Study of ED risk loses the signal by including past users of finasteride

Response to: Hagberg KW, Divan HA, Persson R, Nickel JC, Jick SS. Risk of erectile dysfunction associated with use of 5-α reductase inhibitors for benign prostatic hyperplasia or alopecia: population based studies using the Clinical Practice Research Datalink. BMJ. 2016. doi:10.1136/bmj.i4823

SUMMARY: Hagberg et al, 2016 fails to assess the risk of erectile dysfunction (ED) associated with using finasteride, because past users of the drug were counted as ED cases. A majority (69%) of ED cases in the finasteride (1 mg) group had stopped the drug 91 days or more prior to ED diagnosis or treatment. In addition, the study overlooks implications of finasteride (1 mg) being unavailable through the United Kingdom’s National Health Service.

Hagberg et al compared rates of erectile dysfunction (ED) in men who took finasteride or dutasteride vs. controls, based on records of general practitioners (GPs) in the UK-based Clinical Practice Research Datalink (CPRD). Among other findings, they found no increased risk of ED in men who took finasteride 1 mg for hair loss (alopecia) vs. controls.

There is a major limitation: the analysis included past and recent users of finasteride, as well as current users. Past users are defined as those who stopped taking a finasteride at least 91 days before getting diagnosed or treated for ED. Recent users are those who last took the drug 31–90 days before diagnosis or treatment for ED.

A majority (69%) of ED cases in the alopecia group were past users. Only 25% were current users. Therefore, this study does not address ED as a concurrent side effect of finasteride. Consider a patient who had ED related to finasteride, but stopped six months ago and the ED resolved. He would be counted as a non-case in this study. This is a miss. Moreover, men with finasteride-related ED might discontinue the drug at a higher rate than those without this adverse effect. These cases would be disproportionately classified as non-cases, weakening the signal of drug-related risk.

Table 6 from Hagberg et al, 2016. 69% of the cases of erectile dysfunction (ED) were past users of finasteride, meaning they had discontinued the drug 91 days or more before getting diagnosed or treated for ED.

Just as the methodology misses real signals, it is susceptible to false alarms. Consider a patient who stopped finasteride a year ago, and then developed new ED unrelated to prior use of the drug. He would be counted as an ED case in the finasteride group. This would be a false alarm. In summary, including past users produces both misses and false alarms, muddying the link between the drug and the risk of ED.

Following are additional limitations.

Factors unique to younger men. It is possible that younger men are less likely than older men to seek treatment for ED because of embarrassment and/or less frequent use of health care. They may have been less aware of drugs like sildenafil during the study period (2002–2011), since sildenafil became available in the UK in 1998, and marketing was primarily directed at older men. Moreover, sildenafil was not available on the NHS until 2014. As the authors acknowledge, prescriptions from private sources would not show up in the CPRD. These several factors could make the CPRD an insensitive measure of ED in younger men, masking differences in ED rates between groups. (There were just 13 ED cases younger than age 40 in the alopecia group exposed to finasteride, as shown in Table 6.)

Non-formulary status of finasteride (1 mg). Finasteride (1 mg) was not available on the National Health Service (NHS) during the study period [1] (termed “non-formulary status”). GPs could write prescriptions for finasteride for hair loss, but men would have also sought prescriptions from specialists or private practices. For this reason, as well as embarrassment over using a hair loss treatment, it is possible patients were less likely to seek treatment for finasteride-related erectile dysfunction from GPs whose records make up the CPRD [2].

Also because of the non-formulary status of finasteride (1 mg), it is possible that men in the control group were taking finasteride for hair loss, or had previously taken it. If it were obtained from private sources, it would not be included in CPRD unless the patient reported it to their GP—which they might not want to do because of embarrassment over using a medication for hair loss. Unrecorded finasteride use in the control group could mask ED risk associated with the drug.

Conflicts of interest. A co-author of this study, J. Curtis Nickel, had prior affiliations with Merck, the developer and marketer of finasteride, which were not disclosed. In the 1990s, Dr. Nickel participated in the PROSPECT Study Group and PLESS Study Group, both supporting Merck’s development of Proscar (finasteride 5 mg) for the treatment of benign prostatic hyperplasia [3,4]. Dr. Nickel was also a co-inventor of (U.S. Patent No. 6,403,640) granted in 2002 and assigned to Merck and Temple University (not related to finasteride). Merck researcher Elizabeth Stoner, who was involved in the development of finasteride, was a co-inventor.

This response has been posted to PubPeer. A notification was sent to the corresponding author who has the opportunity to respond there.

This paper is one of three using a similar methodology to assess, respectively, the risk of depression, erectile dysfunction (ED) and gynecomastia associated with 5-ARIs [5–7]. The study of depression risk has a similar limitation: past users of finasteride are included in cases, diluting the signal of drug-related risk. Read a response to that study: Study of depression from prostate drugs loses the signal by including past users


NOTES

  1. Men to pay for anti-baldness drugBBC News. June 7, 2000.
  2. Hagberg et al discuss the potential impact of private prescription of PDE5 inhibitors, but not that of finasteride (1 mg). The use of private sources of finasteride (1 mg) might differentially reduce willingness to seek treatment for finasteride-related ED from a GP, compared to men in the control group with ED not related to finasteride. This is a potential confounding factor in a comparison between these groups.
  3. Nickel JC, Fradet Y, Boake RC, Pommerville PJ, Perreault JP, Afridi SK, Elhilali MM, and the PROSPECT Study Group. Efficacy and safety of finasteride therapy for benign prostatic hyperplasia: results of a 2-year randomized controlled trial (the PROSPECT study). CMAJ. 1996. PMCID:PMC1335066
  4. McConnell JD, Bruskewitz R, Walsh P, Andriole G, Lieber M, Holtgrewe HL, Albertsen P, Roehrborn CG, Nickel JC, Wang DZ, Taylor AM, Waldstreicher J, for the Finasteride Long-Term Efficacy and Safety Study Group. The effect of finasteride on the risk of acute urinary retention and the need for surgical treatment among men with benign prostatic hyperplasia. N Engl J Med. 1998. doi:10.1056/NEJM199802263380901
  5. Hagberg KW, Divan HA, Nickel JC, Jick SS. Risk of incident antidepressant-treated depression associated with use of 5α-reductase inhibitors compared with use of α-blockers in men with benign prostatic hyperplasia: a population-based study using the Clinical Practice Research Datalink. Pharmacotherapy. 2017. doi:10.1002/phar.1925
  6. Hagberg KW, Divan HA, Persson R, Nickel JC, Jick SS. Risk of erectile dysfunction associated with use of 5-α reductase inhibitors for benign prostatic hyperplasia or alopecia: population based studies using the Clinical Practice Research Datalink. BMJ. 2016;354:i4823. doi:10.1136/bmj.i4823
  7. Hagberg KW, Divan HA, Fang SC, Nickel JC, Jick SS. Risk of gynecomastia and breast cancer associated with the use of 5-alpha reductase inhibitors for benign prostatic hyperplasia. Clin Epidemiol. 2017. doi:10.2147/CLEP.S124674