In Reply.
—We thank Drs Baron and Greenberg for appropriately pointing out that our study would have benefitted by adjusting for the age distribution in the parkinsonian population. However, we doubt that our omission of this adjustment carries the dire consequences suggested.First, the Table Baron and Greenberg provide commits the common error of failing to include adjustment for differential mortality. Smokers suffer a mortality of 1.8 times that of nonsmokers,1 so any group of older living persons will therefore be biased against ever-smokers. The gradient in the Table is greatly accentuated by this bias.Second, the prevalence of current smoking among those with Parkinson's disease (PD) is very low—about 6%.2 Therefore, the relevant figures for age adjustment should not be for ever-smokers, but former smokers, because nearly all ever-smokers with PD are actually former smokers. The ever-smoker figures in Baron and Greenberg's Table are actually totals of former