Study design and participants
Role of the funding source
Population characteristics in the Atherosclerosis Risk in Communities study, by quantile
The median length of follow-up was 25 years, during which there were 6283 deaths. The highest risk of mortality was observed in participants with the lowest carbohydrate consumption, in both unadjusted and adjusted models (p<0·001; figure 1; appendix p 8). However, the relationship between carbohydrate consumption and risk of mortality was significantly non-linear (p<0·001), resulting in a U-shaped association, with the lowest observed risk associated with carbohydrate consumption of 50–55% (figure 1). There were corresponding significant differences in mean residual lifespan based on carbohydrate intake (appendix p 2). For example, we estimated that a 50-year-old participant with intake of less than 30% of energy from carbohydrate would have a projected life expectancy of 29·1 years, compared with 33·1 years for a participant who consumed 50–55% of energy from carbohydrate (difference 4·0 years [95% CI 2·6, 5·3]). Similarly, we estimated that a 50-year-old participant with high carbohydrate intake (>65% of energy from carbohydrate) would have a projected life expectancy of 32·0 years, compared with 33·1 years for a participant who consumed 50–55% of energy from carbohydrate (difference 1·1 years [0·1, 2·0]). We did a sensitivity analysis using 50–60% energy from carbohydrate as the comparison group, with similar findings (data not shown). The association of overall carbohydrate intake with cardiovascular and non-cardiovascular mortality is shown in the appendix (pp 3, 4). There were similar results when we used dietary information from Visit 1 and Visit 3 in the sensitivity analysis (appendix pp 5, 6).
We updated a meta-analysis12 published in 2012, by identifying two additional studies that had since been published and that met inclusion criteria, using previously defined methods;13, 24 we also added results from ARIC because they met previously defined inclusion criteria12 (table 2). Including data from the ARIC cohort, there were 432 179 participants in eight cohort studies investigating carbohydrate intake, with 40 181 (9·3%) deaths reported. Because there was significantly lower consumption of carbohydrate in European and North American regions compared with Asian countries, low-income countries, and multinational cohorts (p<0·001), studies fell into two categories in the meta-analysis: North American and European studies (mean carbohydrate intake approximately 50%) that compared low carbohydrate diets with primarily moderate carbohydrate consumption as the reference (figure 2A), and Asian and multinational studies (mean carbohydrate intake approximately 61%) that compared high carbohydrate consumption with moderate carbohydrate consumption as the reference (figure 2B; table 2). The association between carbohydrate consumption and mortality was dependent on the range of carbohydrate intake. Figure 2 illustrates the significantly increased risk of all-cause mortality among participants with low carbohydrate versus moderate carbohydrate consumption (pooled HR 1·20, 95% CI 1·09–1·32; p<0·0001). This relationship remained significant if the ARIC study was excluded from the analysis (1·31, 1·07–1·58; p=0·007). High carbohydrate consumption was associated with a significantly higher risk of all-cause mortality compared with moderate carbohydrate consumption (1·23, 1·11–1·36; p<0·0001; figure 2).
Table 2. Meta-analysis study characteristics
The ARIC and PURE studies were the only two cohorts for which data were published or available about the continuous percentage of energy from carbohydrate. Figure 3 shows the overlapping and continuous relationship between percentage of energy from carbohydrate intake and mortality in these cohorts. By comparison with ARIC, the PURE study13 assessed participants primarily at the high end of the overall range of percentage of energy from carbohydrate consumption (Figure 2, Figure 3). However, the associations between primarily high carbohydrate intake and mortality in the PURE study still fell within the confidence intervals of those observed in ARIC (figure 3).
To explore the association between mortality and the source of fat and protein alternatives to carbohydrate intake, we compared studies that assessed animal-based and plant-based scores, which represented increasing substitution of animal-based or plant-based fat and protein for carbohydrate intake (table 2). Baseline characteristics of the ARIC study population, according to animal-based or plant-based low carbohydrate diet scores, are shown in the appendix (pp 9–11). The plant-based low carbohydrate dietary score was associated with higher average intake of vegetables but lower fruit intake (appendix p 11). By contrast, the animal-based low carbohydrate dietary score was associated with lower average intake of both fruit and vegetables (appendix pp 9, 10). Both low carbohydrate diets were associated with higher fat intake in exchange for carbohydrate, although the plant-based low carbohydrate diet had higher average polyunsaturated fat and lower saturated fat intake compared with the animal-based low carbohydrate diet (appendix pp 9–11). Overall, total protein intake was higher in the animal-based diet (appendix p 9). We determined the five foods that differed most significantly between the highest and lowest quantiles of animal-based and plant-based low carbohydrate dietary score. The animal-based low carbohydrate diet had more servings per day than did higher carbohydrate diets of beef, pork, and lamb as the main dish; beef, pork, and lamb as a side dish; chicken with the skin on; chicken with the skin off; and cheese (appendix p 10). The plant-based low carbohydrate diet had more servings per day of nuts, peanut butter, dark or grain breads, chocolate, and white bread than did higher carbohydrate diets (appendix p 11). Both low carbohydrate diets were lower in average regular soft drink intake (appendix pp 10, 11). In the ARIC cohort and in meta-analysis, increased consumption of animal-based protein and fat instead of carbohydrate was associated with a significant increase in all-cause mortality (p<0·0001; table 3). Alternatively, increased consumption of plant-based protein and fat instead of carbohydrate was associated with a significant decrease in all-cause mortality (p<0·0001; table 3). The animal and plant-based findings were consistent for cardiovascular and non-cardiovascular mortality (appendix pp 3, 4). Sensitivity analysis of plant-specific and animal-specific findings, using dietary information from Visit 1 and Visit 3, yielded similar results (appendix pp 6, 7). Similarly, in the meta-analysis, mortality increased when animal-derived fat and protein were substituted for carbohydrate, and decreased when these substitutions were plant-based (table 3). In the post-hoc sensitivity analysis, we assessed all meta-analyses using a fixed-effects model, with similar findings. Additionally, to minimise the likelihood of reverse causation, we did a sensitivity analysis whereby individuals with cardio-vascular disease, diabetes, or cancer at baseline were excluded from the analyses. These post-hoc analyses also yielded similar results.
Association between diets that substitute carbohydrates for animal-based or plant-based protein and fat with mortality in multiple cohort studies