Article Text
Abstract
Objectives Infectious syphilis (syphilis) is diagnosed predominantly among men who have sex with men (MSM) in the Netherlands and is a strong indicator for sexual risk behaviour. Therefore, an increase in syphilis can be an early indicator of resurgence of other STIs, including HIV. National and worldwide outbreaks of syphilis, as well as potential changes in sexual networks were reason to explore syphilis trends and clusters in more depth.
Methods National STI/HIV surveillance data were used, containing epidemiological, behavioural and clinical data from STI clinics. We examined syphilis positivity rates stratified by HIV status and year. Additionally, we performed space–time cluster analysis on municipality level between 2007 and 2015, using SaTScan to evaluate whether or not there was a higher than expected syphilis incidence in a certain area and time period, using the maximum likelihood ratio test statistic.
Results Among HIV-positive MSM, the syphilis positivity rate decreased between 2007 (12.3%) and 2011 (4.5%), followed by an increasing trend (2015: 8.0%). Among HIV-negative MSM, the positivity rate decreased between 2007 (2.8%) and 2011 also (1.4%) and started to increase from 2013 onwards (2015: 1.8%). In addition, we identified three geospatial clusters. The first cluster consisted of MSM sex workers in the South of the Netherlands (July 2009–September 2010, n=10, p<0.001). The second cluster were mostly HIV-positive MSM (58.5%) (Amsterdam; July 2011–December 2015; n=1123, p<0.001), although the proportion of HIV-negative MSM increased over time. The third cluster was large in space (predominantly the city of Rotterdam; April–September 2015, n=72, p=0.014) and were mostly HIV-negative MSM (62.5%).
Conclusions Using SaTScan analysis, we observed several not yet recognised outbreaks and a rapid resurgence of syphilis among known HIV-positive MSM first, but more recently, also among HIV-negative MSM. The three identified clusters revealed locations, periods and specific characteristics of the involved MSM that could be used when developing targeted interventions.
- SYPHILIS
- GAY MEN
- HIV
Statistics from Altmetric.com
Introduction
Infectious syphilis (primary, secondary and early latent syphilis), hereinafter referred to as syphilis, is a strong indicator for high-risk sexual behaviour. Therefore, an increase in syphilis can be an early indicator of resurgence of other STIs, including HIV. In the Netherlands, approximately 90% of all syphilis and HIV infections at STI clinics are diagnosed among men who have sex with men (MSM).1
Outbreaks of syphilis among MSM are reported worldwide. Recently, investigations of an outbreak in the UK showed an association with geosocial mobile applications, which is also seen in other locations.2–4 These applications are popular, especially among MSM to meet (more) new sexual partners rapidly and easily, possibly internationally. These geosocial applications lead to the formation of riskier sexual networks, more mixing between sexual networks and an increased number of private sex parties involving drug use, which may all contribute to rapid spread of the disease.5 However, the effect of these geosocial applications on the spread of STI is not yet well reported, as there is a lack of evidence showing change in sexual behaviour after switching to geosocial applications from other venues for meeting sexual partners.6
The Dutch surveillance of syphilis mainly consists of regular analyses and reporting of trends.1 Currently, an early detection of specific outbreaks is left to be noticed by alert local healthcare providers, but highly depends on localised cases. An observed increase in syphilis among MSM in the Netherlands recently, reported increases and outbreaks internationally and potential changes in sexual networking created urgency to further investigate and analyse available syphilis surveillance data among MSM. In this study, our primary aim was to explore whether one or more as of yet undetected clusters could be identified within a generalised increased incidence, using space–time cluster analysis, complementary to standard syphilis trend analysis using the Dutch STI/HIV surveillance data from 2007 to 2015.
Methods
Dutch STI surveillance includes data on consultations of all 26 STI clinics. These clinics provide low threshold and free of charge STI/HIV testing and care for high-risk groups, including MSM. The testing guideline stipulates that MSM are at least tested for chlamydia, gonorrhoea, syphilis and HIV. The surveillance records are anonymised and contain epidemiological, behavioural and clinical data,1 including residence locations of the MSM. This study is a secondary analysis of an anonymised database for surveillance research purposes. Therefore, under Dutch law, ethical approval for this study was not necessary.7
We explored trends in the syphilis positivity rate among MSM by HIV status using the Cochran–Armitage test for trends (SAS, V.9.4). We calculated incidences per municipality per year. The cut-offs for the incidence were calibrated (based on Jenks' natural breaks of the average incidence per year) and kept equal for all years to maintain comparability. General population data were obtained from Statistics Netherlands. Cluster analysis on municipality level was performed to detect statistically significant clusters with the space–time scan statistic (no overlap) over the period 2007–2015 (SaTScan software, V.9.0). We assumed the number of cases in each location to be Poisson distributed. SaTScan identifies clusters by imposing a cylindrical window on the map and allows its centre to move over the area so that at any given position, the window includes different sets of neighbouring municipalities. The model tests the null hypothesis, the expected syphilis incidence is constant over the whole area in a certain time period, against the alternative hypothesis that there is an elevated syphilis incidence within, compared with outside the window, using the maximum likelihood ratio test statistic.8 Allowing for overlap in clusters resulted in similar results (data not shown). To clarify the extent in which clusters explain the overall increase in syphilis, we also assessed the increase excluding identified clusters.
Results
Study population
The number of syphilis tests among MSM increased each year (figure 1 and table 1). The median age was higher among known HIV-positive MSM (43 years; IQR: 36–49) than among HIV-negative MSM (36 years; IQR: 27–46) and MSM newly diagnosed with HIV (35 years; IQR: 28–44). Known and newly diagnosed HIV-positive MSM (15.0% and 1.3%, respectively) reported higher number of partners (>10 partners in the last 6 months, respectively: 24.2%, 19.3% vs 14.9%, p<0.001) and STIs in the preceding 2 years (34.4%, 22.1% vs 16.0%, p<0.001) compared with HIV-negative MSM. Among known HIV-positive MSM, 3.0% were bisexual; this was 12.0% among those newly diagnosed with HIV and 19.3% among HIV-negative (p<0.001).
Trends in syphilis
The absolute number of syphilis diagnoses between 2011 and 2015 increased 110% (429 in 2011, 901 in 2015; table 1). The overall syphilis positivity rate displayed a decreasing trend until 2011, afterwards an increase occurred to 2.6% in 2015.
Among HIV-negative MSM, the positivity rate increased only since 2014 (2015: 1.8%). Among known HIV-positive MSM, the positivity rates followed the same pattern as that of the total group. The trend in syphilis positivity among newly HIV-diagnosed MSM decreased between 2007 and 2010, and fluctuated thereafter. All reported trends were significant (p<0.001).
Clusters in the period 2007–2015
We identified three significant clusters in space and time (table 2). The first cluster was located in the south-eastern region of the country (p<0.001). The available epidemiological, behavioural and clinical data showed that 9 out of 10 cases were young, male commercial sex workers, mainly originating from Eastern Europe. None were known HIV-positive, but halve were newly diagnosed with HIV. All cases reported a high number of partners in the past 6 months.
The second cluster was located in Amsterdam (p<0.001). The MSM were older with a median age of 41 years, predominantly non-Dutch. As this cluster spanned over multiple years, changes within this cluster over the years were observed, the proportion of HIV-positive MSM decreased from 69.7% in 2011 to 56.0% in 2015, whereas the proportion of HIV-negative MSM increased from 30.3% to 43.3%. The proportion of MSM who reported having had an STI in the past 2 years increased from 2.6% in 2011 to 40.3% in 2015 (overall: 28.1%). The proportion of MSM notified for an STI varied between the years (overall: 36.6%; min–max: 25.8–41.1%). No other notable differences between the years were observed.
The third cluster was a large cluster in space that included the city of Rotterdam, 12 municipalities in Zeeland and parts of the adjoining provinces Zuid-Holland and Noord-Brabant (p=0.014). Of all 72 cases, 43.1% did not originate from the Netherlands, 62.5% was HIV-negative, 54.2% reported STI symptoms and 37.5% reported having had an STI in the past 2 years. Twenty-two cases (30.6%) were notified for an STI, of whom 17 specifically for syphilis (data not shown). The median number of partners was lower than in the two other clusters.
Beside the clusters and a general increase nationally since 2011, thematic mapping showed that other areas also have higher incidences, but these were not identified as statistically significant clusters. Excluding cases from the largest cluster, national trends showed that positivity rates still increased, although the increase was attenuated and the onset of the increase was delayed to 2013 among known HIV-positive MSM (data not shown).
Discussion
Using SaTScan for the analysis of routine surveillance data, we detected three previously unidentified geospatial clusters, which coincided with a national resurgence of syphilis. These clusters differed on several characteristics, most notably more recently syphilis was diagnosed predominantly among HIV-negative MSM whereas previously this predominantly used to be HIV-positive MSM.
The strength of our approach is that we used readily available systematically collected surveillance data, including a large number of consultations among high-risk MSM over a 9-year period. A limitation of using these data is the absence of details specifically on sexual behaviour and lifestyles. Even though SaTScan is the most commonly used method for cluster analysis,9 a limitations of this method is that larger clusters might obscure smaller ones. However, excluding the larger cluster did not reveal undetected clusters (see diagnostic frequencies in online supplementary figure). Second, long-term clusters could contain several peaks; however, this was not apparent in our clusters. Third, as we do not have population data on MSM, we used population data. We expect that the proportion of MSM among all men is somewhat higher in the bigger cities. Therefore, we expect that the true incidence is lower in areas where the proportion of MSM among all men is larger, and vice versa. This hampers the comparability between regions. However, as our second cluster was large, we expect that adjusting for MSM population instead of male population would still result in a significant cluster. Moreover, it is unlikely that differences in time by region would be affected.
supplementary figure
Strength of this study in relation to other studies is that similar increases in syphilis have been observed in several other countries in Europe, including our direct neighbours Germany and Belgium (although for Belgium only the number of cases were available not positivity rates).10 Most outbreaks are reported among MSM residing in large cities with large MSM communities, which is true for the second Amsterdam cluster, but the first cluster is a more rurally located municipality and the third cluster covers a larger area than just the city of Rotterdam.11 ,12 It has been suggested that this pattern in syphilis epidemiology internationally indicates globalisation of sexual networks among MSM.5 ,13 The geospatial location of our clusters strengthens the suspicion that transmission could have taken place internationally. The first cluster and parts of the third cluster were closely located to the Belgian and/or German border. The second cluster was located in Amsterdam, which has a large MSM community and is a popular (inter)national city to have sexual encounters.
Geosocial mobile applications could affect the rapid spread of the disease, as these applications seem to increase the number of partners, mixing of sexual networks, increase the number of private parties and drug use and international sexual tourism.5 Our findings are in line with most of these associations, suggesting that these applications could have contributed to the resurgence of syphilis in the Netherlands. Notably, and possibly contradicting the geosocial application explanation for the resurgence in syphilis in the second cluster, the number of partners in the past 6 months did not change over time, whereas the proportion of MSM who had an STI in the past 2 years did increase over time. This might suggest that MSM were increasingly part of high-risk sexual networks or performing riskier behaviours, without increasing their number of partners. Another explanation is that the increase in number of sexual partners because of applications already reached its ceiling before the onset of this cluster.
The resurgence of syphilis is of great concern as beside health concerns it may herald increases in other STIs. These increases in, for instance, HCV and HIV could be driven by either high-risk behaviour or with increased transmission risk due to a syphilis infection.14–16 Therefore, syphilis infections could be used by clinicians and policymakers as an indication to offer screening for HCV and the offer of preexposure prophylaxis against HIV among HIV-negative MSM.
Current policies for syphilis prevention focus on partner notification. Healthcare professionals can be alert for syphilis symptoms, when outbreaks are detected, which emphasises the need for real-time geospatial information to be added to routine surveillance methods to be able to respond more timely and efficiently. In addition, collecting behavioural data could be worthwhile. Exploring whether the outbreaks of syphilis are linked to signals of increased risk behaviour, such as the recent emergence of ‘slamming’ (injecting crystal meth) or chemsex.17 Finally, geospatial cluster analyses of the incidence of other STIs could be investigated to test the assumption of syphilis as an early indicator for other STIs among MSM.
The cluster in Amsterdam lasted 5 years and is still ongoing. Removing this cluster from the data showed that it contributed significantly to the observed surge in positivity rates since 2011, but was not solely responsible for the syphilis trends. In conclusion, analysing geospatial clusters and the characteristics associated with MSM in these clusters provided insight into potential transmission patterns and specific locations for targeted interventions.
Key messages
Surveillance showed that syphilis positivity rates among HIV-positive men who have sex with men (MSM) increased since 2011, but more recently, findings also showed an increase among HIV-negative MSM.
Space–time cluster analysis identified three previously undetected syphilis clusters; the cluster in Amsterdam contributed but was not solely responsible for the resurge.
Characteristics associated with MSM in these clusters provided insight into potential transmission patterns and (sexual) risk behaviour.
Space–time cluster analyses provided added value to trend analysis, as it provided insight into important locations and characteristics for targeted interventions.
Acknowledgments
We would like to thank all public health nurses and physicians of the STI clinics for their contribution to the data collection and medical microbiology laboratories for STI diagnostics.
Footnotes
FvA and CdD contributed equally.
Handling editor Jackie A Cassell
Contributors All authors contributed to the design of the study. FvA and CdD led on the data analysis and drafting of the manuscript supported by BHBvB, MABvdS, LCS and HJCdV. Additionally, LCS assisted with the analysis in SaTScan. All authors commented on drafts of the manuscript and approved the final version.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.