A public-private partnership for facility-based tuberculosis screening using chest radiographs, Viet Nam.

OBJECTIVE

To determine tuberculosis prevalence among health-care facility patients and to assess the effectiveness of chest radiography in facility-based tuberculosis screening.

METHODS

Our cross-sectional analysis used individual-level data collected during 2020-2023 from adults presenting at 796 health-care facilities participating in a public-private partnership in 15 provinces of Viet Nam. We constructed tuberculosis care cascades stratified by symptomatic presentation according to the World Health Organization (WHO) four-symptom screening and chest radiography. We reported diagnostic yield, number needed to screen and positivity.

FINDINGS: Among 1 423 818 participants, 22 598 had bacteriologically confirmed tuberculosis. Diagnostic yields were 2.2% (19 289/892 894) among symptomatic individuals with cough, 1.0% (1869/191 677) among symptomatic individuals without cough and 0.4% (1440/339 247) among asymptomatic individuals. Tuberculosis-suggestive chest radiograph results compared with normal results increased bacteriologically confirmed positivity by 9.4 percentage points in symptomatic individuals with cough (17.3%; 18 746/108 650 versus 7.9%; 543/6851), by 4.6 percentage points in symptomatic individuals without cough (12.6%; 1748/13 831 versus 8.0%; 121/1515) and by 7.6 percentage points in asymptomatic individuals (15.0%; 1328/8862 versus 7.4%; 112/1511). Without chest radiography, the number of individuals requiring testing to detect tuberculosis among symptomatic individuals with cough and without a cough would have been 6.5 (1 084 571/165 679) and 5.4 (892 894/165 679) times higher, respectively.

CONCLUSION

Chest radiograph screening can reduce the number of expensive molecular tests used in symptomatic patients and enable tuberculosis diagnosis in asymptomatic patients. Future research should analyse cost and cost-effectiveness of using chest radiography in active and intensified case finding.