#PerProtocol - Twitter Hashtag

almost 4 years ago

Reason No. 56726 to come to @bcm_gihep for fellowship: you can make it to Uchi happy hour after continuity clinic! @NathanLindnerMD @Ashwin_RaoV @BoomBoomKO #IntentionToDrink #PerProtocol #GITwitter

ShaleenVasavada's tweet photo. Reason No. 56726 to come to @bcm_gihep for fellowship: you can make it to Uchi happy hour after continuity clinic! @NathanLindnerMD @Ashwin_RaoV @BoomBoomKO #IntentionToDrink #PerProtocol #GITwitter https://t.co/LZjndD4Pr5

2

24

5

0

Neghae Mawla, MD, FASDIN @NMawlaMD

over 4 years ago

@wasse_m @aishaikh @vandyniyyar @Nephro_Sparks @Asdin @Rush_Nephrology @TerryLitchfield Is there a #datadriven answer? I think it’s all opinion, comfort, convenience driven #perprotocol 😂

1

10

0

Miguel Hernán @_MiguelHernan

about 7 years ago

@epipubs @BUSPH @HarvardChanSPH @AmJEpi @societyforepi Check freely available paper & software to compare effect estimates from randomized and observational studies. The GFORMULA_RCT macro implements the #gformula in @SASsoftware to estimate #perprotocol effects in randomized trials. https://t.co/vk4cWnEkYp https://t.co/eXIDWwdDIS

0

5

3

0

Dr Ellie Murray, ScD

@EpiEllie

over 7 years ago

Since the authors want to know what happens if people *actually* follow these treatment plans, their causal contrast of interest is the #perprotocol effect—that is, the effect of everyone had followed each of the specified treatment plans.

1

6

2

0

Dr Ellie Murray, ScD

@EpiEllie

about 8 years ago

Time is money, so it’s time to move on: next on the list is some data analysis. For this project, I’m using the parametric g-formula to estimate a #perprotocol effect. “Parametric” sure returns some trippy gifs!! 😂🤪

1

3

0

Dr Ellie Murray, ScD

@EpiEllie

about 8 years ago

Nice example of why it’s hard to use instrumental variables for #perprotocol effects in #RCTs: “there is no right number to divide by because the longitudinal nature of the instrument-exposure relationship cannot be captured by one number.” @ja_labrecque_

1

8

3

1

0

Dr Ellie Murray, ScD

@EpiEllie

about 8 years ago

Fantastic thread👇🏽by @ja_labrecque_ explaining his new paper on how to think about Mendelian randomization and instrumental variables when your treatment or exposure varies with time. Important read for anyone interested in #MR, #IV, or #RCT #perprotocol effects!

0

10

2

0

Dr Ellie Murray, ScD

@EpiEllie

about 8 years ago

Check out thread for ur responses to the types of qs we asked #pragmatic trial investigators👇🏽 U ranked higher: incl/excl criteria, power issues, unbiasedness of ITT, and “realness” of #perprotocol Plus, >50% of u have heard of or done post-rand adjustment for #perprotocol!

Dr Ellie Murray, ScD

@EpiEllie

about 8 years ago

Let’s talk about the trialists. We interviewed 5 researchers with expertise in #pragmatic trials, and then sent a survey to all (24) PIs funded by @PCORI under their #pragmatic trial mechanism. Half responded our survey request (n =12).

1

0

Dr Ellie Murray, ScD

@EpiEllie

about 8 years ago

Tho #perprotocol effects are useful to patients & investigators, only ~1/3 trials conducted a per-protocol analysis. Worse, most of these used restriction to adherers without adjustment for confounding (baseline *or* post-rand). That’s not great: these estimates are likely wrong.

1

0

Dr Ellie Murray, ScD

@EpiEllie

about 8 years ago

Despite interest, many of our investigators were not familiar with methods for estimating the #perprotocol effect that require adjustment for post-randomization covariates. How about twitter? Do you know how to estimate #perprotocol effects using post-rand covariates.

1

0

1

0

Dr Ellie Murray, ScD

@EpiEllie

about 8 years ago

All interviewed & 7/12 surveyed PIs *also* planned to estimate #perprotocol effects. Their top reasons👇🏽👇🏽. Which is your top pick?

1

0

1

0

Dr Ellie Murray, ScD

@EpiEllie

about 8 years ago

Patients find subgroup effects, absolute risks, and #perprotocol effects (when they intend to adhere) valuable for shared decision-making. But what do trial investigators think? And how well do #pragmatic trials do at providing this information? Paper: https://t.co/AMhcFEHMCr

EpiEllie's tweet photo. Patients find subgroup effects, absolute risks, and #perprotocol effects (when they intend to adhere) valuable for shared decision-making.

But what do trial investigators think? And how well do #pragmatic trials do at providing this information?

Paper: https://t.co/AMhcFEHMCr https://t.co/ObYp6M9f8A

1

5

4

0

Dr Ellie Murray, ScD

@EpiEllie

about 8 years ago

Curious about when #perprotocol effects might be useful? Here’s a 4 tweet vignette to help you think about it👇🏽👇🏽

Dr Ellie Murray, ScD

@EpiEllie

about 8 years ago

For fun, let’s see how you respond! Imagine you’ve gone to your doctor because of concern over cholesterol levels. She tells you about a new drug which worked just as well as the standard drug in a recent trial, with same cost & and side effects. Which do u choose?

1

0

1

0

Dr Ellie Murray, ScD

@EpiEllie

over 8 years ago

Some thoughts on adherence and #perprotocol effects in the #ProACT trial ht @GermsAndNumbers for the suggestion @NEJM https://t.co/U73dOgBEh9

1

14

4

8

0

Marcio S Bittencourt @MBittencourtMD

over 8 years ago

I gave a talk about why #ITT is less biased than #perprotocol analysis for Med students from @hosp_einstein just yesterday. #CABANA just came as a great example for the class...

1

10

1

0