Reading a study

What A 'Responder' Is: Placebo Response In Chronic Constipation Trials

“Responder” is the word that turns a personal experience into a number, and placebo arms in constipation trials produce a lot of them. This article explains how responders are defined, how large placebo response is in chronic constipation and IBS-C trials, why a “lighter feeling” cannot separate a product from everything else that helped, and what a fair test of a blend would require.

Sourced from the product label and PubMed-indexed research · Last updated September 27, 2026 · 13 min read · Reading a study

What a responder is

Trials in chronic constipation cannot report that a treatment “worked”. They have to turn a messy, personal, day-to-day experience into a yes or no for each participant, so that the yeses can be counted in each arm and compared. The yes has a name, responder, and the rule that makes someone one is written down before the trial starts.

The commonest rules in chronic idiopathic constipation are built on complete spontaneous bowel movements, the CSBM: a bowel movement that comes without a laxative or enema shortly before it and leaves the person feeling that the bowel has emptied. A 2019 meta-analysis in the American Journal of Gastroenterology looked at two of them. One counts a participant as a responder if they average three or more CSBMs a week. The other counts a responder as anyone whose average rises by at least one CSBM a week over their own baseline.

Those two rules do not describe the same event. The first is an absolute bar, and someone who moves from zero to two a week is a non-responder despite a real change. The second is a relative one, and someone who moves from two to three is a responder despite still being below the absolute bar. Neither is wrong. They ask different questions, and the answer to each is different, as the next section shows.

None of this is a criticism of trial designers. A threshold is the price of counting. But it means every “response rate” you read is a statement about a definition as much as about a treatment, and comparing rates from different definitions is like comparing distances measured in miles and kilometres without checking the unit.

How big placebo arms actually are

The part that surprises most people is that participants who receive an inert capsule are counted as responders in substantial numbers. In that 2019 meta-analysis of 23 placebo-controlled trials, placebo response “ranged from 4% to 44%”. Pooled, it was 13 per cent when the endpoint was three or more CSBMs a week, and 28 per cent when the endpoint was a rise of at least one CSBM a week over baseline.

The authors also asked what pushes the number up. A higher baseline CSBM count, older age and a larger proportion of men in a trial were each associated with a stronger placebo response. Several things people might expect to matter did not reach significance: the number of study visits, the length of the study, the year of publication, the number of dropouts and the chance of getting the active drug.

A second analysis, by Chen and colleagues in Clinical and Translational Gastroenterology, went wider: 46 studies with 5,992 constipated patients on placebo. Pooled placebo response was 28.75 per cent (95 per cent confidence interval 23.83 to 33.67), with very high heterogeneity between trials. The important split was by endpoint. Where success was defined as subjective improvement, the placebo response was 41.40 per cent. Where it was defined as improvement in complete spontaneous bowel movements, it was 18.31 per cent, and where it was a composite improvement, 20.35 per cent.

Chen’s group drew a conclusion worth quoting closely: they do not recommend subjective improvement as an endpoint when designing therapeutic trials in constipated patients. Read that as a statement from people who run and analyse these trials about which measure is the leakiest.

Every figure is the response of people who were given an inactive capsule.

The same pill, three different results

Put the numbers above in a concrete picture. Take 100 people constipated enough to enter a trial, and give every one of them the same inert capsule. If success is defined by how they say they feel, about 41 of them count as responders. If it is defined by bowel movements, about 18. If it is a composite, about 20. Nobody received anything but placebo, and the answer to “how many improved” more than doubled depending on the question.

Irritable bowel syndrome with constipation, a neighbouring condition with more elaborate endpoints, shows the same thing more sharply. Barberio and colleagues pooled 17 trials of licensed drugs against placebo in IBS with constipation, with 4,603 patients on placebo. Using the US Food and Drug Administration’s recommended endpoints and counting a response only if it held for at least 6 of 12 weeks, placebo response was 18.9 per cent for the composite, 34.6 per cent for abdominal pain and 30.1 per cent for stool. Raising the bar to at least 9 of 12 weeks, the figures fell to 4.3, 24.5 and 7.7 per cent. Same trials, same patients, and the stool response dropped from 30 per cent to under 8 per cent by asking for consistency.

That analysis assumed dropouts were treatment failures, which is the conservative way to count, and its authors’ message was that future trials should use the recommended stringent endpoints because they produce lower placebo responses. A separate meta-analysis of 73 randomised trials in IBS found placebo response of 27.3 per cent for global improvement, 34.4 per cent for abdominal pain and 17.9 per cent for the composite FDA responder definition, and identified design features linked to higher placebo response: a run-in period of two weeks or less, dosing three or more times a day, a smaller control group, and earlier publication. Its recommendation was a run-in of at least two weeks and dosing once or twice a day.

The practical reading for anyone weighing a supplement is simple. Whenever you see a percentage of people who “improved”, the first question is not how large it is. It is which definition produced it, and what the same definition gives in people who took nothing active.

Regression to the mean and the week you started

Placebo arms contain more than the effect of belief. They also contain everything that would have happened anyway, and in a condition that fluctuates, quite a lot does.

Constipation waxes and wanes. Travel, a week of poor sleep, a change in what you are eating, a stressful month, a course of a medicine that binds the bowel: all move the numbers. And people do not decide to try something at a random moment. They decide when it is bad. That timing matters because of a statistical effect described clearly by Barnett, van der Pols and Dobson in the International Journal of Epidemiology. Regression to the mean, they explain, is a phenomenon that can make natural variation in repeated data look like real change: unusually extreme measurements tend to be followed by measurements closer to the average. It becomes more noticeable as measurement error grows, and when follow-up is examined only in a subgroup picked because of an extreme baseline value.

Apply that to a person. The week they buy something is, by selection, a bad week. The next few weeks are, on average, better than the one that drove them to act, whatever they took. If they started a supplement in the bad week, the improvement gets attached to the supplement.

That is not a failing of the person. It is a property of how variable symptoms behave, and it is precisely why a fair test compares an arm that received the product against an arm that received an inactive look-alike, both selected in the same way, rather than comparing a person’s bad week with their later average.

Even an honest placebo helps some symptoms

One more piece of the picture is uncomfortable for people who assume a placebo only works if you are fooled. Lembo and colleagues ran a six-week, three-arm trial in 262 adults with irritable bowel syndrome: an open-label placebo, in which participants knew the pill was inert, a double-blind placebo, and a no-pill control. Open-label placebo produced significantly greater improvement in IBS symptom severity than no pill, and did not differ significantly from double-blind placebo. The authors concluded that blinding may not be necessary for placebos to be effective.

That was irritable bowel syndrome and not constipation alone, so it should not be stretched too far. But it makes a useful point about the ritual. Taking one small thing at a set time each day, paying more attention to your gut than usual, and expecting to feel different are all real interventions, and they change how symptoms are experienced and reported. Someone who feels better on a daily capsule is not imagining it. What nobody can tell from inside the experience is how much of the improvement belongs to the contents.

What this means for the seller’s supports language

The seller’s wording for SodaMelt is structure-and-function language: it supports healthy digestion, it supports everyday metabolic wellness, and it offers a lighter feeling. The printed carton carries none of those phrases, and nothing the seller says states a responder rate, and none of these phrases names an endpoint that could be counted in a trial.

Look at each against what the meta-analyses say. “A lighter feeling” is a subjective report by construction. It is the category the Chen analysis found most vulnerable to placebo, the one that returned 41 per cent in people taking nothing active. “Supports healthy digestion” could be measured, but only once someone says how: bowel movements per week, a stool-form score, straining, bloating. “Supports everyday metabolic wellness” does not name a measurable outcome at all, which is part of why the psyllium and weight literature discussed elsewhere on this website is a null finding rather than a promise.

Nor is there a shortcut through the ingredients. A search of PubMed-indexed records for the product’s name found no published trial of the formula. The trials that exist are for single ingredients, at doses the blend cannot reach, as set out in Psyllium Husk And The Ten-Gram Question, or for different plant parts, as in The Oat Study This Label’s Claims Actually Rest On. And because the panel prints eleven entries inside one blend with no separate amounts, described in What A Proprietary Blend Actually Hides, nobody can build the whole-product result up from the parts.

An absence of a trial is not evidence that a product does not work. It is the absence of a controlled answer, which leaves the question open. Open is a legitimate place for a question to be, provided nobody treats an unmeasured impression as if it had been measured.

What a fair test of a blend would need

If a group did want to know whether this particular capsule does something beyond placebo for people with constipation, the lessons of the meta-analyses translate into a fairly short specification.

1. A look-alike control. An inactive capsule that matches in size, colour and taste, so participants cannot tell which arm they are in.
2. Random allocation. So the arms start with the same mix of severity, age, sex and expectation. Baseline CSBM count, age and sex all predicted placebo response in the 2019 analysis, which makes imbalance between arms a real hazard.
3. A responder definition fixed in advance. Something countable, such as a set number of CSBMs a week, with abdominal symptoms reported separately, and chosen before anyone sees the results.
4. A stringent, sustained threshold. The IBS-C data show that requiring the response to hold across most of the weeks cut the stool response from about 30 to under 8 per cent in placebo arms. A response that has to persist is harder to produce by chance.
5. A baseline diary of at least two weeks. The 73-trial IBS analysis linked short run-ins to higher placebo response.
6. A control group big enough. The same analysis found smaller control groups were associated with higher pooled placebo response.
7. Counting dropouts as failures. The conservative rule used in the IBS-C analysis, so that people who leave because it did not work are not quietly dropped.
8. The whole product, as sold. Not a single ingredient, and not a different dose, since the result must apply to the capsule a person can actually buy.
9. Both numbers. The share of responders and the average change, so a small real effect is not lost inside a threshold.

One of the design recommendations sits comfortably with this product, and it is worth saying so. The 73-trial analysis found dosing once or twice a day was associated with lower placebo response than dosing three times a day or more, and the directions on this label are one capsule a day. A one-a-day product is, if anything, easy to test properly.

If you try it anyway

Nothing here says a person should not. It says the tools a trial uses can be borrowed, in a small way, by an individual who wants a clearer answer than a general impression.

• Write down a baseline first. For two weeks before the first capsule, note each day whether you had a spontaneous bowel movement, how much you strained and how it felt afterward. Numbers stop the bad week from being remembered as the average one.
• Change one thing at a time. If you begin the capsule the same week you change your diet, your water intake or your medicines, you have lost the ability to say which one moved anything.
• Judge with the same log. Compare the count and the strain against your own baseline range, not against how you remember feeling.
• Expect fluctuation. One good week is inside the normal range for a condition that varies, and only a sustained difference means anything.
• Take the log to a clinician if the problem persists. A record of what actually happened is far more useful to them than a description of how you felt.

Placebo response is not a slur on anyone who reports feeling better. It is a reminder that feeling better is real, common and cheap to produce, so a claim built on it has to be tested in a way that separates the product from everything else that helped.

References