Table 2.
Comparing K8.1 and LANA titres of HIV patients who have developed KS to those who have not developed KS
| K8.1 |
LANA |
||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| GEE |
Random effects1 |
GEE |
Random effects1 |
||||||||||
| Factors | Estimate (95% CI) | p | Estimate (95% CI) | p | Estimate (95% CI) | p | Estimate (95% CI) | p | |||||
| KS patient vs. Control2a | 4.13 | (3.39,4.88) | <0.001 | 4.14 | (3.34,4.94) | <0.001 | 2.29 | (1.10,3.48) | <0.001 | 2.2 | (1.16,3.25) | <0.001 | |
| Years prior to diagnosis2b | 0.02 | (−0.10,0.15) | 0.69 | 0.03 | (−0.11,0.17) | 0.66 | 0.08 | (−0.13,0.29) | 0.48 | 0.006 | (−0.13,0.15) | 0.93 | |
| Case-Time interaction3 | 0.48 | (0.23,0.73) | <0.001 | 0.48 | (0.24,0.73) | <0.001 | |||||||
| Sex (Female vs. Male) | −0.6 | (−1.26,0.51) | 0.07 | −0.53 | (−1.17,0.11) | 0.1 | 0.07 | (−0.79,0.93) | 0.87 | −0.08 | (−0.96,0.79) | 0.85 | |
| Age at diagnosis | 0.02 | (9.75x10−5,0.04) | 0.05 | 0.02 | (−0.002,0.04) | 0.08 | 0.02 | (−0.009,0.05 | 0.18 | 0.02 | (−0.008,0.06) | 0.14 | |
| CD4 count | 200–500 cells per ml | 0.73 | (−0.19,1.64 | 0.12 | 0.71 | (−0.07,1.48) | 0.07 | 0.21 | (−0.97,1.39) | 0.72 | 0.26 | (−0.81,1.33) | 0.63 |
| (vs <=200 cells/ml) | >500 cells per ml | −0.71 | (−1.43,0.02) | 0.06 | −0.68 | (−1.54,0.17) | 0.12 | 0.2 | (−0.85,1.25) | 0.71 | 0.19 | (−0.95,1.33) | 0.74 |
| Random effects4 | Intercept | 2.02 | (1.30,3.16) | 6.99 | (5.26,9.29) | ||||||||
| Slope | 0.005 | (0.001,0.16) | 0.29 | (0.19,0.45) | |||||||||
| Covariance | −0.1 | (−0.27,0.06) | 0.7 | (0.27,1.14) | |||||||||
Serum samples of cases prior to KS diagnosis are compared to samples of control prior to pseudo-diagnosis using linear regression
covariance between the two estimated random effects (base titre and time effect per patient) was found to contribute only to the LANA model but has been retained in the K8.1 model.
due to the significant interaction in the models of K8.1 these factors may be interpreted as (a) the difference in titre between case and control patients at (pseudo)diagnosis (i.e. time=0) and (b) the time trend among patients who did not develop KS.
in the K8.1 models this factor is interpreted as the difference in change in titre between case and control patients over time; not significant in LANA models and therefore not included.
variance estimates