Screening for asymptomatic tuberculosis among adults with household exposure to pulmonary tuberculosis: a prospective observational cohort study

Abstract

Background: More than half of tuberculosis cases detected by community prevalence surveys are classified as asymptomatic. We evaluated yield of symptom and chest radiograph screening of tuberculosis-exposed household contacts in South Africa.

Methods: For this prospective observational cohort study, adult volunteers (aged ≥18 years) with household exposure within the past 6 months to patients with untreated or partially treated pulmonary tuberculosis, identified through local health services, were enrolled at three sites in South Africa (Worcester and Ravensmead, Western Cape Province, and Soweto, Gauteng Province). Household contacts were excluded if they were unlikely to attend study visits, or had conditions interfering with consent or study participation, including psychiatric illness, substance dependence, or incarceration. Systematic screening of tuberculosis symptoms (any duration), chest radiograph (any abnormality indicative of active tuberculosis), and sputum microscopy, Xpert Ultra, and liquid culture were performed. Serum C-reactive protein (CRP) was measured by multiplex bead array. Prevalent tuberculosis was microbiologically confirmed (Xpert Ultra or culture). Symptomatic and asymptomatic tuberculosis were defined as prevalent tuberculosis with and without reported symptoms compatible with tuberculosis. The primary outcome was the diagnostic yield (sensitivity) of microbiologically confirmed pulmonary tuberculosis.

Findings: Between April 22, 2021 and Sept 22, 2022, 979 household contacts were enrolled, 345 (35·2%) male and 634 (64·8%) female, 185 (18·9%) living with HIV and 187 (19·1%) with previous tuberculosis. Prevalent tuberculosis occurred in 51 (5·2%) and was asymptomatic in 42 (82·4%) of 51. Only 13 (31·0%) of 42 asymptomatic people with tuberculosis were sputum-smear positive; eight (61·5%) of these 13 had a low bacillary burden, with smear grades scanty or 1+ (1-99 acid-fast bacilli per 100 fields). CRP did not discriminate healthy household contacts from those with asymptomatic tuberculosis (area under the curve 0·60, 95% CI 0·47-0·73). An abnormal chest radiograph suggestive of tuberculosis was observed in 23 of 41 asymptomatic (sensitivity 56·1%, 95% CI 41·0-70·1) versus eight of nine symptomatic (sensitivity 88·9%, 56·5-98·0) people with tuberculosis. Sensitivity of chest radiograph in combination with symptom screening was 32 (64·0%) of 50 (50·1-75·9) for all prevalent tuberculosis.

Interpretation: More than 80% of confirmed people with tuberculosis among household contacts were asymptomatic; chest radiograph screening missed more than 40% of these. Community prevalence surveys reliant on symptom-based and chest radiograph-based approaches might substantially underestimate the prevalence of asymptomatic tuberculosis in endemic countries.

Funding: Regional Prospective Observational Research for Tuberculosis South Africa through funding from the US National Institutes of Health, the Civilian Research and Development Foundation, and the South African Medical Research Council.

Figure 1:. Sensitivity of symptom and chest radiography screening for tuberculosis with universal sputum microbiological testing irrespective of presence of symptoms or chest radiograph abnormality

Forest plot of the sensitivity of symptom screening (A), chest radiograph (B), and parallel symptom and chest radiograph screening (C) for tuberculosis reported in the literature. Cohorts are listed by year of publication and first author. HHC=household contacts. PDL=people deprived of liberty.

Figure 2:

Study flow diagram

Figure 3:. Venn diagram demonstrating yield of symptom screening and chest radiography for detection of pulmonary tuberculosis among household contacts

*Seven symptomatic household contacts without tuberculosis who did not have a chest radiograph. †One asymptomatic household contact with tuberculosis who did not have a chest radiograph.

Figure 4:. CRP is unable to differentiate asymptomatic tuberculosis cases from household contacts without tuberculosis

CRP concentration measured by multiplex bead array in a subset of 25 asymptomatic and seven symptomatic pulmonary tuberculosis cases, and 126 healthy household contacts, from this cohort, and 607 symptomatic adults presenting to clinic with tuberculosis symptoms (199 symptomatic pulmonary tuberculosis and 408 without tuberculosis). CRP concentrations were censored at 0·1 mg/dL (lower limit) and 150 mg/dL (upper limit) and are shown on a log₁₀ scale. Each dot represents one participant. The midline is the median and the whiskers show IQR. CRP=C-reactive protein.