I am not a neurologist and not research-minded, but a practicing ER doc for >30 yrs. I have practiced through the entire TPA era. ER docs were the first to seriously question over and over again the utility and efficacy of thrombolytics in ischemic strokes. Unfortunately, the pharmaceutical dollars have captured and hold captive the C-suites who make the decisions of clinical practice, not the doctors on the ground like us. That train left the station long ago, and even an entire specialty (EM) couldn't stop it. Every stroke patient is unique and a "fight" to do the right thing. I've disagreed openly with neurologists, and yet they still give the lytic. There is so much spin, and the public just buys it. The neurologist tells us there are more lawsuits for NOT giving the lytic than for withholding it. To intimidate us EPs into giving lytics? To follow blindly the standard of care of the local community? To do what the C-suite has established as standard of care?? I rejoice when I find a contraindication to lytic therapy.
Mandrola is right to remind us that the original NINDS trial had weaknesses — the groups weren’t perfectly balanced, and that matters. But he takes that too far. He treats those flaws as if they invalidate the whole case for thrombolysis. That’s not true.
When you look at the bigger picture — all the trials put together, especially the patient-level meta-analyses — the evidence still shows that clot-busting drugs help if they’re given quickly to the right patients. The benefit is real but time-sensitive, and it comes with bleeding risks.
So the sensible position is not “throw thrombolysis out,” but:
Use it early (the earlier the better).
Select patients carefully (don’t overgeneralise).
Keep auditing your own outcomes (because benefits are modest and risks real).
That’s the balanced truth: not heroic, not nihilistic — just precise, time-critical, and monitored.
As a retired doctor, I agree. As with almost everything, the evidence is nuanced. Each situation is unique. Each patient is unique. Their valuation of the risks and benefits will differ. Accordingly, there is no “right answer”, just “what this patient chose for themself at this point in time, knowing what they knew.”
Unfortunately, where time is of the essence, a 20 minute conversation about the risks and benefits of thrombolysis may work against them. So, what do doctors do? They oversimplify the choice. Instead of shades of grey, it’s presented as black and white. “You get thrombolysis, you’ll get better. No thrombolysis, you’ll stay the same or get worse.” The oversimplification ends up reinforced by the payment systems, the hospital bureaucracy, and the legal/malpractice system.
Sadly, once it’s standard practice, it becomes unethical to set up a trial in which one group doesn’t get thrombolysis.
I don’t know how to create a clearer message than what I’ve been trying to convey for the past 20 years, e.g. https://hbiostat.org/bbr/change. Difference from baseline is not appropriate for ordinal stroke outcome scales!! Do a proper analysis and show us the result. Don’t look for differences at baseline but use pre-specified prognostic factors for adjustment. Use an ordinal regression model in the raw ordinal outcome scales adjusted for baseline. For the baseline version of the outcome scales, adjust flexibly by not assuming linearity.
When in doubt remember: the proper analysis of a parallel group design is to compare parallel groups, not change from baseline.
Another note: median outcomes are not appropriate when there are many ties in the data.
I followed Jerome Hoffman’s analysis as a paid subscriber to one of his publications for many years. He also was an early debunker of the ADA promulgated idea that everyone’s hemoglobin A1c needed to be as low as possible, which was subsequently borne out by multiple trials showing net harm with very aggressive glycemic control efforts.
However, although I thought his criticisms of the thrombolytic treatments for stroke were very well reasoned, as you mention in your post, thrombolytic treatment for stroke has spread widely, and it’s considered heresy to challenge it. In my opinion, part of the reason for that is financial. Hospitals advertise these “stroke centers” and anyone who comes in with any neurological symptoms at all, even if non localizing and extremely unlikely to be stroke, gets at least 3-4 advanced imaging studies and there is a very low threshold for giving thrombolytics and putting someone in the ICU. I’ve seen thrombolytics given for slurred speech in people who were objectively intoxicated with alcohol based on blood test results. And when people return repeatedly with vague neurologic symptoms, the same workup is repeated over and over.
This is a huge money maker for hospitals and others. It’s also often a terrible use of resources and can be dangerous.
I am not a neurologist and not research-minded, but a practicing ER doc for >30 yrs. I have practiced through the entire TPA era. ER docs were the first to seriously question over and over again the utility and efficacy of thrombolytics in ischemic strokes. Unfortunately, the pharmaceutical dollars have captured and hold captive the C-suites who make the decisions of clinical practice, not the doctors on the ground like us. That train left the station long ago, and even an entire specialty (EM) couldn't stop it. Every stroke patient is unique and a "fight" to do the right thing. I've disagreed openly with neurologists, and yet they still give the lytic. There is so much spin, and the public just buys it. The neurologist tells us there are more lawsuits for NOT giving the lytic than for withholding it. To intimidate us EPs into giving lytics? To follow blindly the standard of care of the local community? To do what the C-suite has established as standard of care?? I rejoice when I find a contraindication to lytic therapy.
Mandrola is right to remind us that the original NINDS trial had weaknesses — the groups weren’t perfectly balanced, and that matters. But he takes that too far. He treats those flaws as if they invalidate the whole case for thrombolysis. That’s not true.
When you look at the bigger picture — all the trials put together, especially the patient-level meta-analyses — the evidence still shows that clot-busting drugs help if they’re given quickly to the right patients. The benefit is real but time-sensitive, and it comes with bleeding risks.
So the sensible position is not “throw thrombolysis out,” but:
Use it early (the earlier the better).
Select patients carefully (don’t overgeneralise).
Keep auditing your own outcomes (because benefits are modest and risks real).
That’s the balanced truth: not heroic, not nihilistic — just precise, time-critical, and monitored.
As a retired doctor, I agree. As with almost everything, the evidence is nuanced. Each situation is unique. Each patient is unique. Their valuation of the risks and benefits will differ. Accordingly, there is no “right answer”, just “what this patient chose for themself at this point in time, knowing what they knew.”
Unfortunately, where time is of the essence, a 20 minute conversation about the risks and benefits of thrombolysis may work against them. So, what do doctors do? They oversimplify the choice. Instead of shades of grey, it’s presented as black and white. “You get thrombolysis, you’ll get better. No thrombolysis, you’ll stay the same or get worse.” The oversimplification ends up reinforced by the payment systems, the hospital bureaucracy, and the legal/malpractice system.
Sadly, once it’s standard practice, it becomes unethical to set up a trial in which one group doesn’t get thrombolysis.
I don’t know how to create a clearer message than what I’ve been trying to convey for the past 20 years, e.g. https://hbiostat.org/bbr/change. Difference from baseline is not appropriate for ordinal stroke outcome scales!! Do a proper analysis and show us the result. Don’t look for differences at baseline but use pre-specified prognostic factors for adjustment. Use an ordinal regression model in the raw ordinal outcome scales adjusted for baseline. For the baseline version of the outcome scales, adjust flexibly by not assuming linearity.
When in doubt remember: the proper analysis of a parallel group design is to compare parallel groups, not change from baseline.
Another note: median outcomes are not appropriate when there are many ties in the data.
Thank you for this analysis but left with the question of. ….What do we do??? Watchful waiting would be tough on pt and doc
Not as tough as an intracranial hemorrhage.
I followed Jerome Hoffman’s analysis as a paid subscriber to one of his publications for many years. He also was an early debunker of the ADA promulgated idea that everyone’s hemoglobin A1c needed to be as low as possible, which was subsequently borne out by multiple trials showing net harm with very aggressive glycemic control efforts.
However, although I thought his criticisms of the thrombolytic treatments for stroke were very well reasoned, as you mention in your post, thrombolytic treatment for stroke has spread widely, and it’s considered heresy to challenge it. In my opinion, part of the reason for that is financial. Hospitals advertise these “stroke centers” and anyone who comes in with any neurological symptoms at all, even if non localizing and extremely unlikely to be stroke, gets at least 3-4 advanced imaging studies and there is a very low threshold for giving thrombolytics and putting someone in the ICU. I’ve seen thrombolytics given for slurred speech in people who were objectively intoxicated with alcohol based on blood test results. And when people return repeatedly with vague neurologic symptoms, the same workup is repeated over and over.
This is a huge money maker for hospitals and others. It’s also often a terrible use of resources and can be dangerous.
No treatmet = No payment. Trial replication should be standard of investigative practice, but we know who controls those purse strings.