Abstract
BACKGROUND
In 2018, the World Health Organization commenced a multi-country validation study of the Cepheid GeneXpert for a range of molecular-based point-of-care (POC) tests in primary care settings. One study arm focused on the evaluation of POC tests for screening ‘women at risk’ for chlamydia (CT), gonorrhoea (NG) and trichomonas (TV) in four countries – Australia, Guatemala, Morocco and South Africa.
METHODS
Study participants completed a pre-test questionnaire which included demographics, clinical information and general questions on POC testing (POCT). Two vaginal swab samples (either self-collected or clinician collected) from each patient were tested on the GeneXpert at the POC and at a reference laboratory using quality-assured nucleic acid amplification tests (NAATs).
RESULTS
One thousand three hundred and eighty-three women were enrolled: 58.6% from South Africa, 29.2% from Morocco, 6.2% from Guatemala, and 6.0% from Australia. 1,296 samples for CT/NG and 1,380 samples for TV were tested by the GeneXpert and the reference NAAT. The rate of unsuccessful tests on the GeneXpert was 1.9% for CT, 1.5% for NG and 0.96% for TV. The prevalence of CT, NG and TV was 31%, 13% and 23%, respectively. 1.5% of samples were positive for all three infections; 7.8% were positive for CT and NG; 2.4% were positive for NG and TV; and 7.3% were positive for CT and TV. Compared to reference NAATs, pooled estimates of sensitivity for the GeneXpert tests were 83.7% (95% confidence intervals 69.2-92.1) for CT, 90.5% (85.1-94.1) for NG and 64.7% (58.1-70.7) for TV (although estimates varied considerably between countries). Estimates for specificity were ≥96% for all three tests both within- and between-countries. Pooled positive and negative likelihood ratios were: 32.7 ([CI] 21.2-50.5) and 0.17 (0.08-0.33) for CT; 95.3 (36.9-245.7) and 0.10 (0.06-0.15) for NG; and 56.5 (31.6-101.1) and 0.35 (0.27-0.47) for TV.
CONCLUSION
This multi-country evaluation is the first of its kind world-wide. Positive likelihood ratios, as well as specificity estimates, indicate the GeneXpert POC test results for CT, NG and TV were clinically acceptable for ruling in the presence of disease. However, negative likelihood ratios and variable sensitivity estimates from this study were poorer than expected for ruling out these infections, particularly for TV.
