J Knee Surg 2019; 32(07): 701
DOI: 10.1055/s-0038-1660840
Letter to the Editor
Thieme Medical Publishers 333 Seventh Avenue, New York, NY 10001, USA.

Reliability of Roentgenographic Knee Alignment Measurements in Gonarthrosis: Methodological Issues to Avoid Misinterpretation

Siamak Sabour
1  Department of Clinical Epidemiology, School of Health, Shahid Beheshti University of Medical Sciences, Tehran, Iran
› Author Affiliations
Further Information

Publication History

Publication Date:
13 June 2018 (online)

I was interested to read the article by Huang et al published in the Journal of Knee Surgery, 2018.[1] A reliable method of measuring knee alignment is critical in the preoperative planning of high tibial osteotomy (HTO). There are two commonly used methods for measuring knee alignment: the mechanical axis and the anatomical axis. The authors aimed to determine which method would be more reliable and reproducible. Fifty consecutive patients (50 knees) with medial gonarthrosis were treated with valgus-producing HTO. Four independent investigators assessed preoperative standing lower limb radiography (including hip, knee, and ankle joints) to measure both anatomical and mechanical axes. They applied a generalized estimating equation model to assess the reproducibility. Based on their results, the effect of the investigators was not statistically significant; however, the effect of the measurement method was highly significant (p < 0.0001). They concluded that the estimated coefficient of the anatomical axis method was negative, indicating that the reproducibility of the anatomical axis method was better than that of the mechanical axis.

However, these results are not the most appropriate estimates to assess reliability (reproducibility, repeatability, or precision).[2] Reproducibility is the degree to which measures are free from error and therefore yield consistent results (i.e., the consistency of a measurement procedure). If a measurement device or procedure or method consistently assigns the same score to individuals with equal values, the method is considered reliable. It is crucial to know that, in reproducibility analysis, our approach should be individual based.[2] It is good to know that to assess reproducibility for quantitative variable, intraclass correlation coefficient single measure or Bland–Altman's plot can be applied. Moreover, statistically significant (p < 0.0001) and clinically important are two completely different methodological issues. Therefore, to assess reproducibility, appropriate test as well as correct interpretation should be considered.[2] [3] [4] [5]

In this letter, I emphasized on methodological and statistical issues on reliability analysis. Any conclusion should be supported by the above-mentioned methodological and statistical issues. Otherwise, misinterpretation cannot be avoided.