In avucoh/nPOD: What the Package Does (One Line, Title Case)

Matching methods

The underlying matching is performed by the optmatch R package "oriented to matching of treatment and control groups in observational studies". Specifically, the module imports pairmatch for optimal 1:1 matching. In pairmatch, units from the treatment group are matched with those in a larger control reservoir by minimizing a total distance metric.

For citation and further description, see the reference paper:

Hansen, B.B. and Klopfer, S.O. (2006) Optimal full matching and related designs via network flows, JCGS 15 609-627.

Important notes

We assume the "treatment group" submitted as CohortX is a smaller size subset than the available nPOD cohort match subset. If your group is actually larger than nPOD, matching can fail because then "there is too much competition for certain controls".

Using a small number of the most important parameters will also produce better results than using all parameters possible. Moreover, the selection of parameters must make sense for your purpose. For example, if you are a matching an external cohort of T1D to non-diabetic controls in nPOD, you would match on age and sex but probably not on C-peptide.

avucoh/nPOD documentation built on April 1, 2020, 5:24 p.m.