Reviewed by
Rainer Hilscher
RTI International
“Tipping Points: Modelling Social Problems and Health” by John Bissell and colleagues is the outcome of a conference held at the University of Durham (UK) held in 2012 on “Modelling Social Problems and Health” as part of the Tipping Point Project. It is a complicated book. Complicated in the sense that its interesting individual chapters do not present a coherent whole. With this book the whole is less than its parts because neither the theoretical approach of individual chapters nor their respective research domains add up to a convincing argument.Chapters in this edited book range from mathematical modelling (e.g. chapters 1, 2, 4, 6, and 10) to the descriptive (e.g. chapters 3, 7, 11, and 12). To turn the title on itself I was waiting for the moment where the sequence of chapters would become significant enough to cause a major shift in my impression of this book. No such tipping occurred.
The title promises a book with a strong focus on tipping points and a secondary focus on modelling social problems and health. As such I would have at least expected an introductory chapter on tipping points and possibly a second introductory chapter on modelling in the social and public health domain. Instead we get an introductory chapter that provides a promising sounding frame around summaries of the individual chapters in this book. The reader is then thrown in at the deep end with chapter 1 on “Generalized Compartmental Modelling of Health Epidemics”. To study the transmission of smoking behaviour in a population John Bissell (first listed editor of this book) implements such a generalised compartmental model in chapter 1. He shows that the inclusion of multiple “infectious activities” (his term) leads to a higher number of system steady states (compared to the base compartmental model) resulting in bistability and discontinuous tipping points.
Chapter 2 explores the uncertainty associated with the input data for the smoking model presented in chapter 1 using probabilistic sensitivity analysis. Chapter 3 stays with the tobacco theme but completely switches gear. It provides a descriptive analysis of gender disparities in smoking rates in the UK where the North East stands in stark contrast to the rest of the UK with the reduction in smoking prevalence is much stronger in men than women. The author argues that the social and cultural context is key to understanding why more women smoke in the NE than anywhere else in the UK.
Chapters 4 to 6 leave tobacco behind and delve into modelling medical contexts. Chapter 4 looks at performance monitoring in cardiac surgery and the authors conclude that fitting a prediction model once to data in this domain results in a degrading fit due to calibration drift and propose a dynamic approach (periodic model refitting). This drift, the authors argue, is caused by the inherent dynamic nature of cardiac surgery. Chapter 5 introduces a Bayesian modelling approach to analysing real-time electrocardiogram (ECG) data to continuously monitor the underlying health of a person. Chapter 6 outlines the basic philosophy of mathematical modelling and illustrates this outline with models of drug-eluting stents, blood flow, and a fluorescence capillary fill-device (best-known application is pregnancy test). This chapter discusses how mathematical modelling supports the design of complex products (e.g. by identifying sensitive parameters through sensitivity analysis). Neither chapter 5 or 6 discuss tipping points.
Chapter 7 outlines of a “mathematical theory of social systems” based on the Black Swan concept and complexity aspects of social systems (that are “definitely complex” according to the authors and it aims to answer five key questions that are essential, according to the authors, for a mathematical theory of social systems (relevant complexity features? appropriate mathematical tools? do Black Swans result from interactions? can models be validated? conceptual paths leading to mathematical theory?). This least focused of all chapters in the book refers to a lot of interesting theories and concepts (evolution, complexity, Black Swans, model validation) but ultimately fails to present a convincing, coherent method of modelling social systems.
Chapter 8 presents an agent based model to study the conditions under which a given population evolves locally, spatially separated sub-cultures to become either more homogenous or heterogeneous. This chapter considers collective human decision making as a driver of the evolution of either heterogeneity or homogeneity and it relates these dynamics to policy making and how to make policies more effective.
Chapter 9 presents an introduction to modelling cultural evolution and gene-culture co-evolution. For this chapter, the authors focus on health behaviours in general and spread of self-medication of disease. A second approach to modelling human culture that is discussed in this chapter is derived from infectious disease modelling; here the authors discuss the spread of drinking through a population using an adapted SIR model (susceptible, infectious, recovered). Key to this approach is the basic reproduction number (average number of secondary infected cases generated by the introduction of a single infected individual into a population) with R0 = 1, in the drinking model, constituting a tipping point where a problem drinking culture becomes fixed.
Chapter 10 introduces a compartmental model of trend transmission dynamics (with potential trend followers and trend pioneers). This model replaces the traditional, epidemiology inspired, bilinear recruitment term with a non-linear conformity bias function with the result that the model now exhibits tipping points where niche trends gain sudden mass appeal. Such tipping points are not observed in the traditional model without the conformity bias function.
Chapter 11 and 12 are subtitled “Resilience of tipping points” which sounds pop-scientifically cool but what does it actually mean for a tipping point to be resilient? Chapter 11 is a descriptive chapter about risk and resilience in general and risk assessment of suicide in particular. It advocates the study of contextual factors rather than looking at individual interactions of actors in these contexts. The authors advocate to rather study environments that promote suicides in prison than to focus on individuals and their mental states. This chapter also discusses biases in human decision making without making any possible connections with tipping points. Chapter 12, finally, is a descriptive chapter about risk governance in health care systems (never-events in psychiatric hospital settings and adaptation of health care systems to extreme weather events). It uses a pop-science definition of tipping points (Gladwell) and provides a philosophical discussion of complex systems. This chapter argues that ability to predict the past does not mean we can predict the future after the occurrence of tipping points that cause changed conditions.
Researchers already familiar with the approaches and cited literature in the respective chapters will find novel research (as of 2012; the book is published in 2015) and fascinating applications of that research. Those researchers with a broader interest in understanding and modelling tipping points in social and public health contexts will have a challenging time distilling generalizable knowledge from these chapters. This is particularly true because several chapters seem to have tipping points injected in them as an afterthought (to make them “about” tipping points) rather than as their specific focus.
Return to Contents
of this issue