BACKGROUND: Secondary transmission after point-source outbreaks is an integral feature of the epidemiology of gastrointestinal pathogens such as norovirus. The household is an important site of these secondary cases. It can become the source of further community transmission as well as new point-source outbreaks. Consequently, time-series data from exposed households provide information for risk assessment and intervention. METHODS: Analysis of these data requires models that can address (1) dependencies in infection transmission, (2) random variability resulting from households with few members, and (3) unobserved state variables important to transmission. We use Monte Carlo maximum likelihood via data augmentation for obtaining estimates of the transmission rate and infectious period from household outbreaks with the 3 above features. RESULTS: We apply this parameter estimation technique to 153 infection sequences within households from a norovirus outbreak in Sweden and obtain maximum likelihood estimates of the daily rate of transmission ([Greek small letter beta with circumflex accent] = 0.14, 95% confidence interval [CI] = 0.08-0.24) and average infectious period (1/[Greek small letter gamma with circumflex accent] = 1.17 days, 95% CI = 1.00-1.88). We also demonstrate the robustness of the estimates to missing household sizes and asymptomatic infections. CONCLUSIONS: Maximum likelihood techniques such as these can be used to estimate transmission parameters under conditions of unobserved states and missing household size data, and to aid in the understanding of secondary risks associated with point-source outbreaks.
BACKGROUND: Secondary transmission after point-source outbreaks is an integral feature of the epidemiology of gastrointestinal pathogens such as norovirus. The household is an important site of these secondary cases. It can become the source of further community transmission as well as new point-source outbreaks. Consequently, time-series data from exposed households provide information for risk assessment and intervention. METHODS: Analysis of these data requires models that can address (1) dependencies in infection transmission, (2) random variability resulting from households with few members, and (3) unobserved state variables important to transmission. We use Monte Carlo maximum likelihood via data augmentation for obtaining estimates of the transmission rate and infectious period from household outbreaks with the 3 above features. RESULTS: We apply this parameter estimation technique to 153 infection sequences within households from a norovirus outbreak in Sweden and obtain maximum likelihood estimates of the daily rate of transmission ([Greek small letter beta with circumflex accent] = 0.14, 95% confidence interval [CI] = 0.08-0.24) and average infectious period (1/[Greek small letter gamma with circumflex accent] = 1.17 days, 95% CI = 1.00-1.88). We also demonstrate the robustness of the estimates to missing household sizes and asymptomatic infections. CONCLUSIONS: Maximum likelihood techniques such as these can be used to estimate transmission parameters under conditions of unobserved states and missing household size data, and to aid in the understanding of secondary risks associated with point-source outbreaks.
Authors: Henrik Salje; Justin Lessler; Kishor Kumar Paul; Andrew S Azman; M Waliur Rahman; Mahmudur Rahman; Derek Cummings; Emily S Gurley; Simon Cauchemez Journal: Proc Natl Acad Sci U S A Date: 2016-11-07 Impact factor: 11.205
Authors: A Sarah Walker; David W Eyre; David H Wyllie; Kate E Dingle; Rosalind M Harding; Lily O'Connor; David Griffiths; Ali Vaughan; John Finney; Mark H Wilcox; Derrick W Crook; Tim E A Peto Journal: PLoS Med Date: 2012-02-07 Impact factor: 11.069
Authors: Michael A L Hayashi; Joseph N S Eisenberg; Emily T Martin; Andrew N Hashikawa Journal: J Pediatric Infect Dis Soc Date: 2021-09-23 Impact factor: 5.235
Authors: Jonathan L Zelner; Benjamin A Lopman; Aron J Hall; Sebastien Ballesteros; Bryan T Grenfell Journal: PLoS One Date: 2013-07-24 Impact factor: 3.240