Our study has several limitations. First, the cross-sectional design does not allow inference of a causal relationship between dietary patterns and IR. Thus, the association between these dietary patterns and IR remains to be confirmed
in prospective studies. Second, HOMA-IR can be used as a surrogate measure of IR at the population level, but it is
perhaps less useful on an individual basis. However, this study was based on a nationally representative large sample and we targeted a population without disease condition such as diabetes or hypertension so that our findings would be used to
develop a dietary strategy in preventing diabetes applicable to this population.