Two-test algorithms for infectious disease diagnosis: Implications for COVID-19.

Area of Science:

• Infectious disease diagnostics
• Medical technology assessment
• Public health preparedness

Background:

• Point-of-care diagnostic tests for infectious diseases, including COVID-19, often face limitations in accuracy, complexity, cost, and turnaround time.
• Optimizing the use of diagnostic assays is critical to meet user needs and improve patient outcomes.
• Current diagnostic strategies may not be sufficient for diverse clinical and public health settings.

Purpose of the Study:

• To evaluate the diagnostic outcomes, test utilization, and turnaround times of two-test algorithms compared to singular testing strategies.
• To develop and utilize a simulation model, with a web-based application, for comparing diagnostic performance under various disease prevalence and test characteristic scenarios.
• To assess the impact of a two-test algorithm (Ag-RDT followed by RT-PCR) on sensitivity, specificity, and resource allocation for infectious disease diagnosis.

Main Methods:

• A simulation approach was employed to model diagnostic outcomes for two-test algorithms (RT-PCR and Ag-RDT) versus singular testing.
• A web-based application was developed to visualize and compare diagnostic outcomes based on disease prevalence, sensitivity, and specificity.
• The model was validated using hypothetical COVID-19 prevalence data and performance characteristics of RT-PCR and Ag-RDT.

Main Results:

• A two-test algorithm using RT-PCR on Ag-RDT negative samples showed significant sensitivity gains (27% and 7%) compared to Ag-RDT or RT-PCR alone.
• Specificity gains of 2.9% and 1.9% were predicted when RT-PCR was applied to Ag-RDT positive samples.
• The Ag-RDT followed by RT-PCR algorithm substantially reduced RT-PCR testing resource requirements, particularly in low-prevalence settings.

Conclusions:

• A two-test algorithm combining rapid screening with confirmatory laboratory testing can enhance diagnostic accuracy, expedite results, and conserve resources.
• This strategy may lead to missed cases in high-prevalence settings, necessitating careful consideration of the algorithm's limitations.
• The developed web application serves as a valuable tool for identifying optimal, tailored testing strategies for specific use cases, as demonstrated by its application within the ACT-Accelerator Diagnostics Pillar.