BPC-157 Before And After: What Was Measured, and In What

A BPC 157 before and after is a study design before it is a photograph: one subject compared with an earlier version of themselves, with no control group, no blinding and no pre-registered endpoint. That design has a name in methodology, the uncontrolled pre post comparison, and it has known failure modes that were documented long before anyone applied it to peptides.

This article treats the format as a design and asks what it can carry. The answer is narrow, and it is narrow for reasons that have nothing to do with whether the compound works.

The uncontrolled pre post comparison, described honestly

The design compares a measurement taken at one time with a measurement taken later, in the same subject, after an intervention. Its appeal is obvious: it is cheap, it requires no recruitment, and the subject is perfectly matched to themselves on every variable that never changes.

The problem is everything that does change. Between the two measurements sits the passage of time, the natural course of the condition, whatever else the subject did, and the subject's own expectation. An uncontrolled pre post comparison cannot separate the intervention from any of those, because it contains no group in which the intervention was absent and everything else was the same. That is not a technicality. It is the reason controlled designs exist.

Related reading on this compound: [whether cycling applies here](/research/guides/bpc-157-cycle), [the benefit claims graded by organism](/research/guides/bpc-157-benefits), [what BPC-157 actually is](/research/guides/what-is-bpc-157).

Where the comparison goes wrong

Several distinct mechanisms can produce an apparent improvement in a human self report with no contribution from the intervention, and they are additive rather than alternatives.

Natural history is the largest. Injuries and gut complaints in people often follow a course that ends in improvement regardless of treatment. Anyone who starts an intervention will usually start it when symptoms are at their worst, which is precisely the point from which the average trajectory is upward.

That last effect has a name, regression to the mean, and it is a property of measurement rather than of biology. A value selected because it is extreme tends to be less extreme when measured again.

Co-intervention follows. People who begin a peptide rarely change only that. Those same people also rest, modify load, start rehabilitation, sleep differently or reduce the activity that caused the injury. Any of those can produce the observed change, and the design cannot apportion credit.

Expectancy and observer effects come next. When the person assessing the outcome is the person who chose the intervention and paid for it, the assessment is unblinded in the strongest possible sense.

Then recall. A before state reconstructed from memory after the after state is known is not a baseline. It is a description written by someone who already knows how the story ended.

Then the material itself is unverified. Compounds sold as research chemicals are not assayed by the purchaser, so in most of these reports the substance in the comparison is uncharacterised. Identity, purity and quantity are three separate measurements, and a supplier label asserts rather than demonstrates all three.

Finally, and operating on the whole collection rather than on any single report, there is selection. What reaches a forum or a comments section is not a sample of what happened. People who improved write up their experience; people whose injury behaved as it always would tend to stop reading about the compound and never post at all. The visible record is therefore filtered before anyone reads it, and no amount of careful reading corrects a filter applied upstream. This is why a page of positive accounts and a page of negative accounts would both be uninformative: neither was assembled by a process that could have produced the other.

What was measured, and in which organism

The table sets out how each kind of comparison is actually conducted, which is where the difference between a rat experiment and an online post becomes concrete.

One outcome deserves separate attention, because it is the one people actually care about and the one a self comparison handles worst. The claim is rarely that recovery happened. It is that recovery happened faster than it otherwise would have. That is a claim about a counterfactual, a version of events in which the compound was not used, and nobody observes their own counterfactual. It has to be supplied from somewhere, and in practice it is supplied by expectation: what the person believed the injury would take, formed from a clinician's estimate, a previous injury that was not the same injury, or something read online. Comparing an observed recovery against a remembered expectation is not a measurement, because the expectation was never a measurement either. This is precisely the gap a control group fills, and it is the reason the rat experiments include one.

The rat rows share something the human self report row lacks entirely: a comparison group and an instrument. The bottom row is the one that would answer the question people are asking, and it is currently empty, because no registered human trial of this compound has published efficacy results.

What a BPC 157 before and after would need to be readable

The requirements are not exotic. A documented baseline recorded before the intervention started, using a defined measure rather than a recollection. An outcome measure chosen in advance, so that whichever thing happens to change cannot be selected afterwards as the point. An assessor who does not know what was administered. A comparison, whether a control group or, at minimum, a documented account of what else changed during the period. And an assay confirming the identity and purity of the material used.

An individual account meeting even three of those is far more informative than the usual version, and nothing about the list requires a laboratory. It requires deciding what counts as success before finding out what happened.

It is worth being clear about what such an account would then be worth. It would still be one uncontrolled observation, sitting at the bottom of any evidence hierarchy, unable on its own to establish that the compound did anything. What it would gain is the ability to be wrong in a visible way. A prespecified measure can fail to move. A blinded assessor can report no change. An assayed vial can turn out to contain something other than the label claims. A report that cannot come out negative is not weak evidence, it is not evidence, and the difference between the two versions of a personal account is exactly whether a negative outcome was possible.

Where the human record stands

Human trials of this compound exist and are registered. NCT07803250 is a Phase 1 study in recovery after rotator cuff repair, listed with 30 participants and not yet recruiting. NCT02637284 is a Phase 1 safety and pharmacokinetics study listed with 42 participants, with status unknown.

Those filings are why the claim that this compound has never been tested in people is wrong. The absence of any published efficacy result from them is why the claim that human benefit has been shown is also wrong. Registered, completed and published are three separate states, and there is no marketing authorisation from FDA, EMA or MHRA that would indicate a regulator has reviewed outcome data and reached a conclusion.

The published efficacy record remains rat and cell culture work covering tendon, ligament, muscle and gut outcomes, much of it from a small number of research groups. Those experiments do contain before and after comparisons, conducted with control groups and instruments, in rats. That is what the format looks like when it is done properly, and it is why the rat data carries weight the online version does not.