Longitudinal analysis with different start time and time intervals

sy2426 — Fri, 14 Nov 2025 19:09:44 GMT

Hi,

I have a few questions about longitudinal analysis with different start time and time intervals.

In my study, I have six time points where BMI is measured in two groups of participants:

Time1: Beginning of school year 1.

Time2: End of school year 1.

Time3: Beginning of school year 2.

Time4: End of school year 2.

Time5: Beginning of school year 3.

Time6: End of school year 3.

There is summer break between each end of school year and beginning of next school year (eg. Summer break between time 2 and time 3). Currently, I used mixed model and only included participants who completed all six timepoint:

proc mixed data=data method=ML;

class id time(ref="1") group(ref="1") ;

model bmi = time group time*group/ s outp=predicted residual;

repeated time/type=un subject=id ;

run;

As the study continues with more time add down the road, I want to expand my sample by including 1) participants with different start point, such as including people who only start at time 3; and 2) people who miss one school year in between, such as people who participate in school year 1 and 3, but miss school year 2.

Do you have any tips on how to modify the model?

Thanks!

Re: Longitudinal analysis with different start time and time intervals

Ksharp — Sat, 15 Nov 2025 03:31:24 GMT

It is very likely you want to impute some missing value.
You could check PROC MI + PROC MIANALYZE

https://blogs.sas.com/content/iml/2020/12/02/score-external-logistic-model.html

Re: Longitudinal analysis with different start time and time intervals

Ksharp — Sat, 15 Nov 2025 03:38:03 GMT

Or you could try GEE model which could handle these missing value very well.
PROC GEE
or
PROC GENMOD

topic Re: Longitudinal analysis with different start time and time intervals in Statistical Procedures

Longitudinal analysis with different start time and time intervals

Re: Longitudinal analysis with different start time and time intervals

Re: Longitudinal analysis with different start time and time intervals