The Risk Was in the Chart Before the Drug Was
A short read on what a bleed after thrombolysis (treatment aimed at dissolving the clot causing a stroke) can honestly be called, and the arithmetic both sides skip. A composite of cases I see repeatedly, with no detail drawn from any actual matter.
An expert report lands on your desk. A patient received alteplase (the clot-dissolving drug) within the treatment window, bled into the brain, and had a bad outcome. The report calls the hemorrhage an unforeseeable complication of a properly indicated treatment and notes that no contraindication was present. That last part is accurate.
Is the report right that nothing was knowable? It is right that no exclusion was breached, and wrong that the risk was unknowable. Three validated scores estimate the probability of symptomatic intracranial hemorrhage (bleeding into the brain severe enough to worsen the patient neurologically) from information already recorded before the drug is given. The SEDAN score uses blood glucose, age above 75, a stroke severity score of 10 or more, and two readings from the initial scan: early infarct signs and a dense artery, which is the clot itself visible as a bright vessel. Across its derivation cohort, it separates a hemorrhage rate of 1.4 percent at zero points from 33.3 percent at five. A second score reports 2 percent at zero points against 44 percent above three. None of those inputs require hindsight.
Does that make the report right, that the treatment was a departure? No, and the opposite error starts exactly here. None of these scores are eligibility criteria, and none were designed as such. They discriminate at roughly 0.61 to 0.77, on a scale where 0.5 is a coin flip and 1.0 is perfect, which stratifies groups and does not predict individuals. SEDAN's authors describe it as supporting decisions in high-risk patients; the authors of the second say it needs prospective confirmation before it can be used in clinical decision-making at all.
Consider the mirror image. Another report reads a high score as proof the drug should have been withheld, reasoning backward from the bleed to the decision. In the highest-scoring group, roughly two in three patients did not bleed, and the treatment maintained its expected benefit: in the pooled analysis of nine randomized trials, treatment within three hours left about 10 additional patients per hundred alive and free of disability. Withholding treatment based on a risk score is not the conservative choice. It is a different decision, made on an instrument built for another purpose.
So the question underlying an opinion about post-thrombolysis hemorrhage is narrower than either report makes it appear. Not whether the bleed was foreseeable in the abstract, but what the chart recorded before the drug was given and whether anyone did the arithmetic at the time.
Two things follow for a reviewer. A hemorrhage called "unforeseeable" is a claim about one specific chart, and the inputs that test it are glucose, age, severity, and the initial scan, all recorded before treatment. And a risk score is evidence about the conversation that should have happened, not about eligibility.
The primer in this month's Writings sets this decision in its full taxonomy, with the registry numbers of eligible patients never treated.
— T.B.K
Tobias B. Kulik, MD, FAAN
CorteXion LLC · Scientia et Veritas
The Primer in this month's Writings
Thrombolysis for Stroke, Part II: Where the Decision Fails — cortexion.co/writings/iv-thrombolysis-primer-part-2
Primary source
Strbian D, Engelter S, Michel P, et al. Symptomatic intracranial hemorrhage after stroke thrombolysis: the SEDAN score. Ann Neurol. 2012;71(5):634-641. doi:10.1002/ana.23546
A note on cadence. The CorteXion Brief is monthly, occasionally twice in a month when the docket warrants it. One question per issue, no drip sequences, no scheduled series. If you prefer not to receive future issues, you may unsubscribe below.
