Introduction Extra-genital tests for gonorrhoea (NG) and chlamydia (CT) are important in MSM and increasingly in women as vulvovaginal swabs (VVS) alone can miss infections but rectum, pharynx and urogenital swabs treble the diagnostic cost.Self-sampling is frequently used despite no robust RCT assessing its efficacy. We compared clinician-taken extra-genital samples with self-taken samples analysed both separately and as pooled samples for diagnostic accuracy.

Methods Women and MSM attending a sexual health clinic were recruited. Clinician and two self-samples (analysed separately and pooled) from the pharynx and rectum (plus VVS or FCU in MSM) tested for NG and CT using Aptima Combo 2. Sampling order was randomised. Patient infected status was defined as at least two positive confirmed samples.

Results 1795 (1284 women, 509 MSM) recruited. Overall prevalence: NG 9.0%, VVS/FCU 4.2% (75), rectum 4.6% (83), pharynx 4.0% (72); 9.4% females and 66.7% MSM were VVS/FCU negative. CT 15.4%, VVS/FCU 12.1% (217), rectum 13.9% (249), pharynx 3.5% (63); 13.1% females and 71.8% MSM were VVS/FCU negative. Sensitiviities, specificities, PPVs and NPVs:NG Rectum Clinician: 93.98, 99.94, 98.73, 99.71 NG Rectum Self: 96.39, 99.77, 95.24, 99.82 NG Pharynx Clinician: 93.06, 99.94, 98.53, 99.71 NG Pharynx Self: 95.83, 99.94, 98.57, 99.83 NG Self Pooled: 98.25, 99.94, 99.12, 99.88CT Rectum Clinician: 95.95, 99.87, 99.16, 99.35CT Rectum Self: 97.17, 99.81, 98.77, 99.55 CT Pharynx Clinician: 92.06, 99.94, 98.31, 99.71CT Pharynx Self: 93.65, 99.83, 95.16, 99.77 CT Self Pooled: 96.01, 99.60, 97.79, 99.28. There was no difference between clinician and self-taken extra-genital samples or between self-taken analysed separately or pooled by McNemar test.

Conclusion This is the first RCT showing self-taken extra-genital samples are comparable to clinician-taken and can be analysed accurately as a pooled sample. High levels of infections are missed with just VVS/FCU. Trebling diagnostic costs would be unaffordable for many health systems but a pooled sample has the same cost as the current VVS/FCU.