Table 4. The effect of changing from a Gaussian to a skew-normal on the estimated regression coefficients, with 95% credible intervals, in the slope model.
| Parameter | cVL2 adjusted model 1 | cVL2 adjusted model 4 | ||
|---|---|---|---|---|
| Estimate | 95% CI | Estimate | 95% CI | |
| Female-sex | 23.8 | (12.53, 34.96) | 23.8 | (12.61, 35.12) |
| Baseline age | -6.3 | (-11.57, -1.08) | -6.4 | (-11.61, -1.14) |
| Baseline CD4 count | 82.9 | (77.4, 88.38) | 82.8 | (77.36, 88.29) |
| Baseline log viral load | 7.5 | (2.03, 12.92) | 7.5 | (2.06, 12.91) |
| Time on treatment | 55.7 | (48.83, 62.65) | 55.7 | (48.72, 62.57) |
| Time on treatment-squared | -22.7 | (-24.86, -20.78) | -22.8 | (-24.84, -20.78) |
| cumulative log viral load | -19.6 | (-28.25, -10.93) | -19.4 | (-28.12, -10.74) |
cVL2 –cumulative HIV log10 viral load. All estimates reported as posterior mean and 95% credible intervals.