Background and aims: We aimed to compare the diagnostic accuracy of ankle brachial index (ABI) and toe brachial index (TBI) for peripheral arterial disease (PAD) in a wide spectrum of PAD populations and reference standard tests, and to examine variables influencing heterogeneity in the estimates.
Methods: Systematic searches in EMBASE, MEDLINE, Web of Science and the Cochrane Library databases were performed, from inception to January 2020. Hierarchical summary receiver operating characteristic curves (HSROC) were used to summarize the pooled test performance.
Results: Thirty five (patient-level: 1318 patients, limb-level: 5637 limbs) and nine studies (patient-level: 294 patients, limb-level: 826 limbs) were included in ABI and TBI meta-analyses, respectively. The QUADAS-2 tool identified many studies with high risk of bias, especially in the "patient selection" domain. Pooled estimates for ABI in detecting 50% or greater stenosis were sensitivity = 61% (95% CI: 55-69), specificity = 92% (95% CI: 89-95) and dOR = 16.5 (95% CI: 11.5-23.6). Similarly, TBI yielded sensitivity = 81% (95% CI: 70-94), specificity = 77% (95% CI: 66-90) and dOR = 13.1 (95% CI: 7.0-24.8). In a direct comparison of seven studies jointly analyzing ABI and TBI, TBI showed better overall diagnostic accuracy (16.4 vs 11.0 in dOR) at the expense of sensitivity (82% vs 52%), while specificity (77% vs 94%) performed worse in TBI than ABI. Heterogeneity was large in sensitivity for ABI, with variables as different reference standard tests, smoking habit and PAD prevalence accounting for such variability. Similarly, gender, different index test cut-offs and sample size influenced the heterogeneity in TBI specificity.
Conclusions: Though ABI and TBI showed similar diagnostic performance to diagnose PAD, TBI showed far better sensitivity than ABI, especially in "challenging populations", as those exhibiting calcification.
Keywords: Ankle brachial index; Meta-analysis; Peripheral arterial disease; Sensitivity and specificity; Toe brachial index.
Copyright © 2020 Elsevier B.V. All rights reserved.