BMJ  2005;331:E373 (13 August), doi:10.1136/bmj.331.7513.E373

Following are edited excerpts from the Rapid Responses generated by this article, all of which can be read in their entirety at http://bmj.bmjjournals.com/cgi/eletters/330/7502/1233.—Editor

BMJ USA: Letter

RAPID RESPONSES FROM BMJ.COM

Fairbank et al reported a randomized controlled trial of surgical stabilization of the lumbar spine compared to intensive rehabilitation program in patients with chronic low back pain.¹ The principle of randomization was based on a "gray zone" of uncertainty (on behalf of the treating clinician) as to whether surgical stabilization or an intensive rehabilitation program was the appropriate management (ie, that there was clinical equipoise as to whether the patient should have surgery or non-surgical treatment).

Many spinal surgeons including myself would not feel that this is an area where clinical equipoise is possible. If a patient is thought to be appropriate for an intensive rehabilitation program then that is what should occur. Surgery is contraindicated until all conservative treatments have been exhausted, including an intensive rehabilitation program. A number of spinal surgeons were unwilling to join this trial because of that.

If a spinal surgeon is apparently in clinical equipoise as to surgical/non-surgical management, then it is likely that a patient has either (i) not had a full course of medical management or (ii) has less severe disease. Either of these would bias the trial in favor of the intensive rehabilitation program. Another noteworthy point is that despite the likelihood that the trial is biased in favor of non-surgical treatment 28% of patients randomized for rehabilitation eventually had a surgical procedure.

Koes (p 327) correctly concludes that there is limited evidence that confirms the benefit of surgical stabilization and that appropriate patient selection is difficult. However the primary position is that until a patient has had a full course of non-surgical treatment surgical stabilization is not an appropriate option.

Newcastle General Hospital, Westgate Road Newcastle Upon Tyne, NE4 6BE, UK

We read with great interest the recent contribution by Fairbank et al (May 28, 2005) regarding the surgical versus rehabilitation interventions for chronic low back pain. As a center that successfully utilizes both operative and non-operative techniques in the management of our patients, we hoped to draw conclusions from the paper, which were similar to our own philosophy. Unfortunately, the study has a number of shortcomings that may lead to misinterpretation of the collected data and consequently conclusions which are unsubstantiated.

First, the entry criteria for the study essentially excluded all patients for whom surgery was the initial treatment recommendation. The exclusion of this subset of patients, likely having the highest pretreatment Oswestry scores, introduces an unacceptable bias. By removing these patients, who would conceivably benefit most from surgery and show the greatest benefit from each treatment dollar spent, the authors effectively predetermine the outcome of the study. It is not surprising that the post-operative Oswestry scores barely reached statistical significance.

Second, by employing the intention to treat method of analysis, the study allows and fails to account for one-way crossover of patients. For example, the 48 patients (28%) in the non-operative group who crossed over to the surgical group should be considered failures of rehabilitation. Obviously, if these patients required surgery within the two years of study follow-up, they would have incurred the cumulative cost of both treatment regimens. This could hardly be considered an appropriate control group, nor a cost-effective means of patient care.

Third, the authors admit to an extended period of time for patient accrual into the study (June 1996 to February 2002). In a study that requires a prolonged enrollment time (over 60 months) to enroll patients, we are concerned by the 20% loss to follow-up rate at 24 months. How can one possibly make decisions regarding cost-effectiveness and overall treatment efficacy, if one cannot verify adherence to the patient's prescribed treatment regimen and follow up over the standard two years? Similarly, we are concerned about the length of enrollment. Could the techniques and costs associated with any intervention possibly be standardized over such a period of time?

In conclusion, we applaud the significant effort put forth by Fairbank et al, but also caution against drawing conclusions unsubstantiated by their data.

The Virginia Spine Institute 1831 Wiehle Avenue, Reston, VA

CiteULike    Complore    Connotea    Del.icio.us    Digg    Reddit    Technorati    What's this?