Table 4.
n | Frail (%) | Age-adjusted univariate PR (95% CI)a | Multivariate model 1bPR (95% CI) | Multivariate model 2cPR (95% CI) | |
---|---|---|---|---|---|
HIV-negative (Ref ) | 573 | 8 | |||
CD4+ | |||||
<100 | 82 | 20 | 2.43 (1.37-4.32)* | 2.89 (1.55-5.39)* | 2.02 (1.01-1.11)* |
100–199 | 117 | 14 | 1.36 (0.72-2.59) | 1.46 (0.70-3.02) | 1.05 (0.46-2.38) |
200–349 | 267 | 9 | 1.15 (0.70-1.88) | 1.20 (0.68-2.13) | 1.04 (0.57-1.89) |
350–499 | 255 | 10 | 1.26 (0.77-2.04) | 1.21 (0.67-2.17) | 1.22 (0.65-2.27) |
≥500 | 466 | 6 | 0.77 (0.48-1.24) | 0.84 (0.56-1.61) | 0.93 (0.53-1.66) |
AUC CD4+d | |||||
<100 | 151 | 17 | 2.02 (1.23-3.34)* | 2.66 (0.98-7.22) | 2.18 (0.75-6.32) |
100–199 | 260 | 10 | 1.27 (0.78-2.08) | 1.37 (0.47-3.99) | 1.63 (0.54-4.93) |
200–349 | 379 | 9 | 1.16 (0.74-1.80) | 1.73 (0.62-4.80) | 1.96 (0.66-5.85) |
350–499 | 264 | 6 | 0.74 (0.41-1.34) | 1.03 (0.33-3.22) | 1.19 (0.36-3.95) |
≥500 | 117 | 6 | 0.91 (0.41-2.03) | 1.79 (0.49-5.98) | 2.37 (0.64-8.81) |
CD4+/CD8+e | |||||
≤0.29 | 255 | 14 | 1.62 (1.03-2.56)* | 1.79 (1.08-2.96)* | 1.44 (0.84-2.49) |
0.30–0.46 | 233 | 13 | 1.49 (0.93-2.39) | 1.44 (0.84-2.47) | 1.21 (0.67-2.17) |
0.47–0.63 | 241 | 8 | 1.00 (0.58-1.71) | 1.01 (0.55-1.85) | 0.90 (0.47-1.72) |
0.64–0.93 | 233 | 6 | 0.82 (0.46-1.48) | 0.84 (0.43-1.64) | 0.92 (0.47-1.83) |
≥0.94 | 224 | 5 | 0.66 (0.34-1.28) | 0.79 (0.38-1.61) | 0.78 (0.36-1.69) |
History of AIDS | |||||
Yes | 498 | 12 | 1.55 (1.03-2.34)* | 1.54 (0.96-2.45) | 1.33 (0.79-2.24) |
No | 708 | 7 | 0.95 (0.63-1.44) | 1.12 (0.70-1.79) | 1.04 (0.63-1.71) |
PR, prevalence ratio; CI, confidence interval.
Multivariate model 1, referred to as the partially adjusted model, was generated using logistic regression to generate prevalence ratios. This model adjusted for age (at substudy visit), study site, distance of timed-gait test, race/ethnicity, baseline educational attainment, baseline depressive symptoms, baseline hepatitis C virus, baseline frequency of alcohol consumption, baseline income, baseline BMI, baseline marital status, baseline smoking history, and baseline HIV risk category.
Multivariate model 2, referred to as the fully adjusted model, was generated using logistic regression. In addition to covariates listed in the partially adjusted mode1, the fully adjusted model included sociodemographic covariates collected at the substudy visit. These covariates included concurrent depression, concurrent frequency of alcohol consumption, concurrent income, concurrent BMI, concurrent marital status, and concurrent smoking history.
AUC CD4+, area under the curve; multivariate models were further adjusted for last CD4+ count prior to HAART initiation.
Concurrent ratio of CD4+/CD8+ is based on distribution of HIV-infected women.
p values significant at α = 0.05.