Dynamic Modelling of New TB Diagnostic Strategies

Estimating the impact and cost effectiveness of the GeneXpert tuberculosis assay in low resource, high burden settings
Investigators:

Ted Cohen

Meghan Murray

Hsien-ho Lin

 

Tuberculosis causes considerable morbidity and mortality in resource-poor settings, especially in those areas affected by concurrent HIV epidemics. The treatment of TB is complicated by imperfect diagnostic technology and the need for multiple-drug regimens provided over long duration. Deficiencies in past drug treatment -- inadequate drug regimens and poor adherence -- has led to substantial resistance against the most effective first-line medications in some settings.  

Recent technological developments in the field of TB diagnosis have yielded new tools with potential to address both under-diagnosis and inappropriate care. In particular, the GeneXpert MTB/RIF assay was shown in late 2010 to provide improved accuracy of test results and a rapid turnaround time, as well as providing information on resistance to one of the main 1st-line drugs, rifampicin. GeneXpert is also more expensive and technically challenging than the most commonly used diagnostics, and so decisions must be made as to how this test might best be incorporated into existing diagnostic algorithms. 

This project develops a mathematical model of tuberculosis transmission and progression as a tool for testing the consequences of GeneXpert adoption by national TB control programs. Empirical studies have demonstrated the benefits of GeneXpert over sputum smear diagnosis in terms of sensitivity for both smear-positive and smear-negative TB suspects, and information on rifampicin resistance allows regimens to be better matched to the drug sensitivity profile of a patient’s TB strain. In addition, by allowing rapid assessment, the new diagnostic might offer further benefits in reducing disease progression and loss to follow-up that may occur while waiting for test results, as with sputum culture and conventional drug sensitivity testing. While these effects are relatively well understood on an individual-patient basis, the epidemiology of TB is complex, and available empirical data provide less clarity on the potential consequences of GeneXpert adoption for population level health outcomes including TB incidence and the distribution of drug resistant TB strains.  Mathematical models can be used to incorporate data on test performance in combination with existing knowledge on TB transmission dynamics and disease progression, in order to project the potential impact of widespread GeneXpert adoption before important implementation decisions have to be made.

This project is being undertaken by a study team composed of researchers at CHDS, the Harvard Department of Global Health and Population, the Harvard Department of Epidemiology, and the Institute of Epidemiology at the National Taiwan University. This team is using the model to estimate the potential costs and health benefits of diagnostic algorithms based on the GeneXpert technology in multiple low-resource, high-burden settings. This project will provide information needed by policy-makers and technical experts as they make decisions about how to incorporate new diagnostic technology into TB control programs.