Introduction and objective Cough is predominant in pulmonary tuberculosis and transmits infection, yet it is unclear how rapidly it responds to treatment. We explored changes in objectively-measured cough frequency during TB therapy with respect to other markers of disease.
Method Before or on commencing anti-tuberculous treatment, consecutive patients with pulmonary tuberculosis wore the Leicester Cough Monitor for 24 h, a device comprising a small digital audio recorder and microphone with software for cough detection. Those with a baseline cough frequency greater than the upper limit of normal (c. 100 coughs/24 h [c/24h]) were asked to undergo repeat monitoring during initial hospitalisation (if applicable), and later to coincide with routine clinic attendance or directly observed therapy (DOT).
Results Median baseline cough frequency was 203 (IQR 75–470) c/24 h in all participants (n = 44), and >100 c/24 h in 32 (73%). 22 patients were willing and available to undergo serial cough monitoring (12 current smokers, 18 sputum smear positive disease [12 also with visible lung cavities], and one with HIV). Three had isoniazid-resistant disease; the remainder were fully drug-sensitive. All were eventually treated successfully. None had other respiratory diagnoses.
Cough frequency in the majority declined consistently during therapy with substantial improvements by one week (Figure 1). At 2 and/or 8 weeks, five patients had a higher cough frequency than at baseline. Amongst these slow responders there was initial extensive radiographic change (n = 1), poor drug adherence with ongoing weight loss (n = 1), a paradoxical reaction to treatment with the development of a paraspinal abscess (n = 1), and, in the patient with HIV, persistent sputum smear positivity at 8 weeks with minimal radiographic improvement despite DOT and normal plasma rifampicin levels. One other patient had a highly variable cough frequency during the first 8 weeks of treatment. There was no evidence for an effect of isoniazid resistance, cavitary disease, smear status or smoking on early rates of cough resolution, although there was a trend towards relatively higher cough frequencies in smokers than non-smokers at the end of treatment ( p = 0.100).
Conclusions Objective cough frequency measurement is feasible in tuberculosis and could provide a novel biomarker of treatment response.