Table 3.
Final model of effects between PTSD, PTSD symptom clusters, and pain predicting AUD symptomology. Estimates are displayed as β [95% Confidence Interval].
| Variable | Model 1 PTSD and pain main effects | Model 2 PTSD and pain interaction | Model 3 PTSD symptom cluster and pain main effects | Model 4 PTSD symptom cluster and pain interactions |
|---|---|---|---|---|
| Co-variates | ||||
| Intercept | 21.778 [18.08, 25.48] | 22.19 [18.55, 25.83] | 21.76 [18.08, 25.45] | 22.48 [18.85, 26.11] |
| Age | −0.10 [−0.19, −0.01] | −0.08 [−0.17, 0.01] | −0.10 [−0.19, −0.01] | −0.08 [−0.17, 0.01] |
| Sex | −7.07 [−8.13, −6.01] | −6.91 [−7.90, −5.98] | −6.84 [−8.00, −5.88] | −6.86 [−7.90, −5.81] |
| White (ref) | REF | REF | REF | REF |
| Hispanic | −0.11 [−1.24, 1.02] | −0.03 [−1.15, 1.08] | −0.25 [−1.38, 0.88] | −0.17 [−1.28, 0.95] |
| Black | 1.11 [−0.14, 2.15] | 1.32 [−0.09, 2.35] | 0.87 [−0.37, 2.12] | 0.89 [−0.33, 2.11] |
| Other race | −3.99 [−6.00, −1.98] | −3.61 [−5.59, −1.63] | −4.02 [−6.02, −2.02] | −3.61 [−5.58, −1.64] |
| Combat exposure | 4.70 [3.27, 6.12] | 3.84 [2.41, 5.27] | 4.59 [3.15, 6.02] | 3.54 [2.10, 4.99] |
| PCL-5 total score | 0.10 [0.08, 0.13] | 0.11 [0.09, 0.14] | ||
| Pain | 0.05 [0.04, 0.06] | 0.06 [0.05, 0.07] | 0.05 [0.04, 0.06] | 0.06 [0.05, 0.08] |
| PTSD symptom clusters | ||||
| Re-experiencing | 0.38 [0.23, 0.53] | 0.41 [0.30, 0.52] | ||
| Avoidance | −0.004 [−0.26, 0.26] | 0.04 [−0.20, 0.28] | ||
| Negative cognitions/mood | 0.18 [0.06, 0.30] | |||
| Hyperarousal | −0.19 [−0.35, −0.04] | |||
| Interactions | ||||
| PTSD* Pain | −0.002 [−0.003, −0.002] | |||
| Re-experiencing* Pain | −0.004 [−0.007, 0.000] | |||
| Avoidance* Pain | −0.01 [−0.02, −0.01] |
Note: Ref=reference group for self-reported race/ethnicity covariate (White is reference group). In model 1 we entered main effects of PTSD (PCL-5 scores) and pain (POQ scores) as well as all covariates. In model 2 we entered overall PTSD (PCL-5 scores) interacting with pain (POQ scores) as well as all covariates. In model 3 we entered main effects of PTSD symptom clusters (PCL-5 subscale scores) and pain (POQ scores). In model 4 we entered a taxonomy of interactions between PTSD symptom clusters (PCL-5 subscale scores) and pain (POQ scores). For parsimony, interactions that were not significant were removed. Values above represent our final model of PTSD symptom clusters and pain predicting alcohol use disorder (AUDIT scores).
BOLD = confidence interval does not include 0