Leisure-Time Physical Activity and Mortality: How Is Sausage Made?

To the Editor.—The Finnish Twin Study of leisure-time activity and mortality by Dr Kujala and colleagues¹ has inherent strengths not conveyed by the authors' esoteric statistical analysis, replete with P values and confidence intervals but presenting few operative data. Unfortunately, during recent years medical journals have adopted the practice of presenting derivative data values buttressed by P values and confidence intervals, rather than presenting actual numerical study findings so that readers can discern data patterns relative to reputed determinants.² For example, footnotes in Tables 2 and 4 stating "Adjusted for smoking (pack-years and current smoking habit)" are not adequate substitutes for allowing readers "to see just how sausage is made"—how deaths are distributed in bivariate tables. No sound physician/scientist would have confidence in values derived in a black box and buttressed by confidence intervals. In fact, statistical esoterica are ordinarily aimed at suspending the critical faculty of readers.^3- 4 For the Finnish Twin Study, readers should be enabled to study tabular mortality data patterns relative to the putative determinant (leisure activity), arrayed against the foremost known determinant of premature mortality (pack-years of smoking), so they can judge the relative strength of these determinants, acting separately or in concert, and thus, gain a measured sense of the operative strength of leisure activity in preventing mortality.