Do High Carbohydrate Diets Increase Risk of Death? This Study Explains It

This study looks at the relationship between macronutrient intake and specific health outcomes.

Who and what was studied?

Two weeks ago, a group of researchers published papers on the relationship between macronutrient intake and specific health outcomes.

The study that caused the most controversy investigated associations between macronutrient intake (carbohydrates, protein, and fat) and two outcomes: cardiovascular disease and mortality.

The cohort for this study was 135,335 people between the ages of 35 and 70 living in 18 low-, middle-and high-income countries across five continents.

The cohort was geared more toward low-income and middle-income countries, which is important because those areas are often understudied. The great diversity of its affiliate made this study one of the first of its kind.

Using standardized questionnaires, the authors collected demographic data, such as lifestyle data, socioeconomic status, and health and medication history.

Using a food frequency questionnaire, they also collected dietary data once at the start of the study.


The authors analyzed two primary outcomes (total mortality and major cardiovascular events) and four secondary outcomes (heart attacks, strokes, cardiovascular disease mortality, and non-cardiovascular disease mortality).

They collected these data by tracking participants’ medical diagnoses, abnormal biomarkers, hospitalizations, and autopsies. The median follow-up time of the participants was 7.4 years.

The PURE study is an epidemiological cohort study. It followed 135,335 people between the ages of 35 and 70 living in 18 low-, middle- and high-income countries for about seven years.

The cohort was weighted for low- and middle-income countries. The study aimed to investigate the associations between the consumption of macronutrients and two outcomes: cardiovascular disease and mortality.

What were the results?

The authors organized macronutrient intake into quintiles. The primary sources of carbohydrates were white bread, white rice, cakes, fruits, fruit juices, and sugary drinks – that is, most of the nutrient-poor refined foods.

After adjusting for ten confounders, the researchers found a statistically significant positive association between carbohydrate intake and total mortality.

In general, the higher the carbohydrate intake, the higher the rates of all-cause mortality and mortality from non-cardiovascular diseases at any time.

Those trends were found in Asian and non-Asian regions but were only significant in the non-Asian areas.

The ten confounding factors such as:

  • Energy intake.
  • Physical activity.
  • Waist-hip ratio.
  • Smoking
  • Diabetes.
  • Age.
  • Sex.
  • Education.
  • Socioeconomic level.
  • Urban or rural location.

However, no positive associations were observed between carbohydrate intake and four outcomes:

  • Cerebral infarcts
  • Heart attacks.
  • Major cardiovascular diseases.
  • Mortality from cardiovascular diseases.

Total fat intake was also divided into quintiles, and the researchers found a statistically significant inverse association between fat intake and total mortality. In other words, the higher the fat intake, the lower the risk of death at any time.

This trend held for non-cardiovascular disease mortality and stroke but not for heart attacks, major cardiovascular diseases, and cardiovascular disease mortality.

In Asian and non-Asian regions, higher consumption of monounsaturated fatty acids was inversely associated with total mortality. In comparison, higher consumption of polyunsaturated fatty acids was inversely related to total mortality only in Asian regions.

Those fat intake associations were all significant.

The authors also evaluated the effect of replacing carbohydrates with certain kinds of fats.

Replacing carbohydrates with polyunsaturated fats was associated with a lower risk of total mortality and mortality from non-cardiovascular diseases while replacing carbohydrates with saturated fats was associated with a lower risk of stroke.

After adjusting for several covariates, the authors found a clear association between carbohydrate intake and two outcomes:

1. Total mortality and mortality from non-cardiovascular diseases.

But no associations were found between carbohydrate intake and four other outcomes:

2. Heart attacks, significant cardiovascular disease, and mortality from cardiovascular diseases).

Total fat intake was inversely associated with total mortality, non-cardiovascular disease mortality, and stroke, but not with heart attacks, major cardiovascular diseases, and cardiovascular disease mortality.

What does this study tell us?

These results seem pretty scary: increased risk of death with higher carbohydrate intake? That doesn’t sound very good. But it is essential to consider some of the study’s limitations before concluding.

After adjusting for many confounding factors in their models, the researchers found a clear positive association between higher carbohydrate intake and, at any time, both the risk of total mortality and the risk of non-cardiovascular disease.

Confounders are always to keep in mind, even when making corrections, because multiple unmeasured or unknown variables can affect the results.

The best way to deal with confounders is through randomization, which allows inference of a cause and effect relationship. This cannot be done in observational studies such as the PURE study, but adjustments can account for confounders.

However, this method has its drawbacks: it cannot adjust for unknown confounders; over-tightening can sometimes lead to confounding effects; And the more tests you run, the more likely you are to get a false positive.

Therefore, some of the results may be statistical parasites after adjustment.

However, no positive association was found between higher carbohydrate intake and stroke, heart attack, major cardiovascular disease, or mortality from cardiovascular disease.

This is important because the results are much more specific than the absolute risk of mortality or the threat of non-cardiovascular disease. Therefore, we must consider two possibilities:

  • First, there may not be a dose-response relationship, so there is no causal relationship between carbohydrate intake and specific outcomes. The dose-response relationship is usually a hallmark of causation.
  • Second, it may not have been a dose-response relationship. Still, the small sample size made it difficult to detect, as fewer people experienced those outcomes than people who died (total mortality and mortality from non-cardiovascular diseases).

However, there were clear associations between carbohydrate intake and total mortality and risk of non-cardiovascular disease.

Is there any merit to these associations? Perhaps. Or maybe these results have another cause. Many of the countries included in the analysis consumed most of their carbohydrates from highly refined, nutrient-poor foods.

For example, most people consume most of their calories from white rice in Bangladesh. The poor nutritional status could be contributing to mortality along with a lack of access to specific resources.

Although the authors used four different pieces of data (education, household wealth, household income, and country income level) to fit their models for socioeconomic status (SES),

But what about fat? Does this study suggest that fat is protective?

Unlike carbohydrates, the authors did not compare all fats. They found that replacing a small number of carbohydrates with polyunsaturated fats reduced the risk of total mortality and non-cardiovascular disease mortality.

Typical sources of these fats include fish, nuts, and vegetable oils. These foods are much more nutrient-dense than refined carbohydrate sources. They are also much less available in poor areas.

Which begs the question: is a higher intake of polyunsaturated fats protective? Or is it associated with better nutritional status and unmeasured markers of SES?

The analyzes conducted by the authors of this study also make it a bit difficult to draw firm conclusions at the individual or national level: Data from 18 countries were combined to produce these results, and such a large sample size is a sharp double sword.

On the one hand, it facilitates the detection of natural effects, and the varied population allows generalizations.

On the other hand, by combining data from people who live very different lifestyles, have access to additional resources, and vary in genetic characteristics, the researchers made it more difficult to determine whether those general results apply to specific subpopulations.

In other words, any conclusions drawn from your analysis will hold better for the general world population than for a specific population or individual.

Now, one must remember that this is all speculation based on the limitations of this study and epidemiological research in general.

Many other potential factors could be contributing to the trends we see in the data. Still, even so, the data shows an association between high carbohydrate intake and total mortality and non-cardiovascular mortality.

And it’s pretty impressive that the trends remain after adjusting for so many covariates.

Therefore, even though the effect sizes of this study are not of the same magnitude as those of the epidemiological studies that help infer causal relationships, they are still noteworthy.

Study results are limited by the study design itself. Results may be affected by unaccounted-for confounders, such as unmeasured markers of socioeconomic status.

In other words, poor nutritional status and lack of access to medical resources may explain the high mortality rates.

However, the results of this study should not be directly discounted: despite many adjustments, the associations between carbohydrate intake and mortality remained, making them remarkable.

They may not be as helpful for personalized medicine as for global politics.

The panorama

Despite the methodological limitations of this study, its results are worth pursuing. Many possible factors could be examined.

The poor nutritional status could be one of the reasons that high carbohydrate intake was associated with mortality. Since the PURE study collected blood samples for a separate article, it might be worth examining them for micronutrient deficiencies.

As mentioned above, conducting a randomized controlled trial (RCT) of this scale or length would be difficult. However, a well-designed and powered short-term RCT with surrogate markers could corroborate the results of this epidemiological study.

The authors stated that their study results do not support current World Health Organization dietary guidelines, limiting total fat intake to 30% and saturated fat intake to 10%.

Are the results of this particular study strong enough to reform dietary guidelines as suggested by the authors? Probably not alone.

However, they add to a plethora of studies published in recent decades suggesting that saturated fats may not be as harmful as initially thought and that refined carbohydrates are unhealthy for several reasons.

So between the PURE study and other emerging evidence, lifting some fat and putting one on top of carbs may be worth considering when updating dietary guidelines.

The results of this study are worth following. They could be corroborated by randomized controlled trials or examining different factors, such as micronutrient deficiencies.

They may not be strong enough to reform prevailing dietary guidelines. Still, combined with other evidence, they may help future policies raise your fat limits and lower your carbohydrate limits.