TO THE EDITOR:
We read with interest "Comparison of Intravenous and Oral Antibiotic Therapy in the Treatment of Fractures Caused by Low-Velocity Gunshots. A Prospective, Randomized Study of Infection Rates" (78-A: 1167—1171, Aug. 1996), by Knapp et al. The authors are to be congratulated on the prospective design and the rigorous follow-up, but two aspects of the method and their main conclusion require comment.
First, the main outcome measure is "a strict definition of infection," but that definition was not stated. We have shown that infection rates are substantially higher when assessed on clinical grounds than when a microbiological definition is used1. Second, the follow-up period for the assessment of infection was not stated. Osteomyelitis and infection around an implant can present long after an operation and after apparently uncomplicated healing of the wound. The Surgical Infection Study Group recommended that all infections developing within one year after an operation be audited as postoperative infection2. Finally, the authors found an infection rate of 2 and 1.695 per cent for the two groups and concluded that "there was no significant difference ... between the two groups." The 95 per cent confidence interval of the difference is -3.9 to 3.29 per cent, or approximately a doubling of the infection rate in one direction or the other. The conclusion that the defined population of patients "can be managed with oral antibiotic therapy without an increased prevalence of infection" is a type-2 statistical error. Despite the unenviable frequency of this type of injury in Los Angeles, the sample size is too small to support the main conclusion of this paper.
Andrew J. Kelly, F.R.C.S.; Ian G. Winson: Winford Unit, Avon Orthopaedic Centre, Southmead Hospital, Bristol BS10 5NB, United Kingdom
Dr. Knapp, Dr. Patzakis, Dr. Lee, Dr. Seipel, Dr. Abdollahi, and Dr. Reisch reply:
We welcome the comments of Mr. Kelly and Mr. Winson. However, their statement that our definition of infection was not stated is not valid. On the contrary, we clearly stated: "Treatment was considered to have failed if signs or symptoms of infection developed." This was based on direct clinical evidence, as stated. The only exclusion was the so-called bullet cellulitis, as described in our article. Cultures were not used alone to determine the presence or absence of infection. In fact, we found no association between the results of cultures and the clinical outcome, and we no longer routinely take specimens for culture.
Additionally, they state that "the follow-up period for the assessment of infection was not stated." Our article clearly stated: "The patients were followed until there was both clinical and radiographic evidence of union." To date, there has not been any late presentation of infection in our study group. While there is always the possibility that an infection may present many years later, we believe that this is unlikely.
Their final point that we committed a type-2 statistical error is correct if their infection rates of 2 and 1.695 per cent are considered. However, we reported an over-all infection rate of 2 per cent as well as an infection rate of 2 per cent for both Group 1 and Group 2. From our sample size, we were unable to report a significant result of 1.7 per cent, much less 1.695 per cent. We agree that our sample size is too small. This is the largest study (190 patients) to date on this important subject, and fifteen months were required to accumulate these patients in a prospective, randomized fashion. In order to report a p value of less than 0.05, we would need a study population of more than 18,000 patients. However, we believe that the results would not vary greatly from those reported in our paper.
Thomas P. Knapp, M.D.: Santa Monica Orthopaedic and Sports Medicine Group, 1301 Twentieth Street, Suite 150, Santa Monica, California 90404
Michael J. Patzakis, M.D.; Jackson Lee, M.D.; Peter R. Seipel, M.D.; Karim Abdollahi, M.D.; Robert B. Reisch, M.D.: Department of Orthopaedic Surgery, University of Southern California School of Medicine, 2025 Zonal Avenue, GNH 3900, Los Angeles, California 90033