A recent study on mortality and healthy eating, focused on low-carb and low-fat diets, has been making headlines. The study was a large and long one, with 45% of participants dying during its duration. The headline claims implied that low-carb diets, particularly unhealthy ones, were associated with higher mortality rates, while healthy low-carb diets were linked to lower mortality. On the other hand, all low-fat diets, both healthy and unhealthy, were associated with lower mortality rates.
The study, however, had many flaws. The first flaw of population studies such as these is the use of Food Frequency Questionnaires. The study used a single questionnaire dating back to the mid-1990s, while the participants were followed up for an average of 23.5 years. It is highly likely that diets changed during the follow-up years.
The study's definition of low-carb was an issue, as it often us. Low-carb diets were defined as up to 42% of calorie intake, which is still high carb in the low-carb world. Additionally, the judgment for what constituted healthy vs. unhealthy was arbitrary and subjective, causing discrepancies in the interpretation of data.
Large differences between people in different intake groups were also noticed. Exercise, smoking, and alcohol intake could affect mortality rates, and while the study adjusted for these factors, we can't adjust for an entirely different person - as participants were.
Another issue with the study is the lack of explanation for its findings. The paper failed to explain convincingly why low-carb diets (except plant-based ones) were associated with higher mortality rates. The paper only suggested associations, not causation, yet it inferred causation in several passages in the discussion section.
Given the coverage that this paper received, many people might think that low-fat diets are good (even unhealthy ones) and low-carb diets are bad (unless plant-based and quite high carb). Several issues with the study make its claims invalid, and people should not put much stock in such findings.