Is it appropriate to use fixed assay cut-offs for estimating seroprevalence?

Area of Science:

• Epidemiology
• Immunology
• Biostatistics

Background:

• Population seroprevalence estimation relies on classifying quantitative assay results using fixed cut-offs.
• The selection of these assay cut-offs significantly influences seroprevalence estimates.
• Human parvovirus 4 (HP4) exposure is assessed using serological assays.

Purpose of the Study:

• To evaluate the impact of different assay cut-off choices on seroprevalence estimates for HP4.
• To compare fixed diagnostic assay cut-offs with alternative methods based on mixture modeling.
• To propose improved methods for estimating assay cut-offs in seroprevalence studies.

Main Methods:

• Simulations were used to estimate seroprevalence under various scenarios with a time-resolved fluorescence immunoassay (TRFIA).
• Mixture modeling was employed to estimate assay cut-offs based on component distributions for infected/vaccinated and susceptible individuals.
• Seroprevalence estimates from mixture models were compared to those derived from fixed cut-offs.

Main Results:

• When underlying populations were well-distinguished, most methods yielded accurate seroprevalence estimates.
• Fixed assay cut-offs often produced biased estimates when there was high overlap between population components.
• Mixture model methods also showed bias, particularly when model fit was poor.

Conclusions:

• Fixed assay cut-offs can lead to biased seroprevalence estimates but offer practical advantages.
• Mixture modeling approaches can also yield biased results due to model fit issues.
• Estimating assay cut-offs specifically for seroprevalence studies can reduce bias in population estimates.