The Bristol average that nobody in the group actually had

One reading out of twenty-nine.

That is how often the group I am about to describe logged a Bristol type 4, the smooth, unremarkable stool every gut-health article tells you to aim for. Their average across the whole programme was 3.8. Almost perfect, on paper. And almost none of them were ever there.

Where the 3.8 came from

The data is from a fifteen-day fibre programme I ran with thirty participants. Each morning they logged whether they had a bowel movement and, if they did, its Bristol type. I have written about the dropout and the timing of discomfort before. This is the part I skipped past the first time, because the summary table made it look like there was nothing there.

The summary table said this group, the seven participants recorded as Taeyangin, started the programme at 3.8 and finished at 3.7. No change. I wrote “expectations need managing” next to it and moved on.

Here is what the twenty-nine individual readings actually were:

Type 1–2  (hard)       12 readings   41%
Type 3–5  (middle)      9 readings   31%
    of which type 4     1 reading
Type 6–7  (loose)       8 readings   28%

Twelve hard, eight loose, and very little in between. Average them and you land close to 4. A group split between two ends of the scale produces the same mean as a group sitting comfortably in the middle. The number cannot tell those apart.

It gets narrower when you look at people instead of readings. Only three of the seven logged often enough to have a personal average worth quoting. Those came out at 2.0, 2.7 and 5.6. Two people running hard, one running loose. Not one of them was near 3.8.

The morning with nothing to log

There is a second thing the average hides, and it is bigger.

A Bristol reading only exists on a morning when there was a bowel movement. For this group, 28 of 60 morning check-ins, close to half, recorded none. Those mornings do not drag the average down or up. They are simply absent from it. So the 3.8 was calculated from the good-enough days, and the days with nothing were left out of the arithmetic.

For comparison, the other two groups large enough to look at:

                   no movement    readings in 3–5    exactly type 4
Taeyangin  (7)         47%              31%                3%
Taeeumin   (11)        21%              56%               30%
Soyangin   (7)         27%              88%               63%

The Soyangin group is the opposite picture. Most of their readings were type 4, and individual people barely moved from day to day. The Taeeumin group sits between, but with the widest spread between people: personal averages ran from 1.8 to 5.3 inside the same eleven.

Three groups, three different questions

If you only read the averages, the conclusion is that everyone is roughly fine and roughly the same. Read the spread and each group asks something different.

For the Soyangin readings the question is not really about stool form at all. It was already sitting at 4. Whatever these participants needed from a gut programme, a better Bristol number was not it.

For the Taeeumin group the question is which Taeeumin. Two people in the same category can sit at opposite ends of the scale, and advice pitched at the group average would be wrong for both of them.

For the Taeyangin group the question comes before stool form: whether there was a movement at all. A day with no movement is information, and the standard way of reporting Bristol data throws it away.

How much weight this can bear

Not much, and I want to be specific about why.

The numbers per group are small. Seven, eleven, seven. The Taeyangin pattern rests on three people with enough readings to profile, and 29 readings in total. The single Soeumin participant is left out, as in everything else I have written from this data.

The constitutions were entered by participants at signup rather than assessed as part of the programme. Most of the people who entered one were Korean medicine doctors whose own constitution had already been determined, so these are not quiz results. But I did not re-examine anyone, and a small group of seven does not need many misassigned people to look different.

One thing the numbers did not do is make the Taeyangin group look implausible. English sources tend to describe it as vanishingly rare. That is not what I see, and seven of thirty here is not out of line with it.

What I would still stand behind is the method point, which does not depend on anyone’s type being correct: an average Bristol score near 4 is not evidence that anyone is near 4, and a Bristol average that ignores the mornings with no movement is measuring the easy days.

The order I ask in

In clinic I do not start with the shape. First: do you go every day, without skipping. Then, and only then, is it hard or is it loose.

Which is more or less the argument above, arrived at from the other direction.

The other thing I keep coming back to is food. Stool changes with diet, obviously, so the constitutional pattern probably is not the same everywhere. It will depend on what a culture eats. In Korea the staples are white rice and napa cabbage kimchi. In the Sasang diet tables both of those sit on the list for the Taeyangin. So a Taeeumin who has eaten that way every day for decades is eating somebody else’s food, habitually, for a very long time, and that seems to be where the bowel tendency shows up most clearly. Whether a Taeeumin raised on a different diet would log the same spread, I do not know. I would like to.

And frequency. I am told the Donguibogam says twice a day is about right for someone who eats two meals a day. I have not checked the passage myself. The modern side of it I do take seriously: the longer stool sits in the bowel, the more time gut bacteria have to turn bile acids into the secondary forms that are being studied as a colorectal cancer risk. Nobody has pinned that to a number of hours. But a morning with nothing to log is not a neutral morning, which is the point I keep ending up at.

A better way to keep your own log

If you are tracking your own gut with the Bristol scale, three changes make the record worth more than an average.

Log the mornings with nothing. Write “none” rather than skipping the entry. Over two weeks the count of empty mornings may tell you more than the types on the other days.

Look at your spread before your average. Write down your hardest and loosest reading of the week next to the mean. If they are types 2 and 6, your 4 is not a 4.

And compare yourself with yourself. A population target of type 4 is a reasonable default, but the useful question is whether your own pattern is narrowing or widening over time, and that is a question only your own log can answer.