Social network dynamics and infectious disease propagation 1/30 Social network dynamics and infectious disease propagation Niel Hens www.simid.be www.simpact.org www.socialcontactdata.org FWO Kennismakers, 14 December 2018
Social network dynamics and infectious disease propagation 2/30 Introduction Overview 1 Introduction 2 ERC consolidator grant TransMID 3 Results Household members do not mix at random Holidays and weekends shape influenza epidemics Social distancing of symptomatic individuals Behavioural changes imply different disease dynamics Heterogeneity & frailty in mixing behaviour 4 General conclusion
Social network dynamics and infectious disease propagation 3/30 Introduction Introduction Many infections are transmitted by contact or air Influenza, Varicella, Measles,... The transmission rate β depends on person-to-person contact c the probability of transmission q given a contact but this is not the same for everyone β β(a, a ) = q(a, a ) c(a, a ) Previously: math. convenient WAIFW-structures often informed by serological data (Anderson and May, 1992; Hens et al., 2012)
Social network dynamics and infectious disease propagation 4/30 Introduction Social Contact Surveys Rapoport and Horvath (1961): first social surveys to construct networks for studying the spread of infection Social contacts as proxies of transmission events of airborne infections Wallinga et al. (2006): conversational contacts predictive for age-specific proportion of persons immune against mumps in Utrecht in 1986 and against pandemic influenza in Cleveland in 1957. social contact hypothesis: q(a, a ) = q Several studies conducted since (for a systematic review, see Hoang et al., 2018)
Social network dynamics and infectious disease propagation 5/30 Introduction Who Acquires Infection From Whom? Basics I The Mass Action Principle Assume a S(usceptible) I(nfected) R(ecovered) compartmental model Denote λ the infection hazard or force of infection The Mass Action Principle states that The rate of acquiring an infection is given by the sum of all effective contacts with an infectious individual. In the system of ODEs: # new infections = λ(t)s(t) = βi(t)s(t) di(t) dt where σ is the recovery rate. = βi(t)s(t) σi(t),
Social network dynamics and infectious disease propagation 6/30 Introduction Who Acquires Infection From Whom? Basics II Heterogeneity: Age The age-heterogeneous but time-homogeneous mass action principle: λ(a) = ND ( L ) a β(a, a )λ(a ) exp λ(s)ds da L A with life expectancy L, population size N and mean infectious period D and age A the age of maternal antibody loss The Next Generation Operator The operator that defines the next generation of infected individuals g(a, a ) = ND L β(a, a ) A
Social network dynamics and infectious disease propagation 7/30 Introduction Who Acquires Infection From Whom? Basics III Basic reproduction number R 0 The dominant eigenvalue of this next generation operator (see e.g. Diekmann and Heesterbeek, 2000) Effective reproduction number R e If the proportion susceptible is known: R e (see e.g. Abrams et al., 2014, 2016; Hens et al., 2015) Initial Spread Initial epidemic phase by iterating the next generation operator Identical to the right eigenvector of that operator
Social network dynamics and infectious disease propagation 8/30 Introduction Social Contact Surveys POLYMOD study pilot study: Beutels et al. (2006) main study: Mossong et al. (2008) Social Contacts and Mixing Patterns Relevant to the Spread of Infectious Diseases Joël Mossong 1,2*, Niel Hens 3, Mark Jit 4, Philippe Beutels 5, Kari Auranen 6, Rafael Mikolajczyk 7, Marco Massari 8, Stefania Salmaso 8, Gianpaolo Scalia Tomba 9, Jacco Wallinga 10, Janneke Heijne 10, Malgorzata Sadkowska-Todys 11, Magdalena Rosinska 11, W. John Edmunds 4 1 Microbiology Unit, Laboratoire National de Santé, Luxembourg, Luxembourg, 2 Centre de Recherche Public Santé, Luxembourg, Luxembourg, 3 Center for Statistics, Hasselt University, Diepenbeek, Belgium, 4 Modelling and Economics Unit, Health Protection Agency Centre for Infections, London, United Kingdom, 5 Unit Health Economic and Modeling Infectious Diseases, Center for the Evaluation of Vaccination, Vaccine & Infectious Disease Institute, University of Antwerp, Antwerp, Belgium, 6 Department of Vaccines, National Public Health Institute KTL, Helsinki, Finland, 7 School of Public Health, University of Bielefeld, Bielefeld, Germany, 8 Istituto Superiore di Sanità, Rome, Italy, 9 Department of Mathematics, University of Rome Tor Vergata, Rome, Italy, 10 Centre for Infectious Disease Control Netherlands, National Institute for Public Health and the Environment, Bilthoven, The Netherlands, 11 National Institute of Hygiene, Warsaw, Poland PLoS MEDICINE (WoK: >750 citations to date)
Social network dynamics and infectious disease propagation 9/30 Introduction Social Contact Survey Example: Belgian Contact Survey Part of POLYMOD project Period March - May 2006 750 participants, selected through random digit dialling Diary-based questionnaire Two main types of contact: non-close and close contacts Total of 12,775 contacts ( 16 contacts per person per day) Hens et al. (2009a,b)
Social network dynamics and infectious disease propagation 10/30 Introduction EU mixing patterns common structure converging off-diagonals: parents getting older
Social network dynamics and infectious disease propagation 11/30 Introduction EU mixing patterns
Statistics for Biology and Health Niel Hens Ziv Shkedy Marc Aerts Christel Faes Pierre Van Damme Philippe Beutels A Modern Statistical Perspective Mathematical epidemiology of infectious diseases usually involves describing the flow of individuals between mutually exclusive infection states. One of the key parameters describing the transition from the susceptible to the infected class is the hazard of infection, often referred to as the force of infection. The force of infection reflects the degree of contact with potential for transmission between infected and susceptible individuals. The mathematical relation between the force of infection and eff ective contact patterns is generally assumed to be subjected to the mass action principle, which yields the necessary information to estimate the basic reproduction number, another key parameter in infectious disease epidemiology. It is within this context that the Center for Statistics (CenStat, I-Biostat, Hasselt University) and the Centre for the Evaluation of Vaccination and the Centre for Health Economic Research and Modelling Infectious Diseases (CEV, CHERMID, Vaccine and Infectious Disease Institute, University of Antwerp) have collaborated over the past 15 years. This book demonstrates the past and current research activities of these institutes and can be considered to be a milestone in this collaboration. This book is focused on the application of modern statistical methods and models to estimate infectious disease parameters. We want to provide the readers with software guidance, such as R packages, and with data, as far as they can be made publicly available. Statistics / Life Sciences, Medicine, Health Sciences ISBN 978-1-4614-4071-0 9781461440710 SBH Social network dynamics and infectious disease propagation 12/30 Introduction Serology and social contacts Statistics for Biology and Health Hens et al. (2012): Hands-on with R Varicella zoster virus: Ogunjimi et al. (2009); Goeyvaerts et al. (2010) Modeling Infectious Disease Parameters Based on Serological and Social Contact Data Hens Shkedy Aerts Faes Van Damme Beutels 1 Modeling Infectious Disease Parameters Based on Serological and Social Contact Data Niel Hens Ziv Shkedy Marc Aerts Christel Faes Pierre Van Damme Philippe Beutels Modeling Infectious Disease Parameters Based on Serological and Social Contact Data A Modern Statistical Perspective 5.64 4.21 ( 4.79 5.37 6.07 ) 8.26 8.68 14.08 15.69 W 4 M 1 M 2 MA M 3 C 3 SA C 1 MA L MA MA R ERC consolidator grant TransMID: social contacts and serology
Social network dynamics and infectious disease propagation ERC consolidator grant TransMID A systematic review: Hoang et al. (2018) 14 representative countrywide surveys 13/30
Social network dynamics and infectious disease propagation 14/30 ERC consolidator grant TransMID Summary tables... (more in the review)
Social network dynamics and infectious disease propagation 15/30 ERC consolidator grant TransMID www.socialcontactdata.org Open science approach Collection of datasets and information on social contact surveys made available via Zenodo with doi for the original articles studies: BE, DE, FI, LU, IT, NL, PL, GB, VTN, FR, PE, HK, ZI Relational databases household file (prim.key: hh id) participant file (prim.key: bpart id; foreign key: hh id) survey day file (prim.key: sday id; foreign key: part id) time use file (prim.key: part id; foreing key: sday id) contact file (prim.key: contact id; foreign key: sday id, part id) dictionary file Common file - specific file
Social network dynamics and infectious disease propagation 16/30 ERC consolidator grant TransMID Tools the socialmixr package socialmixr-package by Sebastian Funk (LSHTM) direct connection to the Zenodo contact data repository contact matrix: extracts a contact matrix from survey data dealing with issues as reciprocity and bootstrapping contact matrices ERC social contact data hackaton: smoothing approaches Hens and Wallinga, Wiley StatsRef (2018), Wallinga, van de Kassteele and Hens - HIDDA (foreseen: July, 2019) van de Kassteele et al. (2017) Camarda and Hens (2013); Vandendijck et al. (2018)
Social network dynamics and infectious disease propagation 17/30 Results Overview 1 Introduction 2 ERC consolidator grant TransMID 3 Results Household members do not mix at random Holidays and weekends shape influenza epidemics Social distancing of symptomatic individuals Behavioural changes imply different disease dynamics Heterogeneity & frailty in mixing behaviour 4 General conclusion
Social network dynamics and infectious disease propagation 18/30 Results Household members do not mix at random Household members do not mix at random Households are important units intense contact bridging function Social structure within households Is random mixing a valid assumption? Egocentric data: Potter et al. (2011); Potter and Hens (2013) Flemish household survey: Goeyvaerts et al. (2018)
Social network dynamics and infectious disease propagation 19/30 Results Household members do not mix at random Household members do not mix at random Conclusions Social contact study focusing on contact networks within households Contacts between father and children less likely Mean density decreases with household size during the week Little effect: household reproduction number, attack rates, final size, generation interval,... Limitations Results rely on contact definition Specific simulation setting reflecting influenza transmission Data collection over one day only
Social network dynamics and infectious disease propagation 20/30 Results Holidays and weekends shape influenza epidemics Impact of regular school closure Impact of regular school closure on seasonal influenza epidemics A data-driven spatial transmission model for Belgium S (i) E C I R De Luca et al. (2017)
Social network dynamics and infectious disease propagation 21/30 Results Holidays and weekends shape influenza epidemics Impact of regular school closure (A) simulated incidence profiles for influenza in Belgium; (B) peak time difference; (C) relative variation in epidemic size; (D) relative variation of peak incidence - De Luca et al. (2017)
Social network dynamics and infectious disease propagation 22/30 Results Social distancing of symptomatic individuals Social distancing of symptomatic individuals
Social network dynamics and infectious disease propagation 23/30 Results Social distancing of symptomatic individuals Social distancing of symptomatic individuals Inference on parameters associated with asymptomatic infection based on data from symptomatic cases social contact data from asymptomatic and symptomatic individuals Preferential transmission hypothesis (Santermans et al., 2017) Reduction in total number of cases of 39% (0.30, 0.45) when 50% of individuals would stay home immediately after symptom onset
Social network dynamics and infectious disease propagation Results Behavioural changes imply different disease dynamics Behavioural changes imply different disease dynamics A simple model (no immunological considerations) Distinguishing between symptomatic and asymptomatic infection 24/30
Social network dynamics and infectious disease propagation Results Behavioural changes imply different disease dynamics Behavioural changes imply different disease dynamics Different R0 values, equal for both infections Home isolation for most severe disease (inf 1) Impact on disease dynamics (inf 2)? Hendrickx et al. (2018) 25/30
Social network dynamics and infectious disease propagation Results Behavioural changes imply different disease dynamics Behavioural changes imply different disease dynamics 26/30
Social network dynamics and infectious disease propagation 27/30 Results Heterogeneity & frailty in mixing behaviour Heterogeneity & frailty in mixing behaviour Focus on burden of disease infants & young children elderly
Social network dynamics and infectious disease propagation 28/30 Results Heterogeneity & frailty in mixing behaviour Heterogeneity & frailty in mixing behaviour How about chronically ill people? Pilot study among 10 GP s in Flanders: 23 chronically ill people: 14 chronically ill people experiencing ILI, 25 healthy people; 31 healthy people with ILI, including paired samples chronically ill make fewer contacts; when experiencing ILI even more so but they do make a minimal number of contacts Health-related quality of life (EQ-5D-3L, VAS, see Bilcke et al. (2017)) if in need for care, number of contacts is higher for 60+ if limited in daily activities, number of contacts is lower impact on burden estimation in at risk groups
Social network dynamics and infectious disease propagation 29/30 General conclusion General conclusion Social contact data provide useful proxies for disease transmission events Care should be taken: contacts and effective contacts are likely different (Santermans et al., 2015) However it seems these data provide results predictive for disease transmission a powerful tool for improving models of disease transmission sharpen the questions that need to be asked to inform models Further results: animal-human contacts, time use and contacts,... Future studies: new household studies, new Flemish study,...
Social network dynamics and infectious disease propagation 30/30 General conclusion Acknowledgements This work is part of a project that has received funding from the European Research Council (ERC) under the European Unions Horizon 2020 research and innovation programme (grant agreement 682540 TransMID). The scientific chair of evidence based vaccinology sponsored by handgifts from Pfizer and GSK The Methusalem Funding Program from the Universities of Hasselt and Antwerp VSC Flemish supercomputer centre, FWO All co-authors and colleagues @CenStat, I-BioStat, UHasselt and @CHERMID, Vaxinfectio, UAntwerpen All collaborating centres nationally and internationally We gratefully acknowledge funding from the FWO for various research projects that made this work possible
Social network dynamics and infectious disease propagation 31/30 General conclusion References I Abrams, S., Beutels, P., and Hens, N. (2014). Assessing mumps outbreak risk in highly vaccinated populations using spatial seroprevalence data. Am J Epidemiol, 179(8):1006 1017. Abrams, S., Kourkouni, E., Sabbe, M., Beutels, P., and Hens, N. (2016). Inferring rubella outbreak risk from seroprevalence data in Belgium. Vaccine, 34(50):6187 6192. Anderson, R. and May, R. (1992). Infectious Diseases of Humans: Dynamics and Control. Oxford University Press, Oxford. Beutels, P., Shkedy, Z., Aerts, M., and Van Damme, P. (2006). Social mixing patterns for transmission models of close contact infections: exploring self-evaluation and diary-based data collection through a web-based interface. Epidemiology and Infection, 134:1158 1166. Bilcke, J., Hens, N., and Beutels, P. (2017). Quality-of-life: a many-splendored thing? belgian population norms and 34 potential determinants explored by beta regression. Quality of Life Research : An International Journal of Quality of Life Aspects of Treatment, Care and Rehabilitation., PMID 28349241. Camarda, G. and Hens, N. (2013). Modelling social contact data: a smoothing constrained approach. Proceedings of the 13th IWSM. De Luca, G., Van Kerckhove, K., Coletti, P., Poletto, C., Bossuyt, N., Hens, N., and Colizza, V. (2017). The impact of regular school closure on seasonal influenza epidemics: a data-driven spatial transmission model for Belgium. biorxiv. Diekmann, O. and Heesterbeek, J. (2000). Mathematical Methodology of Infectious Diseases: Model Building, Analysis and Interpretation. West Sussex: John Wiley & Sons Ltd. Goeyvaerts, N., Hens, N., Ogunjimi, B., Aerts, M., Shkedy, Z., Van Damme, P., and Beutels, P. (2010). Estimating infectious disease parameters from data on social contacts and serological status. Journal of the Royal Statistical Society Series C, 59:255 277.
Social network dynamics and infectious disease propagation 32/30 General conclusion References II Hendrickx, D. M., Abrams, S., and Hens, N. (2018). The impact of behavioral interventions on co-infection dynamics: an exploration of the effects of home isolation. biorxiv. Hens, N., Abrams, S., Santermans, E., Theeten, H., Goeyvaerts, N., Lernout, T., Leuridan, E., Van Kerckhove, K., Goossens, H., Van Damme, P., and Beutels, P. (2015). Assessing the risk of measles resurgence in a highly vaccinated population: Belgium anno 2013. Euro Surveill, 20(1). Hens, N., Ayele, G. M., Goeyvaerts, N., Aerts, M., Mossong, J., Edmunds, J. W., and Beutels, P. (2009a). Estimating the impact of school closure on social mixing behaviour and the transmission of close contact infections in eight European countries. BMC Infectious Diseases, 9:187. Hens, N., Calatayud, L., Kurkela, S., Tamme, T., and Wallinga, J. (2012). Robust reconstruction and analysis of outbreak data: influenza a(h1n1)v transmission in a school-based population. American Journal of Epidemiology. Hens, N., Goeyvaerts, N., Aerts, M., Shkedy, Z., Damme, P. V., and Beutels, P. (2009b). Mining social mixing patterns for infectious disease models based on a two-day population survey in Belgium. BMC Infectious Diseases, 9:5. Hoang, T. V., Coletti, P., Melegaro, A., Wallinga, J., Grijalva, C., Edmunds, J., Beutels, P., and Hens, N. (2018). A systematic review of social contact surveys to inform transmission models of close contact infections. biorxiv. Mossong, J., Hens, N., Jit, M., Beutels, P., Auranen, K., Mikolajczyk, R., Massari, M., Salmaso, S., Scalia Tomba, G., Wallinga, J., Heijne, J., Sadkowska-Todys, M., Rosinska, M., and Edmunds, J. (2008). Social contacts and mixing patterns relevant to the spread of infectious diseases. PLoS Medicine, 5(3):381 391.
Social network dynamics and infectious disease propagation 33/30 General conclusion References III Ogunjimi, B., Hens, N., Goeyvaerts, N., Aerts, M., Damme, P. V., and Beutels, P. (2009). Using empirical social contact data to model person to person infectious disease transmission: an illustration for varicella. Mathematical Biosciences, 218(2):80 87. Potter, G. E., Handcock, M., Longini, I. M., and Halloran, E. (2011). Modelling within-household contact networks from egocentric data. Annals of Applied Statistics, 5:1816 1838. Potter, G. E. and Hens, N. (2013). A penalized likelihood approach to estimate within-household contact networks from egocentric data. J R Stat Soc Ser C Appl Stat, 62(4):629 648. Rapoport, A. and Horvath, W. J. (1961). A study of a large sociogram. Behavioral science, 6:279 291. Santermans, E., Goeyvaerts, N., Melegaro, A., Edmunds, W. J., Faes, C., Aerts, M., Beutels, P., and Hens, N. (2015). The social contact hypothesis under the assumption of endemic equilibrium: Elucidating the transmission potential of vzv in europe. Epidemics, 11:14 23. Santermans, E., Van Kerckhove, K., Azmon, A., John Edmunds, W., Beutels, P., Faes, C., and Hens, N. (2017). Structural differences in mixing behavior informing the role of asymptomatic infection and testing symptom heritability. Mathematical Biosciences, 285:43 54. van de Kassteele, J., van Eijkeren, J., and Wallinga, J. (2017). Efficient estimation of age-specific social contact rates between men and women. Ann. Appl. Stat., 11(1):320 339. Vandendijck, Y., Camarda, C. G., and Hens, N. (2018). Cohort-based smoothing methods for age-specific contact rates. biorxiv. Wallinga, J., Teunis, P., and Kretzschmar, M. (2006). Using data on social contacts to estimate age-specific transmission parameters for respiratory-spread infectious agents. American Journal of Epidemiology, 164:936 944.