Table 1.
Basic description of IMASIDA variables used in the analyses.
Table 2.
Cross-classification of the type of censoring of the VL values for the 352 couples. LC = left-censored, IC = interval-censored, UN = uncensored.
Fig 1.
Scatter plot of the couples’ log VL pairs, together with (in red) the pair of averaged values and a two-dimensional 95% ellipse tracing a bivariate normal density contour.
The horizontal and vertical lines are positioned at the log of the values LOD/2 and (LOD + LOQ)/2.
Fig 2.
Scatter plot of the couples’ log VL pairs by ARV status of woman and man, together with a fitted linear regression line (dotted line).
The horizontal and vertical lines are positioned at the log of the substituted values LOD/2 and (LOD + LOQ)/2.
Table 3.
Means and standard deviations (sd) with respective standard errors (se), of the (substituted) log VL values for each of the four combinations of ARV biomarkers.
Table 4.
Comparison of the marginal, equal, shared independent and correlated random-effects models with constant and non-constant .
Table 5.
Parameter and standard error estimates of the independent random-effects model with non-constant .
Fig 3.
Scatter plots of log VL with fitted mean lines for men (left) and women (right), for ARV biomarkers present (blue) and absent (black).
The shaded areas reflect the areas mean ± 2 standard deviation of the fitted normal distributions. The horizontal grey lines at log(40) and log(550) locate the LOD and LOQ log values. The censored data (open bullets) at those LOD and LOQ values were scattered a bit to show their number more clearly.
Fig 4.
Correlation as a function of age of the woman and age of the man, for all four combinations of ARV status.
The black dot refers to the pair of average age of woman and man.
Table 6.
Estimates for for the four combinations of ARV biomarkers, at the averaged age for woman and man.