Table 2.
Overview statistical analyses.
Analysis | Variables | Statistical model | Cohort/data set |
---|---|---|---|
1. Discovery DMPs | Outcome: DNA methylation levels (T0, T1, T2) | Mixed model | PRISMO |
Predictor: PTSD score (T0, T1, T2) | |||
2. Identification DMRs and PRS as determinants | Outcome: PTSD score (T0, T1, T2) | Mixed model | PRISMO |
Predictor: | |||
a. DMR methylation levels (T0, T1, T2) | |||
b. PRS | |||
3. Replication DMRs | Outcome: PTSD score (T0, T1, T2) | Mixed model | MRS |
Predictor: DMR methylation levels (T0, T1, T2) | |||
4. Correlations DMRs, PRS, and PTSD symptoms | a. DMR change score (T0-T1 or T1-T2) x PTSD score (T0 to T5) |
Pearson’s correlation | PRISMO |
b. DMR change score x PRS | |||
c. PRS x PTSD score (T0 to T5) | |||
5. Association DMRs and PRS with PTSD trajectories | Outcome: PTSD trajectory | Multinomial logistic regression | PRISMO (imputed) |
Predictor: | |||
a. DMR change score | |||
b. PRS | |||
6. Prediction model delayed onset PTSD | Outcome: delayed PTSD trajectory (vs. resilient trajectory) | Stepwise backward logistic regression model | PRISMO (imputed) |
Predictors: DMR change score, variables in Table 1 |
Note: DMP = differentially methylation position; PTSD = posttraumatic stress disorder; PRS = polygenic risk score.