Dr Foster's admission data shows a mortality difference in stroke
between units that carry out scans within 48hrs and those that do not. The
authors state this matches the benefit from early scans in clinical
trials, and so claim that this is evidence of a comparable benefit in
clinical practice.
My view is that the data provided prove the opposite, as no allowance
has been made for selection bias, which explains some, and possibly all,
of the perceived benefits.
(Palliative patients will not be referred to regional scanning units,
while those with a good prognosis will be)
The differences in mortality data may be the same size, but it is
poor science to assume that they have the same cause, without considering
alternative hypotheses.
Competing interests:
None declared