PMstep function

Post technical questions about making Pmetrics carry out a particular task.
Post Reply
Posts: 12
Joined: Mon Nov 10, 2014 10:45 am
Location: Barcelona (SPAIN)

PMstep function

Post by isegarra » Thu Apr 27, 2017 3:40 am

Dear all,

I need some input regarding the way PMstep function works and how to interpret its output. In essence, this function will identify possible differences due to the covariates (e.g. sex, weight, a condition, etc). If we have only one covariate the result may be interpreted straight forwardly.
However, how is it interpreted when there are two or more covariates? This could be a scenario including a group of male (M) and female (F) with the presence (1) or absence (0) of a condition which will give four different group populations: M0, F0, M1 and F1. In this scenario, if I want to see the potential effects of covariate sex and covariate condition, PMstep will come out with a p-value for each PK parameter: Is this value taking into account the four groups or it takes only (e.g.) M versus F regardless of the presence or absence of the condition which may affect the outcome differently in M and F?

A possibility may be to make an artificial covariate for each one of the four groups, e.g. cov GR being 1, 2, 3, 4, for each subgroup (M0, F0, M1, F1 groups) and then look at the p-value for cov GR for each PK parameter.

What would be the best way to analyze combined covariates?


Ignacio Segarra, PhD.
Clinical Pharmacy and Pharmacotherapy Research Group. Faculty of Pharmacy.
University of Barcelona.

Post Reply