It is the worst of times. Science is suffering an identity crisis. The world is in dire need of science. Science isn't used to being rushed. "It is a giant and slow churn", said a friend once, "and spews a breakthrough once in a while". Is it possible to make the process faster? That's what everyone is wondering. And praying. And waiting, eagerly. Science isn't used to getting this attention.
"Coronil is 100% effective", said Patanjali folks. "Favipiravir is 88% effective", said Glenmark folks. How to know the truth? Seeking truth has never been easy. Never has it been easy for journalists, scientists, or the common person. In some sciences there are multiple truths. Is medicine one of those sciences? Can there be a single truth in medicine?
I won't use words like epistemology and ontology in this post. (Because I still can't remember which is which). But the question is essentially two:
1. Is there a single truth?
2. Is there a way to know the truth?
I believe medicine is a dangerous subject because of these two questions. Biology is extremely contextual. A drug's effect on a person with any particular infection can be influenced by a thousand factors including - that person's biology, the day, where that person is, what that person is eating, what other medicines that person is taking, the virus that infected them, all the infections they've had in past, other diseases they currently have, the health of their body organs, and so on.
When there are so many things that keep changing, how do we know whether a drug is going to be useful for a person or not? Most of medicine today is an approximation. Many drugs are used because when given to n random people it worked better than it not being given. A gross measurement, if you allow me to call it. Put something in a balance and see which side is hanging lower.
Not that medicine is all guess work. He he. There are some theories. There are some "well-known" pathways. There are some molecules which we understand. There are some we don't. There are some drugs we know act on some molecules in some of these pathways. Sometimes we don't understand some parts of how a drug acts, but we fill in those gaps with the "random" trials as described above.
For example, let us take Paracetamol which is a drug commonly prescribed for fever. And the only drug that many people need during COVID (and Dengue, and many other viral fevers). We don't know how exactly it works. But we have a rough idea on the pathways that it affects. We also have very rich clinical experience in using the drug successfully for fever.
The reason why we don't rely a lot on theory in medicine is that we don't have a lot of theoretical understanding about the biology of our body. We do know a lot. But there are still so many known unknowns. And who knows how much unknown unknowns.
We know a bit about molecules called "interleukins". We seem to know about a molecule we call Interleukin 6. It seems to have a role in acute immune responses. It may very well make sense to somehow block IL-6 to decrease the damage that could be caused by what is called a cytokine storm (which, as it sounds, is a storm that wrecks havoc inside the body) in sick COVID patients.
We seem to know about a class of drugs called monoclonal antibodies. These are molecules (which can be natural or artificial) that target specific kind of molecules. There are some mAbs which seem to be able to target a type of cell called CD6 cells, including Itolizumab.
There seems to be some data that mAbs that act on CD6 can decrease the amount of IL-6.
Now, here is the deal. If Itolizumab can act on CD6 and decrease IL-6 and if IL-6 has a role to play in cytokine storm in COVID, then the inference could be drawn that Itolizumab can help sick COVID patients not die. That's the theory.
But the problem with medicine is that theory doesn't always work. And sometimes what presents as reasonable with our current understanding of the body sometimes becomes dangerous when we actually try it.
As for Itolizumab, Biocon seems to have given it to 20 patients with COVID and moderate to severe respiratory difficulty. And they all seem to have survived. Of the 10 they didn't give it to, three people apparently died. I'm sure they're doing this study on more people at the moment.
According to them this is "statistically significant". I don't have a very deep understanding of statistics. Here, let me do the math.
The way I read it is that based on that data we can be 95% sure that if someone with moderate to severe COVID-19 ARDS takes the drug their chance odds of survival is somewhere between 0.8802 fold to 415.9060 fold the chance odds of their survival without taking the drug.
Didn't I tell you this is the worst of times?
Update: Don't look at my math. That was not the point of this post. Also, my math sucks. Here is why:
At a sample size of 30, the power of this study is like 30% which means it is completely unreliable. I think. I don't know.
Update 2: As per this article, and as per my understanding of beta, if p-value is already acceptable, then it doesn't matter whether beta is high as all that power makes sure is that we don't miss the effect when there is an effect.
But then, am I confusing myself because in this study the effect of the drug is protective? I am 70% sure that the power of this study is not to be worried about.
Update 3: Maybe the contradiction is resolved if we consider this as a type S error.
No comments:
Post a Comment