Figure 5. Likelihood of Subsequent Concordant Results After a Discordant Result.
Lag-1 regression models were developed to determine if a discordant paired result provided predictive ability regarding subsequent pairs of tests. For these analyses, samples that were followed by another tested sample within the same individual were grouped together by specific paired results and are represented by circles (positive: +, negative: −, Ag: pp65 antigenemia). The proportions of discordant samples attaining a subsequent result are labeled (%).
*: No significant difference (p = 0.41) was observed between likelihoods of qPCR-negative/antigen-positive (56.5%) and qPCR-positive/antigen-negative (78.3%) discordant pairs becoming concordantly negative on the subsequent sample.
**: No significant difference (p = 0.47) was observed between likelihoods of qPCR-negative/antigen-positive (17.4%) and qPCR-positive/antigen-negative (8.7%) discordant pairs becoming concordantly positive on the subsequent sample.