A Difference Is a Difference if It Makes a Difference

Abstract

To the Editor: We read with great interest the meta-analysis by Ballantyne et al [1]. The conclusion of the article was, "These meta-analyses confirm that postoperative epidural pain control can significantly decrease the incidence of pulmonary complications." Results were expressed as risk ratios. We wanted to put this into a clinical context and therefore re-analyze the original trials. We did so with comparisons of epidural with systemic opioids, with atelectasis as an end point Figure 1in the original paper). Compared with systemic opioids, the number needed to treat (NNT) for an atelectasis with epidural opioids was -10.7 (95% confidence -6 to -35), for nausea or vomiting was -9.6 (-5.6 to -34), and for urinary retention was 6 (3.7 to 15). Of 359 patients receiving an epidural, 3 had a dural tap (all with thoracic epidurals), and in one each, the catheter slipped out or was impossible to insert. Thus, of 100 patients receiving an epidural with opioids, 9 will not have an atelectasis and 10 will not vomit or be nauseated who would have had these complications had they received systemic opioids. However, 16 additional patients will have urinary retention and another 2 will experience catheter-related complications. The NNT has been known for 10 yr [2]. A way to calculate a 95% confidence interval around the point estimate was proposed [3], and, more recently, the NNT has been put into a wider clinical context [4]. Looking at the NNT data, rational decision making becomes possible. Martin R. Tramer, MD, DPhil Philippe Garnerin, PhD Division of Anaesthesiology; Department APSIC; Geneva University Hospital; CH-1211 Geneva, Switzerland Dr. Ballantyne did not wish to respond.