LAUSR

Objective We evaluated the performance of nucleic acid amplification tests (NAATs) using vaginal specimens in comparison to specimens from the cervix or urine in their ability to detect chlamydia and gonorrhoea infection in women based on patient infection status (PIS). Design Systematic review. Data sources EMBASE and Ovid MEDLINE databases were searched through 3 October 2017. Eligibility criteria for selecting studies We included studies that tested samples from the vagina and ≥1 other site (cervix and/or urine) with ≥2 NAATs for chlamydia and ≥2 NAATs or 1 NAAT and culture for gonorrhoea for each site. Data extraction and synthesis Performance is defined as the sensitivity of a NAAT using a specimen site and PIS of the patient. We assessed risk of bias using modified QUADAS-2. Results Nine publications met the inclusion criteria (eight for chlamydia; six for gonorrhoea) and were narratively reviewed. Pooled summary estimates were not calculated due to the variable methodology and PIS definitions. Tests performed on vaginal specimens accomplished similar performance to cervical and urine specimens for chlamydia (range of performance estimates: vaginal 65%–100%, cervical 59%–97%, urine 57%–100%) and gonorrhoea (vaginal 64%–100%, cervical 85%–100%, urine 67%–94%). Vaginal specimens were estimated to have a performance >80% for chlamydia and gonorrhoea infections in all but one study. Conclusions Performance of the NAATs for chlamydia and gonorrhoea detection using vaginal specimens was similar to that of cervical and urine specimens relative to PIS. As vaginal samples have a higher acceptability and lower cost, the study can support clinical testing guidelines by providing evidence that vaginal samples are a suitable alternative to traditionally used specimens.