844-853 All letters

In your June editorial, “Is Blockade of Pancreatic Lipase the Answer?” (1), you suggest that the success of the 2-y clinical trials of orlistat is diminished by the propensity of obese patients to regain weight. I disagree. The North American Association for the Study of Obesity (NAASO) considers a weight loss of ≥ 5% to be clinically significant (2). In each of the published clinical trials of orlistat, the average weight loss for orlistat-treated patients at the end of the treatment period met or exceeded these guidelines. The efficacy of antiobesity therapies must also be measured by evaluating outcomes other than weight loss. These include positive changes in serum lipids, blood pressure, fasting insulin concentrations, and glycemic control (3, 4). The clinical efficacy of orlistat is greatly enhanced by its effects on these outcomes. As with other treatments for chronic illness, however, efficacy also demands patient compliance with prescribed treatment regimens. The fact that some patients outside the clinical trial environment may have less behavior-modification reinforcement than is provided by a controlled clinical setting should not undermine the potential benefits of orlistat to patients who are motivated to succeed. In addition, you failed to acknowledge the intrinsic design of the study as a weight-regain study rather than a weight-loss study. The editorial suggests that the benefits of weight reduction may be erased over time. Yet, in the same issue of the Journal, Hill et al (5) present data showing that orlistat is significantly more effective than placebo in preventing weight regain. Patients were prescribed a diet of higher energy value to ensure weight maintenance and those who gained weight were asked to maintain the higher weight. Despite the introduction of this bias, after 1 y of treatment with orlistat, nearly one-quarter of patients gained no weight at all or continued to lose weight (5). It is imperative to consider the intention of the initial study design when critiquing the findings of a study.