Introduction Bacterial vaginosis (BV) is the most common cause of vaginal discharge worldwide including India. Syndromic diagnosis and management is used across the country for STIs/RTIs. Aims of the study were 1) to evaluate the prevalence of BV based on Nugent’s, Amsel’s and modified Amsel’s criteria 2) to analyse concordance between self- and physician collected swabs for the three different criteria.

Methods Study included 550 females attending STI/RTI clinics with abnormal vaginal discharge during the study period of January 2015 to May 2016. Self-collection of swabs was done by patients after instructions followed by physician collection under speculum examination. The samples were analysed by Nugent’s, Amsel’s and modified Amsel’s criteria for diagnosing BV.

Results Based on Nugent scoring, 79 (14.3%), 95 (17.3%), 376 (68.4%) patients were found to be BV positive (+), intermediate and BV negative (-) respectively. However, using Amsel’s criteria only 67 (12.2%) patients were observed to be BV+ and 483 (87.8%) were BV-. As modified Amsel’s criteria diagnoses BV+ even when only two out of four Amsel’s criteria are fulfilled, it could detect 96 BV+ cases (17.5%). All the above results were obtained using the physician-collected samples. Sensitivity for diagnosing BV by self-collected samples using Nugent score, Amsel’s criteria and modified Amsel’s criteria was 91.1%, 98.5% and 97.9%, while specificity was 100%, 99.6% and 99.6%, positive predictive value 100%, 97.1% and 97.9% and negative predictive value was 98.5%, 99.8% and 99.6% respectively. High concordance of self-collected samples was established by the Cohen’s Kappa value of 0.890, 0.975, and 0.993 for Nugent scoring, Amsel’s criteria and Modified Amsel’s criteria respectively.

Conclusion Apart from Nugent’s scorimg, Amsel’s and modified Amsel’s criteria were equally consistent methods of diagnosis. Establishing reliability of self-collected samples for diagnosing BV shall pave way for validation of syndromic diagnosis even at peripheral health centres.