org355: Viagra & Dementia

This is an interesting study reported in The Guardian, which hits on many big concepts in epidemiology...

Researchers found that men who were prescribed Viagra and similar medications were 18% less likely to develop the most common form of dementia years later than those who went without the drugs.

1. What is the PICO for this study?

2. Were the researchers measuring incidence or prevalence? What RR figure corresponds to "18% less likely", and what is the name of this RR? Who is in the numerator and who is in the denominator of the RR?

"The effect was strongest in men with the most prescriptions.."

3. What two-word epi concept should immediately jump in your head?

Brauer and her colleagues analysed medical records for more than 260,000 men who were diagnosed with erectile dysfunction but had no evidence of memory or thinking problems.

Just over half were taking PDE5 inhibitor drugs, including sildenafil (sold as Viagra), avanafil, vardenafil and tadalafil. The men were followed for an average of five years to record any new cases of Alzheimer’s.

4a. What is the study design? Dementia is fairly rare, and PDE5 use fairly common - surely a case-control study is more appropriate?

4b. Why not include all men in the study?

5. What are the hypothesised mediation (causal pathways)?

6. There is acknowlegement of known but unmeasured confounders. (I found four in the article...) Explain the implications.

If PDE5 inhibitors do protect against Alzheimer’s, the drugs would be expected to work in women as well as men. “We think it would be very worthwhile to run a trial in a wide group of people,” Brauer said.

7. Do you think a randomised trial is justified? What ethical or methodological limitations do you anticipate?

Re: Viagra & Dementia

by | MADHUTANDRA SARKAR - Monday, 12 February 2024, 2:17 PM

Hi Sujit,

The following are my answers:

1. The PICOs are: 
P (Population of interest): Men newly diagnosed with erectile dysfunction (ED).
I (Intervention or exposure): PDE5 inhibitor drugs, including sildenafil (Viagra), avanafil, vardenafil and tadalafil.
C (Comparison): Men newly diagnosed with ED who were not taking Viagra and similar drugs.
Outcome: Reduction in the risk of Alzheimer’s disease (Neuroprotection).
S (Study design): Cohort study.
S (Setting): UK.

2. The researchers were measuring incidence.
The RR figure which corresponds to “18% less likely” is 0.82.
The name of this RR is hazard ratio.
The numerator includes men with a new diagnosis of ED who were taking PDE5 inhibitor drugs.
The denominator includes men with a new diagnosis of ED who were not taking any PDE5 inhibitor drug.

3. The epi concept applicable here is “biological gradient” or “dose-response relationship”.

4a. The study design is cohort study.
Case-control study is not appropriate here, as a case-control study cannot establish the sequence of events, so chance of reverse causality is there, i.e. Alzheimer’s disease can cause no/low intake of PDE5 inhibitor drugs. Moreover, chance of information bias is also there as the exposure status (use of PDE5 inhibitor drugs) is determined after the outcome (Alzheimer’s disease) has occurred.

4b. All men were not included in the study. Follow-up would have been difficult in that case resulting in selection bias. It is costly also.

5. The causal pathways are:
a) PDE5 inhibitors relax veins and arteries allowing blood to flow more freely, which improves blood flow in the brain and may help protect against Alzheimer’s disease.
b) PDE5 inhibitors raise levels of a compound called cGMP, which may also help to protect brain cells.

6. The unmeasured confounders, such as physical and sexual activity is associated with both exposure (physically and sexually active men were more likely to use Viagra) and outcome (physically and sexually active men had a low risk of developing Alzheimer’s disease), and is not on the causal pathway. As the authors could not account for the levels of physical and sexual activity, residual confounding might have affected their results showing protective effect of PDE5 inhibitor drugs.

7. A randomized trial is not justified here both ethically and methodologically. Sildenafil has prominent side effects. Randomization might be difficult, and there is possibility of selection bias.

Thanks in advance!

Madhutandra

Re: Viagra & Dementia

by | Sujit Rathod - Wednesday, 14 February 2024, 10:57 AM

Well done Madhutandra! Though I'll push back on your points about case-control studies. While recall / temporarily is an issue of particular concern for case-control studies, it's not automatically the case it will be a problem. For this research question, I think there are ways to work around this problem.