Reply

Abstract

We read with interest the article by Perhala et a1 (1) on local infectious complications following total joint arthroplasty in rheumatoid arthritis patients treated with methotrexate (MTX). Readers likely will find it reassuring that the authors detected no significant difference in the rate of postoperative infectious complications in the MTX-treated group versus the non-MTX-treated group.We believe such reassurance is unjustified. The sample size used in the study was much too small to allow detection, with adequate statistical power, of potentially important differences in risks between groups. Given the sample sizes of 110 joints in patients not treated with MTX and 92 joints in patients treated with MTX, and assuming a 5% rate of infectious complications in the non-MTX-treated group and a significance level of 0.05, the power of this study to detect a 10% rate of infectious complications in the MTX-treated group (a relative risk of -2) was just 0.28 (2). This means that even if a 10% rate of complications in patients treated with MTX truly existed, the authors had just a 28% chance of detecting it in their study. The power to detect a 15% rate of complications (relative risk of -3) was 0.67. A power of 0.80 is generally regarded as adequate. Given their sample sizes, the smallest relative risk the authors could have detected with a power of 0.80 was 3.5 (complication rate of 17.5% versus 5%).How many patients would be required to detect realistic, meaningful risks, with adequate power? Assuming a significance level of 0.05, a power of 0.80, and a risk of infectious complications in the non-MTX-treated group of 5%, 474 patients would have to be enrolled in each arm of the study in order to detect a relative risk of 2 (10% versus 5%), and 160 patients would have to be enrolled in each arm in order to detect a relative risk of 3 (15% versus 5%). The number of patients required would be even larger if deep sepsis were the sole complication under study (the authors included superficial infection, hematoma, necrotic eschars, and prolonged wound drainage).These calculations indicate that Perhala et al had little possibility of contributing meaningfully to the debate about infectious complications associated with methotrexate therapy in patients undergoing total joint arthroplasty. The only reassurance we derive from this study, based on a 95% confidence interval for the log odds ratio (3), is that the risk is less than 5-fold. We believe these issues of power and sample size should have been quantified in their report and, more generally, should be primary concerns in the design, reporting, and peer-reviewing of clinical studies. To the Editor:Katz et a1 raise a valid criticism regarding issues of power and sample size of the patient population in o u r recently published study. We compared the local infectious complications following large joint replacement in rheumatoid arthritis patients, utilizing a patient group treated with methotrexate (MTX) versus those never treated with MTX or any other chemotherapeutic...