The 1978 English boarding school influenza outbreak: where the classic SEIR model fails.

Previous work has failed to fit classic SEIR epidemic models satisfactorily to the prevalence data of the famous English boarding school 1978 influenza A/H1N1 outbreak during the children's pandemic. It is still an open question whether a biologically plausible model can fit the prevalence time series and the attack rate correctly. To construct the final model, we first used an intentionally very flexible and overfitted discrete-time epidemiologic model to learn the epidemiological features from the data. The final model was a susceptible (S) - exposed (E) - infectious (I) - confined-to-bed (B) - convalescent (C) - recovered (R) model with time delay (constant residence time) in E and I compartments and multi-stage (Erlang-distributed residence time) in B and C compartments. We simultaneously fitted the reported B and C prevalence curves as well as the attack rate (proportion of children infected during the outbreak). The non-exponential residence times were crucial for good fits. The estimates of the generation time and the basic reproductive number ([Formula: see text]) were biologically reasonable. A simplified discrete-time model was built and fitted using the Bayesian procedure. Our work not only provided an answer to the open question, but also demonstrated an approach to constructive model generation.