In South Asians (Indians, Pakistanis, and Bangladeshis) settled overseas, high rates of coronary disease and non-insulin-dependent diabetes occur in association with central obesity and insulin resistance. To examine whether these disturbances were related to diet, we measured 7-d weighed intakes in 173 South Asian and European men aged 40-69 y in London. In South Asians compared with Europeans, respectively, mean energy intake was lower (9.5 vs 10.8 MJ/day, P < 0.001), total fat intake was lower (36.5% vs 39.2% of energy intake, P = 0.007), starch intake was higher (28.0% vs 21.5% of energy, P < 0.001), polyunsaturated fatty acid intake was higher (8.2% vs 7.0% of energy, P = 0.02), and dietary fiber intake was higher (3.2 vs 2.0 g/MJ, P < 0.001). Elevated serum insulin concentrations at 2 h postglucose were associated positively with carbohydrate intake (P = 0.001) and inversely with alcohol intake (P = 0.006), but not with saturated fatty acid intake. The high coronary risk in South Asian people is not explained by any unfavorable characteristic of South Asian diets.

This FFQ was specifically designed for South Asian women in the UK. Despite the diversity of diets, the FFQ had reasonable validity. The role of diet in breast cancer disease aetiology in this population is being assessed with this instrument.

We investigated the relative validity of an interview-administered FFQ to estimate phytoestrogen intake among South Asian women in the UK. A population-based sample of 108 healthy South Asian women completed random repeated monthly 24-h recalls [with a subsample (n = 58) also providing multiple plasma samples] over a period of 1 y followed by administration of the FFQ. The FFQ produced slightly higher estimates of phytoestrogen intake than the 24-h recalls, but the percentage of women classified into the same +/- 1 quartile by the 2 methods was high for all phytoestrogens (from 81 to 94%) with only a small percentage (<5%) being misclassified into extreme opposite quartiles. Energy-adjusted Spearman correlations coefficients between the estimates obtained by the FFQ and the 24-h recalls were 0.55 for genistein, 0.60 for daidzein, 0.70 for secoisolariciresinol, and 0.63 for matairesinol (all P < 0.001). Spearman correlation coefficients between the FFQ estimates and plasma levels were 0.21 (P = 0.12) for genistein, 0.32 (P = 0.02) for daidzein and 0.10 (P = 0.43) for enterolactone; the corresponding values for the 24-h recalls compared with plasma levels were 0.43 (P < 0.001), 0.40 (P = 0.002), and 0.08 (P = 0.50), respectively. The method of triads was used to estimate the validity coefficients (VCs) between the estimates provided by each assessment method and "true intake." The FFQ had the highest VC for lignans (0.91 vs. 0.73 for 24-h recalls and 0.11 for plasma samples) and satisfactory VCs for both genistein (0.46 vs. 0.95 and 0.45, respectively) and daidzein (0.67 vs. 0.83 and 0.45, respectively). This FFQ is thus a relatively valid tool with which to estimate phytoestrogen intake among South Asian women in the UK.