Reliability and validity of subjective assessment of lumbar lordosis in conventional radiography

Background: Reliability and validity studies of different lumbar curvature analysis and measurement techniques have been documented however there is limited literature on the reliability and validity of subjective visual analysis. Radiological assessment of lumbar lordotic curve aids in early diagnosis of conditions even before neurologic changes set in. Objective: To ascertain the level of reliability and validity of subjective assessment of lumbar lordosis in conventional radiography. Design: A blinded, repeated-measures diagnostic test was carried out on lumbar spine x-ray radiographs. Setting: Radiology Department at Joint Clinical Research Centre (JCRC), Mengo- Kampala-Uganda. Subjects: Seventy (70) lateral lumbar x-ray films were used for this study and were obtained from the archive of JCRC radiology department at Butikiro house, Mengo-Kampala. Results: Poor observer agreement, both inter- and intra-observer, with kappa values of 0.16 was found. Inter-observer agreement was poorer than intra-observer agreement. Kappa values significantly rose when the lumbar lordosis was clustered into four categories without grading each abnormality Conclusion: The results confirm that subjective assessment of lumbar lordosis has low reliability and validity. Film quality has limited influence on the observer reliability. This study further shows that fewer scale categories of lordosis abnormalities produce better observer reliability.