The occurrence of transient neurologic symptoms (transient ischemic attacks, TIA) commonly heralds the later occurrence of a stroke (with its associated non-transient neurologic damage). In 1707 patients with a TIA, a stroke took place in the subsequent 90 days in 14% of the 235 who received warfarin anticoagulation therapy at the time of TIA and in 10% of the other TIA patients (p=0.04). The latter patients generally received aspirin or no specific therapy.
The authors of this study warned that despite the excellent quality of information on treatment received and on outcome events, the study may not provide valid data on the (lack of) efficacy of warfarin anticoagulation in persons with TIA.
What do you believe the is the primary basis for their cautious interpretation?
What kind of stroke are we talking about? Embolic or hemorrhagic or both? Warfarin may prevent embolic strokes but may contribute to hemorrhagic strokes. Maybe the study shows that we have to screen very well before prescribing warfarin. Also, what were the pre-existing conditions or co-morbidities? For example, atrial fibrillation is a big risk factor for embolic stroke. Warfarin is often given for atrial fibrillation. Warfarin is difficult to manage to obtain adequate anticoagulation but not over anticoagulation. Blood levels are hugely affected by diet and concomitant medications administration. Better pt education and tighter controls on blood levels may
ReplyDeleteinfluence subsequent stroke.
Good questions Mom. However, the answer to the question does not rely on the type of stroke and I frankly have no idea! You second question about pre-existing conditions and co-morbidities is even better and on the right track to the answer.
ReplyDeleteRemember, patients who received warfarin had a worse prognosis--why would that be?
Some reaction with something else? The study didn´'t control for other factors?
ReplyDeletePs - I don´t understand what you intelligent people are talking about.
ReplyDeleteAshley's comment that this study didn't control for other factors is also in the ballpark, but not quite there.
ReplyDeleteAny other ideas?
One must control for a variety of potential variables to ascertain the validity of results. What are the algorithmic formulas used for deciding whether to prescribe warfarin or aspirin? For example, age, gender, race, severity of TIA--or type of symptoms presented. Medical variables are also important to consider. Hypothetically, for example, warfarin might be prescribed in the presence of severe atherosclerosis or hypertension, whereas aspirin was more likely to be prescribed in the absence of these risk factors. In such a case, it may be that without Warfarin in the high risk category, the stroke rate would have been much higher. Thus, while seemingly a "bad" choice Warfarin would actually be a life saving choice for many. Matters such as dosage and adherence might be critical. Finally sample size, provider characteristics, and geographic location/ prescriber characteristics may influence. Ah the joys of making sure one does not automatically assume--sometimes mistakenly that correlation and causality are the same thing.
ReplyDeleteJust for the heck of it, let's throw in a confounding variable. Hemorraagic strokes (bleeding into the brain) are much worse than embolic (blockage in a vessel). Especially, with the stroke buster drug TPA that can reduce the severity of an embolic stroke (but should never be given for a hemorragic stroke), one might want to look at % of stroke type attendant to each drug. It may be that, if % or total strokes are higher (but milder) with drug A, but drug B strokes are much more severe or fatal, drug A might be desirable.
Dad--using Mom's account
My brain hurts
ReplyDeleteGood answers, and you're all right! Well, at least Mom, Dad, and the ever so beautiful Ashley get gold stars.
ReplyDeleteEmily gets a black star :(
This is a classic example of what an epidemiologist would call "confounding by indication". What the heck does that mean, you might ask. Well, let's start with confounder. A confounder is a third variable that is associated with an exposure and disease. In this example, patients who received warfarin had a worse case of TIA to start with. The worse severity of disease is what caused them to received warfarin in the first place (here, disease severity is the confounder and warfarin use is the exposure). Disease severity should also be associated with higher risk of stroke.
Therefore, because sicker people were at higher risk for stroke and were more likely to take warfarin, any protective effect of warfarin would be obscured when comparin warfarin and no warfarin patient groups. Back to the phrase "confounding by indication"....I already explained why disease severity was a confounder, we use the phrase "confounding by indication" because the confounder indicated the treatment (exposure).
If we really wanted to evaluate the efficacy of warfarin at protecting against stroke we would want to take a group of patients who had had TIA, and randomly assign them receive warfarin or not receive warfarin and then follow them up for stroke. By using random assignment, disease severity would be balanced in the two groups and we would eliminate this confounding by indication.
I'll share the black star with Emily.
ReplyDeleteOk, so I think we all had the right idea, we just didn't have the right words to say it, like the ever so smart Bubba. The term "counfounder by indication" confounds my brain. It's not a very intuitive phrase.
ReplyDelete