Objective: The development and evaluation of a new chest radiograph reading and recording system (CRRS) for community surveys of tuberculosis (TB) and lung disease.
Design: An experienced pulmonologist read 2608 chest X-rays (CXRs) performed as part of a TB prevalence survey using the newly developed CRRS. The kappa (kappa) for inter-reader agreement was calculated after a second reader reported on a stratified random sample of 810 (31%) of the 2608 CXRs. The kappa for intra-reader agreement was calculated from the repeated reporting of a stratified random sample of 104 CXRs.
Results: The kappa agreement between the two readers was 0.69 (95%CI 0.64-0.74) for abnormalities consistent with TB and 0.47 (95%CI 0.42-0.53) for any abnormalities. The kappa for intra-reader agreement was 0.90 (95%CI 0.81-0.99) for abnormalities consistent with TB and 0.85 (95%CI 0.74-0.95) for any abnormalities.
Conclusion: This standardised method for CXR reading and recording provides satisfactory inter- and intra-reader agreement, making it suitable for surveys of TB and other forms of lung disease in the community. Its use will permit comparisons of results obtained in different surveys.