Abstract Body

Background:

Objectively defining and classifying epidemic or pandemic waves is critical in providing opportunities for timely resource allocation. However, there is no consensus about what composes a pandemic wave despite proposed definitions in the literature. This analysis aimed to identify, apply and characterize wave definition approaches using COVID-19 case data to build towards standardized definitions for improving pandemic preparedness and response.

Methods:

We obtained daily United States (US) case data from February 2020 to March 2022 from the Johns Hopkins COVID-19 data repository. We identified three major definitions of epidemic waves by scoping review. The ‘eR approach’ defined waves as periods where the effective reproduction number (eR) was greater than 1 for at least 14 days. The ‘fold approach’ defined waves as periods where the weekly case rate increased by at least one-fold followed by a decrease by at least one-fold. The ‘threshold approach’ defined waves as periods where the weekly case rate per 100,000 population surpassed 49.99, the US CDC threshold for moderate community transmission. We then compared wave characteristics across definitions.

Results:

The eR approach generated 5 waves with an average length of time between waves of 79 (IQR 47–90) days and average wave duration of 65 (56–73) days. The fold approach generated 10 waves with 25 (7–44) days between waves and average wave duration of 47 (29–54) days. The threshold method produced 2 waves with 62 days between the two waves and average wave duration of 276 (251–301) days. The in-wave average daily case rates per 100,000 population were 24.2 (95% CI 20.9–28.1), 19.7 (17.2–22.5), and 30.4 (28.6–32.3) for the eR approach, fold approach, and threshold approach, respectively.

Conclusions:

This analysis provides novel characterization of various approaches to epidemic wave definition. The fold approach produced the greatest number of waves, likely due to its sensitivity to weekly changes in case rates. The threshold approach produced the highest in-wave average case rate and lowest between-wave average case rate, indicating it may not adequately capture periods of inflection. These findings have public health implications, as overestimating waves may trigger preemptive allocation of limited health resources, and underestimating may result in a reactionary response with poor targeting of treatment and prevention. Further application of the definitions across countries and diseases are needed to build towards consensus.