Epidemiological profile and factors associated with length of stay in patients with acetabular fractures at a level 1 trauma center in Nepal: a retrospective analytical study

Abstract Background Acetabular fractures, predominantly caused by high-energy trauma, pose a substantial healthcare burden in low- and middle-income countries (LMICs) such as Nepal. This study aimed to characterize the epidemiological and clinical profile of patients with acetabular fractures at a tertiary Nepalese center and to identify independent determinants of prolonged hospitalization using multivariable analysis. Methods A retrospective analytical study was conducted using electronic medical records of 591 patients treated for acetabular fractures (ICD-10: S32.4–S32.5) between 2006 and 2023. The primary outcome was a prolonged hospital stay (LOS) of ≥ 3 weeks (≥ 21 days). This was selected as the binary dependent variable based on the median LOS observed in our cohort (18 days) and clinical benchmarks for complex pelvic trauma recovery in LMIC settings. Bivariate analysis using chi-square and Fisher’s exact tests was followed by binary logistic regression to identify independent predictors. Statistical analysis was performed using SPSS v26.0. Results The cohort was predominantly male (78.0%), with a mean age of 38.95 ± 13.97 years. Two-wheeler accidents (37.9%) and pedestrian trauma (24.2%) were the leading injury mechanisms. Posterior wall fractures (45.5%) predominated, and 77.3% had associated injuries. LOS: mean 23.98 ± 19.273 days, median 18 days (range: 2-102 days). In-hospital mortality: n = 12 (2.1%). Overall, 42.8% experienced prolonged LOS (≥ 3 weeks). Binary logistic regression identified two independent predictors: mechanism of injury (overall p = 0.003), specifically 2-wheeler compared to pedestrian (AOR = 1.66, 95% CI: 1.04–2.64, p = 0.033) and Others category (AOR = 3.58, 95% CI: 1.14–11.31, p = 0.030); and presence of associated injuries (AOR = 2.36, 95% CI: 1.45–3.84, p = 0.001). Age, sex, fracture classification, surgical approach, and treatment modality were not independent predictors. The overall omnibus p-value for surgical approach was p = 0.125 (non-significant); nominally significant individual contrasts for specific approach categories reflect sparse-data instability in the reference group and should not be interpreted as independent effects. Model fit: Nagelkerke R²=0.112, Hosmer-Lemeshow p = 0.492. Conclusion Mechanism of injury and presence of associated injuries are independently associated with prolonged hospitalization in acetabular fracture patients in Nepal. Surgical approach and fracture type, while associated bivariately, lost significance after multivariate adjustment, suggesting confounding by fracture complexity. These findings are exploratory and hypothesis-generating; prospective studies incorporating injury severity scoring and continuous LOS modeling are required before translation into clinical guidelines. These findings underscore the need for multidisciplinary polytrauma protocols and targeted road safety interventions in LMIC trauma settings.