To The Editor:
I would like to compliment Dr. Buttermann on his well-conceived and
executed study entitled "Treatment of Lumbar Disc Herniation: Epidural
Steroid Injection Compared with Discectomy"
(2004;86:670-9). This study
supports the commonly held notion that discectomy provides rapid pain relief
and accelerates recovery but does not promise a quantifiably better result in
the long run.
However, the open crossover design muddies the results a bit. Would a
design that prevented crossover, at least for a defined period of many months
to a year (a common clinical scenario in managed-care and Workers'
Compensation systems), have provided the same results?
It is unclear from my reading whether the timetable for the crossover group
was in step with that for the steroid group or the surgical group. If the
values recorded for monthly intervals were measured from the time of the
epidural steroid injection, then these patients are included in some but not
all of the datapoints in Figures 1, 2-A, and 2-B. These patients had at least
some improvement when they opted to cross over, although the degree of that
improvement is unknown. The "n" value for the
"Epidural" group would have also changed during the early
follow-up period.
On the other hand, if the timetable for the crossover group was reset to
the time of surgery, then postinjection follow-up data may not be included in
these figures, where they would have increased the early scores for the
epidural steroid injection group relative to other groups. Also, if those
patients had been retained in the epidural steroid injection group over the
course of an entire year, their true response to epidural steroid injection
would have been seen. Either (1) the patients who had a failure in the
epidural steroid injection group would have continued to experience failure,
elevating the subsequent pain and motor deficit scores throughout the
remaining timepoints and amplifying the clinical efficacy of surgery, or (2)
these patients would have also experienced clinical improvement over the
subsequent months, consistent with previous observations of patients managed
with epidural steroid
injections1.
Since the improvement seen in association with surgical treatment relative
to epidural steroid injection was significant only in the early follow-up
period, and since the improvement seen in association with epidural steroid
injection relative to placebo or untreated controls is also significant only
during the earliest post-treatment
intervals2-5,
it might be argued that the treatment of herniated nucleus pulposus with
discectomy is effective only for providing more rapid recovery, with no
expectation of incremental long-term benefit. It is difficult to argue, from
the data available here, that the crossover group's good final outcome was the
result of subsequent surgical treatment as opposed to the eventual arrival of
healing promised by natural history. Perhaps the author's insight could
clarify this important aspect of the study.
G.R. Buttermann replies:
In his letter, Dr. McLain posed a question about whether our results would
have been different if there had been no crossover group. That is, once a
patient had been randomized to the epidural steroid injection group, he or she
would have been prevented from crossing over to the discectomy group. The
simple answer is that patients would have gone elsewhere for treatment if they
had had a failure of epidural steroid injections and already had had previous
failures of physical therapy, chiropractic treatment, medications, and so
on.
One should note that enrolling patients into this type of study is very
time-consuming. I think that to try to enroll patients into a study without a
crossover group would be extremely difficult, with so many patients opting not
to participate that any results would be invalid because the few enrolled
patients would not be representative of patients with herniated nucleus
pulposus as a whole.
More to the point, I think that in the long term (one year or more), there
would have been improvement in the majority of patients regardless of
treatment. I believe that our treatments improve the quality of life in the
relatively short to moderate-term follow-up period. This study did not
specifically analyze the scenario that Dr. McLain questioned, and thus my
answers are speculative.
Dr. McLain also asked for clarification regarding the crossover group in
the figures illustrating the outcome scores. Clarification of the crossover
group is probably most easily seen in Figure 3, which demonstrates that, as a
group, the patients who considered themselves to have had a failure of
epidural steroid injections still had some mild improvement in their outcome
scores. For the hypothetical case that Dr. McLain mentioned, I think that this
group of patients, had they not been allowed to cross over to discectomy,
would have had higher (worse) scores over the first six to twelve months but
then probably would have had scores similar to the discectomy group at one to
two years. However, this was not addressed in the study and thus my comments
remain speculative.
Finally, my clinical impression is in agreement with the comments made by
Dr. McLain, specifically that surgical treatment provides rapid relief of
symptoms but that in the long term (that is, one to two years later) symptoms
probably would be fairly similar in all groups. Thus, I believe that
discectomy or successful epidural steroid injections provided significant
improvement in the quality of life within the first year after the onset of
symptoms for our patients who had had no improvement after a minimum of six
weeks of nonoperative treatment.
I think that the take-home message of this study is that we can improve
patients' symptoms substantially in the first few months and allow patients to
become functional again in a reasonable amount of time (weeks rather than
years). The goal of many orthopaedic procedures is to get patients functional
in a timely fashion so that they can become productive again and avoid the
financial and emotional hardships of prolonged pain and disability.