The article by Helenius et al., "Harrington and Cotrel-Dubousset
Instrumentation in Adolescent Idiopathic Scoliosis. Long-Term Functional and
Radiographic Outcomes"
(2003;85:2303-9), is an
excellent attempt to compare an older method with a newer method. Changes in
techniques are often made for theoretical reasons without proven benefit. Then
it becomes difficult to perform randomized contemporaneous studies of
efficacy. The authors have indicated some of the limitations of this
historical method of comparison.
However, it should be noted that there were more than four times the number
of lumbar curves (King type-I) in the Harrington group than in the
Cotrel-Dubousset group in the study by Helenius et al. Did this represent a
change in the selection of lumbar curves for anterior instrumentation during
the time when Cotrel-Dubousset instrumentation was available? If so, the
Harrington group may contain a disproportionate number of lumbar curves that
underwent two-stage distraction with posterior instrumentation rather than
anterior instrumentation. This small selection bias could affect the
comparison as a result of the elimination of patients from the
Cotrel-Dubousset group who would no longer be candidates for posterior
surgery. Thus, the improvements seen in this study may reflect changes in
patient selection and improvements in anterior surgery rather than a change in
posterior instrumentation.
As a result of the historical nature of the study design, the studied
groups were somewhat different from one another: the percentage of patients
with a King type-I curve was three times higher in the Harrington
instrumentation group than in the Cotrel-Dubousset instrumentation group (nine
of seventy-eight compared with two of fifty-seven). It is true that anterior
instrumentation (Zielke) alone or together with Cotrel-Dubousset
instrumentation was used for thoracolumbar or lumbar curves during the study
period (between 1987 and 1990). However, the patients treated with anterior
instrumentation were not included in the present study, which may have
produced a small selection bias. In the original article, we stated that if
patients with King type-I curves were excluded from both groups, the number of
patients with abnormal lumbar extension (30% compared with 15%) and trunk
side-bending (56% compared with 36%) was significantly higher in the
Harrington rod instrumentation group (p = 0.039 and 0.026, respectively) than
in the Cotrel-Dubousset instrumentation group. No correlations were observed
between the magnitudes of the thoracic or lumbar curves and the spinal
mobility measurements.
The large and significant difference in the radiographic correction between
the groups does not disappear if patients with King type-I curves are excluded
from both groups (Table I); instead, it becomes even stronger. Thus, on the
basis of our article and the additional data shown here, we do not believe
that the improvements seen in our study were due to patient selection or
improvements in anterior surgery. Instead, we conclude that Cotrel-Dubousset
instrumentation yielded better long-term functional and radiographic outcomes
in patients with adolescent idiopathic scoliosis than did Harrington
instrumentation, even if patients with lumbar curves were excluded.