In reply
Christmas and Moncada raise an interesting issue. Indeed, at 12 weeks, diclofenac sodium was no longer superior to acetaminophen or to placebo (P = .25, analysis of variance). This confirms the "eyeball" test of the Figure (panel A). Nevertheless, the drop in pain elicited by diclofenac at 2 weeks persisted at 12 weeks (P = .002, t test). However, at 12 weeks, by the "eyeball" test again, both the acetaminophen and placebo groups also demonstrated a drop in pain that was not present at 2 weeks, though neither was statistically significant by the t test (P = .13 and P = .42 for acetaminophen and placebo, respectively). We believe that this represents regression to the mean1 and is an argument in favor of performing future, larger, placebo-controlled studies of acetaminophen as we stated in the text. We do not believe that the title should have been "Lack of Efficacy of Acetaminophen or Diclofenac in Treating Symptomatic Knee Osteoarthritis." A large number of placebo-controlled studies of at least 4 weeks' duration have demonstrated the efficacy of diclofenac and nonsteroidal anti-inflammatory drugs in general2- 4; in contrast, as we indicated in the text, only 1 small study has compared acetaminophen with placebo. Minor adverse gastrointestinal adverse effects (dyspepsia) were present in 3 of the 5 diclofenac-treated subjects who dropped out. One acetaminophen-treated subject dropped out because of gastrointestinal causes (dyspepsia), another to "malaise."