How to run latent class analysis with case-controls study?

06 June 2019 3 8K Report

In my research I have two groups of participants: addicts and healthy young people (controls).

I seek to identify latent classes based on genotype polymorphism (categorical), epigenetic changes (continuous) and experience of childhood trauma (continuous) and parental styles (continuous). The assumption would be that different subpopulations have different disease risks that we can measure through different proportions among cases and controls.

Can I do a LCA on a single combined sample of addicts and controls to identify specific subpopulations (latent classes) and latter check for proportions of latent classes among cases and controls?

If so should I include only geno,epi,trauma,parent indicators without group membership variable? Or do I include group variable as indicator also?

In not, should I alternatively run two different LCA each on its own population?

Any other suggestions?

Roberto Melipillan

Hi Darko,

You can follow this approach:

1) Apply LCA to your indicators (polymorphism, epigenetic changes, ,experience of childhood trauma, and parental styles) to identify different latent classes.

2) Use a dummy variable corresponding to the two groups (addicts and healthy) as a predictor of the latent class membership obtained in 1). I think it would be relevant to include some demographics variables as a control variables in this model.

3) When running the model in 2) you need to control for the uncertainty in the class prediction obtained in 1).

4) The literature called the methodology to conduct 3) the "Three-step approach". You can implement this approach using softwares like Mplus or LatentGold.

Best,

Roberto

Darko Roviš

Thanks for the tip Roberto.

So you suggest I run LCA with variables suggested (continuous and ordinal). I don't use Mplus of Latent gold. I have STATA and use gsem procedure for LCA. I don't know what is the procedure for the 3Smodel in stata...

Will take a look at it. Thanks.

Darko Roviš

Can we pick up this discussion further? I took me some time to get my hands (and mind) on mplus.

If my genotype polymorphism is ordinal variable with three values (LL, LS, SS) would you suggest I use three dummy variables or place it in the model as a single ordinal variable? Result I get are quite different…

Also I took a look at the 3step approach you suggested… So if I understand you well, I do LCA mixture model and include auxiliary command to analysis relation of LC to distal outcomes (drug addiction, psychopathology..) and sociodemographics (gender, age..).

How does this look to you?

MISSING ARE ALL (999);

USEVARIABLES care met ctq ll ls ss;

CLASSES = c(7);

AUXILIARY = group gender (DE3STEP);

CATEGORICAL = ll ls ss;

Analysis: Type = Mixture;

Thx.

How to compare two groups with only two measurements?

Enhanced Yellow Fluorescent Protein (Aequorea victoria) DNA sequence?

What kind of fluid could I use for a pitfall trap that also does not invalidate molecular testing?

What's the best way to measure growth rates in House sparrow chicks from day 2 to day 10?

Do we need to write protocol for a second study of a systematic review that has been registered and published?

How to increase a model accuracy ?

Do I need to do linear regression before running mediation analysis using PROCESS MACRO by Hayes ?

Sample dilution for quantitative analysis using ELISA?

Is There Any Feasible Method To Test Fluorescent Compounds Other Than Fluorescence Spectrometers ??

Where can I find life cycle inventory or assessment of non woven materials?