Discussion of Assessing Heterogeneity of Treatment Effect, Estimating Patient-Specific Efficacy, and Studying Variation in Odds ratios, Risk Ratios, and Risk Differences

effect-measures
precision
prediction
interaction
hte
#1

This is a place to discuss this blog article which shows an example of a formal assessment of heterogeneity of treatment effect, shows how to use penalized maximum likelihood estimation to get patient-specific efficacy estimates, and discusses odds ratios, absolute risk differences, and risk ratios.

This topic is also the place to discuss this blog article which shows how personalization of efficacy estimates can result in worse estimates for individuals if HTE is not large.

4 Likes

#2

Thank you for making & sharing this resource!

2 Likes

#3

i wonder then about some journals’ requirement for eg sex-specific estimates of the effect eg Circulation: “Please provide sex-specific and/or race/ethnicity-specific data when appropriate in describing the outcomes of epidemiologic analyses or clinical trials, or specifically state that no sex-based or race/ethnicity-based differences were present.” submission requirments. It’s not exactly clear if they’re alluding to a treatment interaction but i believe they are. As you note in the blog there is typically a serious lack of power, and thus the test it anti-conservative. The forest plot of ORs across subgroups that you describe would be a compelling visual however

regarding the OR, i noticed @timdisher challenged anyone on tiwtter to shake his faith: The odds ratio is the best effect measure of all time…

3 Likes

#4

The age-old NIH requirement was never quite appropriate because of the reason you mentioned Paul. This cannot be done reliably without borrowing of information, e.g., by putting a prior distribution on the interaction effect to assume that the response to tx for males is more like that for females than not. This Bayesian idea of having interactions “half in” and “half out” of the model is discussed well here.

1 Like

#5

I owe my OR love to @f2harrell and most of my day to day work being focused on whether or not HTE exists (as applied to systematic reviews). Will definitely be referring to this/adapting some of these visualizations for my own work.

2 Likes

#6

it’s off topic ie not binary outcomes, but i wonder how one evaluates HTE with a rank-based composite. I’m not sure i’ve seen much literature on that, especially if you want to power on the interaction. In the social science literature they talk about “the adjusted rank transform test”. I’m not familiar with it and wondered if anyone had experience with this, maybe the probability index is used for the visual display and simulations for power estimation at the design stage… power est for composites seem tenuous anyway

0 Likes

#7

Use the same methods as in the blog article but applied to a semiparametric ordinal regression model such as the proportional odds model (generalization of the Wilcoxon test).

1 Like