Quick Summary tl;dr
The authors know as well as anyone the limitations of this study design. In their words: “There are limitations to this study that merit consideration. This study represents observational data and is not a clinical trial; however, randomized trials of low carbohydrates on mortality practical because of the long duration of study required. Another limitation of this study is that diet was only assessed at two time intervals, spanning a 6-year period, and dietary patterns could change during 25 years…” Well said, in my opinion.
Another drawback pertaining to the animal versus plant sources of dietary protein and fat was the lack of differentiation of animal sources. That is, I strongly expect a whole-foods plant-based diet that derives more dietary protein and fat from salmon and oysters to be far healthier than one based on bacon-wrapped cheese dogs. This also likely has a geographical component in that seafood and shellfish are far more accessible in coastal regions, particularly those closer to the equator.
That said, this study was simply not designed to support the headlines: “High carbohydrate diets will kill you! Low carbohydrate diets will kill you! Wait, it’s not the carbohydrates that matter!" and the authors were well-aware of it.
A study published in The Lancet (Seidelmann et al., 2018) led to some confusing headlines such as: “High carbohydrate diets will kill you! Low carbohydrate diets will kill you! Wait, it’s not the carbohydrates that matter!”
This was based on a large and long-term epidemiological study. For many people, this is a red flag because epidemiology does not imply and cannot prove causation. Epidemiology, by design, shows us correlations which may or may not be spurious. In this case, is diet correlated with mortality risk in the long-term?
Let’s dig in.
What Does the Study Actually Show?
Questionnaire Driven Data
Dietary intake was assessed at two time points in the early ‘90’s, and mortality was assessed 25 years later. Many people criticize the food frequency questionnaire, although I think I can fairly accurately recall if I’ve had less than one diet soda this month vs. 6 or more today. This food frequency questionnaire was designed to assess such differences.
Lack of Re-Assessment
People with certain diseases were excluded from the study and if diseases were diagnosed during the study, diet was not-reassessed later in the study. This decision in study design has important implications for data interpretation.
For example, if a 60-year old participant on a low fat diet does a food frequency questionnaire, has a heart attack a year later, adopts a ketogenic diet, does another food frequency questionnaire, then dies of a heart attack a year later — do we attribute the first heart attack to high carb and the death to low carb? No.
Correlation is not causation, and in this context, it is safer to assume the first food frequency questionnaire is more reflective of lifetime carbohydrate exposure and the second food frequency questionnaire reflects something people do when something really bad happens.
In this case, the person had a life-threatening heart attack and responded by radically changing their diet. However, if diet does have a role in the etiology of heart disease (independent of body weight), then the damage incurred by the diet followed for the first 60 years far outweighs that incurred by the diet of the last year.
Were Low-Carb Diets Really Low in Carbs?
The participants were divided into 5 equally sized groups depending on their carbohydrate intake. The lowest carbohydrate group were getting about 37% of their calories from carbohydrates and the highest carbohydrate group were at about 61%.
Other Important Lifestyle Factors
All that said, in this study, everything that could go wrong, went wrong. The low(ish) carbohydrate diet group had the most smokers and people who had quit smoking; the high carb group had the most ‘never-smokers.’
The low(ish) carb group had the highest BMI, most diabetics, and it seems like they may have already been trying to eat healthier to deal with these problems. They reported eating fewer carbohydrates and calories than the higher carb groups. This type of study is not designed to differentiate these variables. Correlation is not causation. We don’t know if the lifestyle choices were in response to or responsible for the pathological conditions.
Another red flag: the correlation of low carb with mortality was reversed when diet included more low carb plant foods than animal foods.
Whole-foods plant-based ketogenic diets are totally possible and assuming you’re getting adequate protein, might be one of the more healthier adaptations of keto… relative to the bacon-wrapped cheese dog version :-)
That said, I’d like to re-emphasize
- this is still an epidemiological study; and
- correlation does not imply and can not prove causation.
Ignoring the animal/plant sources of dietary protein and fat, the researchers determined ~50% of calories coming from carbohydrates correlated with the lowest risk of all-cause mortality in this study.
In my opinion and experience, the animal/plant aspect of this study is more telling. Potentially spurious correlations aside, the specific foods you eat on a particular diet likely matter more the macronutrient ratio. For example, a whole-foods plant-based ketogenic diet leaves little room for processed foods, added sugars, omega-6 and industrial trans fats.
Step 1. Replace some of the starchy carbohydrates in your diet with nuts, avocados, maybe even some edamame.
Do you like this post? Share it with your friends!
Let us know what you think, rate this post!