After literature search, a total of 863 records were identified through the electronic database search. After duplicates were removed, 762 potential publications were left for further assessment. The literature selection generated 34 articles, of which two publications [22, 23] had partially overlapping cohorts derived from the same database (The National Trauma Data Bank). We only included Potter’s study because it had larger sample size with propensity score matching. Finally, 32 studies (10 cohorts [13, 22,23,24,25,26,27,28,29,30]and 22 case series [8, 10, 12, 31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49] of 1577 ET and 6097 OSR patients were included in the quantitative analysis. The PRISMA flow diagram is presented in Fig. 1. Eleven studies [8, 10, 28, 29, 32,33,34, 40, 42, 43, 45] reported outcomes of popliteal artery injury. As for injury types, six articles [12, 25, 30, 31, 39, 40] reported penetrating injuries and four articles [25, 37, 40, 43] reported blunt injuries, others involved mixed types of injury. Study characteristics are presented in Table 2 and Supplemental Table 1.

Study selection flow diagram

The results of the ROBINS-1 tool for cohort studies showed low to moderate risk of bias in most domains of the included studies. Most of the included case series were assessed to be of moderate or high quality, and few were of low quality. The results of quality assessment are shown in Supplemental Tables 2 and Supplemental Table 3.

A total of 19 articles, comprising 2893 patients, reported the major amputation rate after endovascular or open surgical repair of traumatic artery injuries of lower limbs. Data from studies conducted by Branco et al. [26], Abdou et al. [25], and Potter [23] et al. were adjusted for propensity score matching. The pooled results from seven cohort studies suggested that patients who underwent ET had a significantly decreased risk of major amputation than patients who underwent OSR. (OR = 0.42, 95% CI 0.21–0.85; I2 = 34%, Fig. 2A). The pooled incidence major amputation rate was 3% (95% CI 0%–9%; I2 = 85.13%, Fig. 3A) in 828 ET patients and 9% (95% CI 6%–12%; I2 = 87.96%, Fig. 3B) in 5102 OSR patients. The pooled incidence major amputation rate was 3% (95% CI 0%–9%; I2 = 86.40%, Supplemental Fig. 1A) in ET adults and 8% (95% CI 5%–12%; I2 = 81.10%, Supplemental Fig. 1B) in OSR adults. ET had a significantly decreased risk of major amputation than patients who underwent OSR in adults. (OR = 0.47, 95% CI 0.27–0.80; I2 = 34%, Supplemental Fig. 1C).

Forest plot of (A) studies for the difference in major amputation, (B) studies for difference in fasciotomy or compartment syndrome, (C) studies for the difference in mortality, and (D) studies for the difference in length of stay in patients with traumatic lower extremity arterial injury comparing ET vs. OSR. M-H = Mantele-Haenszel; CI = confidence interval; IV = inverse variance; ET = endovascular therapy; OSR = open surgical repair

The pooled estimate for amputation in ET (A) and OSR (B) in patients with traumatic lower extremity arterial injury. ES = estimate proportions; CI = confidence interval; ET = endovascular therapy; OSR = open surgical repair

Among the included studies, a total of 3 studies from 4 cohorts underwent propensity score matching analysis. Subgroup analysis of the propensity score-matched data revealed that patients undergoing ET had significantly lower major amputation risks compared to those undergoing OSR (OR = 0.32, 95% CI 0.14–0.72; I2 = 42%, Supplemental Fig. 1D). We conducted a meta-analysis using all available data from the included studies, comprising a total of 6623 patients. The results indicate that patients undergoing ET had a significantly lower risk of major amputation compared to those OSR (OR = 0.51, 95% CI 0.36–0.73; I2 = 0%, Supplemental Fig. 1E).

Subgroup analysis stratified by injured arteries suggested ET was also associated with a significantly decreased risk of amputation in patients with iliac or femoral arterial injury (OR = 0.15, 95% CI 0.05–0.45; I2 = 0%, Supplemental Fig. 1F), with an estimated amputation rate of 3% (95% CI 0%–12%; I2 = 92.49%) in ET group and 6% (95% CI 3%–9%; I2 = 88.43%) in OSR group. As for patients with isolated popliteal artery injury, the pooled results showed an estimated amputation rate of 5% (95% CI 0%–19%; I2 = 49.72%) in the ET group and 11% (95% CI 5%–18%; I2 = 85.32%) in OSR group.

In the subgroup of patients who suffered from penetrating artery injury, the results showed an estimated amputation rate of 5% (95% CI 4%–7%; I2 = 0.00%) in OSR group. In the subgroup of patients suffered from fracture, the results showed an estimated amputation rate of 10% (95% CI 7%–14%; I2 = 85.57%) in OSR group and 3% (95% CI 0%–10%; I2 = 91.25%) in ET group. Meta-regression analysis suggested that the pooled estimate for amputation was significantly associated with injury severity score (ISS) rather than fracture (ISS: t=-2.52, 95% CI 0.86–0.99, Supplemental Fig. 1G; fracture: t = 1.88, 95% Cl -0.88–7.94). in addition, the funnel plot on amputation rates appeared to be symmetrical on visual inspection, suggesting no publication bias, that could be confirmed statistically with the harbord’s linear regression test (p = 0.949) and egger’s linear regression test (p = 0.840).

A total of 15 studies including 1163 ET patients and 4175 OSR patients reported outcomes of Fasciotomy or compartment syndrome. The pooled estimate of fasciotomy or compartment syndrome rate in the OSR subgroup (23%, 95% CI 13%–36%; I2 = 98.24%) was over two times higher than that of ET group (9%, 95% CI 3%–16%; I2 = 92.92%). In the further meta-analysis of four cohort studies, the results also revealed a significantly higher risk of fasciotomy or compartment syndrome in OSR than ET (OR = 0.31, 95% CI 0.20–0.50, I2 = 14%, Fig. 2B). We conducted a meta-analysis using all available data from the included studies, the results showed a significantly lower risk of fasciotomy or compartment syndrome in ET than OSR (OR = 0.44, 95% CI 0.26–0.74, I2 = 0%, Supplementary Fig. 2A). The pooled estimate of compartment syndrome rate in the OSR subgroup (8%, 95% CI 2%–17%; I2 = 0%) was over two times higher than that of ET group (3%, 95% CI 0%–10% I2 = 93.83%). In the further meta-analysis of cohort studies, the results also revealed a significantly higher risk of compartment syndrome in OSR than ET (OR = 0.36, 95% CI 0.25–0.50, I2 = 0%, Supplemental Fig. 2B). As for nerve injury, this complication was only reported in OSR patients, with an estimated rate of 15% (95% CI 6%–27%; I2 = 93.76%).

Based on 1228 ET and 4285 OSR patients, the pooled estimate of postoperative mortality rate was 4% (95% CI 1%–9%; I2 = 87.58, Supplemental Fig. 3A) in the ET group and 2% (95% CI 1–4%; I2 = 58.67%, Supplemental Fig. 3B) in the OSR group. The in-hospital mortality results from the study by Butler et al. were included and meta-analysis of six cohorts revealed no significant difference in all-cause mortality between ET and OSR groups (OR = 1.11, 95% CI 0.75–1.64, I2 = 31%, Fig. 2C). We conducted a meta-analysis using all available data from the included studies, and the results showed no significant difference in all-cause mortality between ET and OSR groups (OR = 0.96, 95% CI 0.57–1.60, I2 = 82%, Supplementary Fig. 3C). Four cohort studies, comprising 3434 patients, reported the length of stay after ET or OSR. The pooled results suggested ET Patients had a significantly shorter length of stay than patients with OSR (MD=-5.06, 95% CI -6.76 to -3.36, I2 = 65%, Fig. 2D). Significant heterogeneity was noted, and we further performed subgroup analysis stratified by injury types. In patients with penetrating injury, similar results were observed with no heterogeneity (MD=-6.12, 95% CI -7.21 to -5.03, I2 = 0%, Supplemental Fig. 3D).