Does diet map with mortality? Ecological association of dietary patterns with chronic disease mortality and its spatial dependence in Switzerland

We investigated the associations between dietary patterns and chronic disease mortality in Switzerland using an ecological design and explored their spatial dependence, i.e. the tendency of near locations to present more similar and distant locations to present more different values than randomly expected. Data of the National Nutrition Survey menuCH (n 2057) were used to compute hypothesis- (Alternate Healthy Eating Index (AHEI)) and data-driven dietary patterns. District-level standardised mortality ratios (SMR) were calculated using the Swiss Federal Statistical Office mortality data and linked to dietary data geographically. Quasipoisson regression models were fitted to investigate the associations between dietary patterns and chronic disease mortality; Moran's I statistics were used to explore spatial dependence. Compared with the first, the fifth AHEI quintile (highest diet quality) was associated with district-level SMR of 0 center dot 95 (95 % CI 0 center dot 93, 0 center dot 97) for CVD, 0 center dot 91 (95 % CI 0 center dot 88, 0 center dot 95) for ischaemic heart disease (IHD), 0 center dot 97 (95 % CI 0 center dot 95, 0 center dot 99) for stroke, 0 center dot 99 (95 % CI 0 center dot 98, 1 center dot 00) for all-cancer, 0 center dot 98 (95 % CI 0 center dot 96, 0 center dot 99) for colorectal cancer and 0 center dot 93 (95 % CI 0 center dot 89, 0 center dot 96) for diabetes. The Swiss traditional and Western-like patterns were associated with significantly higher district-level SMR for CVD, IHD, stroke and diabetes (ranging from 1 center dot 02 to 1 center dot 08) compared with the Prudent pattern. Significant global and local spatial dependence was identified, with similar results across hypothesis- and data-driven dietary patterns. Our study suggests that dietary patterns partly contribute to the explanation of geographic disparities in chronic disease mortality in Switzerland. Further analyses including spatial components in regression models would allow identifying regions where nutritional interventions are particularly needed.