These studies are all incredibly weak, because they rely on surveys and pluck variables out for correlations. So besides the fact that people almost never remember or accurately report what they eat over many years, it's a dubious correlation that can't be relied on.