np trend or p of trend

[Mirentxu Alberdi]

Hello, I have a problem with "np trend" in adjusted model.
I want to know the association between the consumption of ultra processed food and macrosomia in children, so I am doing a logistic model
In the case of macrosomia and consumption of ultra-processed foods, my independent variable is consumption of ultra-processed foods (categorical qualitative, divided in tertile 1, tertile 2, tertile 3), and my dependent variable would be macrosomia (categorical: yes/no). I want to do a np trend for the unadjusted model and for the adjusted model. I know how to do the np trend for the unadjusted model, but I am having problems with the adjusted model because I don't know how to add the covariate such age for example

This are my commands in Stata description

xi:logistic macrosomia i.upfterciles (unadjusted model) xi:logistic macrosomia i.upfterciles age (adjusted model)
np trend macrosomia, by (upfterciles) (np trend for the unadjusted model)

I don't know how to add age (covariate) for the np trend in adjusted model

[Joseph Coveney]

I assume that you're referring to the official Stata command nptrend. Your syntax for that is incorrect. It's also not clear why you're using the prefix xi:.

Regardless, in order to get a linear trend component of the association that's suggested by fitting a logistic regression model, you can use contrast postestimation command. Try something like the following.

// Unadjusted
logit macrosomia i.upfterciles, nolog
contrast p.upfterciles // Examine the linear component, identified as "(linear)"

// Adjusted
logit macrosomia i.upfterciles c.age, nolog
contrast p.upfterciles