Hacker News Reader

Comment by btilly | Hacker News Reader

You missed the biggest point. Not only was the difference in salt not much, but they would have to have reduced adverse events (eg repeat heart attacks) by over 20% to have had decent odds of detecting it with their actual sample size.

So the real conclusion is that a modest reduction of salt from the recommended low levels likely does not drop the odds of a repeat heart attack by more than 20%.

This is a very, very long ways from the headline that lowering sodium doesn't help heart patients.

And yet, this crap passes as science???

View on HN

It's n = 800 which is way better than most studies, and it's a well designed study. Basic science research is way underfunded in essentially every country.

This is literally how the null hypothesis / statistical significance works; how is this crap science? If the measured effect is too tiny, you cannot conclude that it's a real effect.

I agree that it is better than most studies, and that is part of the problem.

The title as presented here is Lowering sodium intake doesn’t help heart patients.

The impression that people will walk away from that title with is, "If you have heart problems, there is no need to eat a low sodium diet."

This is not a hypothetical interpretation. I first saw this study at https://twitter.com/EricTopol/status/1510292643292491779, and that is exactly the impression that a well-known cardiologist drew. Many others came to the same conclusion.

Now what should we expect?

Their intervention reduced sodium consumption by 415 mg of sodium per day. (According to the least accurate method of measurement - self-reporting. The actual reduction was probably smaller.) If all else was equal, the correlation found at https://www.cdc.gov/salt/research_reviews/sodium_potassium_b... we should expect that to reduce systolic blood pressure by 1.9 mmHg. However not all else is equal, these are all patients receiving medical care. If your blood pressure is too high, you receive more medication. So all else is not equal, and we should expect blood pressure to be approximately equal between the groups, and medication not.

Therefore, to a first approximation, we should expect small differences in blood pressure, larger differences in required medication, and little difference in adverse events.

They designed a study that did not report on average blood pressure between the groups, or what levels of medication were required, and would only notice a large difference in adverse events. And so they failed to find a result that they should not have expected to see, failed to look for the result that they should have expected, and it wound up being reported in a way that undermines public health messaging around sodium reduction.

You may consider this quality scientific research. I don't.

> The title as presented here is Lowering sodium intake doesn’t help heart patients.

Yes, though this is the title posted to HackerNews, so I blame the poster. In the actual link, it's just : Reduction of dietary sodium to less than 100 mmol in heart failure (SODIUM-HF): an international, open-label, randomised, controlled trial

It’s not about repeat heart attacks. The study was in heart failure patients. The main issue is preventing events like exacerbation of heart failure and sudden death.

The study looked for a large effect size. That’s totally fine. I think it is good that non-pharmacological interventions are being studied. Calling it crap is quite unfair.

There was a 27% average decrease in sodium intake. That is a big difference and this is a valid study because these were the recommendations put forth by several organizations like the Cleveland Clinic

https://my.clevelandclinic.org/health/diseases/17072-heart-f....

Having patients suffer through this diet for little changes in outcome is important to know.

> And yet, this crap passes as science???

Have you ever looked at the papers that support the hypothesis that lowering salt consumptions helps heart patients?

I bet you that title is going to cause some people to not get on the dash diet ... when they really need to.

Uhh need to based on what evidence? This isn’t the only study suggests minimal if any improvement with lowered salt intake and the often quoted reasoning for why salt would matter is flawed.

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5179550/. “Graudal et al. (40) observed that sodium restriction (sodium reduction of 42–341 mmol/24 h in studies lasting 4–90 days) was independently associated with increased heart rate in healthy individuals and individuals with hypertension.”

You're quoting one small section of one referenced study in meta-study and leaving out the conclusion of this section that more serious studies are needed to determine the relationship between sodium intake and heart rate (not morbidity due to heart disease, just heart rate).

> Given the discrepancy in results, more studies in this area are required to investigate the relationship between dietary sodium intake and heart rate.

You're cherry picking one small part from a section not really related to the topic at hand and also ignoring the conclusion the study authors draw.

The sentence literally before the one you quote is

> Studies that demonstrated the effects of dietary sodium on heart rate have shown conflicting results.

And then you cherrypick the one example that agrees with your point while ignoring the examples that refute your point.

If you want to say the science is still unsettled, sure. But to pretend that it is settled in the way you want and quote one sentence out of context in a meta-study who's entire point is that the debate is not settled? Come on.

> If you want to say the science is still unsettled, sure.

I agree studies give conflicting results and there is isn’t a clear winner here.

I agree more studies are perfectly reasonable.

However, the default when faced with conflicting information isn’t to put forth effort into changing behavior. That requires justification, which isn’t available.

Consider if millions where micro-dosing Tylenol to avoid dementia and the results where inconclusive, what messaging would you provide? (Stop your wasting money), (do it it’s probably worth it), (might help, might not, we don’t know do what you want). The temptation is to say ‘do it’, the costs are low and the benefits could be significant. However, when there are known useful things people could be doing it’s probably best to stick with the third option.

PS: Ok you could jump on the latest study and constantly flip flop on recommendations, but that just confuses people and erodes trust in the process.

Look man, my comment was saying you can't just cherrypick a single sentence out of context from a long study (especially when the context of that section contradicts your own point). If you want to argue about sodium intake there's other comments in here for that.

The statement you picked made it out that there is no correlation between sodium intake and heart rate. The study you took that statement from was making the point that we can't draw such conclusions either way.

I didn't say anything about whether or not people should flip flop recommendations. But don't irresponsibly cite studies. It defeats your entire argument when you do so.

> The statement you picked made it out that there is no correlation between sodium intake and heart rate.

“was independently associated with increased heart rate in healthy individuals and individuals with hypertension.”

Cherry picking would have been linking to the study that suggested lower salt intake was unhealthy directly. Instead I linked to the meta study to give better context.

It’s great science! Now, insofar as their English and grammar, that leaves much to be desired! That said, I’m more than willing to bet that two things are in play: A. These publications would slough off an application from someone focused on writing faster than a skin from a boiled tomato. B. Most people who “Love science” don’t place value in other areas of learning. STEM doesn’t include “writing eloquently”, to put that mildly.

The science itself seems fine though! Testing, measuring, experimenting. Yes. But either there ought to be an onus on us as readers or them as publishers to make clear what has been done. Otherwise, as I believe you can surmise, there’s no benefit to doing at all!

Good follow up! separating the actual work done vs write up, the title alone gets people to read it/open it. Roger Penrose said he had released a bunch of papers over the more recent years and the only one that got a bunch of feedback was one where there was a small note about extraterrestrial life.

There is another phenomena as well. The odds of a piece of research going viral are directly tied to how surprising it is. Water is wet does not go viral. But it does go viral that trying to exercise self-control tires your brain and makes an impulse buy of a candy more likely.

Therefore the science we encounter goes through a selection bias that picks out the most surprising results. But far too often, "surprising" translates into "unlikely to be true". And so we get a constant stream of intellectual junk food wrapped up in the language of science.

Take this result. It is surprising that low sodium would not help with cardiovascular problems. Why is it surprising? It is because we know two facts:

1. High sodium intake increases blood pressure: https://www.cdc.gov/salt/research_reviews/sodium_potassium_b...

2. High blood pressure is bad for your cardiovascular system: https://www.cdc.gov/bloodpressure/about.htm

And therefore we EXPECT that high sodium intake is bad for your cardiovascular system. Which is why we constantly tell people to reduce salt.

Now we see this research that has a variety of limitations in what it could detect, saying that it found no correlations. We're surprised because we expect a correlation. But now people will walk around saying that science says that salt doesn't matter, ignore salt, and they're probably causing a problem for themselves.

And this brings us to another problem with the research. The known mechanism by which salt is bad for you involves blood pressure. However if you've had a heart attack, your blood pressure is going to be monitored and medications adjusted until your blood pressure is under control. The benefit of less salt will be that you need less medication and will have less side effects. It won't be lower blood pressure and a reduced risk of heart attack! So the primary outcome that should be measured isn't adverse events, it is how much medicine was required!

I had a close relative in congestive heart failure. My understanding of the reason for reducing salt was to avoid the water retention that impairs heart function.

It was sad to see him struggle with food that was less and less appetizing. Salt tastes great :(

I know what that is like.

For some reason, salt stops my blood pressure medication from working. (Several possible diagnoses were explored without success.) This was discovered after I checked in to the ER feeling terrible with a blood pressure around 190/120 and then had to stay for a week. So I'm now on a low salt diet and my blood pressure is now fine.

I don't cheat. And after a while my taste buds kind of reset. What tastes normal to most tastes insanely salty to me, and I get to enjoy the taste of low salt food.

It does suck that I can't simply eat out though.

If they are right that this study only proves that a small reduction in salt intake around the recommended daily level doesn't have a dramatic effect (>20%) on heart attacks, then is this really a useful study? Is it truly worth the money and time spent on collecting all this data, reviewing it etc? Will people keep referring back to it and citing it 5 years from now?

Reply on news.ycombinator.com