Results and discussion

Our literature search found that data describing the use of NAATs in low prevalence settings was reported from three main sources (table 1): GUM clinics serving low prevalence areas3–5; laboratories outside the UK6–9; and from National Chlamydia Screening Programme areas (NCSP) where dual chlamydia and gonorrhoea testing is being offered.10–14 Several different NAAT platforms were used in the studies reviewed. In every study, an initial positive NAAT underwent supplementary testing, although there were differences in the method used. In most laboratories, a positive sample was retested using another NAAT detecting a different target.3 ,4 ,6–10 ,12 Some repeated the test on residual samples using the original assay11 ,13 ,14 which, in fact, demonstrates reproducibility rather than confirmation. In one small study,5 a low PPV of 43% may be explained by the use of culture alone to confirm a positive NAAT. It is difficult to draw comparisons between the studies when the ‘gold standard’ for the diagnosis of infection is the isolation of N. gonorrhoeae, which can have a poor sensitivity. Significantly, many of the studies did not report the number of positive NAATs which were not confirmed, the false positive rates, nor calculated a PPV. One of the NCSP studies found a much lower PPV than expected, of 3.8% for women, and ‘incalculable’ for men.10 However, the authors made the incorrect assumption when estimating the PPV, that sensitivity and specificity may vary by prevalence, which is not substantiated in the context of asymptomatic gonorrhoea testing.

Of five NCSP studies, all but one reported a low prevalence of gonorrhoea infection (less than 1%), and approximately half the individuals with gonorrhoea were coinfected with chlamydia. The one NCSP study which found a much higher rate of gonorrhoea of 4.1% was from Lewisham, an area of South London which probably does not represent a ‘low prevalence’ population, as discussed above.13 The papers from outside the UK described data from community laboratories serving low prevalence populations. The study from Seattle Washington, USA,9 specifically excluded samples from specialist sexual health clinics. The study from Canada8 and the two from New Zealand6 ,7 were based in low prevalence community settings, but did not provide further details on the source of the specimens.

The evidence base to support widespread screening for gonorrhoea in the community is limited and of poor quality. Indeed, the recent National Survey of Sexual Attitudes and Lifestyles (NATSAL), found that the prevalence of gonorrhoea in the UK general population was <0.1%, and those with infection had recognised high-risk factors.15 However, the widespread availability of dual NAAT technology may lead to increased opportunistic testing for N. gonorrhoeae. This should be tempered by the public health need and be restricted to areas with higher prevalence, such as inner cities, or following pilot studies that indicate the usefulness of more widespread testing. We would support the development of an algorithm to calculate local prevalence and PPV in order to help inform whether selected screening should be implemented.