Hennessee et al. Respond

We welcome the comments by Goldstein and Burstyn on our recent publication, in which we discuss considerations for improving reporting and analysis of date-based COVID-19 surveillance data.¹ Goldstein and Burstyn point out the complexities of estimating unknown dates and recommend relevant methods for addressing these. Although we agree that rigorous approaches are needed for inference, our discussion was primarily concerned with identifying a consistent starting point for reporting and analysis (i.e., which date to use) rather than with methods for obtaining a final product for inference. Our use of a simplified approach for estimating infection date was therefore to demonstrate the challenges in interpreting epidemic curves when dates with long reporting lags are used; it was not intended to be a methodological recommendation.

We also point out that clear, straightforward methods used to collect and report case data are essential for public understanding and trust.^2,3 This is especially important in the current environment in which COVID-19 surveillance data are widely consumed, and often questioned, by the public.^4,5 Interpretable approaches are also critical for policymakers and public health professionals who rely on surveillance data for decision-making and are often called upon to explain and defend them.⁶ Efforts to improve surveillance data should therefore be guided by interpretability and transparency in addition to scientific rigor.

Goldstein and Burstyn also discuss the potential for outcome misclassification of COVID-19 cases, and suggest methods for addressing this. Although that was not the focus of our article, we agree that such approaches are important for estimating cases and for comparing disease burden over time. Consistent and clear documentation in the choice of dates used by health departments would indeed benefit such analyses; again, however, we stress the importance of interpretability for how health departments report incident cases. We also support renewed calls to focus on hospitalizations for assessing COVID-19 impact trends, as is highlighted with the rapid spread of the omicron variant.⁷ Hospitalizations are less susceptible to case ascertainment issues and may better reflect the health burden of COVID-19 as it progresses from a pandemic to an endemic disease.

As the United States grapples with the current wave of omicron and prepares for inevitable future variants, we believe a more consistent approach to the choice of dates reported by state and local health departments could improve public comprehension, trust, and decision-making. It would also provide a more consistent starting point for data processing such as that suggested by Goldstein and Burstyn.