# Generalizing population RT-qPCR cycle threshold values-informed estimation of epidemiological dynamics: Impact of surveillance practices and pathogen variability

## TL;DR

This study shows that Ct values from RT-qPCR tests can reliably estimate disease spread in various scenarios, but accuracy depends on pathogen shedding patterns and surveillance practices.

## Contribution

The study provides a systematic evaluation of Ct-based Rt estimation accuracy across diverse surveillance and pathogen conditions.

## Key findings

- Ct-based Rt estimates are generally accurate across diverse surveillance conditions and pathogen shedding patterns.
- Accuracy decreases when detection is biased toward severe patients or during prolonged waves with stable Rt near one.
- Pathogens with non-monotonic viral shedding patterns show lower estimation accuracy.

## Abstract

Population-level viral load distributions, measured by RT-qPCR or qPCR cycle threshold (Ct) values from surveillance testing, can be used to estimate the time-varying reproductive number (Rt) in real-time during COVID-19 outbreaks. However, it remains unclear whether this approach can be broadly applied to other pathogens, sources of virologic test data, or surveillance strategies beyond those specifically implemented during the COVID-19 pandemic in Hong Kong. We systematically evaluated the accuracy of Ct-based Rt estimates using simulated epidemics under different surveillance testing systems and pathogen viral kinetics. Using area under the ROC curve (AUC) to assess accuracy in detecting epidemic growth or decline, we found that case ascertainment rates minimally impacted estimation accuracy, except when detection was heavily biased towards severe patients (AUC: 0.64, 95% CIs: 0.59 - 0.71) or during prolonged waves with stable Rt near one (AUC: 0.54, 0.48 - 0.64), compared to stable detection patterns over time (AUC 0.76, 0.66 - 0.82). By comparing model accuracies across different viral shedding patterns and by parameterizing our model using data from six respiratory pathogens, we found that model performance largely depends on a monotonic viral shedding trajectory following case detection. A pathogen that lacks such shedding pattern – for example, those with a viral peak after onset – exhibited lower accuracy (AUC: 0.58, 0.49 - 0.65). Overall, our findings demonstrate that Ct-based Rt estimation methods are generally accurate across diverse surveillance conditions and pathogen shedding patterns, supporting their practical use as a supplementary tool for timely transmission monitoring while highlighting limitations that warrant further consideration.

Population viral load distributions, often approximated by cycle threshold (Ct) values from RT-qPCR testing, have proven valuable for real-time estimation of transmission rates, enhancing situational awareness during the COVID-19 pandemic. However, a comprehensive framework for applying Ct-based methods in other epidemiological contexts, such as varying levels of surveillance coverage or different circulating pathogens/variants, has yet to be developed. In this study, we evaluated the strengths and limitations of Ct-based epidemic surveillance approaches by simulating a range of scenarios with diverse surveillance coverage reflecting real-life outbreaks and carefully calibrating pathogen viral kinetics using real-world parameters. Our findings demonstrate that Ct-based Rt estimates are generally accurate across a range of surveillance and pathogen conditions, supporting their utility as a supplementary tool for timely epidemic monitoring while also highlighting limitations for consideration in future applications.