This issue of The Journal contains an important article entitled "Resurfacing of the Patella in Total Knee Arthroplasty. A Prospective, Randomized, Double-Blind Study," by Barrack et al. This study is important not only from the standpoint of its findings but also from the standpoint of its methodology.
Most reports in the orthopaedic literature relating to operative procedures are retrospective reviews of series of patients. Indeed, very valuable information can be, and is, obtained from these studies. However, retrospective studies have certain limitations, including, in many instances, the inability to control for many variables because of the lack of specific criteria for inclusion and exclusion. A prospective, randomized, double-blind study, with the careful use of criteria for inclusion and exclusion, allows the investigators to minimize the number of variables that they are analyzing. No word is perhaps more important in the field of statistics than the word random. An item is chosen at random from a population if, in the selection, every such item in the population has the same probability or chance of being selected3. The process of selection itself is critical. The way to obtain statistically dependable results is to choose at random2. The technique of double-blinding the surgeon and the observer also minimizes bias and strengthens the study.
Barrack et al. are to be congratulated for performing a prospective, randomized study. The patients all were evaluated by the same research nurse, independent of physician involvement, in a blinded manner, and all patients were followed for a minimum of two years. The operative procedures were performed by one of three authors at one of three university-affiliated teaching hospitals. All patients had a preoperative diagnosis of severe degenerative osteoarthrosis. The criteria for exclusion included a previous tibial osteotomy or an operation involving the extensor mechanism, a history of septic osteoarthrosis or osteomyelitis, a severe medical disability that limited walking, a disabling disease affecting other joints of the lower extremities, inflammatory arthropathy, and severe deformity. The authors defined a severe deformity as a varus or valgus deformity or a flexion contracture greater than 15 degrees.
The method of randomization is important. The technique used by Barrack et al. involved opening a randomly selected envelope in the operating room, after all femoral and tibial cuts had been made and immediately before the patella was prepared, to determine whether or not the patella was to be resurfaced. The drawing of envelopes or a similar method is the only appropriate way to randomize patients. For example, it is not appropriate to randomize them by hospital number or by the day of the week.
The authors evaluated 137 patients for inclusion in the study. Ninety-seven patients who met the criteria for inclusion were offered enrollment in the study, and eighty-nine patients (121 knees) agreed to participate. It is very impressive that 92 per cent of all eligible patients participated. Because of three early postoperative deaths that were unrelated to the operative procedure, 118 knees in eighty-six patients were available for analysis. The institutional review boards of all of the hospitals involved approved the study protocol, including the consent forms.
All patients received the same posterior cruciate-sparing prosthetic components, and, as previously noted, all operations were performed by, or under the direction of, one of the authors. Knees that were randomized to resurfacing of the patella had insertion of an all-polyethylene (non-metal-backed) patellar component. The goal of the resurfacing technique was accurate reproduction of the preoperative thickness of the patella as determined with the use of calipers. In the knees that did not have resurfacing, a so-called patelloplasty was performed; this included removal of osteophytes, smoothing of fibrillated cartilage, and drilling of eburnated bone. The procedures were carefully documented with intraoperative photographs of the patellar articular surface. The perioperative regimen was the same for all patients, and it included the use of antibiotics as well as prophylaxis against venous thrombosis. Physical therapy was also uniform in the series. All patients had clinical assessments, and their satisfaction was assessed with use of questionnaires.
It is true that Barrack et al. studied a relatively small series, but the variables were limited and the patient population was well controlled. Such studies are especially valuable when one particular variable is being analyzed. The authors have shown that it is possible to perform a prospective, randomized, double-blind study successfully in a population of patients having an operative procedure. I believe that investigators should be encouraged to consider performing such prospective studies if it is at all feasible. Prospective studies dealing with various drug interventions are fairly common. This report demonstrates that such studies are also possible when an operative procedure is involved, particularly if a straightforward, clinically important variable is analyzed.
Charles R. Clark, M.D.
Deputy Editor for Adult Reconstruction