Accounting for Reporting Delays in Public Health Surveillance

Public health surveillance data are rarely complete when first reported. Cases, hospitalisations, and other events take time to be confirmed, entered into databases, and reported to central systems. These reporting delays mean that the most recent data systematically undercount the true number of events, complicating real-time analysis.

This guide is designed with and for state, tribal, local, and territorial (STLT) public health practitioners and modellers in the United States working with delayed surveillance data.

Simple approaches such as discarding recent data, using frozen values, or aggregating by report date are often the first and most accessible options. When delays are long or variable, or when uncertainty quantification is needed, statistical methods can improve estimates by learning from historical reporting patterns. This guide helps determine when simple approaches are sufficient and when statistical methods would be worthwhile.

This guide is intended as a companion to a technical overview of nowcasting methods that covers mathematical formulation and computational properties in detail. Here we focus on practical guidance for choosing and implementing these methods in STLT settings, with emphasis on adoption, communication, and real-world case studies.

What this guide covers

• Challenges that arise when using surveillance data in real time, including reporting delays, data revisions, and site drop-in and drop-out.
• Modelling approaches that can adjust for reporting delays, from chain-ladder approaches to Bayesian hierarchical models.
• A decision tree to help select an appropriate method based on your data characteristics and use case.
• A how-to guide covering the common steps from assessing data through to communicating results.
• Case studies demonstrating practical implementations in surveillance settings.

Who is this for?

This guide is written for two audiences.

Epidemiologists and public health practitioners interested in accounting for reporting delays and other real-time data issues.

Modellers expanding their understanding of reporting delays and the methods available to address them.

Get involved

This is a collaborative effort. We welcome contributions of code examples, case studies, and feedback. See the contributing guide for details.