Reported tuberculosis symptoms: an inadequate classifier of disease state

Around half of people identified with tuberculosis (TB) during prevalence surveys do not report symptoms, prompting increased interest in the public health implications of asymptomatic TB (aTB), as distinct from symptomatic TB (sTB). A recent World Health Organization report proposes investigating the optimal treatment for aTB, stratifying notification data by symptom-status, and publishing incidence estimates based on sTB. We have concerns that the proposed use of aTB/sTB case definitions for public health surveillance, burden estimation, and clinical management is not supported by evidence.Although TB symptoms are linked to disease severity, we present evidence that self-reported symptoms are an inadequate and inconsistent classifier of disease state, heavily influenced by context and substantial interviewer variability, and TB symptoms may frequently have alternative causes. As a consequence, the proposed uses may worsen treatment outcomes, and lead to bias in burden and intervention impact modelling.We advocate for the adoption of more robust measures of disease severity, which better reflect underlying pathophysiology. Future research on the TB disease spectrum should prioritise more objective disease severity metrics, alongside or in place of symptom-based classification.