Dietary patterns capture the overall diet and thereby provide information on how nutrients are consumed in combinations, and have been suggested to be a better method than studying single nutrients. The present study explored the relationship between dietary patterns at baseline and incidence of obesity at 10-year follow-up in women.

A longitudinal study using baseline measurements from 1992–1996, including food intake, medication, heredity, socio-economic status, lifestyle and measured body composition, and follow-up data collected in 2002–2006 including measured body composition.

Data from the Västerbotten Intervention Programme (VIP) in Sweden.

A total of 6545 initially non-obese women aged 30–50 years.

Among women reporting plausible energy intakes, the ‘Fruit and vegetables cluster’ predicted the highest incidence of obesity (OR = 1·76, 95 % CI 1·11, 2·76; P = 0·015) compared with women in the other food pattern groups combined. When adjusting for metabolic factors and BMI at baseline, the risk for obesity in the ‘Fruit and vegetables cluster’ was attenuated to non-significance. In contrast, high intake of fruit per se was associated with a decreased risk of developing obesity (OR = 0·69, 95 % CI 0·51, 0·91; P = 0·010).

Dietary pattern groups identified by cluster analysis are likely to reflect characteristics in addition to diet, including lifestyle, previous and current health status and risk factors for future disease, whereas intake of fruit per se was a stable indicator and less affected by baseline characteristics. These results underscore the need for complementary methods in understanding diet–disease relationships.