How Data is Collected and Interpreted in Clinical Trials

Note: This post comes from one of our favorite readers (he has requested to remain anonymous), who has contributed posts in the past about FDA issues. He provides some insight into clinical trials, and sent this in to answer some of the questions/concerns posted in our comments section about Propecia data.

Finasteride side effects are a hot issue on various forums and on our own site, so hopefully this can clear up some of the hysteria that perpetuates online:

Post by Guest Writer

    A recent post impressed me by referring to Propecia data from a FDA hearing (more updated and accurate information is in the U.S. Product Label) but may have been confused by how data is collected and interpreted in clinical trials. I am a veteran of overseeing clinical trials resulting in drug approvals.

    1. The poster notes that the “rate of sexual side effects is 3.8 percent with finasteride versus 2.1 percent with placebo.” Actually, following the Advisory Committee hearing and in further review of the data, the US Product Label shows that in Year 1, decreased libido was 1.8% for the 945 men receiving finasteride and 1.3% for the 934 men receiving placebo (placebo-adjusted rate of 0.5% or 1 in 200 cases). Similarly, erectile dysfunction was reported in 1.3% of men receiving finasteride and 0.7% of those receiving placebo (placebo-adjusted rate of 0.5% or 1 in 200 cases). Approximately 1.2% of men receiving finasteride and 0.9% of men receiving placebo discontinue the study drug due to sexual side effects. No cases of irreversibility of sexual dysfunction were reported.

    2. The poster asks (given that the reporting was “self-reporting”) whether the results “are scientific”. This latter sentence makes me wonder if the poster wonders if clinical trial data is collected similar to the uncontrolled self-selected questionnaire surveys often cited in other situations. It is not. Unlike self-selected questionnaires, where one does not know the denominator (how many patients are receiving the drug, only those reporting a complaint) or period of observation, a clinical trial is different. In a clinical trial, the physician investigator asks every participating patient at regularly scheduled visits through the entire study (and predetermined) follow-up whether they are having or have had any new or worsened effect in any organ system. By definition, this reporting has to come from the patient. But, it is as scientific as one can get (it is solicited rather than self-selected). In addition, any report of a side effect is attributed to the drug (active vs. placebo) regardless of the causality. That is, if a person gets hit by a car and sustains trauma and is on an active drug, trauma is tabulated under the drug (although an investigator may causally relate it to something else). This conservative approach is taken to capture any possible relationship (i.e., maybe the drug makes someone drowsy and more prone to accidents).

    3. The (implied) gist of the poster’s real concern is whether any weakness of reporting in clinical trials relates to any weakness of reporting possible persistent side effects. It doesn’t, as patients are followed for variable periods (in all clinical trials) after discontinuation of the study drug, including completion of the trial. When an adverse event occurs, investigators are required to follow the patient longer and in most cases until adverse event resolution.

    Instead, the limitation of issues of reversibility vs nonreversibility of any reported side effect have to do with the limitation of study patient numbers and durations. If an event is very rare, it will not occur when several thousand men are assessed but may occur when tens or hundreds of thousands of men are assessed in the postmarketing setting. The best example of this is with the PDE5 inhibitors (sildenafil, tadalafil) for erectile dysfunction. While the causal association between devastating visual events (termed NAION) and these drugs are unclear (because of their rarity), the occurrence of these events in 100-200 of the over 80 million men who take these compounds post-marketing (with zero occurrences reported in the clinical trials) is now in the Product Label as a possible consequence and underscores my point. rare events that may be related to a drug do not show up in clinical trials because of issues with ‘self-reporting”, but due to a smaller patient number and exposure that can occur in the ‘real-world setting” once a drug is approved.

    4. I have not read the comment the poster attributes to the Editor-in-Chief of the Journal of Sexual Medicine. As someone who is asked to review journal articles for submission, my approach would have been to encourage scientists and patients to publish detailed descriptions of this phenomenon (irreversible finasteride-induced sexual dysfunction), excluding other known causes of sexual dysfunction. That would be scientific. To this date, I have yet to see a single case history in the medical literature (including the Journal of Sexual Medicine).

Tags: fda, clinical trials, propecia, finasteride

10 thoughts on “How Data is Collected and Interpreted in Clinical Trials

  1. By the way, what is up with posts disappearing from the comments sections? I’ve seen several topics with only half as many posts as are listed in the comments section. A number of my own posts are gone as well. Am I the only one seeing this?

  2. You aren’t correct on many assumptions you have made and it doesn’t seem like you have actually even read the article.

    The study is predominantly rejected by practitioner’s who have conflicts of interest and have prescribed hundreds even thousands of times. When I showed the study to my prescribing doctor, who is notably not majorly focused in the hair loss industry, he/she responded by accepting it as very strong suggestive evidence despite the ‘limitations’ that are present in the study. However, whenever people like yourself point to these limitations you neglect to mention that there are limitations in literally every single study ever conducted without focusing in on a more nuanced interpretations of the limiting factors.

    Many of the patients interviewed had previously seen Dr. Irwig and this prompted him to collect more data. These patients have no interest in neither seeing hair loss surgeons nor dermatologists which is why these types of doctors have been so quick to reject the casual connection. Urologists and endocrinologists are much more receptive, but do not have much experience or a clear understanding as to the particular mechanisms that have been damaged since this is an orphan disease. This problem poses a mystery to the respectable doctors who have attempted to devise treatments but a huge thorn in the side of hair transplants that do not have any alternative treatments to offer.

    There were definitely medical tests done to look at the problem of sexual side effects but each patient has pursued different types of treatment. The nocturnal tumescence test is pretty worthless as it doesn’t require an inconvenient overnight sleep study to confirm what we already know. After waking up with high pressure, pulsating erections for twenty years (apologies for the description), the erections that are discovered upon waking in the morning are low pressure, weak, and limp. The knowledge of a lack of erections really does nothing to suggest treatments for that matter.

    Dr. Irwig’s data does in fact suggest there is a causal link between finasteride and irreversible impotence and you can find him clearly state this in various public interviews. What he does note is that he cannot specify the frequency with which it occurs. I believe he has stated almost verbatim that he is not sure whether it is happening to one men in a hundred, a thousand, or less frequent.

    You are right that the men who interviewed could have lied, but this also goes for any self-reported clinical study and does not invalidate Irwig’s study. OF note, Dr. Traish also wrote up a more personalized case study in a paper he published on the persistence of finasteride’s sexual side effects in December 2010. It is only one case but is much more in line with what you claim you would like to see.

    Perhaps I will ask the next doctor I visit if he/she will be willing to right up my case report individually describing all of my failed treatments, test results, and exclusions of other causes. Many guys took the drug on separate occasions and have developed progressively worse persistent symptoms each time they took the drug. These cases are particularly illuminating since it is almost completely unlikely they would have an extraneous caused that perfectly correlated w/ their finasteride usage on separate occasions.

    This exact fact pattern for mood effects is what led Dr. Rassman to accept the causality between finasteride and cognitive disturbances. The same fact pattern exists for sexual dysfunction but has not yet been published in any medical journals.

  3. Jeremy,
    You mention that “pharmaceutical companies are basically given carte blanche to design these ‘methodical’ questionnaires that are given to patients at predetermined intervals’. In fact, side effects (“adverse events”) are not collected in clinical trials by questionnaires but by structured open-ended interviews where the subject is asked about any new or worsened development by the investigator or staff. The interviewer also asks not about what the patient is experiencing now but at any time since the last interview.

  4. Dr. L – You are not correct. If you refer to the link I posted above you will see correspondence between the Senior Director of Regulatory Affairs at Merck and a Director of Dermatological Drug Products at the FDA. In this letter, the Merck director refers to how he wants to present data collected from the sexual function questionnaire and somewhat poetically asks the FDA not to keep the contents of the letter confidential.

    As you should know, clinical trials are each designed differently and subject to ‘spin’ from the pharmaceutical companies as is evident from this exchange.

    However, in this case, Merck undeniably used a questionnaire to evaluate the side effects (or as you would euphemistically say adverse events) of its clinical trial subjects.

  5. Jeremy
    As usual, you do not understand the issues of clinical research. The link you provide uses a questionnaire to measure EFFICACY, which is how all sexual function efficacy data is collected. What the poster and Dr L are noting is how SAFETY adverse events are collected. Of course, you are the expert in everything medical and piece together various links that you do not understand,, third party communications, and your own conspiracy theories about pharmaceutical companies to distract from the original excellent point: that the small number of patients and short durations in clinical trials are inadequate to detect extremely rare events.

Leave a Reply

Your email address will not be published. Required fields are marked *