Can visual interpretation of NucliSens graphs reduce the need for repeat viral load testing?

Area of Science:

• Clinical diagnostics
• Virology
• Public health

Background:

• Antiretroviral therapy (ART) adherence monitoring in HIV patients relies on viral load (VL) testing.
• Zimbabwe utilizes NucliSens technology for VL testing at the National Microbiology Reference Laboratory.
• Anecdotal evidence suggests invalid NucliSens graphs (Target Not Detectable) may yield valid results upon retesting, potentially saving resources.

Purpose of the Study:

• To assess the reliability of visual interpretation of NucliSens graphs by laboratory scientists.
• To determine intra- and inter-rater agreement for NucliSens graph interpretation.
• To evaluate the diagnostic accuracy (sensitivity, specificity, predictive values) of NucliSens graphs against repeat VL testing.

Main Methods:

• A cross-sectional study analyzed secondary data from 562 patients.
• Two scientists independently interpreted NucliSens graphs, with assessments repeated after one week for intra-rater reliability.
• Kappa statistics were used to measure agreement; graph interpretations were compared to repeat VL results.

Main Results:

• Near-perfect intra- and inter-rater agreement was observed (Kappa scores 0.96-0.99).
• Agreement between NucliSens graph interpretation and repeat VL results yielded a Kappa score of 0.65.
• The NucliSens graphs demonstrated 71% sensitivity, 92% specificity, 79% positive predictive value, and 89% negative predictive value.

Conclusions:

• Visual interpretation of NucliSens graphs by laboratory scientists is highly reliable.
• The 89% negative predictive value indicates an 11% false negative rate, necessitating caution.
• The potential cost savings from omitting repeat VL testing must be weighed against the clinical risks of underestimating viral load in HIV patients on ART.