Lancet Global Health
Nov 2016 Volume 4 Number 11 e761-e871
Putting numbers on the End TB Strategy—an impossible dream?
Olivia Oxlade, Dick Menzies
In 2015, WHO announced a plan to end tuberculosis by 2035 (their End TB Strategy) and set ambitious intermediate targets to reduce tuberculosis incidence by 50% and mortality by 75% by 2025.1 In The Lancet Global Health, two related papers by Rein Houben2 and Nicolas Menzies3 and their colleagues describe the results of a unique international collaboration between 11 different tuberculosis modelling groups, and public health officials from national tuberculosis programmes. They assessed the feasibility, costs, and epidemiological outcomes of country-specific interventions in India, China, and South Africa, and determined that these 10-year targets could be achievable only in South Africa with a combination of continuous isoniazid preventive therapy for individuals on antiretroviral therapy, expanded facility-based screening for symptoms of tuberculosis at health centres, and improved tuberculosis care. In China and India, important reductions could be achieved, but they fell short of the WHO targets.2 All models that considered costs projected the need for massive and sustained increases in government health spending, to more than three times current levels, although most judged that these interventions could be considered cost-effective. Interestingly, all predicted that patients’ costs would be substantially reduced with most interventions.
This project showed the potential value of two innovative collaborations toward achieving global tuberculosis control. First, this investigation was accomplished simultaneously by several different modelling groups and investigators from a total of ten different countries—in itself a major achievement! The modelling groups worked independently, using their preferred modelling approaches, but with similar parameters and assumptions. Readers will usually want to know if the findings are unchanged when key assumptions are varied in sensitivity analyses, and if results are similar in studies published separately by different groups. We think readers should be sceptical, given the grand scale of assumptions made by the investigators of these two studies. In these Articles,2, 3 results from 11 models are presented together—a sort of uber-sensitivity analysis. The results are quite consistent and provide a coherent message, which we find reassuring. The second innovation was the partnership of these modelling teams with personnel from national tuberculosis programmes, who were responsible for the selection of interventions and helping to estimate their expected effects. This should make the results more applicable and realistic for the countries selected, while also enhancing knowledge translation.
For most health-care professionals, infectious-disease modelling is something of a black box. One can see the input assumptions (ie, what goes in) and the outputs (ie, what comes out), but what happens in between seems close to magic. Given their complexity, to understand any one of the models used in these studies is difficult; to understand the strengths and limitations of all 11 models might be beyond the capacity of most (if not all) readers. So, we must therefore accept a little magic, and rely on a careful review of what goes in, to decide if what comes out is credible. And the assumed inputs are a major limitation of these studies, for although the involvement of national tuberculosis programme officials in selecting interventions and targets was a strength, the actual population-level effect, and costs, of the interventions are unknown.
For example, active case finding through chest radiography was the cornerstone of tuberculosis control for decades in high-income countries,4 and interest in active case finding has been revived recently.5 However, scant published evidence of its effect on outcomes, transmission, or its cost-effectiveness is available,6 and therefore mass screening is not recommended by WHO.6 The true costs of these interventions, when applied at national scale, are also unknown.
Estimations of costs extrapolated from small projects might not be accurate for national-level interventions. For example, the finding that scaling up use of the Xpert RIF/MTB assay might simply reflect better information, since the actual costs for national expansion in South Africa have been carefully measured,7 by contrast with the estimated costs for the other interventions. Even feasibility is uncertain, particularly for population-level interventions such as mass chest radiography and isoniazid preventive therapy in South Africa, or partnerships with the rapidly evolving private sector in India.
Overall, however, we feel the investigators used all currently available information, and did a careful and thorough analysis of innovative approaches for global tuberculosis control. Although further research is required to better understand the epidemiological effects and the enormous health-system expenditures that will be needed to implement these interventions on a large scale, this requirement should not obscure two important messages from these studies. First, the consistent finding of substantial savings for patients is a reminder that reducing the tuberculosis burden is all about reducing the burden on patients. And second, that perhaps the goal of ending tuberculosis is not such an impossible dream.
We declare no competing interests.