After the fun we had with the last bit of science (78 kajillion subjects!), I couldn’t resist taking a peek at this one. (After all, a 16-year study does sound impressive. And very, very expensive.) So, I decided to have a bitch about it. From the abstract (note: all emphases are mine):
“We used data from 2,632 normotensive subjects, ages 30–64 years, in the Framingham Offspring Study.”
Hmm, studying HTN in people who don’t have HTN. Weird, but not a deal-breaker.
“Detailed dietary records were collected…”
Ah, so they didn’t actually measure sodium–specifically via 24-hr urine excretion, which is the gold-standard test for this sort of research. Nor did they even do ‘spot’ urine-sodium measurements, which is the red-headed stepchild of tests in this field. Rather, they just asked people to remember what they ate, and estimated their salt intake on the basis of these dietary self-reports.
OK, that’s not good. But here’s the real kicker:
“[Detailed dietary records were collected] over six days”
Yes, you read that right: This study didn’t collect data for 16 years–it collected data (and notoriously crappy and unreliable data at that) for all of 6 days. (The ‘16 years’ referred to the BP measurements they had on hand–which is nice, but not terribly useful as they had already selected pts who were normotensive to begin with.)
So, this study went on for 16 years the same way the last one had 78M subjects.
“To account for possible confounding of these effects by body size, we used linear regression models to derive nutrient residuals as exposure variables.”
OK, that’s a good thing. Would be misleading to treat 3g of salt intake by a 285# guy the same as 3g salt intake by an 85# woman. But…
“Confounding factors retained in the final models included age, sex, education, height, physical activity, cigarette smoking, and alcohol intake.”
Oh my, that’s a lot of confounding factors.
But setting aside the serious methodological concerns listed above, let’s get to the key finding, to wit:
“After 16 years of follow-up, those with the lowest SBP and DBP levels (129.5 and 75.6 mm Hg, respectively) were those with higher intakes of both sodium and potassium while those with the highest SBP and DBP levels (135.4 and 79.0 mm Hg, respectively) were those with lower intakes of both.”
The key thing to understand here is, this was not a randomized controlled clinical trial (an RCT), which requires random assignment of subjects to experimental conditions. Rather, in an observational study such as this one, subjects assigned themselves to (in this case) their sodium intake level, and this greatly affects how the findings should be interpreted.
Consider: Some individuals reported consuming significantly less sodium than others, yet had higher BPs. Can we think of an explanation for this relationship other than the inference the Terminator wishes to make, ie, that lower sodium causes higher BP? Clearly we can. One very obvious explanation is that the individuals who consumed less sodium elected to do so because they were concerned about their BP, which was at the higher end of the normal range. In contrast, the people whose BP was at the lower end of the normal range were understandably unconcerned about modifying their diet in this regard. So to the extent this speculation is true, the causality arrow runs in the direction opposite of the conclusion drawn by the Terminator–that is, rather than low sodium causing high BP, high BP ‘causes’ low sodium, ie, induces people to adopt a low-sodium diet.
It’s important to note that whether this specific explanation is true is unimportant. What is important is recognizing that, because of weaknesses inherent to its nature, the findings of an observational study such as this are always subject to multiple interpretations. This is why observational studies rarely play a foundational role in the development of clinical guidelines.
tl;dr This study has, like, 78M problems.
(The link:)
https://www.fasebj.org/doi/abs/10.1096/fasebj.31.1_supplement.446.6