The shape of the evidence

2.2.1Why shape rather than results

Readers are usually offered results: a percentage improvement, a satisfaction figure, a number of participants who reported something. Those figures are the least transferable part of any study, because they depend entirely on the design that produced them. A large improvement in an uncontrolled study of a treatment with a strong immediate physical effect and a two week follow up is close to uninformative, whatever the number is.

The shape of a literature, by contrast, tells you what questions it is capable of answering. That is why this reference describes evidence qualitatively and why it publishes no figures from studies it cannot verify in full. The methods used to judge the shape of a body of evidence are not secret, are published by Cochrane and the EQUATOR Network among others, and are written for people who are not researchers.

2.2.2The features that characterise this literature

Small

Cosmetic injectable studies are typically small. Small studies are not worthless, but they are unstable: they produce widely varying estimates, and the ones with striking results are more likely to be published and far more likely to be quoted at you in a clinic.

Often uncontrolled, or controlled within a person

Many studies treat everyone and measure change from baseline. Since this is a treatment whose immediate effect includes local swelling and hydration, and since participants and assessors know who was treated, before and after comparison in an uncontrolled study is a weak design for the question being asked. Split face designs, where one side is treated and the other is not, are better but carry their own difficulties, including the possibility of systemic effect and the fact that a person seeing two sides of their own face is not a blinded assessor.

Short

Follow up is generally measured in weeks to a few months. The claim that matters requires follow up past the point at which the material has certainly been cleared, and comparatively little work reaches that far.

Heterogeneous in outcome measures

Studies use different scales, different imaging, different patient reported instruments and different definitions of improvement. That makes pooling difficult, which is one reason systematic review in this area is harder than it looks, and it makes cross study comparison close to meaningless.

Frequently industry connected

Manufacturer funding, manufacturer supplied product, or authors with commercial relationships are common in this field, as they are across aesthetic medicine. This does not make a study wrong. It is a known and measurable source of bias in the direction of positive findings, and it belongs in the reader's assessment.

Geographically concentrated

A large share of the work originates in a small number of countries where these products were developed and adopted early. Concentration is not a fault, but a literature that has not been replicated across independent groups in different settings is a literature whose findings have not really been tested.

2.2.3What is missing

It is worth being specific about the study that would change this document. It would be randomised. It would have a control arm that is either a sham or an active comparator such as a hyaluronic acid skin booster. Assessment would be blinded, by assessors who did not know the allocation and who were not the participants. Outcomes would be defined before the study started and registered publicly on a trial registry. Follow up would extend well past the clearance window. It would be run by a group without a financial interest in the result, or with that interest fully declared and the analysis independent.

None of that is exotic. It is the ordinary standard for evaluating an intervention, and the tools to do it, including public trial registries such as ISRCTN and ClinicalTrials.gov, are freely available. The absence of such studies is a statement about the commercial incentives in cosmetic medicine rather than about the difficulty of the science.

2.2.4Systematic reviews and why they help less than you would hope

A systematic review is only as good as the studies it can find. Where the underlying literature is small, heterogeneous and largely uncontrolled, a review will say so, and its conclusion will be a call for better trials rather than an answer. That is a genuinely useful output, and it is also frequently misrepresented: a review concluding that a treatment shows promise and requires further study is quoted in marketing as a review showing that the treatment works.

If you want to look for reviews yourself, the Cochrane Library and PubMed are both public and free to search. The PROSPERO register lists systematic reviews that are underway, which is a useful way to see what questions people are currently trying to answer.

2.2.5The fair conclusion

It would be wrong to say that polynucleotide treatment does not work. Nothing in the shape of this literature establishes that. It would be equally wrong to say that it works, and that error is the one being made commercially, at scale, every day.

The defensible position is narrower and less satisfying. There is a plausible mechanism. There is a body of mostly small, mostly short, frequently industry connected work reporting positive findings. There is no independent, adequately powered, blinded, controlled evidence with long enough follow up to establish that the central claim of the category is true. A person can reasonably decide to have the treatment knowing that. What they should not do is decide to have it believing the question has been answered.