Can Nutrition Policy Evaluation be Evidence Based? Examples and Dilemmas

There is much talk about the need for policies to be evidence-based; an admirable objective but not easily achievable, as illustrated in this article with reference to policies to improve diets. The evidence base requires estimates of policy effectiveness (the impact of policy on diets), of the impact of dietary change on health, and an estimate of the cost-effectiveness of the measure. Each is problematic. Randomised Control Trials are not suitable for estimating policy effectiveness in a market environment so alternative statistical approaches have used existing secondary data, but these approaches suffer from a lack of longitudinal data, the need to use data collected for other purposes, and the absence of harmonised data across Europe. Also, there are often insufficient observations to make meaningful assessments for relevant population sub-groups such as the poor, elderly or ethnic minorities. Converting dietary change into health change requires data on dose-response functions that are almost entirely absent, as are reliable estimates of the likely lags between poor diets and poor health outcomes. Cost effectiveness approaches differ between economists and public health professionals; each makes different (if implicit) assumptions about the behaviour of consumers that leads to poor diets and about the need to distinguish between private and social costs. Neither approach is unambiguously correct. Despite all the limitations, the need for policy action is urgent; fortunately the mounting evidence suggests most policies cause a small improvement in diets and that these are cost effective. They are also widely accepted by the public at large.