Merler, P. Poletti, L. Rossi, M. Halloran, I. Longini Jr. Spread of Zika virus in the Americas. Zhang, K. Sun, M. Chinazzi, A. Pastore y Piontti, N. Merler, D. Mistry, P. Bray, M. Zhang, N. Perrotta, D. Paolotti, M. Tizzoni, A. Pastore y Piontti, Q. Zhang, M. Gomes, L. Rossi, C. Poletto, V. Colizza, D. Chao, I. Longing, M. Halloran and A. Chowell and J. Zhang, C. Gioannini, D. Paolotti, N. Perra, D. Perrotta, M.
Quaggiotto, M. Tizzoni, and A. Assessing the impact of travel restrictions on international spread of the West African Ebola epidemic. Poletto, M. Gomes, A. Pastore y Piontti, L. Rossi, L. Bioglio, D. Longini, M. Halloran, V. Colizza, A. Rossi, D. Elizabeth Halloran, A. The Infection Tree of Global Epidemics.
Ferreira Da Costa Gomes, N. Samay, N. Human mobility and the worldwide impact of intentional localized highly pathogenic virus release. Tizzoni, P. Bajardi, C.
Poletto, J. Ramasco, D. These models are important both for developing our understanding of potentially novel disease strains e. For example, models incorporating disease control measures e. Given both the critical role of person to person transmission in the spread of outbreaks e. However, recent epidemiological research has noted a limitation of traditional mathematical models of disease spread: they often do not allow for heterogeneous behavioural responses within a population e.
This emphasis on homogenous behaviour is broadly inconsistent with what we know about human behaviour from decades of psychological research and theory in the context of health-related behaviour change. For example, a meta-analytic review of research involving the Theory of Planned Behaviour a psychological theory of behaviour change, e. It is therefore critical that infectious disease models seeking to incorporate human behaviour do so in a way that realistically reflects its heterogeneous nature.
Before recommendations can be made for how to better operationalise human behaviour in infectious disease models, we need to clearly understand how human behaviour during an infectious disease outbreak is currently modelled. The large scale scoping review presented within this paper represents an attempt to collate and summarise the state of the art concerning the incorporation of behaviour designed to protect oneself against infection within mathematical models of infectious disease spread, for example, vaccination, distancing oneself from other individuals social distancing , condom use, or hand washing.
More specifically, we were interested in developing a detailed understanding of: what diseases and infection prevention behaviours are modelled across the literature; how the behaviour is modelled with an explicit interest in understanding both the mechanism of modelling and the components that contribute to behaviour change , and; what theoretical background is presented to support the modelling of infection prevention behaviour if any.
A wide range of literature drawn from the behavioural sciences is available to assist modellers in developing more realistic models of human behavioural responses to infectious disease outbreaks. For instance, Susan Michie and colleagues worked with health behaviour experts health psychology theorists, health psychologists, and health services researchers to reach a consensus on 12 domains later revised to 14 [ 13 ] that are central to the explanation of behaviour change [ 14 ].
The outcomes of this review are presented and discussed in the context of this available literature, resulting in a series of recommendations designed to help infectious disease modellers to model human behaviour by incorporating insights from the behavioural sciences. We opted to use the scoping study methodology rather than a systematic review methodology as we were not concerned with systematically assessing the quality of all available literature; a task that would befit the explicit use of a systematic review approach.
Instead, we were focused on: a mapping and collating the existing literature to identify current best practice for incorporating human behaviour into infectious disease models, and; b identifying aspects of human behaviour modelling that could be improved through the incorporation of insights from both health and social psychology. Both of these aims are consistent with the use of the scoping review methodology as described in the literature [ 16 , 17 ].
The search strategy contained terms relating to behaviour, infectious disease, and mathematical modelling. All terms were initially developed by the first author. These terms were then reviewed by infectious disease modelling colleagues at Public Health England and Imperial College London to identify any obvious missing terms. All terms were ultimately discussed and agreed with by all members of the primary research team. Iterative development of the search strategy ultimately yielded the final, optimised search strategy.
Time and resource constraints also precluded backward and forward citation searching within included papers. For instance, we did not anticipate the articles would include research participants or any kind of intervention. The a-priori selection criteria simply specified that papers would be included if they presented a mathematical model pertaining to the transmission of an infectious disease within a population, with a particular emphasis on models that present heterogeneous behaviour by agents. As per the scoping review framework, our study selection criteria developed as a function of increasing familiarity with the literature [ 16 ].
In this way we were able to both: a narrow the focus of the review, and; b reduce uncertainty in the selection process. For instance, we initially proposed to include papers identified within review articles as well as grey literature i. The decision to exclude models in which behaviour is exogenously determined i. For the purpose of this review, parental decision-making e. No limits were placed on the publication date of included articles.
These criteria substantially build upon those employed in an earlier review of the same topic specifically, the emphasis on individual, endogenous decision making [ 20 ]. Furthermore, although developed and implemented independently of one another, our criteria share a common impetus with the criteria developed and employed in another recent review of human behaviour within infectious disease models [ 21 ].
Considered together, these criteria therefore allowed us to ensure, as far as possible, that the mechanisms for modelling human behaviour reviewed herein reflect attempts to consider the kind of complex, individual processes underlying human behaviour identified in the psychological literature e. All papers that were selected for inclusion in the review were subjected to a standardised data extraction procedure that was developed by the first author in the first instance, and was agreed by all other authors.
This procedure was revised and extended twice: once during an interim presentation of the review outcomes to a team of infectious disease modellers and behavioural economists on 29th February , and once during a Public Involvement workshop on 30th September Ultimately, the following information was extracted from all papers: authors; date of publication; the type of model used; the disease that is modelled; the behaviour that is modelled; how this behaviour is modelled; whether and how information or awareness spread is modelled; whether and how fading or decaying memory is modelled; whether and what theoretical background for behaviour change is provided; whether and what comparisons there were between the model that incorporates endogenous human behaviour and a control model i.
The extraction criteria are presented in their entirety in the Additional file 1. In the interests of brevity, we have focused on the selection of the data that pertains to our central research questions in the main results section i. All other extracted data is available on request from the first author.
When combined with the 75 papers resulting from the initial PubMed search and the 40 additional records identified through other sources i. Following the removal of duplicates, records were retained for title, abstract, and brief full text in the case of papers about which the first author was uncertain screening.
At this stage, the screening process became iterative as the criteria developed and became more exclusive. The 50 remaining unsure papers were both reviewed by the first author, and were referred to the second and third authors for review. Of these 50 papers, 39 were excluded at this stage, leaving 90 papers for full-text assessment. Following this review process, the exclusion criteria were refined for the final time and were applied by the first author to all papers previously excluded during iteration 2 onwards.
This revealed 28 formerly excluded papers that were re-included for full text analysis. Overall there were papers 90 following unsure paper assessment plus 28 formerly excluded papers subjected to thorough full-text assessment. Following this final, full-text assessment 42 papers were retained and included in the qualitative synthesis. As mentioned in the Introduction, human behaviour is not straightforward to predict. We therefore opted to focus on papers that included complex, individual decision making with regard to the adoption or avoidance of behaviour. A good example of such decision making processes involves a model in which susceptible agents calculate the cost and benefits of both risk-taking and protective behaviours by drawing a random sample of model agents, in conjunction with their previous estimates, to draw inferences about disease prevalence and inform their decision making [ 22 ].
Similarly, Fenichel et al. Similarly, Bhattacharyya and Bauch [ 27 ] present a model of vaccination in which the payoff for an individual vaccinating in a given week is based on the average vaccine coverage across the entire population. Overall, papers of this nature were included as they represent the complex nature of behavioural decision making, albeit with some more population level input.
There were also additional included papers that met the individual level behaviour requirement, but that did not explicitly include the elements of decision making discussed above. This is in contrast to models, excluded as population level, in which individuals behaviour is based on the general, modeller-defined proportion of their contacts who are engaged in that behaviour e. Funk, Gilad, Watkins, and Jansen [ 34 ] instead present a more individual-level model of awareness spreading. It therefore represents a more individual level representation of behaviour than the standard contact — infection models described above.
As detailed previously, 42 papers were retained following full text analysis [ 22 — 29 , 34 — 67 ]. The full citation list of included papers is included in the Additional file 1. In this section we present data relating to the date of publication and discuss aspects of the model design used within the included papers e. The full extracted data is available on request from the first author. Indeed, of the 19 included papers published in —, seven explicitly concerned influenza seasonal or epidemic or an influenza-like infection [ 25 , 28 , 29 , 37 , 42 , 58 , 65 ]. Number of papers presenting infectious disease transmission models with endogenous behaviour change, by year of publication.
The majority of the models described in the included papers did not specify the disease to which they related 22 of 42 papers [ 22 , 23 , 24 , 26 , 27 , 34 , 36 , 41 , 43 , 45 , 50 , 51 , 52 , 53 , 57 , 59 , 60 , 61 , 62 , 63 , 64 , 66 ]. Where models did relate to a specific disease, this was most commonly influenza or an influenza-like-infection [ 25 , 28 , 29 , 35 , 37 , 42 , 46 , 48 , 58 , 65 , 67 ]. A smaller, yet substantial proportion of cases either modelled one or more general behavioural responses 10 cases [ 28 , 29 , 35 , 37 , 39 , 47 , 48 , 49 , 58 , 67 ] or did not specify the behaviour that was being modelled 1 case [ 34 ].
Footnote 2 There was, therefore, very little variation in the type and nature of self-protective behaviour presented in the included papers. A range of different methods were employed across the included papers, with some papers either detailing multiple, combined models, or incorporating components from different modelling methods. It is therefore difficult to precisely quantify the specific models that were most commonly used across all papers.
There are, however, some broad conceptual or methodological similarities that can be outlined.
First, the majority of these papers employed a compartmental model e. Second, network modelling components i. Third, a substantial proportion of the papers explicitly incorporated economic or game theoretic elements within their infectious disease models i. Finally, a common approach to modelling the spread of individual protective behaviour during an infectious disease outbreak was to include more than one model in the analysis [ 24 , 25 , 26 , 27 , 35 , 37 , 46 , 47 , 53 , 54 , 55 , 61 , 62 , 65 , 66 , 67 ]; for instance, a behavioural model e.
As the nature of our review precluded the extraction of detailed PICOS related information, an extensive discussion of comparisons and outcomes was not appropriate. Moreover, the relatively large number of papers included in this review precludes an in depth assessment of the results from all individual studies. Our analysis instead focused on a summary and synthesis of the extracted data related explicitly to the modelling of human behaviour. During data extraction, extensive information concerning the method of modelling human behaviour was collated. In all bar six [ 28 , 29 , 34 , 37 , 46 , 49 ] of the 42 included papers, behaviour was modelled using either a cost-benefit calculation [ 22 , 23 , 26 , 27 , 35 , 36 , 39 , 40 , 41 , 42 , 44 , 45 , 47 , 48 , 50 , 51 , 53 , 56 , 57 , 60 , 61 , 63 , 65 ], behavioural imitation [ 64 ], or an integration of the two [ 24 , 25 , 38 , 43 , 52 , 54 , 55 , 58 , 59 , 62 , 66 , 67 ].
Typically, the cost-benefit calculation involves agents considering the payoff of comparing the utilities associated with engaging in protective behaviour against the utilities associated with remaining susceptible e. These prevalence-based utilities can be based on information from a single modelled infection season e. For example, whole model or local contact infection prevalence can influence the risk of infection e.
Behavioural change models for infectious disease transmission: a systematic review (2010–2015)
The method of incorporating behavioural imitation varied across the models, but commonly involved either a prevalence-based mechanism e. All bar one of the 13 models that incorporated behavioural imitation also incorporated a mechanism of cost-benefit calculation [ 64 ].
- Modeling Influenza Vaccination Behavior Via Inductive Reasoning Games.
- Man in the Mirror (Hold On Be Strong).
- Task Analysis Methods for Instructional Design.
- An Evaluation of the Role of Microbiological Criteria for Foods and Food Ingredients.
- Featured channels?
This either involved incorporating two distinct strategies—one for imitation and one for cost-benefit calculation for example, with the distribution of strategy within the model determined by a static parameter [ 52 ] —or the incorporation of both imitation and cost-benefit calculation together. For example, individuals may select another individual within their model i. The final remaining imitation strategy that did not incorporate both cost-benefit calculations and behavioural imitation models a situation in which individuals can observe the health status of others and are more likely to adopt the behaviour of a healthy person than of an unhealthy person regardless of whether this behaviour is careful or risky [ 64 ].
The remaining six models used a range of different strategies for modelling behaviour, including: information-dependent disease transmission that varies as a function of the number of individuals the information has previously travelled through [ 34 ]; behavioural strategies dependent upon population class e.
Further exploration of the data presented in the included papers was conducted in order to identify the key constructs that contributed to the modelling of human behaviour. This is not intended as an exhaustive list of all constructs that may contribute to behaviour, but instead represents the constructs that were identified as most central to the modelling of human behaviour within each paper.
Similarly, the example references are provided below to highlight each construct, this is not an exhaustive list of all identified papers employing these constructs; more detail can be found in the Additional file 2. These are the most commonly applied constructs within the included literature, and are most typically incorporated as part of the cost-benefit calculation models discussed above.
Some of the included papers also include social norm proxies within their models. First, individuals may identify the number of behavioural adopters within their contacts and modify their own behaviour if this proportion reaches a given threshold [ 28 , 29 ].
- Mathematical and statistical explorations in disease modelling and public health | ICTS.
- Evaluation Essentials: From A to Z.
- Modelling the influence of human behaviour on the spread of infectious diseases: a review..
- The talk of the town - Research - Royal Holloway, University of London?
Second, social norms are also represented as a modification to the payoffs for engaging in protective behaviour, that is, the payoff for engaging in a protective behaviour varies as a function of the number of individuals within the population that are also engaging in that behaviour [ 54 , 55 ]. Behavioural comparison and imitation represent one further key social construct in the modelling of human behaviour.
As outlined above, this can involve comparison with another individual randomly selected from the entire population, e. Similarly, the transmission of information, cultural traits, or awareness from person to person can also constitute a social construct; for example, when this transmission involves an awareness of how many contacts information has passed through [ 34 ], or trait transmission as a function of perceived health status of a contact [ 64 ].
The following additional constructs also contributed to the modelling of health-protective behaviour in the included papers: demographic information e. Every paper included in our analysis presented background literature to support the modelling of protective behaviour. As our primary concern is to explore the extent to which psychological constructs and theories have been incorporated into infectious disease models, all papers were examined to determine whether they cited psychological health behaviour theories e. Close examination of the papers revealed two broad additional classifications of background literature related to human behaviour: economic literature e.
All bar two of the included papers [ 28 , 37 ] contained literature that fit one or more of these criteria; this indicates that we are unlikely to have missed a substantial literature when developing our classification. The number of papers containing each of the three classifications of behaviour change literature is presented in Fig. Footnote 3.
Note: individual papers included in the review may have incorporated more than one of these categories of literature and so may be numerically represented multiple times. Examination of Fig. However, the most commonly cited theory of behaviour change cited by four of the five identified papers [ 44 , 48 , 54 , 55 ] is the Health Belief Model HBM. In the first instance we were simply interested in whether the models detailed in our included papers were applied to, or parameterised by existing data, and if so, which data. In the first instance, 26 papers made reference to either previous literature or data in the parameterisation or validation fit of their models [ 23 , 25 , 27 , 28 , 29 , 35 , 36 , 37 , 38 , 42 , 43 , 44 , 45 , 46 , 48 , 49 , 50 , 53 , 54 , 55 , 56 , 57 , 58 , 59 , 63 , 65 ].
Of these papers, 21 explicitly use data sources within the paper excluding reference to academic papers [ 28 , 29 , 35 , 36 , 36 , 38 , 42 , 43 , 44 , 46 , 48 , 49 , 50 , 53 , 54 , 55 , 56 , 57 , 58 , 63 , 65 ]; our examination of these papers yielded three broad classifications of data sources. First, survey response data is used to accurately model human behaviour; for example, two notable examples make use of a behavioural survey in which participants were asked to list the number of friends from a maximum of 10 that would have to be vaccinated in order for the respondent to consider vaccination to develop individualised social thresholds for behaviour adoption [ 28 , 29 ].
Second, demographic e. Similarly, in [ 37 ], survey responses were used to determine household activity. Third, epidemiological data concerning vaccine uptake is used; for example, UK pertussis vaccine coverage data [ 55 ], and ICONA working group data concerning Italian MMR vaccine uptake data from to [ 43 ]. The majority of the included papers did not focus on a specific disease, although the most commonly modelled was influenza.
Consistent with the disease focus, the most commonly modelled protective behaviour was also influenza-related vaccination. The broad range of models employed in the included papers precludes any firm conclusions regarding best practice for modelling both human behaviour and infectious disease spread. Nevertheless, it was clear from our data extraction that a substantial proportion of papers employed a dual-model method; using compartmental models e.
Moreover, reflecting the importance of social considerations in the modelling of infectious disease spread, a number of models employed social modelling components e. A range of different cognitive and social constructs with an emphasis on the cognitive contributed to the modelling of human behaviour across papers. The cognitive constructs typically focused on the relative costs and benefits of remaining susceptible or engaging in protective behaviour.
These included: perceived or actual risk i. Very few of the included papers made explicit reference to psychological health behaviour theories when discussing human behaviour, relying instead upon literature from behavioural economics and infectious disease modelling. Finally, just under half of the papers included in our review made reference to behavioural data in their modelling. Through synthesising the outcomes of our review with the psychological behaviour change and health protection literatures, we develop three central recommendations for how modellers can ensure that human behaviour is incorporated in their infectious disease models in a realistic and representative fashion: The role of psychological theory; the importance of the social world, and the use of behavioural data.
Broadly speaking, the emphasis on cognitive components within the included papers corresponds well with the psychological literature on health behaviour change. For example, the Integrative Model of Behavioural Prediction includes behavioural belief, perceived risk, normative belief, and efficacy belief components [ 70 ]. The social constructs identified in our review particularly behavioural imitation and social norms are also represented in several psychological theories of behaviour change e. However, as previously noted, very few of the included papers made explicit reference to these theories.
EcoServices: Disease risks
As a point of reference, recent work has identified 83 theories of behaviour change from across the social sciences [ 12 ]. Furthermore, although there is overlap in the constructs used [ 12 ], different models have been designed to reflect contextually-specific predictors of behaviour. For example, the HBM cited most commonly by papers included in this review [ 44 , 48 , 54 , 55 ] was designed to help understand the predictors of preventative behaviour in responses to a health threat [ 12 ], thus making it thoroughly appropriate for application within the current context.
However, both PMT and EPPM were also designed to help understand predictors of behaviour in this context, but with a particular focus on emotional responses i. There is, therefore, a wealth of theoretical literature concerning predictors of behaviour and behaviour change within the social sciences that could be drawn upon to inform the modelling of self-protective health behaviour.
Two papers cited within the current review provide an excellent example of how infectious disease transmission modelling can incorporate a more nuanced representation of human behavioural decision making. Specifically, these models combine statistical modelling specifically logistic regression modelling based on a combination of previous literature and survey data with agent-based modelling techniques, to present detailed models of infectious disease transmission that incorporate the HBM [ 44 , 48 ].
However, there is an inevitable compromise between striving for a realistic presentation of human behaviour, and the requirement and constraints of modelling [ 44 ]. Samenvatting This volume presents infectious diseases modeled mathematically, taking seasonality and changes in population behavior into account, using a switched and hybrid systems framework.
The scope of coverage includes background on mathematical epidemiology, including classical formulations and results; a motivation for seasonal effects and changes in population behavior, an investigation into term-time forced epidemic models with switching parameters, and a detailed account of several different control strategies. The main goal is to study these models theoretically and to establish conditions under which eradication or persistence of the disease is guaranteed.
In doing so, the long-term behavior of the models is determined through mathematical techniques from switched systems theory. Numerical simulations are also given to augment and illustrate the theoretical results and to help study the efficacy of the control schemes. Toon meer Toon minder. Recensie s This book focuses on infectious disease mathematical models, taking seasonality and changes in population behavior into account, using a switched and hybrid systems framework.
This book is strongly recommended to graduate level students with a background in dynamic system or epidemic modeling and an interest in mathematical biology, epidemic models, and physical problems exhibiting a mixture of continuous and discrete dynamics. Hemang B. Panchal, Doody's Book Reviews, April, This book presents a new type of switched model for the spread of infectious diseases.