The answer was in our signatures all along

[dan_bionic]

Great survey Principles!
It reflects the results of studies without being influenced by implant industry or surgeons.
I know there are limitations, for example, if the signatures are not update after a revision, but in general the data shows a great success of penile implants for the treatment of ED.
Cudos to you an thanks to all members of this forum, you all are doing a greta work and help others to find the solution!
Dan

[staphylococcusecoli]

100% agree with Dan. This is an absolutely stellar find by Principles. Dude, you're a super smart man! This table on mechanical failures is more than just enlightening, it provides profound, tangible hope for every brother on this journey. The data is incredible: the fact that 50% of IPP implanted men are still using their original, functioning devices after 20 years is a powerful testament to their durability and reliability. The failures happen only to the 1%. The 2.5 years median time to failure for Titan is just a 1% problem. It also proves a critical point: The reason we don't hear from these 20 year veterans on forums like FrankTalk is simple. They don't have time. They are too busy having sex. Get implanted. Have the best sex of your life. Age is just a number once you're implanted.

[tomas1]

I think the data is encouraging.

I do wonder if some of the posters had an implant and experienced problems, then posted on Frank Talk.
That could add to the failure rate.

I also believe that many guys got an implant without issues and never bothered to post.
All that means the survival rate could be slightly better than the data suggests.

[LetoMan]

This is interesting, but I remain fairly dubious of what value there is in analyzing forum comments (not to be negative, I appreciate the work that went into this). Yes, you extracted a result from the data that matches other methods of study, but what if it had been different? Meaning, whether the methodology is valid should not be determined based on whether the result looks acceptable.

What I mostly think is interesting is that you probably expose at least one bias in how we look at the data obtained in posts. You only count a user once, regardless of how many times they post. But I suspect that there is a strong correlation between number of posts per user and number of revisions per user, as users with more revisions are likely to have simply been around longer, have reasons to come back and start posting again when they get revisions, and generally have something to talk about (I.e., the likelihood of a “I’m getting a revision” post is likely orders of magnitude higher than a “nothing to report, no need for revision” post, which doesn’t even make sense in the first place.) I would be curious to see what the correlation is there, as that would help explain why there is a perception of higher incidence of revision than there actually is.

In that sense, I think the value of the analysis you are engaging in is probably less valuable when it comes to determining actual health outcomes, but likely has value in helping to understand how reported outcomes in places like this forum can affect perceptions.

[Nico_from_belgium]

Great job dude!

Really helpful.

Concerning révision rates, you should just consider the possible death, someone who doesn’t want a revision (age)of the autor or the loss of account. An estimation of the actual age of members could be à good indicator.

[principles]

This is interesting, but I remain fairly dubious of what value there is in analyzing forum comments (not to be negative, I appreciate the work that went into this). Yes, you extracted a result from the data that matches other methods of study, but what if it had been different? Meaning, whether the methodology is valid should not be determined based on whether the result looks acceptable.

What I mostly think is interesting is that you probably expose at least one bias in how we look at the data obtained in posts. You only count a user once, regardless of how many times they post. But I suspect that there is a strong correlation between number of posts per user and number of revisions per user, as users with more revisions are likely to have simply been around longer, have reasons to come back and start posting again when they get revisions, and generally have something to talk about (I.e., the likelihood of a “I’m getting a revision” post is likely orders of magnitude higher than a “nothing to report, no need for revision” post, which doesn’t even make sense in the first place.) I would be curious to see what the correlation is there, as that would help explain why there is a perception of higher incidence of revision than there actually is.

In that sense, I think the value of the analysis you are engaging in is probably less valuable when it comes to determining actual health outcomes, but likely has value in helping to understand how reported outcomes in places like this forum can affect perceptions.

Those are fair concerns, and I do appreciate the pushback.

It’s not that I called the method "valid" merely because it matched legacy studies. It matched because I used the same Kaplan–Meier-with-right-censoring framework that every survivorship paper relies on. That estimator is industry standard for time-to-event datasets precisely because it makes no distributional assumptions. If this forum cohort had a genuinely different revision pattern from historical data, the KM curve would happily show that. Agreement with legacy studies is a consistency check, not the criterion for validity.

As for your second argument, I aggregated the post count for all the users that required zero documented revisions against those with ≥1 using their full forum historical post count:

-No-revision cohort (910 users): median 30 lifetime posts (IQR 10–86)
-Revision cohort (146 users): median 51 lifetime posts (IQR 14.5–142.5)

Both groups contain everything from one-off posters to people with thousands of posts. The Spearman correlation between total post count and revision count is ρ ≈ 0.10 (p ≈ 9*10⁻⁴). Statistically non-zero because the sample is big, but it’s a very weak relationship in practical terms.

Plotted on log-spaced bins, the post-count distributions for “revision” vs “no revision” basically lie on top of each other from the bottom end up to 1,000+ posts. So the intuitive story of “people with revisions post way more, so they dominate what we see” doesn’t match what the data actually show at the user level.

So yes, forum data have clear limitations for estimating population-level risk, and they definitely influence how risk feels. But the specific concern that high-post users with revisions are driving the survivorship curve doesn’t match what the user-level numbers actually show.

post_count_revision_hist.png

post_count_revision_share.png

[RigiconDownUnder]

Rigicon Infla is set to solve these revision problems. Their DIPP (Dynamic IPP) introduces a 4th layer and its entire purpose is to add a new level of durability. This is a fantastic advancement, a true game-changer and IMO designed to build a more resilient implant from the ground up.