Four Days of Data Said the Beetroot Was Making Me Worse. A Month Said Nothing Happened.

This post contains affiliate links to the beetroot shots and the blood pressure band I used. I bought both with my own money. If you buy through one I may earn a small commission at no extra cost to you, and it makes no difference at all to the numbers below, which are what they are.

Nothing here is medical advice. I am not a doctor. I am one man with a wrist band and a spreadsheet, and my blood pressure has been medicated since 2005. If yours is high, talk to your GP.

At twenty to four this morning my blood pressure was 120 over 82, with a heart rate of 62.

I know that because the band took the reading without asking me, while I was asleep, which is the only reason any of this is possible. That reading is one of 710 the band took in August, and August is the month that finishes off the beetroot experiment.

It does not finish the way the early data suggested it would.

What I said I would do, before I started

The rules were written down and published before the first shot, which is the whole point of doing it that way. One Beet It Sport Nitrate 400 a day. A pre-specified signal threshold of 3 mmHg systolic, decided in advance so I could not move the goalposts afterwards. Then a washout period once I stopped, to see whether anything came back.

I ran it for 33 days rather than the registered 30, which I disclosed at the time. The last shot was on 27 July. I have had no beetroot since. Not a shot, not a juice, not one of those vacuum packed things from the salad aisle.

The result on treatment was nothing. 133 over 85 before, 133 over 85 during, against a threshold of 3. A clean null.

The four days that looked like something

Here is where it got interesting, and here is where I nearly got it wrong.

When I pulled the updated July report there were four clean days on the end of it, the 28th to the 31st, with no beetroot in me. Those four days averaged 124.0 systolic across 117 readings. The on treatment period, the 1st to the 27th of July, averaged 128.5 across 831 readings.

Read that quickly and it says something quite dramatic. My blood pressure was three to four points lower off the beetroot than on it. Not a rebound. The opposite of a rebound. If you wanted a headline, there was one sitting right there: the supplement I was testing appeared to be making me slightly worse.

I wrote at the time that it should be treated as noise and that the washout would tell us. This is the washout telling us.

A full month says nothing happened

August was a complete month with no beetroot in it at all. 710 readings. Here is the whole thing next to everything that came before it.

Period	Systolic	Readings
June, baseline, daytime	133	about 400
1 to 27 July, on beetroot	128.5	831
28 to 31 July, four days off	124.0	117
August, full month off	130	710

The 124 is gone. It did not survive contact with a bigger sample. Given a full month rather than four days, the off treatment figure landed at 130, which is one and a half points above the on treatment figure and comfortably inside the 3 mmHg threshold I set before I started.

So there is no rebound, there is no delayed benefit, and there was never a four point drop. There were four days.

Back to precisely where I started

The awake figure for August is 133 over 89, from 483 readings.

The June baseline, before any of this began, was 133 over 85.

The systolic is identical. Not close. The same number. Two months of shots, a month of washout, 1,600 odd readings, and the top number has come back to exactly where it was in June.

The diastolic is four points higher, at 89 against 85, and I am deliberately not making anything of that. Different month, different weather, different weeks. When one number moves four points and the other moves zero, that is usually what noise looks like rather than a finding. If it is still there in October I will say so.

The mistake I nearly made, which is the actually useful bit

This is the part I would want somebody else to take away from it, because the beetroot answer itself is boring.

One hundred and seventeen readings felt like a lot. It is more measurements than most people take in a year with a cuff. It came from a validated device and it was a real, computed average, not a guess. And it was wrong by about six points against what a full month showed.

If I had published that four day figure as a finding in late July, it would have been a good post. It would have had a hook, a surprise and a number. It would also have been rubbish, and somebody might have stopped taking something on the strength of it.

So a new house rule, and it is going in the editorial policy. No period shorter than two weeks gets a number attached to it on this blog. Short windows can be described. They do not get to be evidence.

The gaps, because there are some

Five days in August have no data at all: the 4th, the 5th, and then the 19th, 20th and 21st.

That second gap matters more than the first, because it sits in the middle of the week we spent on Anglesey. That week involved a good deal more walking than usual, two long coastal days, and rather more beer than a normal Tuesday. So the period most likely to have shifted the numbers is partly missing from the numbers. I would rather tell you that than quietly average around it.

The band was calibrated against a cuff twice across the window, once on 27 July and once on 26 August, which brackets the month at both ends. That is the best I can say for its accuracy here, and it is not the same as saying the readings are right.

One thing in the report I cannot explain

The August summary gives a standard deviation of 3 for awake systolic. It also gives a maximum of 153 and a minimum of 111.

A standard deviation of 3 does not describe a set of raw readings spanning 42 points. Something else is being calculated there, possibly the spread of daily averages rather than of individual measurements. I do not know, the report does not say, and I am not going to invent an explanation to make the post tidier.

What I have done is use the means, which are straightforward arithmetic on a large number of readings, and ignore the spread entirely. If anyone from Hilo reads this and wants to tell me what that figure represents, I will update the post and say so.

So, beetroot

Two ways of finding nothing. Nothing while I was taking it, nothing when I stopped.

That is not the same as saying the research is wrong, and I want to be careful here. The published meta analyses do show real average reductions from dietary nitrate. An average is a statement about a population. It was never a promise about a fifty five year old on lercanidipine and ramipril whose pressure is already partly controlled by two drugs that work on entirely different mechanisms.

The most likely explanation for my null is the least dramatic one. There was not much room left to move.

I would not tell anybody not to try it. It is cheap, it is food, and the downside is limited to your wallet and the colour of things. What I would say is that if you try it and you do not measure properly, you will have no idea whether it did anything, and you will probably decide it did.

What I am doing next

The obvious question hanging over every number in this post is whether the band is telling the truth in the first place.

I have worn a Hilo band every day for two years, and for most of that time I have also been calibrating it against an Omron X7 upper arm cuff. That is two years of paired readings between a cuffless wrist device and a proper cuff, taken by the same person under the same conditions, and as far as I can tell nobody has published anything like it.

That is the next one, and it is a far more useful post than this one. It is the question everybody asks before they spend the money, and I am sitting on the data to answer it.

---

Pete Harrison is 55 and lives in Buckley, Flintshire. He has had medicated high blood pressure since 2005 and spends more of the night than he would like reading about it. Things I Learnt at 3am is where he writes honestly about what he has actually bought, tested and measured, including the times the answer is no.