Table 2.
Regression models for predictors of ART adherence intention (N = 148).
| Characteristics | Model 1 | Model 2 | Model 3 | Model 4 |
|---|---|---|---|---|
| PR (95% CI)a | PR (95% CI)a | PR (95% CI)a | PR (95% CI)a | |
| Water source | ||||
| Piped water | Referent | Referent | Referent | Referent |
| Borehole or surface water | 0.73 (0.52–1.02) | 0.77 (0.52–1.14) | 0.80 (0.61–1.05) | 1.06 (0.78, 1.45) |
| HIV transmission knowledge | ||||
| High (12–13) | Referent | Referent | Referent | |
| Low (<12) | 0.59 (0.49,0.70)* | 0.62 (0.49–0.79)* | 0.65 (0.50,0.84)* | |
| Age | ||||
| 18–30 | Referent | Referent | ||
| Above 30 | 1.02 (0.65–1.60) | 1.06 (0.67, 1.68) | ||
| Occupation | ||||
| Employed | Referent | Referent | ||
| Housewife | 0.96 (0.80–1.15) | 0.84 (0.58, 1.23) | ||
| Unskilled labor | 1.01 (0.49, 2.06) | 1.13 (0.61, 2.06) | ||
| Household monthly income | ||||
| ≥3,000 Zambian Kwacha | Referent | Referent | ||
| <3,000 Zambian Kwacha | 1.31 (1.08, 1.61) | 1.18 (0.78, 1.78) | ||
| Place of residence | ||||
| Urban | Referent | |||
| Rural | 0.53 (0.24–1.15) | |||
| Pearson's and Deviance goodness of fit tests both had P = 1.00 for all four models.b |
*
p-value ≤ 0.05.
a
From mixed effects Poisson regression with robust variance estimate (i.e., log-linear model), including a random effect for clinic. PR, prevalence ratio.
b
From fixed effects Poisson regression model (with a fixed effect for clinic, since goodness of fit tests are not implemented for random effects models; model coefficients negligibly different).