Hey #MedTwitter

Here are the basics of survival curve and hazard ratio interpretation. Trust me it's conceptually easy. Dreaded terminology will be decoded.

Ever wondered what a hazard ratio really means? What is a survival function? What is cox proportional hazards?

🧵 Key point #1:

Time to event analysis. We look at the time it takes to experience an event in a particular group.

Three options:
1⃣Have the event
2⃣Don't have the event
3⃣Lost track of participant so we assume they would have done as well as everybody else on average (censor)

Hey #MedTwitter

Here are some basics on propensity score matched analysis

What is this? How does this help with confounding by indication? Why is this not a substitute for randomization?

Tried to keep it brief but have a lot to cover...

🧵 In medicine, we want to know how much does treatment A improve outcomes over treatment B?

Ideally, we run a randomized control trial (A vs. B) which by design will balance known and unknown confounders

But this may not be logistically feasible so what can we do instead?

Hey #medtwitter

Here are the basics on unequal randomization in trials

Why should we do 1:1 instead of 2:1 or 3:1 in phase 3 RCTs?

🧵 First why do we randomize?

Prevent bias which is a major limitation in observational studies:

1⃣ Balance known and unknown confounders
2⃣ Prevent immortal time bias
3⃣ Prevent selection bias
4⃣ Contributes to validity of statistical tests