Reckoning with underreporting—response to Lauck et al, 2024

It’s not clear that medical records would pick up sexual problems in healthy men, especially those taking drugs for hair loss. Moreover, the study includes men using 5-ARIs at the time of sexual dysfunction, so the results do not address why it might persist after stopping these drugs.

Response to: Lauck KC, Limmer A, Harris P, Kivelevitch D. Sexual dysfunction with 5-alpha-reductase inhibitor therapy for androgenetic alopecia: a global propensity score matched retrospective cohort study. J Am Acad Dermatol. 2024. doi:10.1016/j.jaad.2024.03.019

This responds to the pre-proof manuscript.

SUMMARY: Lauck et al propose persistent sexual dysfunction after stopping 5-ARIs might be due to other conditions including obesity, depression and anxiety. In support, they cite a study that is in direct disagreement with this hypothesis. Moreover, it is not clear that the underlying medical records are sensitive to sexual dysfunction in healthy men. As the authors point out, there may not be follow up with the prescriber. Even if there is, stigma and embarrassment might prevent reports of SD. Records might be even less sensitive to SD due to 5-ARIs for two reasons: embarrassment over taking a medication for hair loss, and confounding by specialty since dermatologists may be less attuned to SD and less inclined to record drug side effects. Patients might also be less likely to report SD to dermatologists. In light of these concerns, it comes as no surprise that SD is highly correlated with comorbidities: these individuals are getting more medical attention, from non-dermatologists, so an SD evaluation may be more likely. Lauck et al’s findings may be merely an artifact of insensitivity to SD experienced by healthy men. If the signal of SD is weak or absent, statistics cannot compensate for this. It is critical to establish the quality of source data with respect to SD, as well as SD in the special case of men taking 5-ARIs for hair loss, before ruling out 5-ARIs as the cause of persistent SD. A final, fundamental concern: since the analysis includes men currently taking 5-ARIs, the results cannot shed light on why SD might persist after stopping 5-ARIs.

It is critical to establish the quality of source data with respect to sexual dysfunction before ruling out 5-ARIs as the cause of persistent SD.

Background

The Research Letter reports a cohort study of the risk of sexual dysfunction (SD) in men with hair loss who used 5-alpha reductase inhibitors (5-ARI) vs. controls. The samples are large, with over 10,000 men in each group. In patients who took 5-ARIs and had SD, there was a higher prevalence of obesity, nicotine dependence, diabetes mellitus, hypertension, mood and anxiety disorders (p<0.0001). When patients with any of these conditions were excluded, the association between SD and 5-ARI use was no longer significant.

Are other diseases to blame?

Based on this finding, the authors suggest:

because many of these chronic comorbidities affect patients after 5-ARIs are stopped, these results may contribute to a working hypothesis for persistent SD after 5-ARI discontinuation, a subject debated in literature and media.

This hypothesis for persistent SD depends on the comorbidity arising after a 5-ARI is stopped, but no evidence is provided to support this sequence. The data appears only to show that there was a comorbidity with SD, not the sequence in which they developed.

Moreover, the analysis does not specifically focus on post-5-ARI SD. It includes a diagnosis of SD up to one year after prescription of a 5-ARI. This would include men still taking the drug—for whom it is not “persistent” SD. Therefore these results cannot shed light on the basis of persistent SD in men who stopped taking a 5-ARI.

The authors cite Kiguradze et al, 20171 in support of a hypothesis that comorbidities may explain persistent SD in the 5-ARI group. While that study did find an association of SD with other factors, it noted:

duration of 5a-RI exposure was a more accurate predictor of PED [persistent erectile dysfunction] than all other assessed risk factors except prostate disease and prostate surgery. Among young men with 5a-RI exposure, duration of 5a-RI exposure was a more accurate predictor of PED than all other assessed risk factors.

These results are in direct disagreement with Lauck et al’s suggestion that comorbidities could explain persistent SD after stopping 5-ARIs.

How strong is the signal?

A limitation is briefly mentioned: “underreporting of SD as patients require a follow up visit for documentation.” The authors claim that the limitation is offset by the size of the data set and the statistical methodology:

However, this study analyzes over 23,000 real-world patients with robust external validity while controlling for confounders in a way not feasible in other study types of this size.

The authors refer to “real-world patients,” but the study is in fact based on medical records written by others based on limited encounters with the patients. If underreporting is indeed present, as they authors acknowledge is possible, these records would not have robust external validity.

The matter of underreporting needs careful consideration. There are strong taboos surrounding SD. Patients may not wish to bring it up with their doctor. The doctor may not want to ask about it, fearing offending the patient. Even if it does come up, it may not be recorded in visit notes or diagnosed. There are also special considerations in men taking 5-ARIs for hair loss. The drugs are often prescribed by dermatologists who do not diagnose or treat sexual disorders. Patients may therefore hesitate to report SD to them, and dermatologists may not be interested in hearing about SD. They may be less likely to attribute SD to the drug than other specialties. Finally, dermatologists may have different general attitudes about drug side effects and be less likely to record them than other physicians such as general practitioners. Medical specialty is a potentially significant confounding factor. Lopez-Gonzalez et al reported that medical specialty was the personal and professional characteristic most closely associated with under-reporting of adverse drug reactions.2

If the source data are insensitive to SD, there is a loss of statistical power. And if there is a differential insensitivity to SD in men taking 5-ARIs for hair loss, then the likelihood of detecting a difference between groups is even lower. A null result does not rule out the drug as a cause of SD; it may only mean that SD, and particularly drug-related SD, is not adequately captured in medical records.

Conclusion

The authors propose that persistent SD after stopping 5-ARIs might be explained in terms of comorbidities. Their reasoning has notable limitations: 1) a study they cite contradicts the authors’ hypothesis; 2) they do not show that comorbidities arise after stopping 5-ARIs; 3) the analysis is not limited to patients who stopped taking 5-ARIs, so it is unknown whether SD is “persistent” at all; 4) it has not been established that the source data is sensitive to SD, and specifically to SD in men taking 5-ARIs, where underreporting could plausibly be greater.

The lack of a difference between 5-ARI and control groups in an adjusted comparison may merely reflect the limitations of the source data. Finally, this study design cannot address the basis of persistent SD after stopping 5-ARIs because it does not specifically analyze this patient population.

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


References

  1. Kiguradze T, Temps WH, Yarnold PR, et al. Persistent erectile dysfunction in men exposed to the 5α-reductase inhibitors, finasteride, or dutasteride. PeerJ. 2017. doi:10.7717/peerj.3020 ↩︎
  2. Lopez-Gonzalez E, Herdeiro MT, Figueiras A. Determinants of under-reporting of adverse drug reactions: a systematic review. Drug Saf. 2009. doi:10.2165/00002018-200932010-00002 ↩︎