Diagnostic accuracy of tests for diabetic foot osteomyelitis: a Bayesian network meta-analysis protocol.

Abstract

Objective: The objective of this systematic review is to evaluate and compare the diagnostic accuracy of physical examination, foot imaging, and inflammatory marker tests for the diagnosis of diabetic foot osteomyelitis in adults.

Introduction: Diabetic foot osteomyelitis is a serious complication of diabetic foot ulcers, often leading to amputation, prolonged hospitalization, and increased mortality. Several diagnostic modalities, ranging from bedside examinations to advanced imaging and laboratory biomarkers, are available; however, their diagnostic performance varies widely in terms of accuracy, invasiveness, and cost.

Eligibility criteria: This review will include diagnostic accuracy studies evaluating adults with diabetes and suspected diabetic foot osteomyelitis, comparing index tests (clinical, lab, imaging) against a reference standard (biopsy or clinical follow-up). Eligible studies must clearly define diabetic foot osteomyelitis as a diagnostic outcome and specify which criteria are used to the confirm the diagnosis.

Methods: A 3-step search strategy will be applied. We will search Cochrane Central, Scopus, MEDLINE, ProQuest, and OAISTER. Google Scholar and ProQuest will be searched for gray literature. Studies published from January 2000 to the present will be considered. Two reviewers will independently conduct title and abstract screening, retrieval of full-text studies, full-text review, study selection, data extraction, and risk of bias assessment. Estimates of the individual results will be visualized by coupled forest plots, and the variability of individual results will be visualized using cross-hairs plots. For each index test, a random-effects, bivariate beta-binomial model will be used to pool sensitivity and specificity. Heterogeneity will be assessed using Zhou and Dendukuri's bivariate extension of the I 2 statistic. Publication bias will be investigated using Deeks' funnel plots. Sensitivity and subgroup analyses will be performed. For subgroups of index tests that form networks, we will perform diagnostic test accuracy network meta-analysis using a random-effects, bivariate beta-binomial arm-based model. The certainty of evidence for each diagnostic comparison will be assessed using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach for diagnostic test accuracy.

Review registration: PROSPERO CRD420251085609.