The Reliability of Quantifying Upright Standing Postures as a Baseline Diagnostic Clinical Tool

A scientific exchange is always fruitful when it brings to light issues that deserve more analysis and discussion. We believe that Haas, Betz, Harrison, and Harrison have misunderstood the point of our article,1 which was to assess an individual's ability to reposition themselves in their normal upright standing posture and not interrater or intrarater reliability. The questions about study design and methodological issues would have been better dealt with in a friendly exchange of information between the 2 groups rather than wasting time and paper in an exchange through a letter to the editor, a process that should be used to allow free discussion of interesting and contentious points to the readership, not minutiae or insulting exchanges.

Haas et al1 mention several studies that they claim contradict our conclusions and were omitted from our article. Of the 16 studies they cite, only 1 describes a study design requiring repeated photographic assessment of posture of participants.2 Incidentally, this article found mixed results that were far from “highly repeatable.” The majority of the remaining papers used repeated examinations of photographs by different examiners to test intraexaminer/interexaminer reliability of postural measurements3, 4, 5, 6, 7, 8 but did not include or assess posture repeatability of the subjects at different points in time. The work of Swinkels and Dolan9, 10 does show more repeatable results than our study; however, they used an electromagnetic approach and did not allow their subjects to move between repeats. Further, they tested how well subjects could reposition to different target flexion positions with results demonstrating a wide range of ICCs (0.1 to 0.78) for upright standing. The purpose of our study1 was to investigate the repeatability of upright standing posture and not the people who evaluated it. This necessitated a design that tested subjects on a repeated basis and with only 1 examiner to evaluate the photographs, so as to not confound the results with errors from different examiners. Further, good intraexaminer and interexaminer reliability have been established in both the studies cited by Haas et al and our own work11; therefore, using only 1 examiner would likely contribute very little to the variability observed in our results.

Our use of 14 subjects and 1 examiner was challenged, stating that at least 30 to 60 subjects and 3 examiners are needed for a conclusion about reliability. Again, the point of our article was not to test interexaminer and intraexaminer reliability. We also find it puzzling that studies are cited as criticism of our work that fail to meet the Haas et al criterion.2, 4 Furthermore, generic statements about fixed sample sizes for studies are meaningless and entirely dependent on the variability and magnitude of the effect being studied.

A power analysis only indicates the number of subjects that would be required to detect a significant difference. The analysis of variance tests performed in our study showed no significant differences in any of the variables, indicating low power by its design. When this information is combined with low ICCs and high coefficients of variation, it becomes blatantly clear that the data are highly variable and hence posture repeatability is low, the stated conclusion of our paper.1 Even if additional subjects were collected to achieve sufficient power to detect a significant difference in posture, this would only support the argument that repeats of posture analysis are different and not repeatable between trials.

In all work we perform in our research laboratories, careful scientific control is followed. The suggestions that marker mounting errors, the use of fins, and the movement of markers and fins during walking could account for the results seen in our study are preposterous. The use of surface markers and fins has been used for decades in dynamic analyses and gait studies. Any surface measure to represent internal kinematics contains some error; however, as stated in our article, within-session variability was high, which could not be attributed to many of the error sources speculated.

Autodetection of marker centroids is a digitization process. Depending on the shape of the reflective surface, the location of the centroid does not necessarily fall in the center of a marker. Digitized dynamic kinematic data are filtered to reduce the high-frequency noise added by autodigitization. In other words, there is variability in a computer algorithm detecting a point within a marker that is no different than an operator manually digitizing the marker location. This manual approach was used in a study by members of the Haas et al group funded by BioTonix.12 Regardless of the method of extracting coordinates from an image, it is considered digitization. The inclusion of BioTonix in our article was solely to show that postural analyses were being assessed with software packages. Given the association of at least 2 of the members of the Haas et al group with BioTonix, we took their criticism of our apparently erroneous comments about the BioTonix software very seriously. However, BioTonix does compare angles with the vertical and horizontal (see the BioTonix web site: http://prca.biotonix.com/corp/english/index.html) and nowhere in our article do we state that this is the only measure used by this or any other program. We encourage users of any product to question the fundamental assumptions and theories on which it is based.

It has been shown that visual assessment of spinal posture is not reliable and is inaccurate.13, 14 Because of these findings, it is not surprising that researchers and clinicians are searching for more reliable and accurate ways to quantify posture. However, the literature is contradictory when it comes to determining the reliability of quantitative methods of measuring posture. Although it has been established that most methods have good intraexaminer/interexaminer reliability, at best the verdict is split on whether actual upright standing posture is repeatable.