Abstract

STUDY OBJECTIVE--To examine (a) changes in the shape of the distribution of dietary fat intake as the mean dietary fat intake of the population shifts and (b) implications for setting national dietary goals. DESIGN--Data on the percentage of energy from total fat, saturates, monounsaturates, polyunsaturates, and the P:S ratio were analysed for two dietary intervention trials and six cross sectional dietary surveys. The nutrient distributions from each study were described in terms of the mean, standard deviation (SD), coefficient of variation (CV), and skewness statistic. For the intervention trials statistical parameters were compared for groups who received and did not receive dietary advice. For the cross sectional studies, statistical parameters were compared across groups with different levels of mean fat intake. The implications of the results for setting dietary goals were considered using statistical models. MAIN RESULTS--For most fat fractions there was a positive association between the mean and the SD, and an inverse association between the mean and the CV, indicating that as the mean shifts upwards the SD increases but not in proportion to the mean. This is intermediate between a constant SD and a constant CV model. For a population nutrient goal of a maximum of 15% saturates, the estimated population mean for British women would be 8.4% using the constant SD model and 10.8% using the constant CV model. For saturates and the P:S ratio, a lower mean intake was associated with a greater positive skew in the distribution of reported intakes. For saturates, this is consistent with a group of high fat consumers who fail to reduce their intake as the population mean shifts downwards: a "rearguard effect". Findings for the P:S ratio are consistent with a group of consumers who produced a strong positive skew at low mean intakes, which reduced in size as the mean population intake increases: a "vanguard effect". CONCLUSIONS--These findings provide evidence that the distribution of fat intake may change with the mean to a degree that the setting and monitoring of nutritional goals may need to take account of changes in variance and shape of the intake distribution.