Diagnostic accuracy of presumed ocular tuberculosis criteria: A validation study

• Sensitivity, specificity, and AUROC analyses confirmed the criteria’s ability to reliably distinguish presumed ocular tuberculosis from non-tuberculous uveitis. • The structured use of mandatory and supportive criteria improved diagnostic consistency and reduced misdiagnosis in presumed ocular tuberculosis. • A documented history of prior tuberculosis contact emerged as a significant diagnostic indicator, strengthening case identification beyond laboratory findings alone. The objectives of this study is to evaluate the accuracy of a proposed diagnostic criteria for ocular tuberculosis in endemic region Iraq, evaluating the sensitivity and specificity of these criteria A prospective diagnostic validation study Patients with presumed Tuberculous and non-Tuberculous uveitis at Ibn Al-haitham eye teaching hospital, a tertiary referral center for ophthalmology in Baghdad, Iraq Diagnosis required both compatible ocular features—such as granulomatous anterior uveitis, serpiginous-like choroiditis, or panuveitis—and exclusion of other causes. Supportive criteria included chest CT findings suggestive of TB, positive IGRA, and a history of TB exposure or contact. The response to anti-tuberculous treatment (ATT) was used as the diagnostic reference (gold) standard. Favorable response was defined by resolution of inflammation within 8 weeks of ATT, sustained remission for ≥6 months, and absence of alternative diagnoses during follow-up. The study included 177 patients with uveitis, of whom 79 were ultimately classified as presumed ocular tuberculosis based on treatment response to anti-tuberculous therapy and exclusion of alternative diagnoses. Chronic granulomatous inflammation was the most common anterior uveitis pattern affecting 10.1%, while granulomatous intermediate uveitis was the leading intermediate form in 12.7%. Serpiginous choroiditis predominated in posterior uveitis cases in 6.3%. When at least one supportive criterion was present in addition to the mandatory criteria, the diagnostic framework achieved 100% sensitivity and 70.4% specificity. Requiring at least two supportive criteria improved specificity to 100% but reduced sensitivity to 75.9%. In conclusion, the proposed criteria demonstrated clinical utility with variable accuracy. Using one supporting criterion in addition to the mandatory criteria ensured detection of all TB cases but reduced specificity.