Diagnostic Accuracy of Preoperative High-Resolution Computed Tomography Compared With Intraoperative Findings in Active Squamous Chronic Otitis Media: A Cross-Sectional Study

Background and objectiveHigh-resolution CT (HRCT) is widely used to map cholesteatoma before surgery, but its diagnostic performance differs across temporal bone structures.This study evaluated the diagnostic accuracy of preoperative HRCT and used paired analysis to determine whether HRCT systematically over-or underdetected operative findings in active squamous chronic otitis media. MethodsThis cross-sectional diagnostic accuracy study included 55 patients who underwent temporal bone HRCT followed by cholesteatoma surgery at Bangladesh Medical University from September 1, 2025, to August 29, 2026.Intraoperative findings were the reference standard.Patient-level 2 × 2 tables were used to calculate sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and overall accuracy with exact 95% confidence intervals (CIs).Two-sided exact McNemar tests were used for paired comparisons, with Holm adjustment as a multiplicity sensitivity analysis. ResultsThe mean age of the cohort was 23.02 ± 11.33 years, and 39/55 patients (70.9%) were male.Scutum erosion was present in 55/55 patients (100.0%) on both HRCT and surgery.Overall accuracy was 52/55 (94.5%) for posterior meatal wall erosion and 51/55 (92.7%) for lateral semicircular canal (LSCC) fistula.Incus erosion was detected in 37/55 patients (67.3%) on HRCT and 50/55 (90.9%) at surgery; sensitivity was 36/50 (72.0%) and specificity 4/5 (80.0%) (exact McNemar p = 0.00098; Holm-adjusted p = 0.0049).Tegmen erosion was reported in 28/55 patients (50.9%) on HRCT versus 4/55 (7.3%) at surgery, and facial canal dehiscence in 38/55 (69.1%) versus 11/55 (20.0%); both were significantly overdetected and remained significant after Holm adjustment.Mesotympanic involvement was 48/55 (87.3%) on HRCT versus 39/55 (70.9%) at surgery (p = 0.035), but no disease-extension comparison remained significant after Holm adjustment. ConclusionsHRCT is a useful preoperative surgical roadmap in cholesteatoma, but its diagnostic performance is structure-dependent.It performed well for posterior meatal wall erosion and lateral semicircular canal fistula, while incus erosion was significantly underdetected and tegmen erosion and facial canal dehiscence were significantly overdetected.Exact McNemar analysis appropriately accounts for the paired nature of HRCT and intraoperative observations and should be interpreted alongside structure-specific diagnostic accuracy measures.