Multivariate analysis of KIR/HLA effects on HIV-1 discordant and concordant couple status
| Predictor variables . | OR . | 95% CI . | P . | AIC . |
|---|---|---|---|---|
| Model 1 | ||||
| Index male sex | 0.02 | 0.003-0.19 | < .001 | 63.9 |
| Recipient KIR2DL2/KIR2DL2 | 3.48 | 0.39-31.1 | .264 | |
| Recipient KIR2DL1/C1/C2 + index C1/C1 | 17.6 | 1.80-174 | .014 | |
| Recipient KIR2DL3/KIR2DL3 | 1.21 | 0.26-5.71 | .809 | |
| Recipient KIR2DL3/KIR2DL3 + index C1/C2 | 0.04 | 0.002-0.86 | .039 | |
| Model 2 | ||||
| Index male sex | 0.02 | 0.003-0.20 | < .001 | 62.0 |
| Recipient KIR2DL2/KIR2DL2 | 3.17 | 0.41-24.5 | .269 | |
| Recipient KIR2DL1/C1/C2 + index C1/C1 | 17.3 | 1.80-170 | .014 | |
| Recipient KIR2DL3/KIR2DL3 + index C1/C2 | 0.05 | 0.003-0.82 | .036 | |
| Model 3 | ||||
| Index male sex | 0.02 | 0.003-0.18 | < .001 | 61.2 |
| Recipient KIR2DL1/C1/C2 + index C1/C1 | 14.5 | 1.50-138 | .020 | |
| Recipient KIR2DL3/KIR2DL3 + index C1/C2 | 0.04 | 0.002-0.66 | .025 |
| Predictor variables . | OR . | 95% CI . | P . | AIC . |
|---|---|---|---|---|
| Model 1 | ||||
| Index male sex | 0.02 | 0.003-0.19 | < .001 | 63.9 |
| Recipient KIR2DL2/KIR2DL2 | 3.48 | 0.39-31.1 | .264 | |
| Recipient KIR2DL1/C1/C2 + index C1/C1 | 17.6 | 1.80-174 | .014 | |
| Recipient KIR2DL3/KIR2DL3 | 1.21 | 0.26-5.71 | .809 | |
| Recipient KIR2DL3/KIR2DL3 + index C1/C2 | 0.04 | 0.002-0.86 | .039 | |
| Model 2 | ||||
| Index male sex | 0.02 | 0.003-0.20 | < .001 | 62.0 |
| Recipient KIR2DL2/KIR2DL2 | 3.17 | 0.41-24.5 | .269 | |
| Recipient KIR2DL1/C1/C2 + index C1/C1 | 17.3 | 1.80-170 | .014 | |
| Recipient KIR2DL3/KIR2DL3 + index C1/C2 | 0.05 | 0.003-0.82 | .036 | |
| Model 3 | ||||
| Index male sex | 0.02 | 0.003-0.18 | < .001 | 61.2 |
| Recipient KIR2DL1/C1/C2 + index C1/C1 | 14.5 | 1.50-138 | .020 | |
| Recipient KIR2DL3/KIR2DL3 + index C1/C2 | 0.04 | 0.002-0.66 | .025 |
Data are calculated by binomial logistic regression analysis. P values < .05 are in bold. OR values > 1 are predictive of HIV-1–discordant status, OR values < 1 are predictive of HIV-1–concordant status. Models 2 and 3 are obtained from model 1 by backwards stepwise selection based on the AIC. The model with the lowest AIC shows the best fit.
OR indicates odds ratio; CI, confidence interval; and AIC, Akaike Information Criterion.