In an informative op-ed piece in the New York Times, Michael Alderman, an epidemiologist, questions a government campaign to reduce salt in processed food. His piece raises two (wildly different) questions.
1. Several studies have correlated less salt with worse health. Why? Alderman writes:
Nine [observational] studies, looking at a total of more than 100,000 participants who consume as much sodium as New Yorkers do, have had mixed results. In four of them, reduced dietary salt was associated with an increased incidence of death and disability from heart attacks and strokes. In one that focused on obese people, more salt was associated with increased cardiovascular mortality. And in the remaining four, no association between salt and health was seen.
And in the one experimental study that Alderman knows of, “the group that adhered to a lower sodium diet actually suffered significantly more cardiovascular deaths and hospitalizations than did the one assigned to the higher sodium diet.”
Those are useful facts. Alderman gives a few possible explanations. Here’s another one: Several popular fermented foods, including sauerkraut, buttermilk, miso, and cheese, are high in salt, and fermented foods protect against heart disease. I haven’t read the experimental study Alderman describes but it is unlikely that the two groups in that study ate food that was the same in every way except for salt content. What probably happened is that one group was instructed to choose a low-salt diet and the other group wasn’t. The low-salt group ate less salt in part by avoiding high-salt fermented foods (such as cheese).
2. Alderman writes:
[Observational] research can justify action only when multiple studies produce consistent, robust findings across a wide range of circumstances, as the research on tobacco and lung and cardiovascular health has done.
The puzzle is why he writes like this, which I find irritating. Most of the editorial is good, which makes this lapse especially interesting. I call this black-and-white speak, talking as if something complex was black and white and — always associated with this — people on one side are better than people on the other side. In my professional life, I hear black-and-white speak from some statisticians, who divide analyses into “correct” and “incorrect.” According to them you should analyze your data by following a set of black-and-white rules. Here is a less-irritating version of Alderman’s statement:
Successful public health campaigns have been built on observational studies but in the best-known case — the danger of smoking — the findings were consistent and robust across a wide range of circumstances.
See: no need to moralize. Alderman’s statement, of course, is just one example of something very common.