Cluster analysis is a statistical technique for grouping individuals with similar dietary features. In this study, cluster analysis, was conducted using diet history data from two clinical trials at baseline and then weight loss outcomes at 3 months were reviewed based on these clusters (n=231).

The cluster solution was analysed using defined food groups and the change in these groups over time. Usually, this sort of research method is performed at a population level – so it is novel within the context of an intervention trial.

Two distinct dietary patterns were identified : Subjects in Cluster 1 reported
food patterns characterised by higher intakes of low-fat dairy and unsaturated oils and margarine and were generally more closely aligned to food choices encouraged in national dietary guidelines. Subjects in Cluster 2 reported a dietary pattern characterised by
very high intakes of non-core foods and drinks, higher- and medium-fat dairy foods, fatty meats and alcohol.

At 3 months, Cluster 2 subjects reported greater reductions in energy intake (-5317 kJ; P<0.001) and greater weight loss (-5.6 kg; P<0.05) compared with Cluster 1.

What it means for clinical practice:

Overweight subjects with reported dietary patterns similar to dietary guidelines at baseline may have more difficulty in reducing energy intake than those with poor dietary patterns. This makes life difficult for practitioners! On the other hand, if a diet history is able to identify, quantify and correctly target the the types of non-core foods and drinks this yields large reductions in energy and, as in this study, more successful weight loss.

This is a modified extract from my first paper ‘Baseline dietary patterns are a significant consideration in correcting dietary exposure for weight loss.’ which is now published online in EJCN http://www.nature.com/doifinder/10.1038/ejcn.2013.26 in the Advance Online Publication (AOP) service.