Incorporating social vulnerability in infectious disease mathematical modelling: a scoping review

Abstract

Background: Highlighted by the rise of COVID-19, climate change, and conflict, socially vulnerable populations are least resilient to disaster. In infectious disease management, mathematical models are a commonly used tool. Researchers should include social vulnerability in models to strengthen their utility in reflecting real-world dynamics. We conducted a scoping review to evaluate how researchers have incorporated social vulnerability into infectious disease mathematical models.

Methods: The methodology followed the Joanna Briggs Institute and updated Arksey and O'Malley frameworks, verified by the PRISMA-ScR checklist. PubMed, Clarivate Web of Science, Scopus, EBSCO Africa Wide Information, and Cochrane Library were systematically searched for peer-reviewed published articles. Screening and extracting data were done by two independent researchers.

Results: Of 4075 results, 89 articles were identified. Two-thirds of articles used a compartmental model (n = 58, 65.2%), with a quarter using agent-based models (n = 24, 27.0%). Overall, routine indicators, namely age and sex, were among the most frequently used measures (n = 42, 12.3%; n = 22, 6.4%, respectively). Only one measure related to culture and social behaviour (0.3%). For compartmental models, researchers commonly constructed distinct models for each level of a social vulnerability measure and included new parameters or influenced standard parameters in model equations (n = 30, 51.7%). For all agent-based models, characteristics were assigned to hosts (n = 24, 100.0%), with most models including age, contact behaviour, and/or sex (n = 18, 75.0%; n = 14, 53.3%; n = 10, 41.7%, respectively).

Conclusions: Given the importance of equitable and effective infectious disease management, there is potential to further the field. Our findings demonstrate that social vulnerability is not considered holistically. There is a focus on incorporating routine demographic indicators but important cultural and social behaviours that impact health outcomes are excluded. It is crucial to develop models that foreground social vulnerability to not only design more equitable interventions, but also to develop more effective infectious disease control and elimination strategies. Furthermore, this study revealed the lack of transparency around data sources, inconsistent reporting, lack of collaboration with local experts, and limited studies focused on modelling cultural indicators. These challenges are priorities for future research.

Keywords: Global health; Health disparities; Health equity; Health inequalities; Infectious disease; Mathematical modelling; Social determinants of health; Social vulnerability.

Fig. 1

Example of stratifying overall in compartmental modelling: S’₁ and I’₁ is a set of equations for the susceptible and infectious housed population, respectively. S’₂ and I’₂ is a set of equations for the susceptible and infectious unhoused population, respectively. β₂ represents interactive transmission between the models Source: Romaszko J, Siemaszko A, Bodzioch M, Buciński A, Doboszyńska A. Active Case Finding Among Homeless People as a Means of Reducing the Incidence of Pulmonary Tuberculosis in General Population. Adv Exp Med Biol. 2016;911:67–76. https://doi.org/10.1007/5584_2016_225. PMID: 26,992,399

Fig. 2

Example of stratifying within in compartmental modelling: Behaviour was modelled in two ways: an increase in the proportion of the population vaccinated led to an increase in non-compliance of preventative measures, and an increase in deaths resulted in an increase in compliance of preventative measures. Susceptible (S) and vaccinated (V) compartments also had susceptible and vaccinated non-compliant compartments (S^NC, V^NC, respectively) Source: Source: Gozzi N, Bajardi P, Perra N. The importance of non-pharmaceutical interventions during the COVID-19 vaccine rollout. PloS Comput Biol. 2021: 17(9):e1009346. https://doi.org/10.1371/journal.pcbi.1009346

Fig. 3

Example of influencing or including parameters in compartmental modelling: The transmission rate β is influenced by d_t, a social distancing index between zero and one. D_t comprises measures of civic capital (k), the perceived riskiness of the virus (r), and policy response adopted over the course of the pandemic (p). Included are two civic capital parameters: one based on the internalisation of the externalities (η, belief in other’s well-being) and another on law-abidingness (v) Source: Durante R, Guiso L, Gulino G. Asocial capital: Civic culture and social distancing during COVID-19. J Public Econ. 2021 Feb;194:104,342. https://doi.org/10.1016/j.jpubeco.2020.104342. Epub 2021 Jan 4. PMID: 35,702,335; PMCID: PMC9186120

Fig. 4

Flow-chart of database searching and article screening process

Fig. 5

Number of published articles which included an infectious disease mathematical model with social vulnerability incorporated by publication date (to 10 June 2022, N = 89)

Fig. 6

Study setting(s) of published articles which included an infectious disease mathematical model with social vulnerability incorporated (N = 89). Twenty-one articles were written about a hypothetical “Setting X” and 12 articles were written about a collection of countries or region(s)

Fig. 7

Study setting by mathematical model type in published articles which included an infectious disease mathematical model with social vulnerability incorporated (N = 89)

Fig. 8

Distribution of the social vulnerability indicators (n = 342) by category type in published articles which included an infectious disease mathematical model with social vulnerability incorporated (N = 89)

Fig. 9

Methods of including social vulnerability indicators in infectious disease compartmental models. See “ Methods” section for an explanation of the approaches