Pandemic influenza A (H1N1) 2009 (pandemic H1N1) is spreading throughout the planet. It has become the dominant strain in the southern hemisphere, where the influenza season is underway. Here, based on reported case clusters in the USA, we estimate the household secondary attack rate for pandemic H1N1 to be 27.3% (95% CI: 12.2%-50.5%). From a school outbreak, we estimate a school child infects 2.4 (95% CI: 1.8-3.2) other children within the school. We estimate the basic reproductive number, R 0 , to range from 1.3-1.7 and the generation interval to range from 2.6-3.2 days. We use a simulation model to evaluate the effectiveness of vaccination strategies in the USA for the Fall, 2009. If vaccine were available soon enough, vaccination of children, followed by adults, reaching 70% overall coverage, in addition to high risk and essential workforce groups, could mitigate a severe epidemic.Pandemic H1N1, which first emerged in Mexico in , spread worldwide, resulting in more than 130,000 laboratory-confirmed cases and 800 deaths in over 100 countries by midJuly (1). The global distribution of this novel strain prompted the World Health Organization to declare the first influenza pandemic of the 21st century in June 2009 (2). Initially, most cases were clustered in households (3-6) and schools (7) with over 50% of the reported cases in school children in the 5-18 year old age range. A recent analysis of data from the United States, Canada, the United Kingdom, and the European Union suggests case fatality ratios ranging from 0.20%-0.68% in these regions and a higher case fatality ratio in Mexico of 1.23% (95% CI 1.03%-1.47%) (8).Both pandemic and seasonal influenza cause sustained epidemics in the upper northern hemisphere (above latitude ~ 20°N) and lower southern hemisphere (below latitude ~ 20°S) during the respective late Fall to early Spring months, with epidemics in the more tropical regions (between latitudes ~ 20°S and 20°N) occurring sporadically, but sometimes corresponding to the rainy season. The last influenza pandemic was the Hong Kong A (H3N2) * To whom correspondence should be addressed. longini@scharp.org . 1957 -1958 , caused mid-Summer, 1957, outbreaks in Louisiana schools that were open in the Summer because of the need for children helping with the Spring harvest (11). However, there was no extensive community-wide spread of influenza A (H2N2) in the USA until the Fall of 1957, with the national level epidemic rising in September and peaking in October. Pandemic H1N1 will probably spread in a similar spatio-temporal pattern as previous pandemics, but accelerated due to increased air travel (12). Supporting Online MaterialEstimates of the transmissibility of pandemic H1N1are crucial to devising effective mitigation strategies. Historically, the best characterization of influenza transmissibility has been based on the household secondary attack rate. The household secondary attack rate is the probability (sometimes expressed as a percent) that an infected person in the household will infect a...

No abstract

No abstract

Many epidemic models approximate social contact behavior by assuming random mixing within mixing groups (e.g., homes, schools, and workplaces). The effect of more realistic social network structure on estimates of epidemic parameters is an open area of exploration. We develop a detailed statistical model to estimate the social contact network within a high school using friendship network data and a survey of contact behavior. Our contact network model includes classroom structure, longer durations of contacts to friends than non-friends and more frequent contacts with friends, based on reports in the contact survey. We performed simulation studies to explore which network structures are relevant to influenza transmission. These studies yield two key findings. First, we found that the friendship network structure important to the transmission process can be adequately represented by a dyad-independent exponential random graph model (ERGM). This means that individual-level sampled data is sufficient to characterize the entire friendship network. Second, we found that contact behavior was adequately represented by a static rather than dynamic contact network. We then compare a targeted antiviral prophylaxis intervention strategy and a grade closure intervention strategy under random mixing and network-based mixing. We find that random mixing overestimates the effect of targeted antiviral prophylaxis on the probability of an epidemic when the probability of transmission in 10 minutes of contact is less than 0.004 and underestimates it when this transmission probability is greater than 0.004. We found the same pattern for the final size of an epidemic, with a threshold transmission probability of 0.005. We also find random mixing overestimates the effect of a grade closure intervention on the probability of an epidemic and final size for all transmission probabilities. Our findings have implications for policy recommendations based on models assuming random mixing, and can inform further development of network-based models.

Airborne infectious diseases such as influenza are primarily transmitted from human to human by means of social contacts, and thus easily spread within households. Epidemic models, used to gain insight into infectious disease spread and control, typically rely on the assumption of random mixing within households. Until now, there has been no direct empirical evidence to support this assumption. Here, we present the first social contact survey specifically designed to study contact networks within households. The survey was conducted in Belgium (Flanders and Brussels) from 2010 to 2011. We analysed data from 318 households totalling 1266 individuals with household sizes ranging from two to seven members. Exponential-family random graph models (ERGMs) were fitted to the within-household contact networks to reveal the processes driving contact between household members, both on weekdays and weekends. The ERGMs showed a high degree of clustering and, specifically on weekdays, decreasing connectedness with increasing household size. Furthermore, we found that the odds of a contact between older siblings and between father and child are smaller than for any other pair. The epidemic simulation results suggest that within-household contact density is the main driver of differences in epidemic spread between complete and empirical-based household contact networks. The homogeneous mixing assumption may therefore be an adequate characterization of the within-household contact structure for the purpose of epidemic simulations. However, ignoring the contact density when inferring based on an epidemic model will result in biased estimates of within-household transmission rates. Further research regarding the implementation of within-household contact networks in epidemic models is necessary.