The role of social networks in pandemic influenza transmission in and from schools – an example from the United StatesArchived

Role of social networks in shaping disease transmission during a community outbreak of 2009 H1N1 pandemic influenzaCauchemez S, Bhattarai A, Marchbanks TL, et al.Proceedings of the National Academy of Sciences of the United States (PNAS); January 31, 2011

The importance of schools and school-age children in the transmission and amplification of the 2009 pandemic has been repeatedly demonstrated.[1-3] For example in the UK what terminated its substantial spring-summer wave was the closing of schools for the summer break,  and its autumn-winter wave only took off when the schools re-opened.[2]  School re-opening was similarly important in other European countries where there had been limited on-going transmission through the late spring and summer of 2009. However it was the resumption of schooling in the autumn that was associated with autumn-winter waves that took place well ahead of when seasonal influenza transmission has taken off.[3] Finally the only age-group that has been demonstrated to have experienced an excess of premature influenza related mortality in Europe has been school-aged children of 5-14 years.[4]

However until this publication by Cauchemez et al there have been few examinations of the dynamics of school-focused transmission. Their analysis indicates how social networks inside and outside of school can affect influenza transmission. The ‘micro-study’ was carried in a Pennsylvanian School, USA   as an outbreak investigation by a local investigation team, a European modelling group and the US Centers for Disease Control and Prevention. The setting was a semirural community in the state of Pennsylvania (United States) during the early phase of the 2009 pandemic. Uniquely the team made enquiries about pre-existing social network patterns of the children.  The study revealed strong between-place interactions with back-and-forth waves of transmission between the school, the community and the household. The data were collected from the outbreak that started in an elementary school, spreading thereafter to and from families, and was analysed by telephone surveys and use of modelling methods applied to the observed networks.  The transmission model designed by the researchers included parameters estimated from the data via Markov chain Monte Carlo sampling.

The authors found that, although sitting next to a case or being the playmate of a case did not significantly increase the risk of infection, the structuring of the school into classes and grades strongly affected the spread of the disease. The informal behaviour of the children was also important.  A main finding was that in school boys were more likely to transmit influenza to other boys than to girls, similarly girls tended to transmit influenza to other girls, not boys. This reflected the observed mixing among playmates whereby in these age-groups in this setting boys predominately mixed with boys and girls with girls but girl-boy mixing was uncommon. Other important findings of the study were:

• the presence of abnormally high transmission occurring on specific days of the outbreak (when children mixed intensively for group teaching),
• that late closure of the school (for instance, when 27% of the students already had symptoms) had no significant impact on the spread of the disease, 
• that school-aged individuals (from 6 to 18 years old) facilitated the introduction and spread of the disease in households, 
• that only about 1 in 5 cases aged  > 18 years old was infected by a school-aged household member.

The authors noted the importance of their study because of the scarcity of detailed analyses on transmission and spread of respiratory diseases in school and household settings.

ECDC Comment (21 February 2011):

While there have been some early descriptive reports of school outbreaks, for example a number published in Eurosurveillance from France and the UK [5-9]  this study is unusual in combining modelling, social network theory and ‘shoe-leather epidemiology’.  This gives the study considerable added value for both public health science and modelling. It applies a model to a real network, rather then to one created based on certain assumptions (these can be the number of contacts per person, distance to work, workplace size, household size, etc) or one assumed from some existing database.[10]  As more theoretical  approaches introduce  uncertainties, collecting data on how the social network actually functions is preferable.  Applying theoretical models to this kind of situations but not assuming or knowing the social network structure can be misleading. Other factors, including interactions of different age-group, and the effect of short-distance travel are assumed to be important for understanding an outbreak.  It is apparent that these approaches will not address some issues related to the transmission, like the effects of having infected persons sitting nearby each other in the classroom. The only published method ECDC is aware of takes into account time-use data surveys [11], although these will not be able to be applied equally effective to the outbreak in a specific place like the one in the study.

The study is most useful for policy planners as it not only estimates the transmission rates for community, school, grade, class and household interactions but it also suggests some methods for at least mitigating transmission, like not holding classroom group events that enhance close contacts during outbreaks. Of note is the intense boy to boy and girl to girl interactions among ‘fourth graders’ (aged 9-10 years) but the rare girl to boy interaction which seemed to result in almost separate if over-lapping waves of transmission among boys and girls.  At present we do not know as after the initial burst of descriptive studies in Europe in 2009 there seems to have been a dearth of subsequent published analytic school outbreak studies. ECDC would welcome details of studies it has missed or when new studies appear. Comments on this scientific advance should be sent to influenza@ecdc,europa.eu.