Based on the analysis, we conclude that handling below detection data carefully is crucial to obtain more accurate estimates and also correct interpretation of the results.
We then censor the viral load data at detection limit 25, producing totally 23percentt of below detection data which are imputed at 25 as observed in a real experiment.
However, the Bayesian method fails to handle with below detection data so that it can not deal with case C. In addition, we also tried to fit case C with the Gaussian quadrature method.
To circumvent this difficulty, within linear mixed-effects model setting, Hughes (1999) used Monte Carlo integration to evaluate the E-step by repeatedly sampling from the conditional distribution of below detection data.
At 6 hours after infection, cytokine levels were below detection (data not shown).
Results were similar when the four subjects were included in analysis, and a value of 20 pg/mL was imputed for the baseline glucagon values below detection (data not shown).
We describe and compare statistical methods appropriate for concentrations that are below detection limits (non-detect data) and conclude that regression on order statistics provided the most rigorous analysis of our results, particularly for trace elements.
Also, the below detection limit data are missing and can thus be averaged along with random effects at the E-step.
First, the data below detection limit is completely hold out before fitting model, named as 'case A'; second, the below detection limit data are retained but assumed 25 as the true value, named as 'case B'; third, the below detection limit data are considered as censored data, named as 'case C'case
From one point of view, both individual parameter θ i and below detection limit data can be treated as missing data.
