Type 2 Diabetes–Prevention Diet and All-Cause and Cause-Specific Mortality

A Prospective Study

Chun-Rui Wang; Tian-Yang Hu; Fa-Bao Hao; Nan Chen; Yang Peng; Jing-Jing Wu; Peng-Fei Yang; Guo-Chao Zhong


Am J Epidemiol. 2022;191(3):472-486. 

In This Article


In this large prospective multicenter study with a mean follow-up of up to 13.6 years, we found that greater adherence to a type 2 diabetes–prevention diet, as indicated by higher dietary diabetes risk-reduction score, was associated with lower risks of death from all causes, cardiovascular disease, and cancer. Subgroup analyses further found that sex, smoking status, and alcohol consumption were effect modifiers of the observed associations between dietary diabetes risk-reduction score and risks of death from all causes, cardiovascular disease, and/or cancer.

Many previous studies in nutritional epidemiology focus on the roles of individual foods or nutrients in health outcomes. However, considering the potential antagonistic or synergistic effects among dietary components and the fact that individuals always consume a variety of foods simultaneously in their daily life, the health effects of a given dietary pattern may be different from the sum of its individual components.[31] Therefore, dietary pattern evaluation possibly could provide a better understanding for the roles of diets in health outcomes. In fact, the advantages of analyzing the dietary pattern in the field of public health are increasingly being recognized. For example, the 2015 Dietary Guideline Advisory Committee made its dietary recommendations based on dietary patterns rather than individual foods or nutrients.[33] A growing number of studies have shown favorable associations of healthy dietary patterns with mortality risk.[34] For example, Patel et al.[35] recently found that adherence to Dietary Approaches to Stop Hypertension, the Alternate Healthy Eating Index, or the Mediterranean diet was associated with a decreased risk of all-cause mortality. In this secondary analysis of the PLCO Cancer Screening Trial, we have assessed, to our knowledge for the first time, the role of the type 2 diabetes–prevention diet in the risk of mortality and we found that adherence to this dietary pattern was associated with reduced risks of all-cause and cause-specific mortality. Our findings are consistent with those from previous studies[34,35] and extend the favorable associations between healthy dietary pattern and mortality to a type 2 diabetes–prevention diet. Thus, our findings deepen our understanding of the role of dietary exposures in relation to the risk of type 2 diabetes in determining mortality risk. Meanwhile, our findings suggest that increasing intakes of cereal fiber, polyunsaturated fatty acids, coffee, and nuts while decreasing intakes of carbohydrates, trans-fatty acids, red and processed meat, and sugar-sweetened beverages may be helpful for improving longevity, which is particularly significant in that dietary behavior can be modifiable and unhealthy diet is a leading cause of mortality in the US population.[3] In addition, our findings highlight the importance of adhering to a healthy dietary pattern in improving health outcomes and provide some supporting evidence for the recommendation of adhering to a healthy eating pattern by the 2015–2020 US dietary guidelines.[36]

In this study, we observed inverse associations with all-cause or cause-specific mortality for the ratio of polyunsaturated to saturated fatty acids and intakes of cereal fiber, nuts, and coffee, a positive association for sugar-sweetened beverages, and a null association for glycemic index, which are consistent with the results of previous studies.[10–14,16,37] However, our study revealed a null association between red and processed meat consumption and cardiovascular mortality, which is inconsistent with a recent prospective cohort study showing a significant positive association (for tertile 3 vs. tertile 1, HR = 1.33, 95% CI: 1.19, 1.49).[38] The inconsistency may be due to the difference in study population; that previous study was conducted among UK adults aged 40–69 years.[38] It is also possible that the positive association of red and processed meat consumption with cardiovascular mortality observed in the previous study[37] was due to incomplete adjustment for known confounders,[15] such as physical activity. In addition, our study observed a null association of trans-fatty acid intake with all-cause mortality, which is consistent with the results from a prospective study in a British working population (per 1-standard-deviation increase, HR = 1.07, 95% CI: 0.98, 1.18)[39] but is inconsistent with those from several studies showing a positive association.[11,40,41] The exact reasons for the above phenomenon are unclear and may be attributable to the differences in study population, methodology, and/or the extent of adjustment for potential confounders. Hence, more studies are needed to investigate the associations of intakes of red and processed meat and trans-fatty acid with all-cause and cause-specific mortality.

Interestingly, our study observed that the inverse association with cardiovascular mortality was more pronounced in women than in men, while the inverse association with cancer mortality was more pronounced in men than in women, indicating that sex is an outcome-specific effect modifier in our study setting. The exact reasons for this observation are unclear; it may be related to hormonal differences between the sexes. As almost all women in this study were postmenopausal, estrogen-level difference between men and women is not expected to be a major driver for this observation. Instead, testosterone-level difference between the sexes may be a key inducer. Indeed, observational studies have found that testosterone replacement therapy is associated with an increased risk of cardiovascular events[42] but a decreased risk of aggressive prostate cancer;[43] thus, the relatively high testosterone level in men may attenuate the inverse association of the dietary diabetes risk-reduction score with cardiovascular mortality but strengthen the inverse association with cancer mortality. In addition, our subgroup analyses found that the inverse association of dietary diabetes risk-reduction score with cancer mortality was more pronounced in current or past smokers or participants with heavy alcohol consumption, suggesting that diabetes-prevention diet may have interactions with smoking and alcohol drinking in biological pathways. In fact, a prospective cohort study also showed that the inverse association of adherence to Dietary Approaches to Stop Hypertension diet with the risk of all-cause mortality was more pronounced in smokers than in nonsmokers.[44] Of note, we cannot rule out a possibility that the above-mentioned interactions are chance findings, although they are biologically possible. Therefore, our findings from subgroup analyses warrant further investigation.

Although the specific mechanisms underlying the inverse associations of the type 2 diabetes–prevention diet with risks of all-cause and cause-specific mortality remain to be explored, intuitively, this dietary pattern possibly exerts its mortality benefits through its individual components. Human and experimental studies have suggested that polyunsaturated fatty acids are capable of improving insulin resistance.[45,46] Moreover, coffee has been found to inhibit inflammatory responses, possibly by reducing the expression of inflammation-related genes[47] and the release of inflammatory mediators.[48] In addition, nut consumption has been found to be associated with attenuated oxidative stress,[49] which may be through the modulation of nuclear factor-kB and nuclear factor erythroid 2–related factor 2/heme oxygenase-1 pathways.[50] Collectively, these facts suggest that the inverse association of the type 2 diabetes–prevention diet with mortality may be explained by improved insulin resistance and decreased levels of inflammation and oxidative stress. Nevertheless, it is also possible that mortality benefits of adhering to this diet are mediated, at least partly, by potential interactions among individual components of the diet.

Our study has several limitations. First, food consumption information used for the construction of dietary diabetes risk-reduction score was evaluated once at baseline in our study. As dietary habits can change over time, food consumption evaluation at 1 time point may result in nondifferential bias. Nonetheless, it has been suggested that the approaches using baseline diet data only in general yield a weaker association than do these using the cumulative averages.[51] In addition, in this study, nutrient intake was assessed with the DHQ, a self-administered food frequency questionnaire. However, this questionnaire did not contain the essential information that was required to accurately calculate intakes of some nutrients. For example, some trans fats are artificial and added into processed food products. Thus, the content of trans fats in a food product may depend on the brand of the product. However, the DHQ did not contain this information. Hence, nutrient assessment by the DHQ might be subject to measurement errors. Second, death certificates were employed to obtain the underlying cause of mortality in our study. Of note, the cause of mortality from death certificates may be misclassified in some circumstances.[52] Hence, our findings on the association of dietary diabetes risk-reduction score with cause-specific mortality might be susceptible to misclassification bias. Moreover, the validity of mortality assessment in the PLCO Cancer Screening Trial has not been confirmed, raising some concerns on the accuracy of outcome ascertainment. Third, in our study, all participants were US adults between the ages of 55 and 74 years; moreover, 90.9% of participants were non-Hispanic White, 36.6% were college graduates, and 51.0% were current or past smokers. Therefore, our findings may not be generalizable to other populations. Fourth, as shown in Table 4, not all dietary components were associated with all-cause or cause-specific mortality. However, when constructing diabetes risk-reduction score, we assumed that each component contributes equally to the score. Thus, the score used in our study may not precisely reflect the actual role of each dietary component in the real world. Finally, as with any observational study, our results might be influenced by residual confounding due to unmeasured or unrecognized confounders, although a wide range of potential confounders was controlled for. In addition, it should be acknowledged that, based on our findings, the causal association of adhering to a diabetes-prevention diet with mortality risk cannot be established, given the observational design of our study.

In conclusion, the dietary diabetes risk-reduction score is inversely associated with the risks of death from all causes, cardiovascular disease, and cancer in this US population. These findings suggest that adherence to a type 2 diabetes–prevention diet may serve as an attractive strategy for improving longevity. Future studies should clarify the relevant biological mechanisms and validate our findings in other populations.