Letters to the Editors

Trends in fat consumption I read the paper ‘Trends in individual fat consumption in the UK 1900-1985’ by A. M. Stephen and G. M. Sieber (1994) with interest. This is an important subject for anyone concerned with possible links between fat consumption and chronic disease, especially in view of the intense educative activity about dietary fats that is now being encouraged. Try as I may, I cannot reconcile the conclusions of the authors with the data as actually presented. The first and most crucial point concerns the time at which the consumption of fat, expressed as a percentage of dietary energy, started to decline. The authors state variously that this ‘began in the mid-1970s’ (Results, p. 780) or ‘mid to late 1970s’ (Discussion, p. 783). Information on this point is given in three forms. The Tables give data in 10-year groupings, which gives only a rough indication of trends. Fig. 1 is better because all data points are plotted. (Incidentally, the legend to their figure and the Abstract state that data are taken from ninety-seven studies, Table 1 lists ninety-five studies, whereas Table 2 clearly shows that only eighty-seven were included. The points on Fig. 1 are difficult to distinguish in some places, but add up to something more like eighty-seven than ninetyseven.) The spread of points in Fig. 1 for the 1970s and 1980s indicates that in any one year the range of values spanned about 7 or 8 percentage points (for example in 1980, highest 44 YO, lowest 37 %). This suggests that the variability in results from different studies was too great to demonstrate that a 1.2% decline from 1979 to 1985 (Table 3, males) was significant. MAFF Food Survey data show that fat intakes as a percentage of energy have remained constant at about 42% from 1968 to 1992. Although the authors’ decision to avoid household consumption data is understandable, the MAFF data, while perhaps not being an accurate reflection of individual intakes, should provide a reasonable view of trends with time. The principal visual indication of a downward trend after the mid1970s is Fig. 2, the derived plot to which I will return later. The second point concerns the authors’ enthusiasm to show that regional trends in fat consumption may account for regional differences in coronary heart disease (CHD) : something that has not been demonstrated so far in the UK. Data are cited in Table 5 and Fig. 3 as evidence that the trend in fat consumption in South East England is declining while that in Scotland is continuing to rise. The authors fall short of relating this directly to changes in CHD mortality in the two regions, but they imply this by the remark : ‘With the UK studies, however, there were a large number from both Scotland and the South East, two areas with different mortality rates’. Table 5 shows Scottish fat intakes (% energy) in 1960-69, 197&79 and 1980-85 as 40.0,41.3 and 38.1 and in South East England as 38.8, 40.8 and 40.8 in the same periods. The graphical representation (Fig. 3) shows a quite different story, a continuing rise in Scotland and a marginal fall in South East England. While I am not in a position to dispute the authors’ mathematical representation of their data with weighted quadratic regressions, I suggest that few reasonable people presented with the different forms of the data in this paper would accept this as compelling evidence for substantially different trends in the two regions. From the public health standpoint, an important message the authors wish to give is that the decline in CHD mortality observed in the UK relates well to changes in fat consumption in South East England and in Scotland. One’s interpretation clearly depends not only on when fat intake and CHD mortality are judged to have peaked, but also what