Age-specific social mixing of school-aged children in a US setting using proximity detecting sensors and contact surveys

Directly observed social contact patterns among school children in rural Gambia

INTRODUCTION

School-aged children play a major role in the transmission of many respiratory pathogens due to high rate of close contacts in schools. The validity and accuracy of proxy-reported contact data may be limited, particularly for children when attending school. We observed social contacts within schools and assessed the accuracy of proxy-reported versus observed physical contact data among students in rural Gambia.

METHODS

We enrolled school children who had also been recruited to a survey of Streptococcus pneumoniae carriage and social contacts. We visited participants at school and observed their contact patterns within and outside the classroom for two hours. We recorded the contact type, gender and approximate age of the contactee, and class size. We calculated age-stratified contact matrices to determine in-school contact patterns. We compared proxy-reported estimated physical contacts for the subset of participants (18 %) randomised to be observed on the same day for which the parent or caregiver reported the school contacts.

RESULTS

We recorded 3822 contacts for 219 participants from 114 schools. The median number of contacts was 15 (IQR: 11-20). Contact patterns were strongly age-assortative, and mainly involved physical touch (67.5 %). Those aged 5-9 years had the highest mean number of contacts [19.0 (95 %CI: 16.7-21.3)] while the ≥ 15-year age group had fewer contacts [12.8 (95 %CI: 10.9-14.7)]. Forty (18 %) participants had their school-observed contact data collected on the same day as their caregiver reported their estimated physical contacts at school; only 22.5 % had agreement within ±2 contacts between the observed and reported contacts. Fifty-eight percent of proxy-reported contacts were under-estimates.

CONCLUSIONS

Social contact rates observed among pupils at schools in rural Gambia were high, strongly age-assortative, and physical. Reporting of school contacts by proxies may underestimate the effect of school-age children in modelling studies of transmission of infections. New approaches are needed to quantify contacts within schools.

Comparing Contact Tracing Through Bluetooth and GPS Surveillance Data: Simulation-Driven Approach

BACKGROUND

Accurate and responsive epidemiological simulations of epidemic outbreaks inform decision-making to mitigate the impact of pandemics. These simulations must be grounded in quantities derived from measurements, among which the parameters associated with contacts between individuals are notoriously difficult to estimate. Digital contact tracing data, such as those provided by Bluetooth beaconing or GPS colocating, can provide more precise measures of contact than traditional methods based on direct observation or self-reporting. Both measurement modalities have shortcomings and are prone to false positives or negatives, as unmeasured environmental influences bias the data.

OBJECTIVE

We aim to compare GPS colocated versus Bluetooth beacon-derived proximity contact data for their impacts on transmission models' results under community and types of diseases.

METHODS

We examined the contact patterns derived from 3 data sets collected in 2016, with participants comprising students and staff from the University of Saskatchewan in Canada. Each of these 3 data sets used both Bluetooth beaconing and GPS localization on smartphones running the Ethica Data (Avicenna Research) app to collect sensor data about every 5 minutes over a month. We compared the structure of contact networks inferred from proximity contact data collected with the modalities of GPS colocating and Bluetooth beaconing. We assessed the impact of sensing modalities on the simulation results of transmission models informed by proximate contacts derived from sensing data. Specifically, we compared the incidence number, attack rate, and individual infection risks across simulation results of agent-based susceptible-exposed-infectious-removed transmission models of 4 different contagious diseases. We have demonstrated their differences with violin plots, 2-tailed t tests, and Kullback-Leibler divergence.

RESULTS

Both network structure analyses show visually salient differences in proximity contact data collected between GPS colocating and Bluetooth beaconing, regardless of the underlying population. Significant differences were found for the estimated attack rate based on distance threshold, measurement modality, and simulated disease. This finding demonstrates that the sensor modality used to trace contact can have a significant impact on the expected propagation of a disease through a population. The violin plots of attack rate and Kullback-Leibler divergence of individual infection risks demonstrated discernible differences for different sensing modalities, regardless of the underlying population and diseases. The results of the t tests on attack rate between different sensing modalities were mostly significant (P<.001).

CONCLUSIONS

We show that the contact networks generated from these 2 measurement modalities are different and generate significantly different attack rates across multiple data sets and pathogens. While both modalities offer higher-resolution portraits of contact behavior than is possible with most traditional contact measures, the differential impact of measurement modality on the simulation outcome cannot be ignored and must be addressed in studies only using a single measure of contact in the future.

Colonization and transmission of Staphylococcus aureus in schools: a citizen science project

Aggregation of children in schools has been established to be a key driver of transmission of infectious diseases. Mathematical models of transmission used to predict the impact of control measures, such as vaccination and testing, commonly depend on self-reported contact data. However, the link between self-reported social contacts and pathogen transmission has not been well described. To address this, we used Staphylococcus aureus as a model organism to track transmission within two secondary schools in England and test for associations between self-reported social contacts, test positivity and the bacterial strain collected from the same students. Students filled out a social contact survey and their S. aureus colonization status was ascertained through self-administered swabs from which isolates were sequenced. Isolates from the local community were also sequenced to assess the representativeness of school isolates. A low frequency of genome-linked transmission precluded a formal analysis of links between genomic and social networks, suggesting that S. aureus transmission within schools is too rare to make it a viable tool for this purpose. Whilst we found no evidence that schools are an important route of transmission, increased colonization rates found within schools imply that school-age children may be an important source of community transmission.