 |
 |
 |
|
| home | CME | help | feedback | subscriptions | archive |
|
© 1998 The Journal of Bone and Joint Surgery, Inc.
CorrespondenceTO THE EDITOR: "A Meta-Analysis of the Efficacy of Non-Operative Treatments for Idiopathic Scoliosis" (79-A: 664–674, May 1997) by Rowe et al. is an example of the shortcomings of this method of literature review. The first paragraph of the introduction indicated that this was a study of adolescent idiopathic scoliosis. However, one of the two studies on the Milwaukee brace4 dealt strictly with the treatment of juvenile idiopathic scoliosis. Only one study6 was used to represent the natural history of scoliosis. All of the bracing data were compared to only that study as a control. The so-called control study included only girls who were between the ages of ten and fifteen years, with chronological age being the only indicator of maturity. Only single curves with an apex between the eighth thoracic and first lumbar vertebrae were included. Thus, the comparison group of untreated patients serves as a questionable control group for this meta-analysis. Each factor analyzed (age, hours of daily wear, type of brace, and criterion for failure) was assessed independently. For example, hours of wear was evaluated independent of age. Seven of the studies that recommended twenty-three hours of brace wear1,2,5,7,10-12 included patients who were at or near skeletal maturity. Thus, there is a bias in favor of better results with twenty-three hours of brace wear because skeletally mature patients were included in these studies but not in the more current studies of part-time brace wear3,6,8. Rowe et al. reviewed only two studies on the Milwaukee brace. One study was published twenty-one years ago1 and one, sixteen years ago4. In Table II, the study by Kahanovitz et al.4 is listed as a twenty-three-hour brace-wear study. However, the content of that paper is reflected in its title, "The Part-Time Milwaukee Brace Treatment of Juvenile Idiopathic Scoliosis." The committee interpreted the results of these two papers with use of meta-analysis and concluded that the Milwaukee brace is 99 per cent successful and that twenty-three hours of brace wear is 97 per cent successful. Those conclusions alone should alert the clinician to the possibility of an error in methodology. In Table II, the 100 per cent rate of success reported by Edmonson and Morris1 is credited to the Milwaukee brace. Rowe et al. stated that "the number of failures of treatment in each study was determined by calculating the total number of patients who had unacceptable progression of the curve, ... could not comply with or tolerate treatment, or who had an operation." If these criteria had actually been applied to the study by Edmonson and Morris, then the rate of success for the Milwaukee brace would not have been 100 per cent as reported. Edmonson and Morris started with 125 patients, but only the fifty-two who had a successful result were included in the meta-analysis. Of the original 125 patients, only sixty-five completed the recommended program. Therefore, sixty patients "could not comply with or tolerate treatment" and should have constituted a failure group in the meta-analysis. Six patients had an operation and definitely should have been reported as failures by Rowe et al. They were not. In addition, Edmonson and Morris only reported averages for the entire group of fifty-two patients who successfully completed treatment and who had a personal follow-up evaluation. This method of analysis is invalid because averages do not indicate the number of patients who do or do not have progression. Therefore, this study should have been excluded from the meta-analysis. Instead, it is the foundation of the conclusion by Rowe et al. that "the outcome associated with the Milwaukee brace was significantly better than that associated with all other types of braces and that of treatment with observation only (p < 0.0001 for all comparisons)." The validity of this p value is obviously erroneous. In fact, the meta-analysis performed by Rowe et al. is not valid according to accepted standards of meta-analysis. Quality scores were assigned to the studies reviewed, but they were not used to accept or reject studies or to determine the weight of each study. Instead the weight of each study was determined on the basis of sample size alone. The proportions of success or failure were simply based on the number of patients in each study without any adjustments being made for the quality score. Three of the references were unpublished data, including the doctoral thesis by Styblo10. This citation alone accounts for 20 per cent of the bracing patients who were reviewed. I am surprised that unpublished data, which cannot be corroborated, were utilized to this degree. Rowe et al. stated that "nineteen studies were rejected ... because they contained insufficient data regarding treatment and follow-up or they lacked data regarding the completion of treatment." However, the reasons for exclusion did not seem to be applied evenly. Both of the studies on the Charleston bending brace3,8, including the report of 139 patients by my colleagues and me8, should have been excluded from the analysis because they are preliminary studies of patients who are still being managed. Our study is even identified as a preliminary report in its title. Instead of excluding our study, Rowe et al. extracted the data on the forty-four patients who had completed treatment. In the Discussion section of our report, my co-authors and I stated that the "final results of this study will not be known until all patients have passed skeletal maturity. During the period of this study, 44 patients have completed treatment. Thirty-five of these patients did not require surgery while 9 patients (20%) have required surgery. This 20% incidence of surgery in the completed treatment group may be misleading because curves that are `malignant' or truly progressive often require early surgery. Successful brace patients do not complete treatment for several years, so it is possible that the percentage figure for patients requiring surgery may decrease as longer follow-up is obtained." The statement quoted here is the only comment in our preliminary study8 regarding the forty-four patients who had completed treatment at that time. I ask Rowe et al. to explain how they derived and reported a 55 per cent rate of failure for those forty-four patients on the basis of that quotation in a published, peer-reviewed paper8. I have all the raw data from that study and can prove that this meta-analysis is invalid. In fact, our long-term study has now been published9 with a 34 per cent rate of failure instead of the 55 per cent rate extracted by Rowe et al. I submit that the study by Rowe et al. is not a valid meta-analysis but rather a poor review of the literature with confusing statistical manipulations. Perhaps a retraction is warranted.
Charles T. Price, M.D.: The Nemours Children's Clinic, P.O. Box 568908, Orlando, Florida 32856-8908
Dr. Rowe, Dr. Bernstein, Dr. Riddick, Dr. Adler, Dr. Emans, and Dr. Gardner-Bonneau reply: Dr. Price has listed several processes that may hamper meta-analysis research. The Scoliosis Research Society should not be seen as endorsing any brace regimen or type over another. Our meta-analysis was performed to determine the nature of current knowledge regarding treatment of idiopathic scoliosis with a brace. We offered this report as a snapshot of the current literature. We hope that this study will be used as a starting point for future research on brace therapy. We are divided on the usefulness of including preliminary results in meta-analysis. The original version of our manuscript raised this possible conflict in the Discussion. However, this argument was not accepted by the editors and was dropped during the editorial process. The real answer regarding the effectiveness of part-time treatment regimens may be provided by Dr. Price and other researchers when their definitive studies are published. Interactions of age and hours of treatment may have affected the results of the meta-analysis. It cannot be determined from the literature whether this interaction is real. There is no more evidence for the interaction than there is against it. Furthermore, there may be other interactions that we were unable to analyze on the basis of the data in our sources. For example, the type of curve was not classifiable in our abstraction. This factor most likely influences the effectiveness of bracing. Other variables that were not analyzed included possible cultural effects, the attitude of the attending physician toward bracing, ethnic effects, and many other possible contributing factors. Clear final answers would be possible after controlled, randomized studies of bracing. We thank Dr. Price for his useful ideas and constructive comments and look forward to reviewing his updated research results.
Dale E. Rowe, M.D.; Daryle Gardner-Bonneau, Ph.D.: Michigan State University/Kalamazoo Center for Medical Studies, Linda Richards Building, 1000 Oakland Drive, Kalamazoo, Michigan 49008 Saul M. Bernstein, M.D.: Department of Orthopaedic Surgery, Southern California Orthopaedic Institute, University of Southern California, 6815 Noble Avenue, Van Nuys, California 91405 Max F. Riddick, M.D.: Jewett Orthopaedic Clinic, 515 West Highway 434, Suite 210, Longwood, Florida 32750 Federico Adler, M.D.: Surgery Service, Kansas City Veterans Administration Medical Center, 4801 East Linwood Boulevard, Kansas City, Missouri 64128 John B. Emans, M.D.: Department of Orthopaedic Surgery, Children's Hospital, 300 Longwood Avenue, Boston, Massachusetts 02115
References
  CiteULike  Connotea  Del.icio.us  Facebook  Technorati  Twitter What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| home | CME | help | feedback | subscriptions | archive | site index |
|
Stanford University Libraries' HighWire Press (TM) assists in the publication of JBJS Online. |