Modelling the impact of social protection on tuberculosis: the S-PROTECT project

Background

Tackling the social determinants of Tuberculosis (TB) through social protection is a key element of the post-2015 End TB Strategy. However, evidence informing policies are still scarce. Mathematical modelling has the potential to contribute to fill this knowledge gap, but existing models are inadequate. The S-PROTECT consortium aimed to develop an innovative mathematical modelling approach to better understand the role of social protection to improve TB care, prevention and control.

Methods

S-PROTECT used a three-steps approach: 1) the development of a conceptual framework; 2) the extraction from this framework of three high-priority mechanistic pathways amenable for modelling; 3) the development of a revised version of a standard TB transmission model able to capture the structure of these pathways. As a test case we used the Bolsa Familia Programme (BFP), the Brazilian conditional cash transfer scheme.

Results

Assessing one of these pathways, we estimated that BFP can reduce TB prevalence by 4% by improving households income and thus their nutritional status. When looking at the direct impact via malnutrition (not income mediated) the impact was 33%. This variation was due to limited data availability, uncertainties on data transformation and the pathway approach taken. These results are preliminary and only aim to serve as illustrative example of the methodological challenges encountered in this first modelling attempt, nonetheless they suggest the potential added value of integrating TB standard of care with social protection strategies.

Conclusions

Results are to be confirmed with further analysis. However, by developing a generalizable modelling framework, S-PROTECT proved that the modelling of social protection is complex, but doable and allowed to draw the research road map for the future in this field.

Electronic supplementary material

The online version of this article (10.1186/s12889-018-5539-x) contains supplementary material, which is available to authorized users.

How can mathematical models advance tuberculosis control in high HIV prevalence settings?

Existing approaches to tuberculosis (TB) control have been no more than partially successful in areas with high human immunodeficiency virus (HIV) prevalence. In the context of increasingly constrained resources, mathematical modelling can augment understanding and support policy for implementing those strategies that are most likely to bring public health and economic benefits. In this paper, we present an overview of past and recent contributions of TB modelling in this key area, and suggest a way forward through a modelling research agenda that supports a more effective response to the TB-HIV epidemic, based on expert discussions at a meeting convened by the TB Modelling and Analysis Consortium. The research agenda identified high-priority areas for future modelling efforts, including 1) the difficult diagnosis and high mortality of TB-HIV; 2) the high risk of disease progression; 3) TB health systems in high HIV prevalence settings; 4) uncertainty in the natural progression of TB-HIV; and 5) combined interventions for TB-HIV. Efficient and rapid progress towards completion of this modelling agenda will require co-ordination between the modelling community and key stakeholders, including advocates, health policy makers, donors and national or regional finance officials. A continuing dialogue will ensure that new results are effectively communicated and new policy-relevant questions are addressed swiftly.

On diffusion processes with variable drift rates as models for decision making during learning

We investigate Ornstein-Uhlenbeck and diffusion processes with variable drift rates as models of evidence accumulation in a visual discrimination task. We derive power-law and exponential drift-rate models and characterize how parameters of these models affect the psychometric function describing performance accuracy as a function of stimulus strength and viewing time. We fit the models to psychophysical data from monkeys learning the task to identify parameters that best capture performance as it improves with training. The most informative parameter was the overall drift rate describing the signal-to-noise ratio of the sensory evidence used to form the decision, which increased steadily with training. In contrast, secondary parameters describing the time course of the drift during motion viewing did not exhibit steady trends. The results indicate that relatively simple versions of the diffusion model can fit behavior over the course of training, thereby giving a quantitative account of learning effects on the underlying decision process.