Curiosity Over Pride (FYI: To comment, send an e-mail to

Sunday, March 21, 2010

The Butterfly effect, Part III... The collective is not of one mind

... This is Chapter III in a multi-chapter series. If you happen to be coming to this tale in the middle, please begin here, with Part I.

So we left with the idea that in order to statistically find small effects from treatments, very large study populations are often required. Further, in the issue of tPA for stroke in 2000, 7 studies showed no benefit but an 8th and larger study did. And on the basis of this single but more "powerful" study, tPA was approved by FDA.

Further, as I asked you earlier to walk in the shoes of an emergency physician trained to interpret data and studies such as these, you should immediately recognize that the success of tPA in NINDS only suggested that the absolute benefit of tPA for stroke was probably quite small. For if there was a benefit to using tPA, still the benefit couldn't be very large or it likely would have surfaced in the seven earlier but smaller studies. The fact that it took a very large study (powered with enough patients) to find it suggested from the beginning that the effect was not very important.

Later in this post I will share with you the actual statistical outcomes of what happens to patients who are given tPA for stroke. You can decide for yourself whether we were justified in looking at the issue this way back in 2000.

Still, NINDS was a glimmer of hope.

So I now want you to shift your viewpoint from just mine- a single emergency physician- to the entire emergency medicine physician collective. And if you were unaware, the collective medical mind is anything but unified, and this disunity exists for very good reason. For the right thing to do is rarely obvious and we all have learned that statistics can be just that: statistics. Effects witnessed today in one or more studies often vanish under closer scrutiny as we come to learn we were fooled yet again by randomness.

Many physicians in the collective are cautious when new drugs are approved, especially when prevailing data suggests a drug's benefit is small. And as I said earlier, you can decide for yourself whether these physicians were justified in being conservative when I share what we now know 10 years later.

So the approval of tPA by FDA, as any physician with 2 neurons could have predicted, led to a fracturing of the collective medical mind as some physicians (myself amongst them) decided to give tPA to their patients while others opted for caution and took a watch and wait approach. I will not bore you with the details of these disagreements, though they legendary amongst my peers, but let me say they were quite robust.

But here is the key point of this entire post, all the while this debate was happening, the simple fact remained that the FDA's decision to approve tPA for stroke was akin to a butterfly flapping its wings. And while emergency medicine physicians debated these issues within the confines of their own private little world with what they thought were well defined boundaries, a lot of other people in the health care system were also effected by the ripple created by the butterfly's flap. And these other people, working in their own private worlds both outside and unbeknown-st to the emergency medicine community, were also taking notice of the ripple. And this ripple was beginning to make them stir and make worlds suddenly collide.

I shall end Chapter III here and will discuss the beginnings of these boundary collisions in Chapter IV as I have time to write it. But before I proceed, I want to make one last point re: tPA and stroke. Please understand that tPA can only be used on 2% of stroke patients. Sadly, it cannot be used on the other 98% or it will harm them for reasons I will not go into in the interest of time.

Please remember this 2% figure as you continue reading future chapters.

And, as I promised, I will now share what we now know about the cost-benefit of tPA in stroke. Looking at tPA from the perspective of a patient with a stroke, sadly this is what I tell my patients and families all too often:

1. If you have a stroke and DO NOT get tPA, you have a 40% chance you will have a good recovery within 3 months
2. If you have a stroke and receive tPA, you have a 52% chance for a good recovery within 3 months- this means there is a 1 in 8 chance the drug will benefit you if you receive it and a 7 in 8 chance it will not
3. If tPA is used, you have a 1 in 16 chance of developing a hemorrhage in your brain (6% absolute chance of hemorrhage related to tPA)
4. Despite the increased chance of bleeding, there is no evidence tPA make any difference before 3 months.
5. Your risk of death from stroke is approximately the same whether you receive tPA or not.

In other words, if an emergency physician treats 100 stroke patients who meet tPA criteria, e.g.- arrive within 3 hr, and have no absolute or relative contraindications, 12 will have absolute benefit and 6 will have a serious intracranial hemorrhage. And 2 patients will have minimal or no deficit for every 1 patient that gets a hemorrhage. We help a few patients (1 in 8) but we harm slightly fewer. It is not zero-sum, but some definitely lose as a few more win.

So I ask you, is this a wonderful drug?

And having shared this data with you, I have a few questions for you:

1. Is tPA a big deal?
2. Do you feel there is a lot of benefit for the risk?
3. If you had a stroke and qualified for tPA, would you take it?
4. Was the suggestion that the benefit was small in 2000 subsequently proved correct?
4. Were the physicians who took a cautious approach justified knowing what we know now?


Thai said...

For comments

Dink said...

"I will not bore you with the details of these disagreements, though they legendary amongst my peers, but let me say they were quite robust."

Ah, to have been a fly on a wall..

Two percent. Two percent.

You've presented quite a bit of food for thought. I hadn't realized that it could only be used on 2% of stroke pts. And that a pt's 3-month outlook only changed from 40% to 52% for a good recovery. I was under the impression that if 1) you met the thresholds (time, etc.)and 2) lived through it, you were likely to have a great recovery chance (i.e. over 52%).

Hmmm. Must recalculate perceptions...

My paternal grandfather had around seven TIAs before he passed away. Strokes freak me out. I liked the idea that tPA was around.

Hell, I still like it and would still probably go ahead with it. I assume some fine-tuning is being tested so it can be delivered more specifically (more laser, less shotgun)?

Of course, I'm blathering. Still incorporating the new data you presented. Good work with the thought-provoking posts!

Debra said...

God is still at work, throwing the dice, whatever you want to call him, and even if you don't want to call him.
To a certain extent, I feel sorry for the doctors and the patients in this context.
I don't think people REALLY understand the way statistics work. (Particularly when you see how much problems we are still having with... multiplication, for example...)
And of course the real problem remains that people are looking for TRUTH. And truth is all too often perceived as an ABSOLUTE, true in all situations, true for all people.
The world does not work that way, even though we are DESPERATE for it to.
To me, it is VERY debatable whether this drug is really of any use, and it is STILL entirely possible that what are perceived as the benefits from it, in the cases in which it has been used, have not resulted from... other sources, for example.
It is not for nothing that I go on in this saloon about the importance of the placebo effect.
Not at all.
"Data" ? Not a.. WORD I particularly like either.
Thinking about your posts, Thai, has given me an itch to address the thorny problem of just WHAT science is...

Debra said...

Correct "much" to "many" in the above post.

Debra said...

Oops, I meant comment. Sometimes I mix my.. words up.

Thai said...

Most stroke patients are either unaware when there stroke symptoms first began (say they were asleep and awoke with the symptoms or they occur in a region of the brain that affects perception, etc...) and so we are unable to determine if they are within what is now a 4.5 hour window. Or they have a contraindication that throws the risk-benefit calculations decidedly in the direction of drug risk outweighing drug benefit. Or they simply cannot reach the hospital in time as they are alone and paralyzed when they have the stroke. Or they...

When the reasons we can't give the drug all play out, only 2% of stroke patients are eligible.

So the reality of tPA in stroke today is that it is on average a small benefit on a small number of patients. If you are lucky enough to be one of the few it helps, wonderful. But remember, the math is 0.125 x 0.02= 0.25% of all stroke patients will benefit from this therapy.

Again, it is good where it is good, but for the other 99.75% of patients, nada.

Remember this idea of absolute benefit (which is very different than relative benefit) as the story of chaos continues as it is important to put into context everything else that happened thereafter in terms of "what on earth are we doing? Have we all lost our marbles?"

I hope it will explain why I see health care as a bubble and why we are all responsible for this bubble... But I am letting the story get ahead of me for a moment.

Deb, there are still physicians who think the data can still explained by placebo, etc... but they are smaller and smaller in number.

Either way, very few physicians I know have looked at the big picture enough to admit that tPA is no miracle drug in the way cholesterol or blood pressure meds have changed things.

Dink said...

"Boundary conditions were about to intersect."

I look forward to this part of the tale too, but I'm still mesmerized by the tangential subject of the tPA itself.

"Either way, very few physicians I know have looked at the big picture enough to admit that tPA is no miracle drug in the way cholesterol or blood pressure meds have changed things."

But but but!!! Insulin is both a "miracle" (I presume your use of the word translates into large numbers of lives improved)and a potential killer. And heparin. I heard that warfarin can be purchased at Home Depot (in the rat poison section). And adenosine....whoo doggies!!

Within current scope tPA can only be used on a select few, but that's true of many cures. Its mechanism of action (as I understand it) is perfect for the problem if we can get it to the right place in enough time.

Anyway, this is your post and I want you to tell it as intended (so please feel free to ignore my distracting comments).

Thai said...

I'll answer whatever other questions you have but I am not sure I understand what they are.

Dink said...

"I'll answer whatever other questions you have but I am not sure I understand what they are."

Bad form on my part. I was protecting my emotionally-pregnant construct instead of listening to new data. Too attached to the hypothesis. Please proceed with part IV.

Thai said...

May be a while, kind of busy of late

The Most Fabulous Objects In The World

  • Hitchhiker's Guide To The Universe trilogy
  • Lord of the Rings trilogy
  • Flight of the Conchords
  • Time Bandits