Are Pain Meds Giving Me Dementia? Part 1: Wrong Question.

A public peer-review

This is Part 1 of a two-part series evaluating a recent scientific study on medication for low back pain and dementia. Today, we’re going to talk about why this study asks the wrong question. In Part 2, we’ll talk about why it (probably) gets the wrong answer even for the question it asks. Subscribe so you don’t miss out!

This week a newly published scientific study claimed to show evidence that gabapentin use among chronic low back pain patients can cause dementia. As someone with chronic (upper) back pain, that headline is scary! Do I really have to choose between managing my pain now and protecting my memory in the future?? Anyone would struggle to make that choice.

But, I’m not just anyone. I’m an epidemiologist. One with special training in the best state-of-the-art methods for assessing causal relationships directly from the people that developed many of those methods. So I don’t rely on headlines. I evaluate research. And I’m going to help you evaluate it too.

The paper in question is called Risk of dementia following gabapentin prescription in chronic low back pain patients and it was published in a journal called Regional Anesthesia and Pain Medication which the journal of the American Society of Regional Anesthesia and Pain Medication.

Back in March, I wrote an explainer outlining 7 steps for reading a scientific paper (you can read that here if you missed it!), but I’m not going to follow those steps with this paper. I started to, but then I realized there was something I had missed in that list. And it’s one of the major flaws in this paper.

We need an 8th step¹: “What causal question is this study asking?”.

Despite some very careful wording (“associated”, “risk”), this paper is attempting to ask a question about causation. Which means that before we can decide whether the researchers have done a good job answering that question, we need to figure out what the exact question they are asking is!

I’ve touched on this issue before, but I want to get more into the details of what specifically to look for. Because the number one problem with this study is that they don’t actually provide enough information for us to figure this out.

3 Tips for Making Sense of Health News

What causal question is this study asking?

It’s tempting for many people to answer a question like this by saying “well, the authors looked at this group of people and then compared these things”. But what the authors did isn’t necessarily what they were trying to do. To figure this out, we have to read their introduction. Some key points the authors make are:

• Gabapentin use is increasing for low back pain patients, because it has a low potential for addiction.

• Biologically, gabapentin works by altering signals in the brain. But brains are important, so what if this leads to problems in the brain later on in life?

• There are conflicting results from other studies about a potential relationship between gabapentin and dementia.

So, the authors are ultimately trying to answer the question: “if a person takes gabapentin for (low back) pain management, will that cause them to develop dementia?”

But this still isn’t a complete answer to “what causal question is this study asking”, because we need more details. A good causal question includes the specific population we are interested in, the specific treatment options we are comparing, and the specific outcome we are measuring (including the time frame for when it might happen).

This is where things start to fall apart for this study. Because, even though the specific population is pretty clear (adults with chronic low back pain), and the outcome is pretty clear (diagnosis of one of several types of dementia within 10 years), the treatment options are not.

If I wasn’t familiar with gabapentin, I might not have noticed this problem so quickly, because for many medications it’s pretty common to see researchers divide people up into groups that received the medication and groups that didn’t². But I am familiar with gabapentin, and so I know that there is a very wide range of possible dosages.

The very first thing I checked when I read this paper was what dosage of medication they were looking at. They don’t say. In fact, they don’t even tell us the average dosage that the patients in their study take.

And that is a big problem. Because, even if gabapentin can cause dementia, it’s very unlikely that this happens with any dosage. Even the authors don’t seem to think that would happen. How do we know? Because they also compare people based on the number of gabapentin prescriptions they get.³

Number of prescriptions can help you assess duration of exposure, but it’s not necessarily a good proxy of amount of exposure. To actually help us understand whether there’s a biological effect of gabapentin on dementia, we need to know exactly how much gabapentin we are talking about. This paper doesn’t tell us.

Worse, they don’t even give us a clear answer to how many prescriptions they are comparing. In the methods section, they define the treated group as people having 2 or more gabapentin prescriptions, and then immediately in the next sentence define the “Gabapentin” group as having 6 or more gabapentin prescriptions. Which one is the actual definition? It’s not clear. That’s a problem.

So bottom line, this study seems to be trying to answer the causal question “would dementia diagnosis be more common if people with low back pain received ‘many’ gabapentin prescriptions of any dosage compared to if they received no gabapentin prescriptions?”.

But as a patient with back pain, the questions I want to know the answer to are: “would a dementia diagnosis be more common if I were to take a higher versus a lower dose of gabapentin?” or “would a dementia diagnosis be more common if I were to take my specific dose of gabapentin for a longer or shorter time period”?

And neither of those questions is asked in this study.

Bottom line: This study is being reported on as providing us with information about whether or not gabapentin causes dementia. But it doesn’t do that. One problem is that they don’t actually ask a clear or useful question. And if you ask the wrong question, there’s very little chance of getting the right answer.

In Part 2 of this series, I’ll explain why they probably get the wrong answer even for the poorly defined question they do ask. Subscribe so you don’t miss out!

1. Well, we need 8 steps total, but if I were listing all of them out, I would add this new one as step #4 between “What do the authors thing the study is about?” and “Read the methods”.

2. Despite being common, this is actually a very bad way to study medications. In Part 2 of this series, I’ll address this issue more.

3. They do this in a weird way though. The main analysis compares people with zero gabapentin prescriptions to people with 2 or more (or possibly with 6 or more) prescriptions. The secondary analysis compares people with 3 to 11 prescriptions versus people with 12 or more prescriptions. Which means that people with 1 or 2 prescriptions are completely discarded from this analysis. Why? They don’t give a reason.