Longitudinal changes in clinical symptoms and signs in children with confirmed, unconfirmed, and unlikely pulmonary tuberculosis

Abstract

Background: The paucibacillary nature of paediatric pulmonary tuberculosis (PTB) makes microbiological diagnosis difficult and limits the usefulness of microbiology for assessing treatment efficacy. Clinical response to treatment has thus been used by clinicians to monitor disease activity, as well as by researchers in clinical case definitions of intrathoracic TB to differentiate those with unconfirmed PTB from those with other lower respiratory tract infections (LRTIs). There is, however, limited data on the expected pattern and timing of resolution of symptoms and signs, and whether this does indeed differ between those with PTB and those without. Objectives: To longitudinally investigate clinical response to TB treatment in children treated for PTB, to compare this to the clinical course of children with other LRTIs, and to identify factors associated with persistence of symptoms and signs. Methods: This study is a secondary analysis of data collected prospectively in a TB diagnostic study from 1 February 2009 to 31 December 2018. We enrolled children ≤15 years with features suggestive of PTB. Study participants were categorized into 3 groups according to NIH consensus definitions; confirmed PTB, unconfirmed PTB and unlikely PTB. Children were followed at 1 and 3 months after enrolment. Those with confirmed or unconfirmed TB were also followed at 6 months. At enrolment and follow-up symptoms of PTB were recorded using a standardized questionnaire and physical examination was done including anthropometry and respiratory parameters. Data were analysed using STATA version 16.1. The effect of potential predictors of persistence of symptoms and signs was explored with univariable and multivariable logistic regression modelling. Results: Two thousand and nineteen children were included in this analysis, 427 (21%) with confirmed PTB, 810 (40%) with unconfirmed PTB, and 782 (39%) with unlikely PTB. Symptoms resolved rapidly in the vast majority of participants. At 1 month, 9.2% (129/1402) of all participants who had a cough and 11.1% (111/999) of those with loss of appetite at baseline reported no improvement in these symptoms. At 3 months this declined to 2.0% (24/1222) and 2.6% (23/886) respectively, with no differences between the groups. Clinical signs persisted in a greater proportion of participants. At 3 months, tachypnoea persisted in 56.7% (410/723) of participants. Abnormal auscultatory findings (including wheeze, crackles, reduced breath sounds or abnormal breath sounds) similarly persisted in almost a third of participants, with greater proportion in the confirmed group (37.1%) than unconfirmed (23.0%) and unlikely (26.2%) groups (p=0.002). Children living with HIV and those with abnormal baseline chest radiographs had greater odds of persistence of signs or symptoms (including cough, loss of appetite, abnormal auscultatory findings, or no weight improvement if underweight at baseline). No features of clinical response differentiated those with PTB from those without. Conclusion: Symptoms resolved rapidly in the majority of children investigated for PTB whilst clinical signs took longer to resolve. The timing and pattern of resolution of symptoms and signs cannot differentiate those with PTB from those without – and is thus not a suitable parameter for confirming disease classification in paediatric TB research.