A couple months back, JAMA published Jeff Saver’s time-to-tPA association, derived from the Get With the Guidelines-Stroke Registry. My critique of this article remains essentially the same, but this past issue JAMA published an edited version as a response to Dr. Saver.
While many of you will find the content behind the JAMA paywall, the basic gist of Saver’s response to my letter is:
- tPA-treated stroke mimics are not included in the GWTG-Stroke registry during the study period. It is an optional reporting field.
- The tiny bit of data from original NINDS showed rapidly-improving patients from the placebo group were equally distributed between 59 to 90 minutes and 91 to 180 minutes (3/145 vs. 4/167).
- The onset-to-treatment-time and outcome association was consistent between NIHSS 0-14 and NIHSS 15-42, and therefore any effects from TIAs must be insignificant.
In summary: they don’t really know how many TIAs/stroke mimics/aborted strokes there were, but they can draw lines around some numbers to support their conclusion.
These are reasonable responses from their standpoint. I would point out, however, the NINDS group made specific efforts to exclude “rapidly improving” symptoms from inclusion, probably resulting in extreme extra care in preventing enrollment of TIAs. This population is unlikely to be generalizable to current tPA-happy practice across medicine. They do not, unfortunately, address the problem of “aborted stroke” coding, which further received a shout-out on Twitter from Dr. Hsia:
@emlitofnote hit the mark in @JAMA_current with citing our averted #stroke paper: #tPA-treated does not equal #stroke http://t.co/6aE7bFlm7k
— Amie Hsia (@DCStrokeDoc) November 8, 2013
I’ll try and get a peek at the GWTG-Stroke data myself, but with the literature documenting anywhere from 2% to 31% stroke mimic treatment rate – with the actual true rate probably exactly in the middle of that range – it’s critically important to uncover risks and costs of inappropriate tPA administration. If the GWTG-Stroke registry is a true quality device, it needs to do a better job of measuring adverse events.
Or, you could end up like Houston Texans’ coach Gary Kubiak, who received tPA last week for a stroke mimic (currently reported as having had a TIA – which I find dubious, but I have no personal knowledge of the case).
“Acute Ischemic Stroke and Timing of Treatment”
http://www.ncbi.nlm.nih.gov/pubmed/24193085