Nursing-Related Risk Characteristics and Identification of High-Risk Patients With In-Hospital Lower-Extremity Deep Vein Thrombosis After Lower-Limb Fracture Surgery: A Single-Center Retrospective Study

AIM: Lower-limb fractures already contribute ≥5 Caprini points for venous thromboembolism (VTE) risk, limiting the ability of the conventional >5 cutoff to discriminate individual risk within this population. We evaluated whether an admission Caprini score ≥9, preoperative immobilization ≥72 h and severe postoperative day 1 (POD1) activity dependence are associated with in-hospital lower-extremity deep vein thrombosis (DVT), and whether POD1 activity provides additional within-group risk stratification beyond Caprini scoring. METHODS: We conducted a single-center, retrospective, 1:3 matched case-control study of adults undergoing surgery for lower-limb fractures at Ganzhou Hospital-Nanfang Hospital, Southern Medical University between January 2022 and July 2024. Of 783 patients screened, 71 were excluded, leaving 712 in the source cohort (61 DVT cases and 651 ultrasound-negative, matchable controls). Cases and controls were matched 1:3 without replacement by age (±5 years), sex, and fracture site, yielding 61 matched sets (244 patients). After matching, non-matching variables were singly imputed using a random forest algorithm for the primary analysis. To assess uncertainty associated with this approach, a separate sensitivity analysis generated 20 completed datasets from the pre-imputation matched records using a stochastic chained-equation multiple-imputation procedure. Regression estimates across the 20 datasets were combined using Rubin’s rules. Associations were estimated using conditional logistic regression stratified by matched set and reported as odds ratios (ORs) with 95% confidence intervals (CIs). Incremental within-set concordance was corrected using 2000 cluster bootstrap resamples. The primary analysis, complete-case analyses and multiple-imputation sensitivity analysis were conducted separately. RESULTS: Matching variables were well balanced after matching, with standardized mean differences (SMD) of 0.02 for age and 0 for sex and fracture site, although several non-matching characteristics remained imbalanced. Three factors were positively associated with in-hospital DVT: admission Caprini ≥9 (OR 2.43, 95% CI 1.28–4.61), severe POD1 activity dependence (OR 3.89, 95% CI 1.38–10.94), and preoperative immobilization ≥72 h (OR 2.55, 95% CI 1.14–5.67). Compared with patients without any core risk features, the OR increased to 2.81 (95% CI 1.46–5.38) among those with one feature and to 6.28 (95% CI 1.57–25.03) among those with two features. Adding severe POD1 activity dependence to Caprini ≥9 was associated with a modest increase in corrected within-set concordance from 0.589 to 0.633, corresponding to a difference of 0.044 (bootstrap SE 0.022, 95% CI 0.003–0.088). Sensitivity analyses generally preserved the direction of the main associations, although the Caprini estimate was attenuated after additional adjustment for trauma-related variables and D-dimer. CONCLUSIONS: Admission Caprini ≥9, preoperative immobilization ≥72 h, and severe POD1 activity dependence were each associated with in-hospital lower-extremity DVT. POD1 activity assessment was associated with a modest improvement in within-set risk ranking beyond Caprini stratification. This finding should not be interpreted as evidence of established clinical utility or improved patient outcomes; however, it may help guide nursing attention toward prophylaxis delivery, activity support, and symptom surveillance. Prospective multicenter studies are needed to determine its clinical relevance.