Three studies (641 participants comprised 589 with diabetes and 52 controls) were included in the review. The authors stated that all studies showed methodological limitations that could have inflated estimates of sensitivity or specificity. Limitations to methodological quality were noted in footnotes to the results table: two studies did not clearly describe their selection criteria, one of these studies did not account for withdrawals and the other study may have interpreted test results without blinding to other results; and the third study did not fully describe the index test and reference standard.
Sensitivity estimates ranged from 40.9% (95% CI 36% to 46%) to 93.1% (95% CI 77% to 99%) in patients with diabetes (three studies). Sensitivity in control participants was 77.0% (95% CI 72% to 81%; one study).
Specificity estimates ranged from 94.9% (95% CI 86% to 99%) to 100% (95% CI 63% to 100%) in patients with diabetes (three studies). Specificity in control participants was 68.3% (95% CI 58% to 77%; one study).