Hidden conflicts of interest in the medical literature. Case study: strokes.

Getty Images.

In May, JAMA Neurology published an opinion by a respected stroke neurologist. It was really, really weird.

The author argued that patients having strokes should receive certain treatments regardless of when their symptoms began. That was disturbing because all of the studies establishing how well (or poorly) these acute stroke treatments work have always been limited to strict, carefully delineated time windows. The writer felt that should all be tossed out. That was jarring to anyone versed in the stroke literature.

The very notion of inclusion and exclusion criteria for any treatment is that if a patient cannot benefit, the remaining outcomes are either nothing (best case scenario) or injury (and possibly substantial ones). Some of these treatments work well, but some of them (the “clot busting drugs” in particular) are on shaky ground. Meanwhile, it’s known that when these drugs are not given within a certain time window from the start of symptoms, their harms outweigh the benefits.

I had to respond. (Hello, it’s me.)

You know me. I can’t let bad ideas stand uncontested. So, I wrote a letter to the editor—which, unsurprisingly, the journal declined to publish. (They offered that I could post it as a Comment on the JAMA Neurology webpage. No thanks. I decided to post it here for you all instead. See below.)

The upshot of my response was that it was inexplicable for anyone—let alone a stroke expert—to suggest there were no possible harms from stroke treatments when there are well-documented ones in patients after certain times.

I didn’t have to reach too far to make that point, in fact. As you’ll read below, I merely needed to quote an earlier article written by the exact same author of the opinion JAMA Neurology had just published.

Here’s the letter. I’ll add some more commentary on the other side…

Delays in stroke care alter risk-benefit assessments.

Jeremy Samuel Faust MD MS

To the Editor,

In “Time Windows Between Symptom Onset and Treatment—An Outdated Myth,” Dr. Caplan mentions four stroke trials in which the inclusion criteria had strict time windows for the receipt of acute stroke interventions such as thrombolysis or thrombectomy.¹ He argues that limiting these treatments solely to patients fitting the time-gated inclusion criteria of those studies could mean missing the opportunity to help similar patients whose “last known well” time would have meant exclusion from those trials. Regarding those trials, he states “All had time limits for the treatments studied. None extended their observations beyond the time limits for inclusion.”

I was surprised that Dr. Caplan omitted mention of IST-3, a randomized control trial which enrolled more participants than these others combined, and which had drug supplied by Boehringer Ingelheim, the study drug’s manufacturer.² Infamously, the primary outcome (“alive and independent at 6 months”) was not met in IST-3. However, its “success” has often been resuscitated by focusing on secondary outcomes from patients divided into three bins, by time to randomization. The outcome assessing efficacy of alteplase in the first 3 hours of the 6-hour inclusion window favored the drug. Meanwhile, efficacy during the 3-4.5 hour and >4.5-hour time windows showed no statistical difference from null values (albeit, uncomfortably close to harm in the 3-4.5-hour interval). If it is acceptable to draw conclusions by culling data from favorable secondary outcomes, then doing so for null findings is equally valid. Applying this principle to the >3-hour window would leave only the known harms of intracranial hemorrhage, found in IST-3 and in every important trial of thrombolysis of which I am aware. In short, “What’s the harm?” That’s easy. The documented ones.

My word against one of the most respected stroke neurologists ever may not hold weight here, data analysis notwithstanding. Therefore, I refer you to a source far more reputable than my own for support in this matter:

“The potential benefits and risks do vary considerably with time in patients with stroke. For example, thrombolysis and clot retraction have less potential benefit and more risks the longer time has elapsed after neurological symptom onset.” –Louis Caplan.³

Jeremy Samuel Faust MD MS

Brigham & Women's Hospital Department of Emergency Medicine

Mass General Brigham Division of Health Services Research

Assistant Professor of Emergency Medicine, Harvard Medical School

Funding: None. Disclosures: No relevant conflicts.

References

1.         Caplan LR. Time Windows Between Symptom Onset and Treatment—An Outdated Myth. JAMA Neurol. Published online May 28, 2024. doi:10.1001/jamaneurol.2024.1370

2.         The benefits and harms of intravenous thrombolysis with recombinant tissue plasminogen activator within 6 h of acute ischaemic stroke (the third international stroke trial [IST-3]): a randomised controlled trial. The Lancet. 2012;379(9834):2352-2363. doi:10.1016/S0140-6736(12)60768-5

3.         Caplan L. Caplan-Fisher Rules. Stroke. 2021;52(5):e155-e159. doi:10.1161/STROKEAHA.121.035017

Now, I’ll admit there was a subtle burn in here. Mentioning the IST-3 trial (and naming its funder) was something of a dig at the author of the original essay. I knew from prior experience that the author had taken quite a lot of money from the very company that made the stroke drug (alteplase) that IST-3 had studied. (Alteplase was the drug that used to be the standard of care for acute strokes; in the US and elsewhere, including my ER, tenecteplase has replaced it in recent years.) I felt like this was a glaring conflict of interest. I also noticed that no such conflict was declared in the essay. Why? Because the JAMA Network only requires disclosures of funding from a drug company received by an author within 3 years. This author, now well into his mid-to-late 80s, had indeed not taken money from the company in five years or so. The most recent $63,111 he had accepted had been in 2017-2018. So, he did not need to declare this conflict, according to JAMA policy.

Well, I thought it was relevant, myself. Not relevant enough to make a stink about it overtly in my response, but enough to make the conspicuous mention of IST-3. I felt that if someone was going to make such a pro-pharma argument, any major past relationships remained relevant for longer than three years.

Things only got worse for the pro-pharma argument.

There was one wrinkle, though. I expected that if the journal published my letter, the author might come back with something about how the safety profile of tenecteplase (the clot-busting stroke drug that has recently become the standard of care for major strokes presenting within 3 hours of symptoms) compared somewhat favorably to the previous standard (alteplase, the one studied in IST-3.)

Well, right around the time JAMA Neurology politely declined to publish my letter, a new piece of research came out in The Lancet—the leading medical journal in Europe (and perhaps overall). Researchers had studied whether patients with less severe strokes might benefit from tenecteplase. Here’s what the authors concluded: “There was no benefit and possible harm from treatment with intravenous tenecteplase. Patients with minor stroke and intracranial occlusion should not be routinely treated with intravenous thrombolysis.”

Long story short, the idea that we should just give these stroke drugs to anyone who might be having a stroke, regardless of symptom severity or timing, is a poorly reasoned one.

Meanwhile, the good news is that modern procedures in which specialists physically go into large arteries serving the brain and physically remove stroke-causing clots (called thrombectomy) represent an important leap forward in stroke care.

---

👋Made it this far? I hope you’re learning something! If so, please consider supporting independent analysis like this by upgrading to Inside Medicine premium. Thanks!

---

How often are the conflicts of interest omitted in medical journals?

This all led me to wonder: How often are relevant conflicts of interest like the one in this case omitted because, technically, they’ve exceeded the journals’ sunset policy on disclosures?

I brought this up in a MedPage Today editorial meeting. It led to this excellent piece, reported and written by the great Cheryl Clark. It turns out that journals vary greatly in their policies, and there is basically no mechanism for enforcing the policies (i.e., it’s the honor system, folks). I don’t think that’s terribly reassuring.

The meaning of a message is always influenced in some measure by who the messenger is. While experts with financial conflicts of interest can of course still be correct on the merits, their entanglements past and present should be taken into account. I think we deserve to know these things, without having to hire investigative reporters.

Questions? Comments? Please discuss in the Comments!