The vision for this discussion paper is that we should work towards providing mathematical and statistical tools that can be used to inform and support highly effective, locally tailored interventions for these highly spatially heterogeneous infections. To be effective these analyses need to arise as part of an effective co-operation and collaboration between local stakeholders – community representatives, healthcare workers, (national) politicians and epidemiologists (modellers). It requires modelling teams working locally (sub-nationally), but located within regional centres of excellence, ideally tied to local laboratory (surveillance) capacity. This also requires that modellers overcome hurdles associated with communicating quantitative sciences to non-specialists, an issue which affects many aspects of public health, not just NTDs.
More local modelling has to fit with the roles of local authorities and community leaders., as well as integrating with current frameworks. All the components of health systems will move in one way or another once any change is effected on any of the building blocks (human resources, information, financing, governance, service delivery, etc.). Any move to create local capacity for NTD modelling should aim to strengthen other health system components. As data analysis and synthesising and modelling are generally useful to health care evaluation and planning, such integration is likely to prove beneficial. To some extent, the structure of such as system would depend on how NTD is situate within the organisation.
This network would not be addressing a philosophical or theoretical need. Programme managers are seeking to design the best intervention for their population and regularly ask for guidance on how they should design complex interventions in their situation. For example
-
How well is my intervention working?
-
How much longer will I need to hold it in place?
-
What combination of interventions should I be using where?
-
What resources and costs will I need to secure now and in the future?
-
What synergies are available between different interventions?
-
What are the opportunity costs of different interventions?
-
What is the impact on patient access and costs?
-
How large is the undiagnosed epidemic?
Of course, although data analysis/synthesis and modelling can address the effectiveness aspects of these questions, economic evaluation is required to address the cost aspects. This is a major undertaking in itself, and much of what we have written about modelling could also be written about economic evaluation: it is context sensitive and under-resourced, so that supra-national results are having to be used to make local decisions (for example,[22]).
Programme managers look to the World Health Organization (WHO) and other international bodies for guidance, but often there is a need to tailor to local need. For example, the decision of whether to introduce an expensive diagnostic depends on local cost-effectiveness[23]. Local capacity would also address problems of analysis being done far from the point of need, the need to keep data and analysis close to the populations to which they apply and for the analysis to be an active discussion targeted at improving the local intervention rather than for academic impact. Modellers would also need to have a clear understanding of local realities, rather than be led by pre-conceived ideas of how processes work.
The key to modelling, whatever the scale, is that it is based on accurate data. Data are always collected locally, although frequently then collated at a larger scale with much of the information potentially lost. A likely major advantage of moving modelling towards the point of data collection is that the perceived value of the data increases, which consequently drives up quality, completeness and timeliness, benefitting all levels of analysis.
Another key argument for local modelling is that NTDs are highly dependent on local ecologies and behaviours. Consequently, as national and regional control of NTDs is achieved, there will be “hot spots” of infection (as has been seen for malaria[24]) for which general interventions are insufficient or inappropriate. Following control and then local elimination, intervention programs maybe halted, but on-going well-designed surveillance and monitoring will be required to maintain these gains. Ultimately all NTD elimination will be local, so developing a local capacity now is both providing the opportunity to improve the design of current interventions and improving preparation for the “final mile” and true elimination. This requires not only good transmission modelling, but also better models of health systems. If this can be done for NTDs it can also strengthen capacity and health systems with positive effects for public health more generally.