Background: The microbiome of built environment surfaces is impacted by the presence of humans. In this study, we tested the hypothesis that analysis of surface swabs from clinic toilet/bathroom yields results correlated with sexually transmitted infection (STI) notifications from corresponding human populations. We extended a previously reported study in which surfaces in toilet/bathroom facilities in primary health clinics in the Australian Northern Territory (NT) were swabbed then tested for nucleic acid from the STI agents Chlamydia trachomatis, Neisseria gonorrhoeae and Trichomonas vaginalis. This was in the context of assessing the potential for such nucleic acid to contaminate specimens collected in such facilities. STIs are notifiable in the NT, thus allowing comparison of swab and notification data.
Methods: An assumption in the design was that while absolute built environment loads of STI nucleic acids will be a function of patient traffic density and facility cleaning protocols, the relative loads of STI nucleic acids from different species will be largely unaffected by these processes. Another assumption was that the proportion of swabs testing positive for STIs provides a measure of surface contamination. Accordingly, ``STI profiles'' were calculated. These were the proportions that each of the three STIs of interest contributed to the summed STI positive swabs or notifications. Three comparisons were performed, using swab data from clinics in remote Indigenous communities, clinics in small-medium towns, and a single urban sexual health clinic. These data were compared with time and place-matched STI notifications.
Results: There were significant correlations between swab and notifications data for the both the remote Indigenous and regional data. For the remote Indigenous clinics the p values ranged from 0.041 to 0.0089, depending on data transformation and p value inference method. Further, the swab data appeared to strongly indicate known higher relative prevalence of gonorrhoeae in central Australia than in northern Australia. Similarly, the regional clinics yielded p values from 0.0088-0.0022. In contrast, swab and notifications data from the sexual health clinic were not correlated.
Discussion: Strong correlations between swab and notifications were observed. However, there was evidence for limitations of this approach. Despite the correlation observed with the regional clinics data, one clinic yielded zero positive swabs for C. trachomatis, although this STI constituted 25.1% of the corresponding notifications. This could be ascribed to stochastic effects. The lack of correlation observed for sexual health clinic data was also likely due to stochastic effects. It was concluded that toilet/bathroom surface swab sampling has considerable potential for public health surveillance. The approach may be applicable in situations other than primary health clinics, and for targets other than STIs.