Utility of Rt epidemic trend categories for influenza surveillance: alignment with observed emergency department trends across two influenza seasons in the United States

CDC has estimated the time-varying reproduction number (Rt) for influenza in the United States since 2023, translating the posterior probability that Rt exceeds one, P(R_t>1), into epidemic trend categories (growing, likely growing, not changing, likely declining, declining). If Rt reliably indicates infection dynamics, growing or declining trends should precede corresponding changes in influenza-related emergency department (ED) visits, accounting for the lag between infection and care-seeking. However, because these categories reflect confidence in the direction of change rather than magnitude, their alignment with the corresponding trends ED activity measures is unclear. We compared weekly Rt epidemic trend categories to observed ED trends from the National Syndromic Surveillance Program across the 2024/2025 and 2025/2026 influenza seasons, using three thresholds to define observed trends at different magnitudes (≥0.1% difference, ≥0.5% difference, ≥5% relative change between consecutive weeks), comparing Rt trends to observed trends the following week. We assessed categorical alignment and calibration between P(R_t>1) and observed trends. Alignment was moderate overall, strongest for growing and declining categories under the 5% relative change threshold (61% aligned) and weakest under the 0.5% difference threshold (38.2% aligned); calibration was similarly best under the 5% relative change threshold. Misalignment concentrated around epidemic inflection points and periods of low influenza activity. Rt epidemic trend categories showed moderate, threshold-dependent alignment with observed ED activity, reflecting a combination of real-time model error and misalignment between direction-based and magnitude-based surveillance metrics. Defining categories based on Rt magnitude could improve correspondence with size-based change indicators and strengthen their utility for real-time monitoring.