Much is made of retrospective comparisons derived from registry data in acute ischemic stroke. The most egregious of these attempt to match a subgroup of tPA-treated folks with a control group not receiving tPA.
This publication, from an Austrian stroke registry, pulls 890 patients from 54,917 with mild (NIHSS <5) symptoms to perform a comparison matched on many clinical features. In their matched comparison, 41% of tPA-treated patients were better off, 30% were identical on mRS, and 29% favored non-treatment. Therefore, these authors declare there is an overall shift towards favorable outcome with tPA.
What is wrong with drawing anything but hypothesis-generating conclusions from this data (or, any retrospective, observational data)? While the authors control for a few factors, there are many pre-existing comorbid conditions influencing treatment or non-treatment with tPA. For example, a patient with advanced metastatic cancer will almost certainly be excluded from treatment with tPA – and certainly have poor 3-month functional outcomes – but that comorbid state cannot not be captured by this registry. Then, just as we saw in IST-3, patients who are treated with tPA frequently receive more vigilant initial stroke care. Adherence to clinical pathways as well as the control of hyperglycemia, fever, and swallowing dysfunction have profound effects on subsequent death and disability far exceeding the effect size supposed for tPA. This registry, taking place over the course of a decade of evolving stroke care, cannot control for follow-up treatment and therapy.
This is, essentially, why all retrospective, observational comparisons are almost guaranteed to favor the tPA cohort.
Just as an added note – for anyone considering responding directly to the authors in the pages of Stroke – you will be charged $35 for typesetting for an electronic-only publication which then resides behind a paywall. Traditional scientific journals are for profit, not for science – that’s where FOAMed comes in!
“Thrombolysis in Patients With Mild Stroke: Results From the Austrian Stroke Unit Registry”
http://stroke.ahajournals.org/content/early/2014/01/30/STROKEAHA.113.003827.abstract