To determine the efficacy of gait analysis in the treatment of ambulation problems in cerebral palsy, the authors set out to assess the consistency of both the quantitative data and recommendations for treatment from different motion analysis laboratories. To accomplish this, they sent 11 children to four separate gait analysis centers and then compared the clinical data and treatment recommendations from these centers. They found that gross inaccuracies existed in the clinical data, with the sagittal plane data being the most reproducible and the transverse plane the least. Linear measurements and EMG also showed large variations. Treatment recommendations were similar in only 2 of the 11 patients and were widely disparate in the other 9. From this they concluded that traditional methods of physical examination and observational gait analysis were more reliable, and that methods other than gait analysis were needed to determine the outcomes of treatment of gait problems in cerebral palsy. The authors speculated about but did not attempt to determine the reasons for the disparate data. They concluded, despite good evidence to the contrary, that inconsistencies of treatment recommendations were inexorably linked to inaccuracies in the clinical data. On the basis of that assumption, however, Dr. Noonan and his colleagues proposed to “throw the baby out with the bath water.” I was present at the Pediatric Orthopaedic Society meeting at which this paper was first presented, and on the basis of the enthusiastic response of the audience I could not help but feel that this represented the feelings of most of them as well. In my personal view, gait analysis has been received by the pediatric orthopedic community at large with all the enthusiasm of “ants at a picnic.” This paper could be critiqued from many aspects. In fact, I was privileged to have had access to the reviewer’s critiques of the manuscript, which were excellent. I wish these could have been published as well, as I do not feel that the authors have adequately answered them. For example, according to Table 2, the variation of the clinical range of motion in hip flexion/extension was 28°, compared with 14° in the kinematic examination in Table 4. Yet, to lessen the magnitude of error of the clinical examination, the authors chose to compare the former to total passive and the latter to total active range of motion. Rather than pick at flaws in methodology, however, I will accept the authors’ premises that variation of gait analysis data is currently too large to be acceptable and that conclusions reached on the basis of that data vary widely. Nevertheless, it is not valid to assume, as the authors seem to have done, that just because you install a gait analysis laboratory and turn it on means that you will get good data, or that somehow you will be endowed with knowledge of how to treat cerebral palsy. Gorton et al. have shown that system accuracy between 12 Shriners Motion Analysis Laboratories is <1% (3). All systems were able to determine kinematic angles of an instrumented rod within 3° and find a known distance within 3 mm. The test subject was shown to be repeatable both within and between test days if marker placement was controlled, but had differing results between test sessions by different clinicians within and between sites. In addition, variables such as walking speed and toe-off time will influence the magnitude, pattern, and timing of joint angle data. These will cause right or left shifts of the kinematic curves such that large variations in magnitude of the data might occur at specific points in the gait cycle (particularly those areas of the curve around the initial contact and toe-off points). The authors noted this when they stated, “on average in at least 5% of the gait cycle, the difference between these four laboratories could exceed 8° to 34°.” Thus, we know that the major source of gait laboratory errors relate to marker placement, definition of the points of toe-off and initial contact, and gait velocity. When I am interpreting gait laboratory data, I tend to look at “patterns of motion data,” and patterns are not disrupted by these shifts. However, this does not excuse the need for better quality control. No technique, be it MRI or gait analysis, is so good that it can’t be undermined by poor quality control. Good laboratories pay close attention to this. For example, we have just completed a study in our laboratory that demonstrates intertherapist, intersession, and intertrial errors of <4°, and with further attention to quality control and dynamic joint-centering techniques, we should be able to improve things further (6). With respect to the treatment recommendations, Skaggs et al. have shown that when identical data was Address correspondence and reprint requests to James R. Gage, M.D., Gillette Children’s Specialty Healthcare, 200 East University Avenue, St. Paul, MN 55101 (e-mail: jgage@gillettechildrens.com). From Gillette Children’s Specialty Healthcare, St. Paul, Minnesota. Journal of Pediatric Orthopaedics 23:290–291 © 2003 Lippincott Williams & Wilkins, Inc., Philadelphia
[1]
S. Rethlefsen,et al.
The Effect of Preoperative Gait Analysis on Orthopaedic Decision Making
,
2000,
Clinical orthopaedics and related research.
[2]
R M Kay,et al.
Impact of Postoperative Gait Analysis on Orthopaedic Care
,
2000,
Clinical orthopaedics and related research.
[3]
Michael H Schwartz,et al.
Comprehensive Treatment of Ambulatory Children With Cerebral Palsy: An Outcome Assessment
,
2004,
Journal of pediatric orthopedics.
[4]
Joyce P Trost,et al.
Measurement and management of errors in quantitative gait data.
,
2004,
Gait & posture.
[5]
S. Rethlefsen,et al.
Variability in Gait Analysis Interpretation
,
2000,
Journal of pediatric orthopedics.
[6]
J. Gage,et al.
The role of gait analysis in the treatment of cerebral palsy.
,
1994,
Journal of pediatric orthopedics.
[7]
S. Õunpuu,et al.
Alterations in surgical decision making in patients with cerebral palsy based on three-dimensional gait analysis.
,
1997,
Journal of pediatric orthopedics.