Inferior vena cava filter placement in patients with acute closed lower-limb and pelvic fractures and deep-vein thrombosis: A multicenter retrospective cohort study

Abstract Inferior vena cava (IVC) filter placement is widely used at Chinese trauma centers to prevent pulmonary embolism in fracture patients with deep-vein thrombosis, yet multicenter evidence evaluating its clinical effectiveness, safety, and cost-effectiveness in this population is lacking. To evaluate the association between IVC filter placement and pulmonary embolism, mortality, filter-related complications, and cost-effectiveness in patients with acute closed lower-limb and pelvic fractures complicated by lower-extremity deep-vein thrombosis. Multicenter retrospective cohort study conducted at 3 major level I trauma centers in China using electronic health records. Participants were adult patients with acute closed lower-limb or pelvic fractures and confirmed lower-extremity DVT. Of 1570 eligible patients, 685 received IVC filter placement within 48 to 72 h of admission and 885 did not. Retrievable IVC filter placement within 48 to 72 h of emergency department admission versus standard care without IVC filter. The primary outcome was pulmonary embolism incidence, defined as a composite of symptomatic and asymptomatic PE; symptomatic PE, a component of the composite, was reported separately as a key secondary outcome. Other secondary outcomes included all-cause mortality at 90 days and 1 year, filter-related complications, and the incremental cost-effectiveness ratio. Treatment effects were estimated using multivariable regression and, as the primary adjusted analyses, inverse probability of treatment weighting (IPTW) and 1:1 propensity-score matching, with a 72-hour landmark analysis to address immortal time bias. Among 1570 patients (mean age, 51.3 years; 60.3% male), baseline characteristics were well balanced (maximum standardized mean difference 0.006 after IPTW). The composite PE rate was lower in the filter group (6 of 685 [0.88%] vs. 36 of 885 [4.07%]; absolute risk reduction, 3.19% [95% CI, 1.68–4.78]; number needed to treat, 31 [95% CI, 21–60]). The association was consistent across multivariable adjustment (odds ratio [OR], 0.26 [95% CI, 0.12–0.50]), IPTW (OR, 0.18 [95% CI, 0.06–0.39]), propensity-score matching (OR, 0.17 [95% CI, 0.06–0.43]), and a 72-hour landmark analysis excluding early events (OR, 0.17 [95% CI, 0.06–0.48]); the E-value for the multivariable estimate was 6.92. Symptomatic PE occurred in 0.29% versus 2.37%. Vital status was known for all patients, and no difference in mortality was detected at 90 days (7.2% vs. 7.8%; risk difference, − 0.64% [95% CI, − 3.28 to 2.00]) or at 1 year (10.9% vs. 11.5%; risk difference, − 0.57% [95% CI, − 3.72 to 2.58]). Device-related complications occurred in 181 of 685 patients (26.4%); 147 (21.5%) were minor imaging findings and 34 (5.0%) were clinically significant. Filter retrieval was attempted in 80.3% with 93.5% success at a median dwell time of 45 days (IQR, 36–59). Imported devices accounted for 675 of 685 filters (98.5%), Denali (Bard/BD) alone for 493 (72.0%). Mean total cost was ¥102,657 (US$14,418) per patient versus ¥68,375 (US$9,603), giving an incremental cost of ¥34,281 (US$4,815) and an incremental cost-effectiveness ratio of ¥1,074,021 (US$150,846) per PE prevented. The incremental QALY gain was negligible (+ 0.0004 [95% CI, − 0.0154 to + 0.0166]) and the probability of cost-effectiveness was 0% at willingness-to-pay thresholds up to 10 times GDP per capita. In this multicenter retrospective cohort study, IVC filter placement was associated with a lower incidence of pulmonary embolism, but no difference in mortality was detected. Device-related complications were common but predominantly minor. The intervention was not cost-effective at any plausible willingness-to-pay threshold, largely because of the high price of imported IVC filter devices in the Chinese market. These findings do not support routine IVC filter placement in fracture patients with DVT.