what is STATA's command for "conditional" logistic regression in "multilevel" models?

[Danah Abdul]

I am trying to assess the association between a binary outcome and multiple independent variables.

The study design is a risk-set case-control study. I am hence using "conditional" logistic regression to develop "multivariable" model as the controls are matched with cases based on the time factor (risk-set).

However, I am also matching on general practice, so would like to adjust for clustering at the general practice level (level two) using "multilevel" modelling. Can I use conditional logistic regression in mixed/multilevel models? if yes, which command is used for that in STATA?

Does the multilevel model adjust for similarity between cases and controls (intra-cluster similarities) induced by matching? In other words, would using the "melogit" command in STATA alone bypass having to use "conditional" regression strategies?

[Clyde Schechter]

As far as I know, there is no analog to conditional logistic regression in multilevel models.

The way I would handle your situation is to have a three-level model. The individual observations are nested in the matched tuple (level 2) which are in turn nested in the general practice.

[Danah Abdul]

Clyde Schechter
The three-level model makes a lot of sense!
Thank you very much and have a great day.