Pathogen seasonality and links with weather in England and Wales: a big data time series analysis

Cherrie, Mark P. C.; Nichols, Gordon; Iacono, Gianni Lo; Sarran, Christophe; Hajat, Shakoor; Fleming, Lora E.

Infectious disease data from England and Wales were collected by Public Health England (PHE) (formerly the Health Protection Agency and before that the Public Health Laboratory Service) through a voluntary reporting system, whereby hospital laboratory records are transferred to regional epidemiology units, processed and added to the LabBase2 national surveillance database [14] . To avoid duplication by organism and patient, each record has a unique identifier called the Organism Patient Illness Record (OPIE). If a record is sent with the same patient and organism information within 14 days (26 weeks for Mycobacterium spp.), then these cases are merged to ensure a single OPIE for the entire duration of the episode. The Second Generation Surveillance System (SGSS-formerly LabBase2) voluntary national surveillance database holds records on 12,904,446 reportable human infectious cases spanning from the 1st week in 1988 to the 2nd week in 2015 for 344 root organisms and 2014 serotypes. Pathogen counts were recorded at a weekly level in the database. The analysis for individual serotypes was restricted to complete years, from 1989 to 2014, with serotypes greater than 854 cases (above quartile one, i.e. top 25% in terms of total count), as a time series model could not be automatically estimated with fewer cases (n = 277). We aggregated the data to a monthly level and linked with national meteorological data held on the Medical and Environmental Data Mash-up Infrastructure project (MEDMI) platform [15] . The analysis was performed at a national scale due to multiple factors at a local level that act as noise to obfuscate the relationship between infectious disease and weather [16] .