Table 3. Posterior probabilities for selection of predictive variables in a model for the association between participant characteristics and increased internal or common carotid intimal–medial thickness using Bayesian model averaging.
| Predictor | Common cIMT (%) | Internal cIMT (%) |
|---|---|---|
| Male | 18.4 | 0 |
| Age | 100 | 100 |
| African–American | 100 | 0 |
| Hispanic | 50.8 | 3.6 |
| Systolic blood pressure | 97.4 | 8.8 |
| Diastolic blood pressure | 95.2 | 8.8 |
| Current smoker | 0 | 90.6 |
| Past smoker | 0 | 0 |
| Diabetes | 0 | 0 |
| Total cholesterol | 14.3 | 0 |
| HDL cholesterol | 84.9 | 5.8 |
| Injection drug use | 0 | 0 |
| AIDS by OI alone | 0 | 2.7 |
| Duration of HIV infection | 2.6 | 0 |
| HCV infection | 0 | 0 |
| CD4 (baseline) | 0.8 | 17.5 |
| CD4 (nadir) | 0 | 5.6 |
| Undetectable viral load | 1.8 | 9.9 |
| C-reactive protein | 0 | 3.1 |
| Interleukin-6 | 0 | 0 |
| Fibrinogen | 0 | 0 |
| Antiretroviral medications (duration of use) | ||
| Lamivudine | 0 | 0 |
| Zidovudine | 0 | 0 |
| Stavudine | 0 | 0 |
| Tenofovir | 51.3 | 3.3 |
| Didanosine | 0 | 0 |
| Abacavir | 0 | 0 |
| Emtricitabine | 0 | 0 |
| Zalcitabine | 0 | 0 |
| Ritonavir | 17.6 | 0 |
| Indinavir | 0 | 5.2 |
| Nelfinavir | 0 | 4.3 |
| Lopinavir | 0 | 0 |
| Saquinavir | 0 | 0 |
| Atazanavir | 0 | 0 |
| Amprenavir | 0 | 0 |
| Fosamprenavir | 5.0 | 0 |
| Efavirenz | 0 | 0 |
| Nevaripine | 0.9 | 0 |
| Delavirdine | 0 | 0 |
The results in this table are multivariate and reflect the weighted average of the probability that a variable will be included in one of the models with competitive goodness of fit. cIMT, carotid intimal–medial thickness; HCV, hepatitis C virus; HDL, high-density lipoprotein; OI, opportunistic infection.