The intra- and inter-examiner reliability of the radiographic assessment of the cervical lordosis

Abstract

Aim: To determine the intra- and inter-examiner reliability of the radiographic assessment of the cervical lordosis of asymptomatic adult males.

Participants: Eighty lateral plain film radiographs of the cervical spine of asymptomatic males aged 18-45 years (taken in a previous study) were utilised for this study. However, due to the obstruction of the C7 vertebral body by the trapezius muscle, the examiners were unable to assess the CL on all 80 plain film radiographs. Three examiners took part in the study viz. Examiner One who was a qualified chiropractor with three years of clinical experience, Examiner Two who was a qualified chiropractor with six years of clinical experience and Examiner Three who was a chiropractic master’s student.

Methodology: The initial set of assessments of the CL using the C1-C7 and C2-C7 modified Cobb methods was completed by Examiner One and captured on an Excel spread sheet for Round One. The procedure was then repeated for Examiners Two and Three. The process was repeated for the second set of assessments (Round Two). Each examiner was given a maximum of two weeks to complete their assessments for each round. The data was statistically analysed using SPSS 22.0 and Stata 13. Descriptive data was presented in tables as mean and standard deviation at a 95% confidence interval while intra- and inter-examiner reliability was determined using the Kappa coefficient.

Results: The mean (± SD) CL values obtained by each examiner using the C1-C7 modified Cobb method for Round One was: Examiner One: 45.6˚ (± 10.4˚) (n = 70), Examiner Two: 44.0˚ (± 11.0˚) (n = 75) and Examiner Three: 43.8˚ (± 12.0˚) (n = 72). The mean (± SD) CL values obtained by each examiner using the C1-C7 modified Cobb method for Round Two was: Examiner One: 46.7˚ (± 10.7˚) (n = 72), Examiner Two: 43.3˚ (± 11.1˚) (n = 74) and Examiner Three: 43.8˚ (± 11.5˚) (n = 72).

The mean (± SD) CL values obtained by each examiner using the C2-C7 modified Cobb method for Round One was: Examiner One: 15.9˚ (± 9.2˚) (n = 72), Examiner Two: 22.6˚ (± 9.7˚) (n = 75) and Examiner Three: 17.2˚ (± 9.7˚) (n = 72). The mean (± SD) CL values obtained by each examiner using the C2-C7 modified Cobb method for Round Two was: Examiner One: 16.3˚ (± 9.4˚) (n = 72), Examiner Two: 20.5˚ (± 9.0˚) (n = 74) and Examiner Three: 16.9˚ (± 9.2˚) (n = 72).

The intra-examiner reliability obtained by each examiner using the C1-C7 modified Cobb method for Round One and Round Two was: Examiner One: K = 0.16, Examiner Two: K = 0.11 and Examiner Three: K = 0.16. The intra-examiner reliability obtained by each examiner using the C2-C7 modified Cobb method for Round One and Round Two was: Examiner One: K = 0.21, Examiner Two: K = 0.04, Examiner Three: K = 0.22.

The inter-examiner reliability obtained by each examiner using the C1-C7 modified Cobb method for Round One and Round Two respectively was: Examiner One vs Examiner Two: K = 0.03; K = 0.09, Examiner One vs Examiner Three: K = 0.19; K = 0.15, Examiner Two vs Examiner Three: K = 0.03; K = 0.08. The inter-examiner reliability obtained by each examiner using the C2-C7 modified Cobb method for Round One and Round Two respectively was: Examiner One vs Examiner Two: K = 0.00; K = 0.01, Examiner One vs Examiner Three: K = 0.19; K = 0.11, Examiner Two vs Examiner Three: K = 0.02; K = 0.05.

There was a significant difference in the intra-examiner findings for both the modified Cobb methods (p < 0.05). Using the C1-C7 modified Cobb method, there was a significant difference in the inter-examiner reliability findings between all three examiners for both rounds (p < 0.05). There was no significant difference in the inter-examiner findings of the CL using the C2-C7 modified method between Examiner One versus Examiner Two for Round One (p = 0.33) and Round Two (p = 0.23) but there was a significant difference in the findings between Examiner One versus Examiner Three (p < 0.05) and between Examiner Two versus Examiner Three (p < 0.05) for Round Two only.

Conclusion: The results of this study are in agreement with those of a previous study which reported that the C1-C7 modified Cobb method over-valued the magnitude of the curve while the C2-C7 modified Cobb method under-valued the curve. A significant difference in the intra-examiner findings suggests that recall bias did not significantly affect the assessments while inter-examiner findings suggest that experience and skill of the examiners as well as assessments that require drawing of lines and measuring of angles might lead to differences in the results obtained. Further studies which would utilise a large number of digitised radiographic images from both asymptomatic and symptomatic individuals are required to confirm the findings of this study.