Study population and data source

This is a retrospective cohort study to assess the effectiveness of a nationwide ban on raw beef liver provision at restaurants using interrupted time-series analysis (ITSA). Detection of EHEC from a clinical specimen (either symptomatic or asymptomatic) is a notifiable condition in Japan, and the law requires an immediate report to public health officials after diagnosis. Asymptomatic infection is defined by detection of EHEC from stools of asymptomatic patients, and is not necessarily restricted to outbreak investigation. The law does not specify the required microbiological methods to isolate EHEC, but the screening by selective culture media (Sorbitol-MacConkey agar) is commonly used with genotypic or phenotypic confirmation of verotoxin production. HUS is defined by the presence of the triad of hemolytic anemia, thrombocytopenia, and acute renal failure, and the law requires mandatory reporting if O antigen agglutinating antibody, anti-verotoxin antibody, or verotoxin (either genetically or phenotypically) was detected [16–18]. Local governments collect and submit case-based data via the National Epidemiological Surveillance of Infectious Diseases (NESID) system to MHLW [16]. Data are open to the public and retrievable from the website of the National Institute of Infectious Disease (NIID) in Japan (https://www.niid.go.jp/niid/en/), and we extracted national weekly data from January 2008 to December 2017 for the number of cases of both asymptomatic and symptomatic EHEC infections, of HUS, and death. We calculated cases per million Japanese population based on the annual demographics of Japanese population provided by Statistics Japan (http://www.stat.go.jp/data/jinsui/index.html).