The use of reportable disease data for monitoring and surveillance of sexually transmitted infections (STIs) is a foundation of public health in most jurisdictions.1 However, although case reports are the basis of most reporting systems, reliance on case data alone has limitations; for example, the inability to analyze trends beyond basic demographic characteristics, lack of data for clinically diagnosed STIs, and inability to describe trends among key populations, such as gay/bisexual men who may not be well defined in surveillance data.
Building on disease registries, several options have been proposed to enhance the robustness of surveillance systems. These include prevalence monitoring as part of routine screening programs and population-based sampling,2 such as HIV behavioral surveillance that has used a variety of sampling methods to determine risk behaviors among gay/bisexual men.3 Sentinel surveillance, in which disease monitoring is done at select sites when population-level estimates are not available, is another approach that has been implemented in many settings. Several global examples underscore the critical role that sentinel surveillance can have in public health programs. In England, diagnoses of STI in genitourinary medicine clinics and other community-based settings have been used as a supplementary data source for their annual surveillance trends.4 In the United States, enhanced data from its STD Surveillance Network have been used to reduce the amount of missing case characteristics from gonorrhea case reports.5 Furthermore, the Gonococcal Isolate Surveillance Project was established to monitor trends in antimicrobial resistance for gonorrhea from sentinel STD clinic sites and regional laboratories throughout the US.6
In British Columbia (BC), we have a centralized reporting system for STI surveillance data that falls under the provincial mandate of the BC Centre for Disease Control (BCCDC). We sought to enhance this foundation by establishing a sentinel surveillance system using data from public health STI clinics that use a common electronic medical record (EMR) system. We believe that this system would greatly enhance our ability to monitor disease rates on an ongoing basis and evaluate public health interventions in BC. A recent evaluation showed that the EMR system was useful in providing additional data that could fill in gaps with missing patient demographics and risk factors.7 In addition, the data could provide routine surveillance for diseases that are based on clinical diagnosis, such as genital herpes and warts, which are currently not notifiable communicable diseases in the province.
The new surveillance platform would be useful for monitoring trends in high-risk groups that are more responsible for ongoing propagation of STI and identifying the emergence of new infections in these populations.8 In particular, BC has seen an increase in infectious syphilis cases among gay/bisexual men, of whom a majority are known to be HIV positive. The increase has been steady since the early 2000s, and apart from a short period of decline starting in 2008, the number of cases has been increasing since 2010.9 It remains to be seen whether the rise in syphilis cases coincides with an increase in other infections, which would be important to assess given the syndemic relationship of HIV and STI.10 Within this context, our objective was to describe disease rates for syphilis and other STI among gay/bisexual men using the sentinel surveillance system that we have developed in BC.
MATERIALS AND METHODS
The STI Information System (STI-IS) is an electronic charting system used by STI clinics across the province of BC. These clinics are publicly funded and provide sexual health care to clients for screening and diagnosis of STI. To establish the sentinel surveillance system, we included 3 clinics in Vancouver and street outreach programs operated by BCCDC. In addition, we identified higher-volume clinics (≥75 visits per year) located in other regions of the province that use the STI-IS for charting and patient care: Fraser (3 clinics), Interior (4 clinics), Northern (2 clinics), and Vancouver Island (3 clinics). We included high-volume clinics to reduce the likelihood of artificially inflating rates due to small-volume clinics that may enter only test-positive cases into the EMR.
Clients who were seen between January 1, 2000, and December 31, 2013 (inclusive), were included in the analysis. The following variables were extracted from STI-IS: client ID, date of visit, clinic site, age, ethnicity, sex, sexual orientation, test, and diagnosis. Gay/Bisexual men were defined as male clients whose sexual orientation was identified as homosexual or bisexual at the time of testing, based on the sex of sexual partners. Cases were defined as clients who tested for a particular STI and had a relevant diagnosis for that STI (see Table, Supplemental Digital Content 1, https://links.lww.com/OLQ/A100, which describes the test and diagnosis values from STI-IS that were used for each STI). HIV data were based on a value of “HIV positive” in the diagnosis field of the EMR system, analogous to the other STI. The system is linked to laboratory test results for each client.
Our analysis was restricted to gay/bisexual men, and we calculated diagnosis rates and incidence density for various STI: chlamydia, gonorrhea, syphilis, genital herpes, genital warts, hepatitis C (HCV), and HIV. Data for HIV were only included starting in 2004, a year after HIV became a reportable disease. The diagnosis rate was calculated by dividing the number of unique cases by the number of unique clients who tested for the STI each year. For chlamydia and gonorrhea, specimens from all anatomic sites were included together. A client was counted only once for a given year even if there were multiple diagnoses or visits within the year. Because diagnoses of genital herpes and warts are made based on clinical examination and not test results, the denominator was the number of unique clients who visited the STI clinic for any reason.
Incidence Density Estimation
Incidence density was estimated among clients with at least 2 tests and whose first test result was negative. Incident cases were defined as clients with a documented change in status from a negative to positive test result, whereas nonincident cases were those who had persistently negative test results throughout their visits. Only the first case for each client was included in the analysis. The incidence density was calculated by dividing the number of unique incident cases by the total person-time contributed by unique clients who tested for the STI, per year. For incident cases, the person-time contribution spanned the time from the first negative test in STI-IS to the first positive test result. For nonincident cases, the person-time contribution spanned the time from the first negative test to the last negative test result. The person-time contribution was then allocated to the relevant years for each client. For instance, if a client contributed 2.5 years from January 2010 to June 2012, 1 person-year would be allocated to years 2010 and 2011, whereas 0.5 person-year would be allocated to 2012.
The incidence density was calculated in this manner for chlamydia, gonorrhea, and syphilis. For genital herpes and warts, estimating incidence rates was not feasible given that diagnoses are not test based. Because the exact time of conversion is less clear for HIV and HCV, we used a method that has been established to calculate incidence density for HIV among repeat testers under the assumption that the probability of conversion is uniformly distributed across the entire seroconversion interval, or the time from a last negative to first positive test.11,12 Based on this method, only the numerator changes from the previous incidence calculation because a fraction of the seroconversion interval, rather than 1, is allocated to the relevant years depending on the proportion of the interval that fell within each year. As an example, a client whose seroconversion interval spans 1.5 years from July 2005 to December 2006 would contribute 1/3 case to 2005 and 2/3 case to 2006. If the entire seroconversion interval lies within the same year, the client contributes 1 to the numerator. For all incidence density estimates, years 2000 and 2013 were excluded due to the presence of a truncation bias that can lead to artificially inflated rates in the end years, as these years tend to capture clients with shorter intertest intervals (i.e., less person-time contribution) who may be at higher risk.13
All diagnosis rates and incidence density were estimated on an annual basis, and 95% confidence intervals were calculated using exact methods. Confidence intervals and person-time contributions were calculated using the “iri” and “stptime” commands, respectively, in Stata (Version 13; College Station, TX). We performed χ2 tests for nonlinear trend and used Poisson regression to examine whether HIV coinfection was associated with incident syphilis and other STI. Our multivariable models also included age, ethnicity (white or nonwhite), clinic location (in or outside Vancouver), and disease-specific annual test volumes (proxy for increased testing), to evaluate their effects on incidence rates. We did not use a model selection process to determine which variables were included in the final model, as we wanted to assess the independent effect of these factors after adjusting for more rather than fewer variables. The Poisson analysis was performed on individual-level data and included person-years as the offset parameter. Analyses were conducted using Stata 13.
Because the data come from an electronic charting system that is used for clinical and disease surveillance purposes under the stewardship of the BCCDC, ethics approval was not required. However, approval was obtained from the regional health authorities for inclusion of their data for analysis.
There were 47,170 visits made by 16,399 unique gay/bisexual men in our sentinel STI clinics during the time period 2000 to 2013 (Table 1). The median number of visits per client over the entire study period was 5 (interquartile range, 2–10), whereas the median number of visits per client each year ranged from 3 to 6 visits annually. Most (96%) of client visits occurred in Vancouver, which included BCCDC outreach sites and clinics. The median age was 34 years (interquartile range, 27–43 years). Most clients were classified as white (63%), followed by Asian (16%), Hispanic (5%), and South Asian (3%). Gay/Bisexual men were most often tested for syphilis at clinic visits (80%), followed by testing for HIV (75%), gonorrhea (72%), and chlamydia (69%). Only 22% of visits involved HCV testing. The prevalence of HCV and HIV in our client population was 1% and 3%, respectively, based on test results or self-reported history during the period of analysis.
The overall diagnosis and incidence rates were highest for chlamydia and gonorrhea. Although diagnoses of chlamydia have remained stable over time, there was a spike in cases in 2012, but this increase was not sustained in the following year (Fig. 1). The incidence density for chlamydia also increased beyond the long-term range in 2012. Diagnoses of gonorrhea, however, fluctuated in the early 2000s but have been on a gradual and steady rise over the long-term trend. Similarly, the incidence of gonorrhea has been increasing since 2008 (Fig. 2).
Syphilis diagnoses increased steadily until 2008 and decreased over the next 2 years before rising again in 2010 to approach the highest rates of the long-term trend. Incident cases of syphilis showed the same trend as its diagnosis rate, with a steady rise before 2008 followed by a decrease and then an upswing after 2010. Diagnoses of new HIV cases have remained constant over time, whereas incidence rates have decreased gradually since 2004 with a sporadic rise in 2011 that was not sustained the following year. Hepatitis C experienced a drop in diagnoses during the period 2000 to 2004, which has been maintained in subsequent years. Although only 15 incident cases of HCV among gay/bisexual men were identified during the study period, increases were observed in the latter years. Rates of genital herpes and warts have remained steady, with a gradual increase for genital herpes and a gradual decrease for genital warts.
From the multivariable Poisson regression models, the adjusted incidence rates for each STI are shown in Table 2. Hepatitis C was excluded from this analysis due to the low number of cases. Moreover, all 15 HCV cases involved HIV-infected individuals. Clients with HIV infection were more likely to have a new gonorrhea and syphilis diagnosis. Clients seen at clinics based in Vancouver were more likely to be diagnosed as having chlamydia, gonorrhoea, and HIV. Using those 60+ years of age as the reference group, we found that younger ages were associated with higher number of chlamydia, gonorrhoea, and HIV diagnoses. There were no differences in incidence between white and nonwhite clients.
By using a clinical management system that contains longitudinal testing data for gay/bisexual men who visited provincial STI clinics, we were able to calculate diagnosis rates and incidence density. Our findings show that the rates of syphilis and gonorrhea are on the rise. The trends for syphilis from our sentinel surveillance system mirror those from provincial surveillance. The rising number of syphilis cases in BC has been shown to be part of a long-term epidemic rather than caused by distinct outbreaks, due to similar characteristics among cases over the phases of the epidemic curve.9 A study from San Francisco also found that their syphilis epidemic followed a similar pattern.14
Gonorrhea cases have been rapidly increasing for several years and preceded the rise in syphilis cases. This trend is particularly concerning given the emergence of drug resistance in BC.15 Our Poisson regression analysis showed that HIV infection was associated with higher incidence rates for both gonorrhea and syphilis, suggesting a common mode of transmission that may be contributing to ongoing transmission among gay/bisexual men.16 The increasing trends for gonorrhea and syphilis are similar to those reported in England using surveillance data from their genitourinary medicine clinics, where gay/bisexual men made up the highest proportion of gonorrhea and syphilis diagnoses.4
Increases in disease rates may be due to different practices or temporal changes that can be described but not controlled for in the analysis.17 The spike in diagnosis and incidence density of chlamydia in 2012 may reflect a pilot study where all gay/bisexual men were routinely offered pharyngeal and rectal nucleic acid amplification testing in the Vancouver clinics. This was done to increase case finding for lymphogranuloma venereum18; routine screening of multiple pharyngeal and rectal sites increased diagnosis rates and impacted the incidence of chlamydia and gonorrhea but did not improve lymphogranuloma venereum case detection (unpublished data). Chlamydia diagnoses dropped in 2013 after the pilot study ended. Although the incidence of chlamydia for 2013 remains incomplete, we are projecting a return to the historic baseline level.
As diagnosis rates of chlamydia, genital herpes, and genital warts have been relatively stable, this suggests that there have not been changes in overall sexual behaviors among gay/bisexual men in our clinic population. Rather, the increase in gonorrhea and syphilis cases could reflect different sexual networks involving HIV-positive men or different routes of transmission (e.g., oral sex). There is a need to investigate these hypotheses to inform broader sexual health interventions beyond targeted disease-specific interventions for gay/bisexual men, such as expanded testing for STI among HIV-positive men.19 Rates of HIV were also stable in our population, which could be due to the impact of the provincial strategy to better link HIV-positive individuals to antiretroviral therapy to reduce the population burden of HIV. However, the low prevalence of HIV-positive men in our sample precluded the ability to stratify disease rates by HIV status, likely due to the large volume of testing occurring outside STI clinics that would not be captured. A 2011 survey among gay/bisexual men in BC showed that 13% of respondents reported being infected with HIV, ranging from 3% in northern BC to 26% in Vancouver.19 Efforts are underway to link other data sources to better determine HIV infection and extend our sentinel sites to other clinics accessed by HIV-positive men.
Although the number of new HCV cases was small, the incidence density seems to be increasing among HIV-positive gay/bisexual men in the latter years of our study period. However, the rise in seroconversions within the most recent years may also be attributed to an increase in HCV screening among gay/bisexual men starting in 2011. Other studies in Europe and Australia have also reported similar trends among HIV-positive individuals.20–23 There is mounting evidence suggesting that HCV is sexually transmitted among gay/bisexual men.24,25
A surveillance system based on administrative data is inexpensive and efficient for estimating and tracking incidence density. However, our sentinel surveillance data have some limitations. The rates among an STI clinic population are expected to be higher than those in the general population given the increased risk of infections in those attending STI clinics. The use of retrospective cohorts can also bias incidence rates because they will increase as the analytic period gets shorter due to an artifact of using health administrative data.13 In addition, we did not include behavioral data or provider type because the quality of this data in the system has not yet been assessed. Validation of this type of data would be key to understanding and interpreting incidence trends among important subgroups.
Disease rates may also vary with changes in test sensitivity or changes in the client population over time (see Table, Supplemental Digital Content 2, https://links.lww.com/OLQ/A101, which describes trends in the composition of our sample of gay/bisexual men). In our sample, the proportion of gay/bisexual men tested for any STI increased over time, likely due to increased awareness among clients and providers. The proportion of clients with HIV infection decreased, whereas there were no substantial changes in the age composition of our population. Because our study population was overrepresented by Vancouver-based clients, our findings may not reflect the pattern of infections for gay/bisexual men living in rural or remote areas. Furthermore, the use of data from STI clinics would not capture patients who get testing in primary-care settings. Approximately one-third of all syphilis and HIV cases reported in the province among gay/bisexual men during the study period were captured by our sentinel surveillance system.
Additional limitations were related to the incidence density calculations. The reason for seeking care and the risk of STI are not independent, and our clinic population will be at higher risk compared with the general population. The more likely the visit was related to the probability of disease, the greater the bias toward increasing incidence density. We used the more complex analysis of calculating incidence density (compared with the simpler cumulative incidence approach) to take into account the at-risk time for each client and to use a consistent framework for all STI. Using time at risk rather than individuals provides more informative estimates, which is especially pertinent for HIV and HCV where disease may be asymptomatic and a prior negative test result is needed to ascertain an incident case because duration of at-risk time is not discrete.
We only included the first diagnosis for each client. Because repeat infections are common for bacterial STI,26,27 the incidence density will be underestimated for episodic infections. Using the data from our sentinel surveillance system to calculate rediagnosis and coinfection rates will be an area of future work. Also, the restriction to repeat testers with a first negative test resulted in substantial data loss; 42% to 45% of the data remained for most STI, whereas the proportion was lower for HCV (25%). However, one study that estimated the incidence of HCV among HIV-positive individuals showed that rates were similar between the standard approach that we have used and a method that allows for inclusion of all patients.28
Overall, our study demonstrates the usefulness of implementing a sentinel surveillance approach to describe STI rates among gay/bisexual men using a STI clinic population. Future work will improve the current analysis by including multiple infections for each client, evaluating trends using differing analytic period lengths,13 and increasing the amount of data for analysis by including first test positives.28 Sentinel surveillance data can help characterize infection trends in key groups, such as gay/bisexual men, to target prevention efforts and to evaluate the impact of interventions. This has been done most recently for evaluating the effect of mass HPV vaccination programs on the decline of genital warts.29,30 In the future, data from our sentinel clinic sites will be incorporated into data warehouses developed by the BCCDC to permit ongoing, real-time linkages of surveillance, clinical, and laboratory data for analysis and reporting. Thus, the new sentinel surveillance platform will support STI monitoring in BC to inform future policies and reduce the burden of disease in the province.
1. Taylor M, Gilbert M. Evaluation of the Chlamydia Infection Surveillance System in BC, 1998–2007. Vancouver: BC Centre for Disease Control; 2009: 1–34.
2. Douglas JM, Fenton KA. STD/HIV Prevention Programs in Developed Countries. In: Holmes KK, Sparling PF, Stamm WE, et al., eds. Sexually Transmitted Diseases. 4th ed. New York: McGraw-Hill; 2008: 1767–1785.
3. Paquette D, De Wit J. Sampling methods used in developed countries for behavioural surveillance among men who have sex with men. AIDS Behav 2010; 14: 1252–1264.
4. Sexually Transmitted Infections and Chlamydia Screening In England, 2013. London: Public Health England; 2014: 1–18.
5. Sexually Transmitted Disease Surveillance 2012. Atlanta: Centers for Disease Control and Prevention; 2013.
6. Schwarcz SK, Zenilman JM, Schnell D, et al. National surveillance of antimicrobial resistance in Neisseria gonorrhoeae
. The Gonococcal Isolate Surveillance Project. JAMA 1990; 264: 1413–1417.
7. Rank C. Evaluation of the Sexually Transmitted Infections Information System to Assess Trends in Gonorrhea Among Clinic Attendees, 2000 to 2009. Vancouver: BC Centre for Disease Control, 2012: 1–39.
8. Thomas JC, Tucker MJ. The development and use of the concept of a sexually transmitted disease core. J Infect Dis 1996; 174 (suppl 2): S134–S143.
9. Hottes T, Lindegger M, Consolacion T, et al. Infectious Syphilis Among Gay, Bisexual and Other Men Who Have Sex With Men in British Columbia, 2003 To 2012. Vancouver: BC Centre for Disease Control, 2013: 1–16.
10. Singer MC, Erickson PI, Badiane L, et al. Syndemics, sex and the city: understanding sexually transmitted diseases in social and cultural context. Soc Sci Med 2006; 63: 2010–2021.
11. Calzavara L, Burchell AN, Major C, et al. Increases in HIV incidence among men who have sex with men undergoing repeat diagnostic HIV testing in Ontario, Canada. AIDS 2002; 16: 1655–1661.
12. Kitayaporn D, Uneklabh C, Weniger BG, et al. HIV-1 incidence determined retrospectively among drug users in Bangkok, Thailand. AIDS 1994; 8: 1443–1450.
13. Kent CK, Chaw JK, Kohn RP, et al. Studies relying on passive retrospective cohorts developed from health services data provide biased estimates of incidence of sexually transmitted infections. Sex Transm Dis 2004; 31: 596–600.
14. Bernstein KT, Stephens SC, Strona FV, et al. Epidemiologic characteristics of an ongoing syphilis epidemic among men who have sex with men, San Francisco. Sex Transm Dis 2013; 40: 11–17.
15. Hottes TS, Lester RT, Hoang LM, et al. Cephalosporin and azithromycin susceptibility in Neisseria gonorrhoeae
isolates by site of infection, British Columbia, 2006 to 2011. Sex Transm Dis 2013; 40: 46–51.
16. Wolitski RJ, Fenton KA. Sexual health, HIV, and sexually transmitted infections among gay, bisexual, and other men who have sex with men in the United States. AIDS Behav 2011; 15 (suppl 1): S9–S17.
17. Merrill R. Introduction to Epidemiology. 5th ed. Burlington, MA: Vancouver: Jones & Bartlett Learning; 2010.
18. Lindegger M, Hottes T, Gilbert M, et al. Lymphogranuloma Venereum in British Columbia, 2004 to 2011. Vancouver: BC Centre for Disease Control, 2012: 1–11.
19. HIV, Stigma and Society: Tackling a Complex Epidemic and Renewing HIV Prevention for Gay and Bisexual Men in British Columbia. Provincial Health Officer’s 2010 Annual Report
, Victoria, BC: Ministry of Health, : 2014.
20. Blaxhult A, Samuelson A, Ask R, et al. Limited spread of hepatitis C among HIV-negative men who have sex with men in Stockholm, Sweden. Int J STD AIDS 2013; 25: 493–495.
21. Jin F, Prestage GP, Matthews G, et al. Prevalence, incidence and risk factors for hepatitis C in homosexual men: Data from two cohorts of HIV-negative and HIV-positive men in Sydney, Australia. Sex Transm Infect 2010; 86: 25–28.
22. Schmidt AJ, Falcato L, Zahno B, et al. Prevalence of hepatitis C in a Swiss sample of men who have sex with men: Whom to screen for HCV infection? BMC Public Health 2014; 14: 3.
23. van de Laar T, Pybus O, Bruisten S, et al. Evidence of a large, international network of HCV transmission in HIV-positive men who have sex with men. Gastroenterology 2009; 136: 1609–1617.
24. Bradshaw D, Matthews G, Danta M. Sexually transmitted hepatitis C infection: The new epidemic in MSM? Curr Opin Infect Dis 2013; 26: 66–72.
25. Gorgos L. Sexual transmission of viral hepatitis. Infect Dis Clin North Am 2013; 27: 811–836.
26. Brunham RC, Pourbohloul B, Mak S, et al. The unexpected impact of a Chlamydia trachomatis
infection control program on susceptibility to reinfection. J Infect Dis 2005; 192: 1836–1844.
27. Ogilvie GS, Taylor DL, Moniruzzaman A, et al. A population-based study of infectious syphilis rediagnosis in British Columbia, 1995–2005. Clin Infect Dis 2009; 48: 1554–1558.
28. van der Helm JJ, Prins M, del Amo J, et al. The hepatitis C epidemic among HIV-positive MSM: Incidence estimates from 1990 to 2007. AIDS 2011; 25: 1083–1091.
29. Ali H, Donovan B, Wand H, et al. Genital warts in young Australians five years into national human papillomavirus vaccination programme: National surveillance data. BMJ 2013; 346: f2032.
30. Donovan B, Franklin N, Guy R, et al. Quadrivalent human papillomavirus vaccination and trends in genital warts in Australia: Analysis of national sentinel surveillance data. Lancet Infect Dis 2011; 11: 39–44.