Estimates of the basic reproduction number for rubella using seroprevalence data and indicator-based approaches

Area of Science:

• Epidemiology
• Infectious Disease Modeling

Background:

• The basic reproduction number (R0) is crucial for understanding and controlling infectious diseases.
• Estimating R0 often relies on country-specific seroprevalence data, but this is unavailable for many regions.
• Existing methods for estimating R0 in data-scarce settings, such as regional averaging, have unclear reliability.

Purpose of the Study:

• To assess the feasibility of predicting rubella's R0 using demographic, economic, education, housing, and health indicators.
• To compare the predictive performance of indicator-based models against regional averaging for R0 estimation.

Main Methods:

• Calculated R0 for rubella across 98 settings.
• Correlated R0 with 66 different indicators using Pearson, Spearman, and Maximum Information Coefficient.
• Trained a random forest regression model and compared its performance against simple linear regression and regional averaging using 4-fold cross-validation.

Main Results:

• Rubella R0 was typically low (<5 in 81 settings).
• R0 showed weak correlations with indicators, strongest with educational attainment and household indicators.
• Random forest and linear regression models did not outperform a simple regional averaging approach in predicting R0.

Conclusions:

• Predicting rubella's R0 using available indicators is not straightforward due to weak correlations.
• A regional averaging method offers comparable reliability to indicator-based approaches for estimating R0 in data-limited settings.
• Findings suggest regional averaging may be a practical alternative for estimating R0 and disease burden for other infections lacking seroprevalence data.