Abstract:Background: Under-reporting and, thus, uncertainty around the true incidence of health events is common in all public health reporting systems. While the problem of under-reporting is acknowledged in epidemiology, the guidance and methods available for assessing and correcting the resulting bias are obscure. Objective: We aim to design a simple modification to the Susceptible - Infected - Removed (SIR) model for estimating the fraction or proportion of reported infection cases. Methods: The suggested modification involves rescaling of the classical SIR model producing its mathematically equivalent version with explicit dependence on the reporting parameter (true proportion of cases reported). We justify the rescaling using the phase plane analysis of the SIR model system and show how this rescaling parameter can be estimated from the data along with the other model parameters. Results: We demonstrate how the proposed method is cross-validated using simulated data with known disease cases and then apply it to two empirical reported data sets to estimate the fraction of reported cases in Missoula County, Montana, USA, using: (1) flu data for 2016-2017 and (2) COVID-19 data for fall of 2020. Conclusions: We establish with the simulated and COVID-19 data that when most of the disease cases are presumed reported, the value of the additional reporting parameter in the modified SIR model is close or equal to one, so that the original SIR model is appropriate for data analysis. Conversely, the flu example shows that when the reporting parameter is close to zero, the original SIR model is not accurately estimating the usual rate parameters, and the re-scaled SIR model should be used. This research demonstrates the role of under-reporting of disease data and the importance of accounting for under-reporting when modeling simulated, endemic, and pandemic disease data. Correctly reporting the "true" number of disease cases will have downstream impacts on predictions of disease dynamics. A simple parameter adjustment to the SIR modeling framework can help alleviate bias and uncertainty around crucial epidemiological metrics (e.g.: basic disease reproduction number) and public health decision making.

Bias in the estimated reporting fraction due to vaccination in the time-series SIR model

A simple modification to the classical SIR model to estimate the proportion of under-reported infections using case studies in flu and COVID-19

Estimating efficacy of measles supplementary immunization activities via discrete-time modeling of disease incidence time series

Statistical implications of relaxing the homogeneous mixing assumption in time series Susceptible-Infectious-Removed models

Combining clinical and diagnostic surveillance to estimate the burden of measles disease: a modeling study

Quantifying indirect and direct vaccination effects arising in the SIR model

Estimating Unreported COVID-19 Cases with a Time-Varying SIR Regression Model

Coupling an Individual Adaptive‐decision Model with a SIRV Model of Influenza Vaccination Reveals New Insights for Epidemic Control

Modeling the effects of adherence to vaccination and health protocols in epidemic dynamics by means of an SIR model

Ecological inference for infectious disease data, with application to vaccination strategies

A discrete-time model with vaccination for a measles epidemic

Estimating Unreported COVID-19 Cases in the United States Based on Time-Varying SIR Model

Estimating Unreported COVID-19 Cases in the United States Based on the Tvsiru Model

On the Analytical Solution of the SIRV-Model for the Temporal Evolution of Epidemics for General Time-Dependent Recovery, Infection and Vaccination Rates

Time-Since-Infection Model for Hospitalization and Incidence Data

Challenges in Estimating Time-Varying Epidemic Severity Rates from Aggregate Data

Impact of periodic vaccination in SEIRS seasonal model

Quantitative analysis of a fractional order of the $ SEI_{c}\, I_{\eta} VR $ epidemic model with vaccination strategy

Estimating community-wide indirect effects of influenza vaccination: triangulation using mathematical models and bias analysis