To identify predominant dietary patterns in four African populations and examine their association with obesity.

Cross-sectional study.

We used data from the Africa/Harvard School of Public Health Partnership for Cohort Research and Training (PaCT) pilot study established to investigate the feasibility of a multi-country longitudinal study of non-communicable chronic disease in sub-Saharan Africa. We applied principal component analysis to dietary intake data collected from an FFQ developed for PaCT to ascertain dietary patterns in Tanzania, South Africa, and peri-urban and rural Uganda. The sample consisted of 444 women and 294 men.

We identified two dietary patterns: the Mixed Diet pattern characterized by high intakes of unprocessed foods such as vegetables and fresh fish, but also cold cuts and refined grains; and the Processed Diet pattern characterized by high intakes of salad dressing, cold cuts and sweets. Women in the highest tertile of the Processed Diet pattern score were 3·00 times more likely to be overweight (95 % CI 1·66, 5·45; prevalence=74 %) and 4·24 times more likely to be obese (95 % CI 2·23, 8·05; prevalence=44 %) than women in this pattern’s lowest tertile (both P<0·0001; prevalence=47 and 14 %, respectively). We found similarly strong associations in men. There was no association between the Mixed Diet pattern and overweight or obesity.

We identified two major dietary patterns in several African populations, a Mixed Diet pattern and a Processed Diet pattern. The Processed Diet pattern was associated with obesity.