LETTERS Entacapone and Motor Fluctuations in Parkinson’s Disease Jack E. Riggs, MD There are some confusing details in the recent Annals article by the Parkinson Study Group (PSG), concerning the use of entacapone.’ First, the y axis of Figure 2 is labeled “Log (Total Daily Levodopa Dosage)” and goes from 7.04 to 7.36. As “Log” indicates base 10, this means the total daily dose range of levodopa indicated on the y axis goes from 10,964,782 to 22,908,677 mg, an obvious misintent. If the PSG actually meant “Ln,” which indicates base e, then the total daily dose range of levodopa indicated on the y axis would be from 1,141 to 1,572 mg. But this range does not make sense either if the average total daily dose of levodopa, from Table 1, was 752 mg in the placebo group and 791 mg in the entacapone group. If the PSG did not mean a logarithmic scale, but rather meant t h e y axis number times 102 to indicate the total daily dose of levodopa, then the dose range indicated on t h e y axis went from 704 to 736 mg. But then this range cannot be right either if the average daily total dose of levodopa dropped by 12% (or -100 mg) in the entacaponetreated group as stated in the article. Second, the PSG indicated that the overall treatment effect (the amount of “on” time per day) of entacapone over placebo was increased 5% or “approximately 1 hdday.” But if the patients at baseline averaged approximately 10 hours of “on” time per day, then the net beneficial effect of entacapone should be 30 minutes per day, not 60 minutes per day. For the patient, this result is the single most practical and meaningful finding of the study. Moreover, the difference between 30 minutes and 60 minutes of improved mobility is no small matter to a patient facing difficult cosdbenefit choices. Finally, the PSG found that only patients with less than 55% of “on” time during their awake hours at baseline had a significant beneficial effect from entacapone. Of the 103 patients treated with entacapone, how many patients fell into this subgroup? Department of Neurology, West Virginia University Health Sciences Center, Morgantown, WV Reference 1. Parkinson Study Group. Entacapone improves motor fluctuations in levadopa-treated Parkinson’s disease patients. Ann Neurol 1997;42:747-755 Reply Michael McDermott, PhD, and Karl Geburtz, M D , on behalf of the Steering Committee We thank Dr. Riggs for his careful reading of our manuscript. (1) Dr. Riggs is correct that there is an error in Figure 2. The term “Log” was used to indicate the natural logarithm (base e). Each patient’s total daily dosage was erroneously multiplied by the number of doses that the patient took per day. W e redid the analyses, using the correct total daily dosage, and obtained very similar results regarding the entaca- pone/placebo comparisons. The average total daily dosage remained 12% (- 100 mglday) lower in the entacapone group than in the placebo group (95% confidence interval, 5-18%; p = 0.0004) across weeks 8, 16, and 24. The revised figure, using the correct data, is virtually identical to Figure 2 in the article, except for the scale on the y axis (figure available). (2) Subjects averaged approximately 60% “on” time while awake (10 hours of nearly 17 awake hours). Treatment with entacapone raised this to approximately 65% “on” time (10.84 of nearly 17 awake hours), so the increase was nearly 1 hour on average. (3) The number of subjects in the subgroup having less than 55% “on” time at baseline was 72 (36 entacapone and 36 placebo). It is noteworthy that the subgroups were chosen so that they would contain approximately equal numbers of subjects, not because 55% was the threshold above which patients did not receive significant benefit from entacapone (see below). It should also be noted that the interaction between treatment and baseline percent “on” time was analyzed by using baseline percent “on” time as a continuous variable. This analysis yielded the quoted p value of 0.0002. The results in Table 3 represent a simplified description of those results. It is much easier to describe the differential treatment effects in this way than to show the nonparallel regression lines of change in percent “on” time versus baseline percent “on” time for each treatment group. The point that we wanted to illustrate is that the more “off” time one has at baseline, the more benefit can be gained by adding entacapone. It is not accurate to say that “the PSG found that only patients with less than 55 percent of ‘on’ time during their awake hours at baseline had a significant beneficial effect from entacapone.” There is nothing magic about 55% “on” time (as noted above); however, it is fair to say (as we did) that the treatment effect is particularly prominent in this subgroup. The baseline percent “on” time above which the benefit of entacapone is no longer significant was not determined by these analyses. Note that in Table 3, the effect in the group with baseline “on” time of 55-70% is nearly 4 points. The lack of statistical significance in this subgroup should not be overinterpreted, as only one-third of the subjects are in this subgroup, and there is greatly reduced power to detect statistical significance of treatment effects within subgroups. Although the confidence interval is consistent with no treatment effect within this subgroup, it is also consistent with a 10-point treatment effect. In other words, Dr Riggs is erroneously interpreting a nonstatistically significant finding as a negative finding (rather than an inconclusive finding). Departments of Neurology and Biostatistics, University of Rochester Medical Center, Rochester, NY 292 Copyright 0 1998 by the American Neurological Association

1/--страниц