The most widely accepted primary outcome in
major stroke trials is either “alive and excellent recovery (modified Rankin
Scale [mRS] 0 to 1)” or ”alive and independent (mRS 0to 2)”. I have heard from one
expert in this field that US FDA only accepts mRS 0 to 1 when evaluating a drug
for stroke.
Sometimes, the researchers may present "mRS 2 to 6" or "mRS 3 to 6". In fact, they are presenting the same thing with a different point of view.
Sometimes, the researchers may present "mRS 2 to 6" or "mRS 3 to 6". In fact, they are presenting the same thing with a different point of view.
For such a dichotomized definition of outcome,
it is hard to achieve a positive result. On the contrary, if we change the
primary outcome as “ordinal shift of mRS scores”, the statistical power will
increase and thus it becomes easier to achieve a positive result. (想想：學生「一定要考90-100分才算好」vs. 「成績只要有進步就算好」就可以明白) As I
know, many stroke trialists are interesting in such a methodology issue.
For example, Wardlaw has indicated in her Cochrane review of stroke thrombolysis
that, "Although the confidence intervals for poor outcome defined by mRS 3
to 6 and 2 to 6 overlap for analyses of all thrombolytic drugs and of just
rt-PA, these data suggest that choosing mRS 2 to 6 as the primary outcome may
provide a more positive trial result. Heterogeneity is present for poor outcome
defined as mRS 2 to 6 as for poor outcome defined as mRS 3 to 6. This suggests
that none of these dichotomous outcomes is specifically robust and that a more
cautious estimate of overall thrombolysis and of rt-PA effect, such as the
ordinal shift analysis, is wise. Note that some individual trials ’wobble’ from
being positive to not positive in going between the mRS definitions (some go
one way and some the other) but overall the trend is to more positive results
with mRS 2 to 6."
Other functional outcomes used in previous
stroke trial includes: Barthel index, Glasgow outcome scale, or even National
Institutes of Health Stroke Scale. Normally, one may expect that a patient with
a mRS score 0 to 1 should also get a Barthel index 90 to100. However, that is not
always the case in our previous experience. Although mRS and Barthel index are
both measuring the functional outcome, they in some extent are measuring
different aspect of neurologic function in a patient. Besides, the difference
between mRS score 0 vs. 1 may not equal that between mRS score 4 vs. 5. I think
it would be the same as for the Barthel index. Those points also puzzle most
stroke trialists in Taiwan, even in other part of the world.
