How does date-rounding affect phylodynamic inference for public health?

Area of Science:

• Epidemiology
• Genomic Surveillance
• Computational Biology

Background:

• Phylodynamic analyses utilize pathogen genome sequences and sampling dates for public health genomic surveillance.
• Patient confidentiality concerns lead to reduced date-resolution (e.g., month, year) for sampling dates.
• This date-rounding can introduce bias into epidemiological parameter inference.

Purpose of the Study:

• To provide a practical guideline on when date-rounding biases phylodynamic inference.
• To assess the impact of date-resolution reduction on epidemiologically important parameters.
• To offer solutions for safer sharing of sampling date data.

Main Methods:

• Analysis of diverse empirical and simulated pathogen genomic datasets.
• Evaluation of bias in epidemiological parameters under varying date-resolutions.
• Investigation of factors influencing bias, including tree priors and substitution rates.

Main Results:

• Date-rounding biases epidemiological parameter inference, with bias direction varying by parameter, dataset, and tree prior.
• Bias is compounded by lower date-resolution and higher pathogen substitution rates.
• Bias decreases with longer sampling intervals, making the guideline most applicable to emerging datasets.

Conclusions:

• Reduced date-resolution in pathogen sampling data can significantly bias phylodynamic analyses.
• Understanding and quantifying this bias is essential for accurate genomic surveillance.
• Future solutions should balance patient confidentiality with data utility, potentially using methods like uniform random number translation for date sharing.