Study Reveals Link Between Depression, Epilepsy, and Treatment Failure

Dr. Sam Terman (00:00):

My name is Dr. Sam Terman. I'm an adult epileptologist at the University of Michigan. We did a study. The objective was to understand the differences in time to treatment failure among patients with newly treated epilepsy in those with depression versus those without depression. Patients with epilepsy have increased chronic comorbidities of all types, but one really important type of comborbidity is mental health disorders. Depression and anxiety are known to be increased in patients with epilepsy. It's a very common comorbidity and very clinically important entity that epileptologists are always attuned to. Patients with depression face many health challenges, including complex comorbidities, socioeconomic differences and disparities in terms of their access, their beliefs, their diagnoses. And so we wanted to understand, does the time until treatment failure differ according to patients with versus without depression?

(01:02):

So the methodology was that we used Medicare data. We started with about 2 million patients with epilepsy codes of any sort, and we filtered down to those that are newly treated, meaning they're starting an anti-seizure medicine with no recent anti-seizure medicine before that. And then we ended up with about 90,000 patients as our total study sample size. About one-third of them had a depression code and about two-thirds did not, which we defined based on ICD-10 codes in the year before the first anti-seizure medication billing.

(01:36):

The first set of outcomes indciated that patients with depression have increased incidence of basically everything that we looked at: heart disease, lung disease, strokes. They were more likely to be female and have minority racial status and also have additional Medicaid. So many differences between these populations in terms of probably their disease severity, their access to care, their comorbidities, et cetera. But the reason that we did this study was then that we looked at the time until treatment failure, and we looked at that in a couple ways. One way was how many months was it until that first line of therapy was essentially given up on?

(02:18):

And so our main endpoint was when is the first line of therapy changed to the second line of therapy, meaning when was there a switch, an addition, a discontinuation of whatever they started off on, which we defined based on their pharmacy fills. So we said if they start a line of therapy, are given anti-seizure medicine, and then based on the pharmacy field, they have at least a 60-day gap. It's not a perfect test, but it's an indication to us that something about that first medication either didn't work or the patient decided to drop it or they couldn't fill it again, or they added, or they switched. And so we had this composite measure of treatment failure, meaning that something about that first medication didn't work out. And time until the first line of therapy was abandoned or added or had to switch or something to that nature was shorter for those with depression compared to without depression.

(03:13):

In our multi-variable logistic regression, we found that the odds of treatment failure with the first line of therapy was 40% higher for those with depression versus without depression. I think it reinforces things that we likely already suspect, but we should make sure that we're attuned to. For example, clinicians can miss depression. It's known that we should be using structured screening tools to pick up on mood disorders that we might otherwise miss. In claims, we have ICD codes, which frankly probably misses a lot of patients with depression. And so it's just another reason that we should be attuned to patients with depression.

(03:50):

Another reason is that, and in our future work, we'll follow up on this, is that we really should monitor these patients. When we start a medication, there's certain patients that we probably should be watching really closely, maybe having more frequent visits, check-ins, or at least that initial counseling saying, "Hey, you're a group that might be at higher risk of something going wrong," either because of lack of tolerance or lack of seizure efficacy or trying to optimize what is it about this group that may cause these patients to stop their initial drug first. I think from a research perspective, we need to look into reasons for this to disentangle.

(04:26):

Is the association that we're finding because certain medications are the offenders or is it because certain comorbidities are really contributing or is it socioeconomic disparities that really make one group more likely to persist on their initial therapy than another one? I would be really interested if there were certain medications, certain anti-seizure medicines such as the mood stabilizers that are less likely to have this disparity and certain ones that may be more likely. So that may have therapeutic implications about which optimal therapy will give patients the maximum likelihood of succeeding? What does that look like? What is that drug and for whom?

(05:02):

There's been at least one other study and a really excellent poster that also touched on the role of depression and epilepsy care at the AES meeting. I think each of these studies are synergistic and looking at a different aspect. Our study was particularly about does depression confer an increased risk of treatment failure? So that's one angle, but for sure there's a lot of interest at this meeting about mental healthcare more generally in this population. I hope our future work will further disentangle the reasons for this difference and to ultimately end up with the best treatment for patients with and without depression.

(05:38):

In our future work, we hope to specifically understand the role of different anti-seizure medicines in mediating or modifying this relationship. For example, medications differ by their mechanism of action. We know we have sodium channel blockers, we have GABAergic medications, we have synaptic vesicle antagonists, and some are more favorable for mood and some are less favorable for mood. We'll parse out this association according to mechanism of action in addition to adjusting for other comorbidities that could contribute to the patient's medical complexity and potentially pose barriers to adherence such as stroke and dementia and disability reason for Medicare entitlement, along with a host of other factors to get down to what's driving this relationship.

(06:29):

It's wonderful to see the diversity of scientific inquiry and discovery at meetings like this, bringing people together from diverse disciplines and geographies to all contribute towards epilepsy care.