How do I test dyad concordance using a nested model in PROC GLIMMIX?

Mollylamb — Sat, 15 Feb 2020 14:13:28 GMT

Hello SAS community,

I have a dataset that contains 175 families, with up to 5 members per family. There are up to 2 sexually-linked couples in each household (for example, one household might have grandma-grandpa, mom-dad, and a kid. The next household might have mom-dad and two kids. the next household might just have a dingle person. etc.). I have a household ID, and individual ID, and an ID ('pairs') that is the same for sexually-linked couples within a given household (1-1 or 2-2). All of the people in the household that are not sexually linked to other people in the household have a 3, 4, or 5 in the 'pairs' column.

Here's an example of some of my dataset:

household	individual ID	pairs	disease
1	10010	3	1
1	10012	4	0
1	10013	5	1
2	10020	1	1
2	10021	1	1
3	10030	3
4	10040	1	1
4	10041	1	1
5	10050	3	1
5	10052	2	1
5	10053	2	0
6	10060	3	1
6	10062	2	1
6	10063	2	1

Our research question is: do the sexually linked people in the household have higher concordance of disease status (binary outcome) compared to all of the non-sexually linked dyads in the house? We want to determine if sexual relationship with a person with disease is a risk factor for disease transmission.

I am using a Glimmix model (SAS version 9.4 (TS1M1):

proc glimmix data = mydatset ASYCORR ASYCOV;

class household pairs;

model disease = /dist = bin link = logit;

random household(pairs) ;

run;

My question: how do I get this model to output the info to answer the research question? Or am I already outputting it and just don't know how to interpret it?

Re: How do I test dyad concordance using a nested model in PROC GLIMMIX?

tellmeaboutityo — Fri, 13 Mar 2020 19:31:19 GMT

First you should create a new binary variable, sex_partners. The 1-1 and 2-2 cases should receive one code (e.g. 1, "Y"), and the 3, 4, 5 should receive the alternate code (e.g. 0, "N"). This is your independent variable. The other variables are nesting variables.

Then I would try:

proc glimmix data = mydatset ASYCORR ASYCOV;

class household pairs sex_partners;

model disease = sex_partners /dist = bin link = logit oddsratio solution;

random intercept / subject=pairs(household);

run;

topic How do I test dyad concordance using a nested model in PROC GLIMMIX? in Statistical Procedures

How do I test dyad concordance using a nested model in PROC GLIMMIX?

Re: How do I test dyad concordance using a nested model in PROC GLIMMIX?