Science of Nutrition or lack thereof

I keep seeing stories about salt intake on the news. It appears to be traced back to an article, "Projected Effect of Dietary Salt Reductions on Future Cardiovascular Disease" published in the New England Journal of Medicine this year. Since reading Good Calories, Bad Calories, I've increasingly come to believe that the science of nutrition is mostly junk. A lot of what we believe to be true isn't actually supported by data. That doesn't necessarily mean it's false, we just don't know. A simple example is fat or more specifically saturated fat. The fat causes heart disease hypothesis was still being debated in the 60s. Seemingly a consensus was formed, then lots of money and time was spent proving the consensus correct, then the data came back inconclusive. By the time the data is in though, the consensus is already formed and no one seems to care about the data. It's easy to explain away even large and comprehensive studies, because the effects are very hard to measure and control.

I found an article about salt by the same author as above here. It talks about the last 30 years of the salt debate and how problematic the findings have been. It's hard to imagine the study above has fixed the problems, but I don't have access to the full text so I don't know. The study says that a reduction in salt would save hundreds of thousands of lives. Here is my guess at the quick and dirty problems with that:

The study is likely doing one of two things:

1. Assuming a relationship from other studies, ie. reducing salt lowers blood pressure X amount and extrapolating from that. Thus, I know salt leads to blood pressure and blood pressure leads to deaths, so I take what I know about blood pressure in the population reduce it and my model tells me how many lives I've saved. The problem is that I'm assuming salt leads to blood pressure, which is a claim that is hard to prove especially in controlled studies.

2. More likely though, I think they are using their database to estimate the effect of salt on either blood pressure and mortality. I just don't think they could get away with the previous option. The problem with this is adequately controlling for the people with heavy salt intakes. Suppose one group of people eat McDonald's every day. Well, McDonald's is bad for you and it's high in salt, but it's not clear the salt is what's bad for you. It could be fat, trans-fat, carbs, calories, lack of nutrients, who knows. It's not hard for me to believe that processed foods are bad for people, but it is hard to imagine a large panel data correcting for the salt in processed foods or something else.

Even if salt causes high blood pressure, which causes death, it's not clear lowering salt as a public policy would actually save any lives. Suppose processed foods and fast foods have to become less salty, the makers probably aren't adding salt just for the hell of it, the foods will probably become less tasty. So what will the manufacturers do? Will they add more fat or more sugar? It's possible the foods actually become less healthy. Increasingly I'm becoming convinced that this is what happened during the "fat is bad" 70s until now. Although, I'm not sure I'm right; it's at least defensible that the push to encourage Americans to eat less fat was actually counter-productive.

I think it's actually very useful to infer what will be in studies before reading them. I will probably never follow nutrition closely enough to have a fully informed opinion. It's quite possible that this study actually has new and informative information that actually helps settle the long debate. If that is in fact the case, I need to make sure to take future big news story studies more seriously and trust the ability of the establishment to find the truth a bit more over time. If the study adds nothing new, and just repeats old mistakes, it's further evidence that nutrition is junk science. What I don't want to do, is read such studies and use whatever it says to confirm my biases. All studies have problems, so it's easy to slightly alter your hypothesis or to undermine the study, so that it seems to work in your favor.

The reason all this hits close to home is that Economics suffers from many of the same problems Nutrition does. We make inferences about complex systems with lots of theory and little data. We have lots of trouble controlling for certain effects, and have very few useful controlled experiments. All of this is worst in my field, macro. When I hold macro to the same standards I hold nutrition, it's very hard to articulate what I "know" about macro.

Update: I still haven't found the full-text of the article, but I found some slides by the first author. There are two studies cited on the slides. One is just another meta-study attacking two other meta-studies. The second, though, is more intriguing, and the full text is available. It's a somewhat controlled experiment of 412 people on both a healthy and normal diet. The study had people on both diets consume high, medium and low diets and the study shows some reduction in blood pressure. What's weird is people on healthy diets didn't get much benefit than people on the control diet. Why should that be? The study doesn't say how salt was added to the diet either, which could be important. But if the second study is robust and replicable, I'm down.

Another interesting thing from the slides is that several countries are already launching into anti-salt campaigns. The U.K., for instance, has been successful in decreasing salt over 10 years, but it didn't say, if people were dying of heart disease and stroke less than before.