P336: Vaccination-Era Serial Intervals Enhance Real-World Surveillance and Forecasting During a US Measles Outbreak
Poster Presenter
Rosemary A Martoma
Physician-Researcher
Boston Children's Hospital | Harvard Medical School United States
Objectives
To empirically estimate vaccination-era measles serial intervals during a contemporary US outbreak and evaluate their impact on real-time estimation of the effective reproduction number (Rt) and short-term forecasting to support operational public health decision-making.
Method
Rash-onset dates from laboratory-confirmed measles cases during the 2022 central Ohio outbreak were analyzed. Epidemiologically verified infector–infectee pairs were used to estimate serial intervals and assess effects on real-time effective reproduction number (Rt) and 1–3-day incidence forecasts.
Results
Ninety cases were reported during the 2022 central Ohio outbreak; 59 transmission pairs were identified, and 29 met inclusion criteria for analysis. The overall mean rash-to-rash serial interval was 11.2 days (95% CI: 9.9–12.6). Serial intervals varied by transmission context, with significantly shorter intervals during clustered transmission within an underimmunized community (9.4 days; 95% CI: 8.0–10.7; n=14) and significantly longer intervals during broader community-wide spread (13.0 days; 95% CI: 11.0–15.0; n=15), yielding a mean difference of 3.6 days (95% CI: 1.2–6.0).
Incorporating empirical vaccination-era serial intervals into modeling pipelines reduced early-phase inflation of the effective reproduction number (Rt), defined as the average number of secondary infections caused by a primary case at time t, relative to models using historical pre-vaccination assumptions. In short-term forecasting, empirical intervals improved accuracy across 1–3-day horizons. For 1-day-ahead forecasts, RMSE values were lower under empirical assumptions (1.83–1.84) compared with historical assumptions (1.90–1.94); at 3 days, RMSE remained lower (2.11–2.12 vs 2.26–2.32). Forecast skill was highest using clustered empirical intervals, whereas historical household means provided no improvement. Findings were robust to resampling and confirmed the hypothesis that empirical vaccination-era intervals improve early-phase inference and forecasting performance relevant to public health operations.
Conclusion
Vaccination-era measles serial intervals were significantly shorter during clustered transmission and longer during community-wide spread. Using empirical intervals improved Rt estimation and short-term forecasting accuracy relative to historical assumptions, providing more reliable situational awareness during outbreak response.
These findings support incorporation of contemporary transmission parameters into real-time modeling pipelines used by public health agencies, health systems, and regulatory stakeholders. More accurate forecasts may strengthen vaccination planning, improve resource allocation, and enhance real-world data and real-world evidence generation for immunization policy in settings where historical parameters no longer reflect current epidemiology.
