Comparing viral load metrics and evaluating their use for HIV surveillance

- The value of (in-care) viral load (VL) metrics for HIV surveillance was assessed.
- Data from 20,740 HIV patients registered in the Dutch ATHENA-cohort was used.
- Mean in-care VL showed the strongest association with new HIV diagnoses.
- VL metrics can identify subgroup differences in viral suppression.
- VL metrics might be used for predicting numbers of new diagnoses in subsequent years.

SummaryObjectivesTo investigate the value of in-care viral load (ICVL) and other viral load (VL) metrics for HIV surveillance by comparing time trends and associations with numbers of new HIV diagnoses.MethodsData from 20,740 HIV patients registered in the Dutch ATHENA-cohort between 2002 and 2013 were used. We compared: six ICVL metrics (i.e. mean of the mean/first/last/highest log VL, median of the median log VL, first log VL for newly diagnosed combined with mean log VL for all others), log VL at diagnosis, proportion of patients with transmission risk (>400 copies/ml) or suppressed VL (≤200 copies/ml). Subgroup differences were assessed using Kruskal-Wallis and chi-square tests. Negative binomial regression was used for studying associations between VL metrics and numbers of new diagnoses 1-4 years later.ResultsMost ICVL metrics showed similar decreasing trends over time. Differences in covariables were found for all VL metrics. Mean ICVL showed the strongest association with new diagnoses: a decrease of one log unit in mean ICVL was associated with a 21% decrease in new diagnoses two years later.ConclusionsVL metrics may be of value for enhancing HIV surveillance by identifying subgroup differences in impact of treatment on viral suppression, and by predicting numbers of new diagnoses in subsequent years.