INT J TUBERC LUNG DIS 20(10):1358–1363 Q 2016 The Union http://dx.doi.org/10.5588/ijtld.16.0186 Reader variability and validation of the Timika X-ray score during treatment of pulmonary tuberculosis B. A. Thiel,* † C. M. Bark,* †‡ J. G. Nakibali, † F. van der Kuyp, ‡ J. L. Johnson* † *Tuberculosis Research Unit, Department of Medicine, Case Western Reserve University and University Hospitals Case Medical Center, Cleveland, Ohio, USA; † Uganda-Case Western Reserve University Research Collaboration, Kampala, Uganda; ‡ Division of Infectious Diseases, MetroHealth Medical Center, Case Western Reserve University, Cleveland, Ohio, USA SUMMARY SETTING: Chest radiographs (CXRs) are widely used for diagnosing pulmonary TB and assessing response to therapy. The Timika X-ray score has been proposed as a tool for measuring disease severity and predicting treatment outcome. OBJECTIVE: To evaluate inter- and intra-reader agree- ment of Timika scores and assess the ability of the score to predict microbiologic outcome at 2 months. DESIGN: Analytical validation study. Disease severity was measured by two readers using pretreatment radiographs and follow-up films taken at 2, 6 and 12 months after the start of treatment among 110 human immunodeficiency virus negative adults with pulmonary TB. One fourth of the films were reread to assess intra- reader agreement. RESULTS: The two-component Timika score had high inter- and intra-reader agreement (intraclass correlation (ICC) inter ¼ 75%, ICC intra . 0.81). Baseline Timika score was associated with positive month 2 smear (P ¼ 0.0004) and culture status (P ¼ 0.03). The average Timika score declined significantly over the course of successful treatment. CONCLUSION: The Timika score showed good inter- and intra-reader agreement and a significant association with microbiological outcomes after 2 months of treatment. The results of this study strengthen the evidence supporting the use of the Timika score for measuring disease severity on CXR. KEY WORDS: radiography; clinical trial; reproducibil- ity; agreement; score CHEST RADIOGRAPHS (CXRs) are widely used for the evaluation of patients with possible tuberculosis (TB) and for assessing response to anti-tuberculosis treatment. Indicators of disease severity, such as the presence of cavitary disease or bilateral disease, have been correlated with the risk of treatment failure and relapse and residual disability and pulmonary function after treatment. 1–4 Many schemes for grading CXR severity of disease have been proposed; however, most have substantial intra- and inter-observer variability, limited sensitivity, or have not been properly validated in diverse populations. 5–9 The lack of a reproducible, standardized, validated scale for grading the severity of disease on CXR of patients with TB and monitoring response to treatment is a recognized gap in the methodology for conducting clinical trials of new TB drugs and drug combinations. Ralph et al. recently proposed a two-component numerical score for grading the severity of TB on CXR. 10 Their numerical score, coined the ‘Timika X- ray score’, was developed as a pragmatic tool that could be used in clinics and field sites without complex aids. The numerical score included two components that were summed: the presence of any cavitary disease (40 points) and the estimated percentage of the amount of both lung fields involved by any tuberculous lesion (0 to 100 points). Possible scores ranged from 0 to 140. The Timika X-ray score correlated well with measures of disease severity, such as baseline sputum smear grade and 2 month smear status, which are accepted correlates of bacteriologic clearance during treatment. 11–14 To evaluate the findings of Ralph et al. and to assess whether they are generalizable to other patient populations, we reread and analyzed serial CXRs from a phase 2 trial conducted in Uganda to evaluate reader variability and predictive ability of the Timika X-ray score. STUDY POPULATION AND METHODS Study population CXR and clinical data from 110 Ugandan human immunodeficiency virus (HIV) negative adults aged Correspondence to: Bonnie A Thiel, Tuberculosis Research Unit, Case Western Reserve University School of Medicine, Room E-202 School of Medicine, 10900 Euclid Avenue, Cleveland, OH 44106-4984 USA. Fax: ( þ 1) 216 368 0105. e-mail: bat2@case.edu Article submitted 4 March 2016. Final version accepted 6 June 2016.