Dynamic impact of social stratification and social influence on smoking prevalence by gender: An agent-based model

Model Choice for Quantitative Health Impact Assessment and Modelling: An Expert Consultation and Narrative Literature Review

BACKGROUND

Health impact assessment (HIA) is a widely used process that aims to identify the health impacts, positive or negative, of a policy or intervention that is not necessarily placed in the health sector. Most HIAs are done prospectively and aim to forecast expected health impacts under assumed policy implementation. HIAs may quantitatively and/ or qualitatively assess health impacts, with this study focusing on the former. A variety of quantitative modelling methods exist that are used for forecasting health impacts, however, they differ in application area, data requirements, assumptions, risk modelling, complexities, limitations, strengths, and comprehensibility. We reviewed relevant models, so as to provide public health researchers with considerations for HIA model choice.

METHODS

Based on an HIA expert consultation, combined with a narrative literature review, we identified the most relevant models that can be used for health impact forecasting. We narratively and comparatively reviewed the models, according to their fields of application, their configuration and purposes, counterfactual scenarios, underlying assumptions, health risk modelling, limitations and strengths.

RESULTS

Seven relevant models for health impacts forecasting were identified, consisting of (i) comparative risk assessment (CRA), (ii) time series analysis (TSA), (iii) compartmental models (CMs), (iv) structural models (SMs), (v) agent-based models (ABMs), (vi) microsimulations (MS), and (vii) artificial intelligence (AI)/machine learning (ML). These models represent a variety in approaches and vary in the fields of HIA application, complexity and comprehensibility. We provide a set of criteria for HIA model choice. Researchers must consider that model input assumptions match the available data and parameter structures, the available resources, and that model outputs match the research question, meet expectations and are comprehensible to end-users.

CONCLUSION

The reviewed models have specific characteristics, related to available data and parameter structures, computational implementation, interpretation and comprehensibility, which the researcher should critically consider before HIA model choice.

Systems Thinking and Complexity Science Methods and the Policy Process in Non-communicable Disease Prevention: A Systematic Scoping Review

BACKGROUND

Given the complex determinants of non-communicable diseases (NCDs), and the dynamic policy landscape, researchers and policymakers are exploring the use of systems thinking and complexity science (STCS) in developing effective policies. The aim of this review is to systematically identify and analyse existing applications of STCS-informed methods in NCD prevention policy.

METHODS

Systematic scoping review: We searched academic databases (Medline, Scopus, Web of Science, EMBASE) for all publications indexed by 13 October 2020, screening titles, abstracts and full texts and extracting data according to published guidelines. We summarised key data from each study, mapping applications of methods informed by STCS to policy process domains. We conducted a thematic analysis to identify advantages, limitations, barriers and facilitators to using STCS.

RESULTS

4681 papers were screened and 112 papers were included in this review. The most common policy areas were tobacco control, obesity prevention and physical activity promotion. Methods applied included system dynamics modelling, agent-based modelling and concept mapping. Advantages included supporting evidence-informed decision-making; modelling complex systems and addressing multi-sectoral problems. Limitations included the abstraction of reality by STCS methods, despite aims of encompassing greater complexity. Challenges included resource-intensiveness; lack of stakeholder trust in models; and results that were too complex to be comprehensible to stakeholders. Ensuring stakeholder ownership and presenting findings in a user-friendly way facilitated STCS use.

CONCLUSION

This review maps the proliferating applications of STCS methods in NCD prevention policy. STCS methods have the potential to generate tailored and dynamic evidence, adding robustness to evidence-informed policymaking, but must be accessible to policy stakeholders and have strong stakeholder ownership to build consensus and change stakeholder perspectives. Evaluations of whether, and under what circumstances, STCS methods lead to more effective policies compared to conventional methods are lacking, and would enable more targeted and constructive use of these methods.

Agent-Based Modelling of Health Inequalities following the Complexity Turn in Public Health: A Systematic Review

There is an increasing focus on the role of complexity in public health and public policy fields which has brought about a methodological shift towards computational approaches. This includes agent-based modelling (ABM), a method used to simulate individuals, their behaviour and interactions with each other, and their social and physical environment. This paper aims to systematically review the use of ABM to simulate the generation or persistence of health inequalities. PubMed, Scopus, and Web of Science (1 January 2013-15 November 2022) were searched, supplemented with manual reference list searching. Twenty studies were included; fourteen of them described models of health behaviours, most commonly relating to diet (n = 7). Six models explored health outcomes, e.g., morbidity, mortality, and depression. All of the included models involved heterogeneous agents and were dynamic, with agents making decisions, growing older, and/or becoming exposed to different health risks. Eighteen models represented physical space and in eleven models, agents interacted with other agents through social networks. ABM is increasingly contributing to our understanding of the socioeconomic inequalities in health. However, to date, the majority of these models focus on the differences in health behaviours. Future research should attempt to investigate the social and economic drivers of health inequalities using ABM.

Health system efficiency in OECD countries: dynamic network DEA approach

BACKGROUND

In recent years, measuring and evaluating the efficiency of health systems has been explored in the context of seeking resources to ensure the sustainability of 'countries' health and social systems and addressing various crises in the health sector. The study aims to quantify and compare the efficiency of OECD health systems in 2000, 2008, and 2016. The contribution to research in the field of efficiency in the healthcare system can be seen in the application of Dynamic Network Data Envelopment Analysis (DNDEA), which help us to analyse not only the overall efficiency of the healthcare system but analyse the overall efficiency as the result of the efficiencies of individual interconnected areas (public and medical care area). By applying the DNDEA model, we can realise the analysis not only within one year, but we can find out if the measures and improvements taken in the healthcare sector have a positive impact on its efficiency in a later period (eight-year interval).

METHODS

The analysis focuses on assessing the efficiency of the health systems of OECD countries over three periods: 2000, 2008, and 2016. Data for this study were derived from the existing OECD database, which provides aggregated data on OECD countries on a comparable basis. In this way, it was possible to compare different countries whose national health statistics may have their characteristics. The input-oriented Dynamic Network Data Envelopment Analysis model was used for data processing. The efficiency of OECD health systems has been analysed and evaluated comprehensively and also separately in two divisions: public health sub-division and medical care sub-division. The analysis combines the application of conventional and unconventional methods of measuring efficiency in the health sector.

RESULTS

The results for the public health sub-division, medical care sub-division and overall health system for OECD countries under the assumption of constant returns to scale indicate that the average overall efficiency was 0.8801 in 2000, 0.8807 in 2008 and 0.8472 in 2016. The results of the input-oriented model with the assumption of constant returns to scale point to the overall average efficiency of health systems at the level of 0.8693 during the period. According to the Malmquist Index results, the OECD countries improved the efficiency over the years, with performance improvements of 19% in the public health division and 8% in the medical care division.

CONCLUSIONS

The results of the study are beneficial for health policymakers to assess and compare health systems in countries and to develop strategic national and regional health plans. Similarly, the result will support the development of international benchmarks in this area. The issue of health efficiency is an intriguing one that could be usefully explored in further research. A greater focus on combining non-parametric and parametric models could produce interesting findings for further research. The consistency in the publication and updating of the data on health statistics would help us establish a greater degree of accuracy.

Agent-Based Modeling for Implementation Research: An Application to Tobacco Smoking Cessation for Persons with Serious Mental Illness

Background

Implementation researchers have sought ways to use simulations to support the core components of implementation, which typically include assessing the need for change, designing implementation strategies, executing the strategies, and evaluating outcomes. The goal of this paper is to explain how agent-based modeling could fulfill this role.

Methods

We describe agent-based modeling with respect to other simulation methods that have been used in implementation science, using non-technical language that is broadly accessible. We then provide a stepwise procedure for developing agent-based models of implementation processes. We use, as a case study to illustrate the procedure, the implementation of evidence-based smoking cessation practices for persons with serious mental illness (SMI) in community mental health clinics.

Results

For our case study, we present descriptions of the motivating research questions, specific models used to answer these questions, and a summary of the insights that can be obtained from the models. In the first example, we use a simple form of agent-based modeling to simulate the observed smoking behaviors of persons with SMI in a recently completed trial (IDEAL, Comprehensive Cardiovascular Risk Reduction Trial in Persons with SMI). In the second example, we illustrate how a more complex agent-based approach that includes interactions between patients, providers and site administrators can be used to provide guidance for an implementation intervention that includes training and organizational strategies. This example is based in part on an ongoing project focused on scaling up evidence-based tobacco smoking cessation practices in community mental health clinics in Maryland.

Conclusion

In this paper we explain how agent-based models can be used to address implementation science research questions and provide a procedure for setting up simulation models. Through our examples, we show how what-if scenarios can be examined in the implementation process, which are particularly useful in implementation frameworks with adaptive components.

Agent-based Modeling in Tobacco Regulatory Science: Exploring 'What if' in Waterpipe Smoking

Objectives

Waterpipe tobacco smoking (WTS) is an emerging public health crisis, particularly among youth and young adults. Different from the use of other tobacco products and e-cigarettes, WTS tends to be a social activity occurring among friends or persons associated with social networks. In this paper, we review a potential strategy for WTS-related research.

Methods

As a bottom-up computational model, agent-based modeling (ABM) can simulate the actions and interactions of agents, as well as the dynamic interactions between agents and their environments, to gain an understanding of the functioning of a system. ABM is particularly useful for incorporating the influence of social networks in WTS, and capturing people's space-time activity and the spatial distribution of WTS venues.

Results

Comprehensive knowledge of WTS-related behaviors at the individual level is needed to take advantage of ABM and use it to examine policies such as the interaction between WTS and cigarette smoking and the effect of flavors used in waterpipe tobacco. Longitudinal and WTS-specific surveys and laboratory experiments are particularly helpful to understand WTS basic mechanisms and elicit individual preferences, respectively.

Conclusions

We argue that the uniqueness of WTS makes ABM a promising tool to be used in WTS-related research, as well as understanding use of other tobacco products.

Introduction to the Theme Issue on Dynamics of Health Behavior: Revisiting Systems Science for Population Health

Systems science can help public health professionals to better understand the complex dynamics between factors affecting health behaviors and outcomes and to identify intervention opportunities. Despite their demonstrated utility in addressing health topics such influenza, tobacco control, and obesity, the associated methods continue to be underutilized by researchers and practitioners addressing health behaviors. This article discusses the growth of systems science methods (e.g., system dynamics, social network analysis, and agent-based modeling) in health research, provides a frame for the articles included in this themed issue, and closes with recommendations for enhancing the future of systems science and health behavior research. We argue that integrating systems sciences methods into health behavior research and practice is essential for improved population health and look forward to supporting the evolution of the field.

Social influence of e-cigarette smoking prevalence on smoking behaviours among high-school teenagers: Microsimulation experiments

The prevalence of electronic cigarette (e-cigarette) use has rapidly increased among young people, while conventional cigarette use has decreased in this age group. However, some evidence suggests that e-cigarette use is likely to induce conventional cigarette smoking. The present study explored the social influence of the prevalence of e-cigarette use in the peer network and in the general population as a potential mechanism by which e-cigarette use affects adolescents’ overall smoking behaviours. For this purpose, we developed an agent-based model in which young agents repeatedly choose to smoke conventional cigarettes and/or e-cigarettes, or to remain non-smokers. The choice is based on the agent’s evaluation of the utility derived from smoking and attitude towards smoking (‘openness’), which is influenced by smoking prevalence in the agent’s peer network and in the broader society. We also assumed a ‘crossover’ effect between the different types of smoking. The model was calibrated with United States National Youth Tobacco Survey data to reflect real-world numbers. We further simulated the prevalence of different types of smoking under counterfactual scenarios with different levels of openness and crossover effects. The models developed successfully reproduced actual prevalence trends in different types of smoking from 2011 to 2014. Openness to smoking is associated with a dramatic increase in e-cigarette smoking and especially in dual smoking, which cancels out the decline in sole conventional smoking. Larger crossover effects are associated with a higher prevalence of conventional smoking. The simulation results indicate that the social influence of the prevalence of e-cigarette use may influence young people to initiate or continue conventional cigarette smoking. Assessing the impact of e-cigarettes in the general population as a ‘healthier’ alternative to conventional smoking may require carefully monitoring trends in young people’s smoking behaviours.

Associations of social relationships with curative and preventive dental care use among young and middle-aged adults: Evidence from a population-based study in Japan

OBJECTIVES

Social relationships are important for diffusion of information and behaviours. Access to preventive dental care is more difficult than access to curative dental care under the Japanese universal healthcare system. Our objective was to examine whether social relationships were differentially associated with curative and preventive dental care use in Japan.

METHODS

A questionnaire survey was conducted between 2010 and 2011 among residents aged 25-50 years in Japanese metropolitan areas. Valid responses were obtained from 1919 men and 2207 women. Social relationships included social networks (membership of organizations and number of close ties) and social support (informational support and instrumental support). Poisson regression analyses with robust variance estimators were conducted to examine associations between each social relationship variable and curative dental care use or preventive dental care use, adjusted for covariates.

RESULTS

While 38.4% of men and 42.0% of women used curative dental care, 22.9% of men and 32.5% of women used preventive dental care in the past year. No measures of social relationships were associated with curative dental care use among men and women, except the number of close ties among men. By contrast, all measures of social relationships were associated with preventive dental care use among men; the multivariate-adjusted prevalence ratios (95% confidence intervals) of the highest compared with the lowest level of social relationships were 1.58 (1.18-2.13) for membership of organizations, 1.58 (1.24-2.00) for the number of close ties, 1.41 (1.10-1.82) for informational support and 1.30 (1.01-1.68) for instrumental support. Among women, no measures of social relationships were associated with preventive dental care use.

CONCLUSIONS

Social relationships were associated with preventive dental care use among men but not among women, while these were not associated with curative dental care use.

Complex adaptive systems: a new approach for understanding health practices

This article explores the potential of complex adaptive systems (CAS) theory to inform behaviour change research. A CAS describes a collection of heterogeneous agents interacting within a particular context, adapting to each other's actions. In practical terms, this implies that behaviour change is (1) socially and culturally situated; (2) highly sensitive to small baseline differences in individuals, groups, and intervention components; and (3) determined by multiple components interacting 'chaotically'. Two approaches to studying CAS are briefly reviewed. Agent-based modelling is a computer simulation technique that allows researchers to investigate 'what if' questions in a virtual environment. Applied qualitative research techniques, on the other hand, offer a way to examine what happens when an intervention is pursued in real-time, and to identify the sorts of rules and assumptions governing social action. Although these represent very different approaches to complexity, there may be scope for mixing these methods - for example, by grounding models in insights derived from qualitative fieldwork. Finally, I will argue that the concept of CAS offers one opportunity to gain a deepened understanding of health-related practices, and to examine the social psychological processes that produce health-promoting or damaging actions.

Socioeconomic Inequalities in Oral Health Among Unmarried and Married Women: Evidence From a Population-Based Study in Japan

Background

Socioeconomic inequalities in oral health have been reported in developed countries, but the influence of marital status has rarely been considered. Our aim was to examine marital status differentials in the association between socioeconomic status (SES) and oral health among community-dwelling Japanese women.

Methods

From 2010 to 2011, a questionnaire survey was conducted among residents aged 25–50 years in Japanese metropolitan areas. Valid responses were received from 626 unmarried women and 1,620 married women. Women’s own and husbands’ educational attainment and equivalent income were used to assess SES. Self-rated “fair” or “poor” oral health was defined as poor oral health. Multiple logistic regression analysis was conducted to examine which SES indicators were associated with oral health.

Results

The prevalence of poor oral health was 21.1% among unmarried women and 23.8% among married women. Among unmarried women, equivalent income was not associated with oral health, but women’s own education was significantly associated with oral health; the multivariate-adjusted odds ratio of poor oral health among those with high school education or lower compared to those with university education or higher was 2.14 (95% confidence interval, 1.19–3.87). Among married women, neither women’s own nor husbands’ education was associated with oral health, but equivalent income was significantly associated with oral health, particularly among housewives; the multivariate-adjusted odds ratio of poor oral health among those in the lowest compared with highest income quartile was 1.57 (95% confidence interval, 1.08–2.27).

Conclusions

These findings indicate that marital status should be considered when examining associations between SES and oral health among Japanese women.

Agent-Based Modeling in Public Health: Current Applications and Future Directions

Agent-based modeling is a computational approach in which agents with a specified set of characteristics interact with each other and with their environment according to predefined rules. We review key areas in public health where agent-based modeling has been adopted, including both communicable and noncommunicable disease, health behaviors, and social epidemiology. We also describe the main strengths and limitations of this approach for questions with public health relevance. Finally, we describe both methodologic and substantive future directions that we believe will enhance the value of agent-based modeling for public health. In particular, advances in model validation, comparisons with other causal modeling procedures, and the expansion of the models to consider comorbidity and joint influences more systematically will improve the utility of this approach to inform public health research, practice, and policy.

Distinct association between educational attainment and overweight/obesity in unmarried and married women: evidence from a population-based study in Japan

Background

Associations between education and obesity have been consistently reported among women in developed countries, but few studies have considered the influence of marital status and husbands’ education. This study aimed to examine differences in the association between education and overweight/obesity by marital status and to determine the contribution of husbands’ education to overweight/obesity among community-dwelling Japanese women.

Methods

A questionnaire survey was conducted from 2010 to 2011 among residents aged 25–50 years in Japanese metropolitan areas. Of 2145 women who agreed to participate and completed the survey, 582 were unmarried and 1563 were married. Overweight/obesity was defined as body mass index ≥25 kg/m². Multiple logistic regression analysis was conducted to determine whether women’s or their husbands’ education was associated with overweight/obesity after adjusting for age, work status, and equivalent income.

Results

The prevalence of overweight/obesity was 11.9% among unmarried women and 10.3% among married women. Women’s own education was significantly associated with overweight/obesity among unmarried women but not among married women. The multivariate-adjusted odds ratio of high school education or lower compared with university education or higher was 3.21 (95% confidence interval: 1.59–6.51) among unmarried women. Among married women, husbands’ education was significantly associated with overweight/obesity: women whose husbands’ educational attainment was high school or lower had significantly higher odds of overweight/obesity than did those whose husbands had a university education or higher (1.67, 95% confidence interval: 1.10–2.55). Among married women whose educational attainment was college or higher, women whose husbands’ educational attainment was high school or lower had a significantly higher risk for overweight/obesity when compared with women whose husbands’ educational attainment was college or higher.

Conclusions

Associations between women’s own education and overweight/obesity varied by marital status, and husbands’ educational level was important for married women’s overweight/obesity. These findings indicate that the social influences bound to educational background affect women’s overweight/obesity.

Modeling the Effects of E-cigarettes on Smoking Behavior: Implications for Future Adult Smoking Prevalence

BACKGROUND

Electronic cigarette (e-cigarette) use has increased rapidly in recent years. Given the unknown effects of e-cigarette use on cigarette smoking behaviors, e-cigarette regulation has become the subject of considerable controversy. In the absence of longitudinal data documenting the long-term effects of e-cigarette use on smoking behavior and population smoking outcomes, computational models can guide future empirical research and provide insights into the possible effects of e-cigarette use on smoking prevalence over time.

METHODS

Agent-based model examining hypothetical scenarios of e-cigarette use by smoking status and e-cigarette effects on smoking initiation and smoking cessation.

RESULTS

If e-cigarettes increase individual-level smoking cessation probabilities by 20%, the model estimates a 6% reduction in smoking prevalence by 2060 compared with baseline model (no effects) outcomes. In contrast, e-cigarette use prevalence among never smokers would have to rise dramatically from current estimates, with e-cigarettes increasing smoking initiation by more than 200% relative to baseline model estimates to achieve a corresponding 6% increase in smoking prevalence by 2060.

CONCLUSIONS

Based on current knowledge of the patterns of e-cigarette use by smoking status and the heavy concentration of e-cigarette use among current smokers, the simulated effects of e-cigarettes on smoking cessation generate substantially larger changes to smoking prevalence compared with their effects on smoking initiation.