MEASURES OF ASSOCIATION AND BENEFIT Clinicians and research scientists are usually more familiar with, and tend to favour, measures of association such as relative risks (or odds ratios) and measures of impact such as the population attributable risk (PAR), which is the proportion of a disease that, assuming causality, could be avoided if that risk factor were removed. However, the size of a benefit, whether to the individual or to the population, is better measured in terms of actual risks or risk differences, because those measures also take into account the actual likelihood of the disease outcome. For example, exclusive breast feeding may cause a greater reduction in cases of coeliac disease (44%) than of asthma (23%), but because asthma is much more common, the numbers of children who would avoid asthma is far greater than for coeliac disease. In this issue of the Archives, Akobeng and Heller have utilised a new calculation of population impact, which was recently described by their group in Manchester, to demonstrate the effect of breast feeding on the numbers of cases with asthma, obesity and coeliac disease that could potentially be avoided. While the phrase ‘‘population impact number of eliminating a risk factor over a time period’’ (PINER-t) is unlikely to roll off the tongues of community midwives, health visitors or paediatricians, the result is easy to interpret as ‘‘the number of cases of a disease that might be avoided by eliminating the risk factor’’. It is also conceptually straightforward (in essence it is simply the PAR multiplied by the total number of cases in a population). Figure 1 graphically illustrates the PIN-ER-t results from Akobeng and Heller’s paper, and shows its relationship with the PAR.