Statistical methods to guide clinical decision making in TB

Tuberculosis is the leading cause of mortality due to an infectious disease globally. The fight against this epidemic is hampered by underreporting and underdiagnosis. Identified TB cases require prompt treatment initiation to prevent onward transmission of the disease among other benefits. Starting from symptoms in a patient, physicians will decide to start or withhold TB treatment, or otherwise invest more resources performing further medical investigations. In practice, when confronted with diagnostic uncertainty, the clinician will consider potential harm and benefit of offering versus withholding treatment. However, this could result in a high rate of treatment initiation on non-TB cases or withholding treatment from TB cases. Thus in order to make informed decisions a better understanding of evidence-based clinical decision making is required.
The European and Developing Countries Clinical Trials Partnership (EDCTP) is funding a 4-year African-European TB TRIAGE+ project aimed at assessing the accuracy, impact and cost-effectiveness of CAD4TB and C-reactive protein (CRP) as triage tools for community-based tuberculosis (TB) screening in rural and semi-rural Lesotho and KwaZulu-Natal, South Africa. (https://tbtriage.com/) Within this project, Institute of Tropical Medicine (ITM) leads the “Clinical decision making in TB” work package. It is within this work package that this PhD project is anchored.
The primary aim of the PhD project is to gain a better understanding of clinical decision making in the diagnosis of TB. To accomplish this, the project will answer these specific objectives: (1) determine the diagnostic accuracy of all available information, including symptoms, diagnostic tests and risk factors, (2) determine the therapeutic threshold of TB (i.e the probability of TB that needs to be exceeded to start treatment), (3) determine the probability of TB in patients with given sets of symptoms, diagnostic test results and background characteristics, and (4) develop a clinical decision making tool, incorporating all diagnostic tools that are studied in the TB TRIAGE+ project in the setting of rural Lesotho and South Africa.
By combining and analysing the information we collect in the TB TRIAGE+ project, we foresee that we can obtain accurate estimates of the prevalence of TB, diagnostic accuracy of clinical symptoms for TB, and therapeutic threshold for starting TB treatment. Those can subsequently be used in the development of an evidence-based clinical decision tool for TB. With such a tool, clinicians will be able to determine if a new patient with given demographics and symptoms reaches the therapeutic threshold to start treatment.