A new study came out this week suggesting use of a particular class of drug after age 65 was associated with dementia. Here’s what you need to know.*
The study is retrospective, meaning it took people who developed the disease of interest and then looked backwards at their medications. Retrospective studies are prone to a number of problems, the biggest one being that even young people with healthy memories are crap at giving you their drug history over the past 10 years, and this is a study of people with dementia. The researchers dodged this by using an HMO database of the subjects complete medical history, which is a neat trick. The second problem is that retrospective studies can easily end up being painting the bulls-eye after they’ve fired the arrow. Mere chance dictates that if you track enough traits, any random subset of a population is likely to have something more in common with each other than with the rest of the population. If you use the traditional bar of statistical significance (5% chance of results arising by chance), checking 20 traits gives you an expected value of 1 false positive. To be fair, this study has a much higher significance level, and the effect was dose dependent, which is a very good sign that it’s legit. The authors heavily imply they deliberately studied anticholinergics rather than shotguning it, but without preregistration there’s no way to be sure.
Anticholinergics come in two forms: antimuscarinics, and antinicotinic. Short version: these work on different types of neuroreceptors, which live in different parts of the body and do different things . Every example drug they give is an antimuscarinic and of the classes of drugs they list, many have no antinicotinic members. Even if they technically included antinicotinics in the analysis, they would be such a small portion of the sample that their effect could be overwhelmed. So I don’t think you can apply this study to drugs like bupropion, which is an antinicotinic.
I don’t like the way they calculated total exposure at all. Essentially they counted the normally recommended dose of any medication as One Standardized Daily Dose. But those dosages vary wildly (even the examples they give span an order of magnitude), as do the particular drugs’ ability to cross the blood-brain barrier. The drugs are prescribed for a huge variety of causes, and what’s sufficient to stop incontinence has nothing to do with what’s sufficient to slow Parkinson’s. This oversight may cancel out with the fact that they created buckets of dosages rather than do a proper linear regression, in the sense that low-def pictures cancels out bad skin.
The obvious question is “but maybe the same thing that drove people to need anticholinergics increases the likelihood of dementia?” This study has a much better retort for that than most, which is that anticholinergics were prescribed for a variety of causes, and it’s unlikely they all correlate with dementia. I find that explanation extremely satisfying, except that they only evaluated the drugs as a single unit. Antidepressants make of over 60% of the total SDDs taken. The next most common is antihistamines at 17%. But since more than 60% of the population took at least one SDD, it seems likely that those were taken intermittently, as opposed to the constant drip of antidepressants. This leaves open the possibility that the entirety of the effect they attributed to anticholinergics was in fact caused by tricyclic antidepressants alone- and that the real culprit was depression. The obvious controls were to evaluate the anticholinergics separately, and to compare rates of dementia among TCA treated patients with those treated with other antidepressants.
The subtler version of this question is “what if anticholinergics prolong life, giving you more time to develop dementia?” I don’t see anything where they checked for that either way. They did ask for people’s perception of their own health, and that was negative correlated with TSDD, but if TSDD is correlated with depression it’s hard to know how to interpret that.
For all those criticisms, this is an amazingly strong result for a medical study**. No one study can prove anything (even if i think they had the data to do more than they did). It definitely merits further investigation (ideally some with animal models, so we can do the causality experiments that would be super unethical in humans), and maybe even behavior change in the meantime, although a lot of the drugs studied are already obsolete or second line. Plus it another piece of data that will help us figure out how to fight dementia, and that makes me really hopeful.
*Read: here’s what I learned.
**Yes, this should worry you.