1/ To #reporters, #policymakers and #execs out there, here is something you absolutely need to know about interventions that claim to reduce hospitalizations or healthcare costs:
Many of these studies are misleading.
Many of these studies are misleading.
2/ To understand why, we need to- deep breath- do a little science. I’m going to quickly walk through a concept called regression to the mean.
goo.gl/images/mmHBgd
goo.gl/images/mmHBgd
3/ Most things in life - blood sugars, hospitalization rates, fantasy football stats- do not remain constant. Instead they go up and down 
4/ If you pick a group of people who are outliers (ie those with high hospitalization rates or high blood sugars) and do absolutely nothing at all, their outcomes will improve over time.
It’s the statistical law of gravity.
It’s the statistical law of gravity.
5/ Lots of studies in healthcare start with a group of outliers, put them through an intervention and then say: Wow! We reduced hospitalizations by 50%!
Nope. Just regression to the mean.
Nope. Just regression to the mean.
6/ I constantly see studies - and headlines and huge policy decisions- based on before/after studies without any comparison group.
So we are investing in “solutions” that probably don’t really work.
So we are investing in “solutions” that probably don’t really work.
7/Look for comparison groups or ideally randomized trials. If your not sure whether to trust a study- ask me or any other research expert.
This matters...a lot.
This matters...a lot.
@threadreaderapp unroll please
