative phases. Early mortality is more closely associated with comorbidity burden, intermediate mortality with inflammatory markers, and late mortality with indicators related to physiological reserve and nutritional status. Given the retrospective design, these findings should be interpreted as associations rather than causal effects, and triage-related bias/residual confounding cannot be excluded.