Increasing Syphilis Detection Among Patients Assigned Male at Birth Screened at a Boston Community Health Center Specializing in Sexual and Gender Minority Health, 2005–2015

Abstract

Background 

United States syphilis rates have increased to levels last seen in the 1990s. We examined syphilis epidemiology of patients attending a Boston community health center specializing in sexual and gender minority health.

Methods 

We performed a retrospective cohort study of all patients assigned male at birth screened with rapid plasma reagin from 2005 through 2015. We developed an algorithm to identify new infections and used repeat cross-sectional analysis to assess temporal trends in syphilis diagnoses. We also performed longitudinal analysis to calculate syphilis incidence using a Cox proportional hazards model that accounts for multiple infections over time.

Results 

Eighteen thousand two hundred eighty-two patients had a total of 57,080 rapid plasma reagins, 1170 (2.0%) tests met criteria for syphilis. Adjusted syphilis diagnoses increased from 1.2% to 1.9%, recurrent syphilis diagnoses increased from 0.04% to 0.3% during the study period. Black and Hispanic/Latinx patients, patients aged 35 to 44 years, gay/bisexual patients, cisgender men, and human immunodeficiency virus (HIV)-infected patients and those who became HIV-infected during the study period were more likely to test positive for syphilitic infection in repeat cross-sectional analysis. Among 6199 patients screened more than 1 time over 21,745 person-years, there were 661 new syphilis cases (3.0% annual incidence; 95% confidence interval [CI], 2.8% to 3.2%). Compared with those aged 14 to 24 years, patients 45 years or older were less likely to experience syphilis. New HIV infection was associated with increased risk of incident syphilis (adjusted hazard ratio, 2.87; 95% CI, 1.61–5.13). Virally suppressed HIV-infected patients were less likely to experience incident syphilis (adjusted hazard ratio, 0.69; 95% CI, 0.55–0.87).

Conclusions 

The high incidence of syphilis among patients assigned male at birth disproportionately affected young patients, black and Hispanic/Latinx patients, gay/bisexual patients, cisgender men, and those with new or chronic HIV infection. Syphilitic reinfection rates increased over time.

In Brief

For patients assigned male at birth seen at a Boston clinic over a decade, we found disproportionate syphilis cases in young patients, black and Hispanic/Latinx patients, gay/bisexual patients, cisgender men, and those with new or chronic human immunodeficiency virus infection; syphilitic reinfections have significantly increased.

The resurgence of syphilis is a significant public health concern.^1–4 The Centers for Disease Control and Prevention (CDC) reports that, from 2000 to 2015, the annual rate of new primary and secondary (P&S) syphilis cases more than tripled from 2.1 to 7.5 per 100,000 United States population.¹ This trend has disproportionately affected men, particularly men who have sex with men (MSM). As of 2016, the CDC reported that the rate of P&S syphilis in males was 9.7 times that in females with 60% of all P&S syphilis cases in the United States occurring among MSM.^1–3 Focused syphilis screening and prompt treatment are critical in addressing this epidemic.^4–6

The goals of the current study were to define trends in syphilis diagnoses over time and identify sociodemographic predictors of syphilitic infection among an open cohort of patients who are assigned male at birth accessing primary care at a Boston community health center nationally recognized for its care of gender and sexual minorities.⁷ The health center is not only the largest primary care provider of MSM and people living with human immunodeficiency virus (HIV) in New England but also diagnoses the greatest number of bacterial sexually transmitted infections (STIs) than any other site in the region.⁸ We reviewed standardized electronic health records to examine our community health center's syphilis epidemiology.

METHODS

Study Design and Population

We conducted a retrospective, electronic chart review of all patients who were assigned male at birth, aged 14 years or older, and screened for syphilis at a federally qualified health center in Boston between October 1, 2005, and September 30, 2015.

Outcome of Interest

Syphilis, defined by rapid plasma reagin (RPR) testing, was the outcome of interest for this study. We classified diagnosis of syphilis in 3 ways: (1) a 4-fold rise in RPR titer after successful treatment for a prior syphilitic infection at least 3 months ago; (2) a new positive titer of at least 1:4 after previous negative RPR testing; or, (3) an RPR titer of 1:16 or greater without any previous RPR testing in the electronic health record. Evaluation of confirmatory treponemal testing was not included in this algorithm and all classifications were restricted to testing performed at our community health center. We reviewed charts of 15% of patients at random to assess the performance of this algorithm using clinician diagnosis of syphilis as the gold standard. The sensitivity of the algorithm was 73.9%, specificity was 99.9% with a positive predictive value of 95.9%, and a negative predictive value of 99.4%.⁹

Covariates

We extracted data regarding age, race/ethnicity, sexual orientation, gender identity, and HIV status from the medical record. We created 3 categories for HIV status: (1) patients who tested negative during the study period and never had a positive test; (2) patients who were known to be HIV-positive at their first syphilis testing visit; and, (3) patients who tested HIV-positive after testing HIV-negative during the study period. We also examined HIV viral loads as a time-varying covariate and a categorical variable. HIV-positive patients were classified as having a detectable or undetectable viral load at each testing visit. We defined undetectable as a viral load less than 200 copies/mL. HIV-positive patients were further classified as having consistent viral suppression (all viral loads undetectable during the study period) or not (more than 1 detectable viral load during the study period). Patients who were never undetectable (259 [10.9% of total HIV-positive population] contributing 533 RPRs [3.4% of the total RPRs contributed by HIV-positive patients]) were classified as inconsistently virally suppressed due to their relatively small numbers in the data.

There was a significant amount of missing data regarding gender identity and sexual orientation of our patients. These details were often not consistently documented nor could be readily extracted from the electronic medical record (EMR). Despite longitudinally compiling gender identity and sexual orientation across all patient visits during the study period, a large amount of missing data remained. Of note, sexual orientation was documented in the EMR by self-reported orientation (heterosexual, gay, bisexual) or self-reported sex of sexual partners (male, female). For purposes of analysis, patients who identified as heterosexual were grouped with those who identified as men who have sex with women; those identified as gay were grouped with MSM; and those who identified as bisexual were grouped with men who have sex with men and women. We present statistical analysis of these variables with patients missing this data designated in a separate category.

Statistical Analyses

Initially, we treated the patients tested for syphilis in each year-long interval during the study period as a cross-sectional sample. From 2005 to 2015, the percentage of patients who tested for syphilis who were nonwhite increased from 23% to 31% and who were HIV-negative increased from 65% to 81%. The percentage of those who were 45 or older increased from 25% in 2005 to 2006 to 29% in 2014 to 2015. Because the change in demographics over time could introduce bias in syphilis diagnosis rates, we calculated race-, age-, and HIV status–adjusted syphilis diagnosis rates using multivariable Poisson regression with robust standard error estimation to account for the changing case mix over time and correlation between cross-sectional samples. We adjusted all models for number of tests contributed by each patient. We used the same techniques to calculate rate ratios (RRs) and 95% confidence intervals (CIs) comparing diagnosis rates between time intervals.¹⁰ We included interactions between time, HIV status, age, and race/ethnicity to determine time trends among groups of patients defined by sociodemographic characteristics and HIV status. Finally, we assessed predictors of overall and recurrent syphilis diagnosis from entrance into the study period to the conclusion of the observed decade.

For the longitudinal analysis, we included patients with 2 or more RPR tests spaced greater than 30 days apart. Time at-risk began with the first negative RPR. The same RPR algorithm was used to define syphilitic infection. For those with an RPR meeting criteria for an incident syphilis, we assumed that infection occurred at the midpoint between the previous RPR that was either negative or did not meet criteria for syphilis and the RPR that did meet criteria for incident syphilis. For the base analysis, we assumed that patients returned to the at-risk pool immediately after diagnosis and treatment (continuous risk intervals). We compared this analytic approach to an approach that assumed that patients were not at risk for reinfection for a 14-day period after diagnosis and treatment (discontinuous risk intervals). Because the conclusions from each approach were the same; the base analysis is presented here. Patients were censored at their last testing visit during the study period.

We calculated the incidence rate and cumulative incidence of syphilis for the overall cohort and for groups of patients defined by demographic characteristics and HIV status. We used the Prentice-Williams-Peterson modification to the Cox proportional hazards model to account for multiple failures (ie, recurrent syphilis episodes) and to estimate bivariable and multivariable hazard ratios and 95% CI.^11–14 We “ordered” the risk for syphilis, classifying patients as being at-risk for their first syphilis episode, as being at risk for a first recurrent infection, and as being at risk for additional recurrent infections. Patients could advance sequentially through these groups over the follow-up period to reflect a change in the hazard of syphilis based on prior infections. This classification served as a stratification variable in our Cox models. We used Schoenfeld residuals to assess the proportional hazards assumption for each model; none of the models violated this assumption.

All statistical testing used a P < 0.05 level of statistical significance. We used STATA 14.2 (StataCorp, College Station, TX) for all analyses.

RESULTS

During the study period, 18,282 patients had 57,080 unique visits for syphilis testing. The patients were 70.3% white, 6.9% black, 6.9% Hispanic/Latinx, 5.2% Asian or Pacific Islander, and 6.0% multiracial. Twenty-four percent were ages 14 to 24 years, 34.4% were ages 25 to 34 years, 21.8% were ages 35 to 44 years, 15.1% were ages 45 to 54 years, and 5.9% were 55 years or older. Forty-three percent identified as gay, 11.4% identified as heterosexual, and 3.1% identified as bisexual; of note, sexual orientation was not documented in 42% of patients screened during this time. Forty-three percent of patients identified as cisgender men, 2.3% as transgender female, and 0.4% as nonbinary or gender nonconforming; 53% of patients did not have data regarding their gender identity. Eighty-five percent of all patients screened were HIV-uninfected at the time of syphilis screening.

Patients tested a median of 2 times during the study period (mean 3.1 screenings; range: 1 to 42 screenings). Forty-nine percent had 1 testing visit during the study period while 16% had 2 visits, 9% had 3 visits, 6% had 4 visits, and 20% had 5 or more visits. The number of testing visits increased over the study period (Table 1).

TABLE 1: Characteristics of Individuals Assigned Male at Birth Who Tested for Syphilis at a Federally-Qualified Community Health Center, Boston, 2005–2015

Time Trends in Syphilis Diagnoses

Overall, 1170 (2.0%) of the 57,080 RPR tests met criteria for syphilis. Annual syphilis diagnosis rates adjusted for age, race, HIV status, and number of visits increased from 1.2% in 2005 to 2006 (95% CI, 1.1%–1.4%) to 1.9% in 2014 to 2015 (95% CI, 1.7–2.1%, test for trend P < 0.001, Fig. 1A). Among HIV-negative patients, adjusted syphilis diagnoses rates increased from 0.9% in 2005 to 2006 (95% CI, 0.7–1.1%) to 1.3% in 2014 to 2015 (95% CI, 1.1–1.5%, test for trend P < 0.001). Among HIV-positive patients, these rates increased from 2.5% (95% CI, 2.0–3.0%) to 3.6% (95% CI, 3.0–4.1%, test for trend P < 0.001, Fig. 1B). Syphilis diagnoses increased in those 14 to 24 years of age and those 45 years and older (Supplemental Fig. 1a, https://links.lww.com/OLQ/A349). With regard to race and ethnicity, syphilis diagnoses increased in all racial and ethnic groups with the exception of Hispanic/Latinx patients (Supplemental Fig. 1b, https://links.lww.com/OLQ/A350).

Figure 1: Longitudinal analysis of trends in syphilis diagnoses at a Boston Community Health Center 2005–2015. Shown on the Y-axis is the proportion of RPR tests indicating overall syphilis cases (A), overall syphilis cases by HIV status (B), recurrent syphilis cases (C), and recurrent syphilis cases by HIV status (D).

Twenty-two percent of all syphilis cases over the study period were considered recurrent infections. Adjusted recurrent syphilis diagnosis rates increased from 0.04% (95% CI, 0.03–0.06%) to 0.3% (95% CI, 0.2–0.4%) between 2005 to 2006 and 2014 to 2015, respectively (test for trend P < 0.001, Fig. 1C). Recurrent syphilis diagnoses increased from 0.02% (upper 95% CI, 0.05–0.2%; 95% CI, 0.1–0.3%; test for trend P < 0.001) among HIV-negative patients and from 0.1% (95% CI, 0.08–0.2%) to 0.9% (95% CI, 0.6–1.2%, test for trend P < 0.001, Fig. 1D) among HIV-positive patients. Recurrent syphilis diagnoses increased among patients 25 to 54 years of age and among all race/ethnicity groups (Supplemental Figs. 2a and 2b, https://links.lww.com/OLQ/A351).

Predictors of Syphilis

In bivariable analyses, patients aged 35 to 44 years (compared to those aged 14–24 years); black, Hispanic/Latinx, and multiracial/multiethnic patients (compared to white patients); gay/bisexual patients (compared to heterosexual patients); cisgender men (compared to transgender and gender nonconforming [TGNC] patients); and HIV-positive patients and those who seroconverted (compared to HIV-negative patients) were more likely to be diagnosed with syphilis. In multivariable analysis, patients 55 and older (compared with 14–24 year olds) were less likely to be diagnosed with syphilis while those tested after 2012 (compared with those tested in 2005–2006); black, Hispanic/Latinx patients (compared with white patients); gay/bisexual patients (compared with heterosexual patients); cisgender men (compared with TGNC patients); and HIV-positive patients and those who seroconverted (compared with HIV-negative patients) were more likely to be diagnosed with syphilis (Table 2).

TABLE 2: Predictors of Overall Syphilis Test Positivity and Test Positivity Representing Recurrent Syphilis Among Individuals Assigned Male at Birth Who Tested for Syphilis at a Federally Qualified Community Health Center, Boston, 2005–2015

Among HIV-positive patients, we evaluated the association of viral suppression on syphilis diagnosis in 2 ways. First, we modeled viral suppression as a time-varying covariate. There were 312 (2.8%) of 11,053 positive tests at which HIV-positive patients had an undetectable viral load, whereas there were 120 (4.5%) of 2668 positive tests at which HIV-positive patients had detectable viral loads. In a multivariable model, viral suppression at time of syphilis testing was associated with a 37% decrease in the likelihood of syphilis test positivity (95% CI, 21%–51%, P < 0.001). Second, we classified HIV-positive patients as having consistent versus inconsistent viral suppression over the follow-up period. Of the 7975 tests contributed by patients who were consistently suppressed, 221 (2.8%) were positive; of the 7860 tests contributed by those were not consistently suppressed, 345 (4.4%) were positive. In a multivariable model, those with consistent viral suppression were 26% less likely to test positive for syphilis (95% CI, 11%–39%, P = 0.002).

Predictors of Recurrent Syphilis

Due to relatively small numbers of syphilis diagnoses representing recurrent infection during the initial year-long intervals of the study period, we modeled time as a continuous variable. Recurrent infections increased 26% over the observed decade in bivariable and multivariable models. In bivariable analyses, patients aged 35 to 54 years (compared with those aged 14–24 years); black patients (compared to white patients); gay/bisexual patients (compared with heterosexual patients); and those who were HIV-positive or seroconverted (compared with HIV-negative patients) were more likely to experience recurrent syphilis. In multivariable analysis, only black race and HIV status predicted syphilitic reinfection.

There was no association between an undetectable viral load as modeled as a time-varying covariate on the risk of a recurrent syphilis diagnosis (absolute RR = 1.02, 95% CI, 0.60–1.73, P = 0.928). However, HIV-infected patients with consistent viral suppression were significantly less likely to have a repeat positive syphilis test than those with inconsistent viral suppression (aRR = 0.51, 95% CI, 0.34–0.75, P = 0.001).

Longitudinal Cohort of Repeat Testers

There were 6199 patients with negative syphilis serology at baseline with 1 or more follow-up tests at least 30 days apart. Median follow-up time was 3 years (range: 30 days, 9.7 years) with a total of 21,745 person-years of follow-up time. Similar to the repeat cross-sectional samples, cisgender, HIV-negative, gay/bisexual patients aged 25 to 54 years tested more recently comprised the majority of person-time.

There were 661 new syphilis cases during this time for an overall incidence of 3.0% (95% CI, 2.8%–3.2%). Compared with latter time intervals, incidence was greatest during the 2005 to 2008 time intervals, after which it decreased and stabilized (Table 3). Patients 14 to 24 years of age were more likely to have incident syphilis compared to those 45 years of age and older. Black patients were more likely than white patients to acquire syphilis; this comparison was statistically significant in bivariable, but not multivariable, models. Gay/bisexual patients and cisgender men were more likely to have incidence syphilis thank heterosexual and TGNC patients, respectively. HIV-positive patients and those who seroconverted during follow-up were significantly more likely to experience incident syphilis than HIV-negative patients.

TABLE 3: Incidence and Predictors of Syphilis in a Longitudinal Cohort of Individuals Assigned Male at Birth Who Tested for Syphilis at a Federally-Qualified Community Health Center, Boston, 2005–2015

Among patients who were HIV-negative at entry into the longitudinal cohort, a new HIV infection was associated with an almost 3-fold increase in the risk of incident syphilis (adjusted hazard ratio [HR], 2.87; 95% CI, 1.61–5.13) in the multivariable model in which HIV seroconversion was a time-varying covariate. Among HIV-positive patients, those with consistent viral suppression over the study period were less likely to experience incident syphilis when compared with those who were inconsistently virally suppressed (adjusted HR, 0.69; 95% CI, 0.55–0.87). There was no association between an undetectable viral load as modeled as a time-varying covariate on the risk of incident syphilis.

Cumulative incidence of syphilis at year 9 of follow-up was 29% (95% CI, 25–33%; Fig. 2A). Cumulative incidence was 15% (95% CI, 11%–21%) among HIV-negative patients, 40% (95% CI, 36%–46%) among HIV-positive patients, and 67% (95% CI: 53%–80%) among those who seroconverted during follow-up (log-rank P < 0.001; Fig. 2B). Nine-year cumulative incidence was 29% for white (95% CI, 25%–33%), Asian/Pacific Islander (95% CI, 18–44%), and Hispanic/Latinx patients (95% CI, 22–37%) and 31% for black patients (95% CI, 25%–39%) and patients identifying as multiracial, multiethnic, or other (95% CI, 20%–45%; log-rank P = 0.003; Supplemental Fig. 3a, https://links.lww.com/OLQ/A352). Among 14- to 24-year-olds, cumulative incidence was 18% (95% CI, 12–28%) through year 7 of follow-up, after which the numbers became too small to calculate an estimate of risk. Cumulative incidence was 28% (95% CI, 22%–34%) among 25- to 34-year-olds, 30% (95% CI, 27%–35%) among 35- to 44-year-olds, 30% among 45- to 54-year-old (95% CI, 23%–38%), and 20% (95% CI, 11%–21%) among patients 55 years and older (log-rank P < 0.001; Supplemental Fig. 3b, https://links.lww.com/OLQ/A353). Finally, among patients who were inconsistently virally suppressed, cumulative incidence was 47% (95% CI, 43%–53%) compared with 29% (95% CI, 24%–35%) among those who were consistently virally suppressed over the follow-up period (log-rank P < 0.001: Fig. 2C).

Figure 2: Cumulative incidence of syphilis diagnoses at a Boston Community Health Center 2005–2015. Shown on the Y-axis is cumulative incidence of syphilis overall (A), syphilis by HIV status (B), and syphilis by viral suppression (C).

DISCUSSION

Among patients assigned male at birth tested for syphilis at a large community health center in the Northeast, we detected a modest but significant increase in syphilis diagnoses between 2005 and 2015. This upward trend was more pronounced in HIV-infected patients, those aged 25 years and older, gay/bisexual patients, cisgender men, and patients who identified as white, black, or Asian/Pacific Islander. Compared with white patients, black and Hispanic/Latinx patients were more likely to be diagnosed with syphilis. Recurrent syphilis diagnoses markedly increased during the observed decade, particularly among black patients and HIV-positive patients and those who seroconverted. Overall, 22% of all positive diagnoses during the study period represented repeat infections. Several sites within the United States and Canada have seen similar trends in repeat syphilis cases. A San Francisco study evaluating syphilitic infections between 2001 and 2002 among MSM found 6.7% of cases had a repeat infection within 1 year and that HIV-infected MSM are higher risk for syphilitic reinfection.¹⁴ Another California study examining rates of early syphilis between 2002 and 2006 found that although the annual proportion of MSM with repeat syphilis cases did not change over 2 years, HIV infection was still found to be a significant risk factor associated with reinfection.¹⁵ A recent study from Alberta, Canada examining HIV-infected patients found over a 3-fold increase of syphilis repeat infections between 2009 and 2016.¹⁶

The increased detection of recurrent infections over the study period may reflect an increase in screening during the same time. Multiple studies have demonstrated that aggressive screening of high-risk asymptomatic patients can be used to achieve syphilis control.^17–19 These screening practices, particularly of MSM and HIV-infected male patients, have been adopted by our community health center. Over a third of the patients in the current study had 3 or more RPR visits over the course of the study period and the average number of syphilis testing visits per male patient per year increased from 1.2 to 4.2 between 2005 and 2015.

The HIV-infected patients and those who seroconverted during the study period were significantly more likely to test positive for syphilis and reinfection. In addition, among HIV-infected patients, the proportion of syphilis diagnoses that represented recurrent infection increased 8-fold (0.2% to 1.6%, test for trend P < 0.001) during the observed decade. There are several factors which may contribute to the increased syphilis risk of HIV-infected patients. First, an increased baseline prevalence of syphilis within the sexual networks of HIV-infected MSM may mean that each sexual encounter is associated with greater likelihood of being exposed to syphilis. Second, serosorting is a well-studied practice in the MSM community, that is, preferential condomless anal sex with partners of the same serostatus.^15,20,21 Although serosorting among HIV-infected sex partners can be a strategy to combat HIV transmission, it may increase syphilis and reinfection in this high-risk group. Third, as HIV-infected patients are more likely to be linked to care than their HIV-uninfected counterparts, they are more likely to be screened more frequently for syphilis and other STI. Finally, it is possible that HIV infection affects the transmissibility and susceptibility of syphilis. It is thought that HIV infection can suppress the proinflammatory pathway responsible for clearing Treponema pallidum, which can enhance a person's risk for infection and reinfection.^22,23 Although some studies have been concerned about the negative effect of certain antiretroviral agents on this proinflammatory response,^23,24 more recent data shows that patients not using antiretroviral therapy were at increased risk of syphilis.¹⁶

Among HIV-positive patients, we found that those with detectable viral loads during time of screening or inconsistent viral suppression over the duration of the study period were more likely to have syphilis. Not only are these patients more susceptible to syphilis, they are more likely to transmit both syphilis and HIV to their sexual partners. Patients with detectable viral loads and syphilis should be a priority population for more frequent STI screening and intensive efforts to achieve durable viral suppression. Those who became HIV-infected during the study had nearly 3 times higher risk of incident syphilis than their HIV-negative counterparts with a cumulative incidence of 67%. Prior studies from Washington State and New York City have established the association with HIV acquisition following a rectal STI or early syphilis diagnosis.^25,26 In addition, increased STI incidence has also been observed after acute HIV infection in a young MSM cohort from North Carolina.²⁷ The HIV seroconversion should be a time for intensive STI screening, particularly of syphilis.

Black patients, although accounting for 6.9% of all patients screened, had disproportionately higher test positivity indicating syphilitic infection and reinfection; these results are mirrored in national data. As of 2016, the CDC found the rate of primary and secondary syphilis in black men was 8.3 times that of their white counterparts.¹ A recent study also found that black men were at increased odds for repeat syphilitic infections.¹⁵ These findings can partially be explained by sexual networks of black men, particularly black MSM, who may be more likely to report sexual partners of the same race, assortative mixing.^28,29 This subsequently can lead to a tighter sexual network and consequently amplify the spread of syphilis and other STIs.

Patients assigned male at birth who identified as gay or bisexual were more likely to be diagnosed with a syphilitic infection or reinfection and more likely to have incident syphilis during the study period. Additionally, cisgender men were more likely to be diagnosed with syphilis and incident syphilis over the observed decade. Despite the significant amount of missing sexual orientation data, our findings among gay and bisexual men were consistent with national data and have been well studied across North America.^{1–3,5,8,14–16} However, our findings with regard to cisgender men compared with TGNC patients are more difficult to interpret. Although STI surveillance data on TGNC persons are limited, a recent meta-analysis estimated the HIV prevalence in US transgender women at 21.6% compared to the 0.3% in the general US population.³⁰ The 2015 STD Treatment Guidelines include transgender persons as a special population for regular risk assessments and STI screenings.⁴ The lack of gender identity information available in the EMR for over half of our patients certainly contributed to sampling bias and may not provide a full picture of syphilis diagnoses in our TGNC patients. This study underscores the need for consistent documentation of gender and sexual identity to better address the challenges of these marginalized populations.

The findings of our study should be interpreted within the context of additional limitations. Our classification of syphilis was based on an algorithm involving individual RPR results. Although this algorithm performed well on a randomized chart review with 99.9% specificity, its maximum sensitivity remained at 73.9%. The algorithm may present an underestimation of the true number of syphilitic infections and reinfections. However, because our algorithm was also restricted to testing performed at our facility and that our gold standard of clinician diagnosis is not always reliable for latent disease, we must also consider the effect of ascertainment bias on the true number of syphilis diagnoses. As much of our data were extracted from electronic records, we were unable to closely examine certain variables. We also did not collect any information about sexual behaviors with regards to condom versus condomless anal sex, number of sexual partners, or history of other STIs as they were not readily extractable from patients' records. These additional variables would better define the syphilis risk group for whom interventions should be aimed. Lastly, the adoption of preexposure prophylaxis occurred at the end of our study period and could not be adequately assessed as a separate variable.

Despite these limitations, our study found that repeat syphilis diagnoses dramatically increased between 2005 and 2015 at a Boston community health center. We also identified young patients, black and Hispanic/Latinx patients, gay/bisexual patients, cisgender men, and those with new or chronic HIV infection as key risk groups for both syphilitic infection and reinfection over this observed decade. As this epidemic continues to grow, we anticipate initial and repeat syphilis episodes to be increasing public health concerns. Developing strategies to recognize those at greatest risk, screen them regularly, and retain them in care will be critical for any future interventions aiming to establish syphilis control.