Incidence of syphilis infection and syphilis-related care utilization among adolescents and young adults living with HIV

Abstract

Background.

Incidence of syphilis has been rising in recent years and disproportionately affects young adults, racial/ethnic minority men, and people living with HIV. This study describes patterns of syphilis infection and syphilis-related care utilization among adolescents and young adults living with HIV (AYALH) in Philadelphia.

Methods:

We conducted a retrospective cohort study of AYALH receiving care at an adolescent-specialty clinic who received a syphilis test and/or benzathine penicillin for syphilis treatment from 2011–2018 (N=335). Syphilis incidence rates were calculated by baseline demographic characteristics and by calendar year. Recurrent survival analysis was used to explore how demographic and neighborhood-level factors were associated with incident syphilis and syphilis-related care utilization.

Results:

Syphilis-related care was provided 145 times and there were 109 episodes of confirmed syphilis among 83 unique participants between 2011 and 2018. The overall syphilis incidence rate was 13.50 (95% CI: 10.9–16.5) cases per hundred person-years. Participants assigned male sex at birth had higher hazards of infection (HR: 6.12, 95% CI: 1.53–24.48), while older participants (HR: 0.64, 95% CI: 0.58–0.72) and those living further from the clinic had lower hazards of infection (HR: 0.97, 95% CI: 0.94–1.00). Race, insurance status, neighborhood diversity index, and neighborhood social disadvantage index were not associated with hazard of infection or syphilis-related care utilization.

Conclusions:

Our study found high incidence of syphilis infection among a cohort of AYALH. Integrating comprehensive STI prevention services into HIV care and improving syphilis prevention services in communities with high syphilis rates should be a priority in future intervention work.

INTRODUCTION

The incidence of syphilis, a sexually transmitted infection (STI) that can be detected with routine screening and effectively treated with antibiotics, has been steadily rising over the last two decades.¹ In the United States in 2019, 38,992 cases of primary and secondary syphilis were diagnosed (11.9 cases per 100,000 people). Young adults experiences some of the highest rates of infection.¹ People living with HIV also experience disproportionate rates of syphilis infection, comprising 45% of syphilis cases among MSM in 2017.² Syphilis infection increases the risk of HIV infection among HIV-negative individuals, and the risk of forward transmission in people living with HIV, by damaging the anogenital skin and mucous membranes and increasing circulating antigen presenting cells which may serve as targets for HIV infection.³

While the increasing rates of syphilis among key populations has been well documented,⁴ there is scant data on syphilis incidence among adolescents and young adults living with HIV (AYALH). A study of syphilis rates among adults living with HIV also found an upward trend in syphilis cases between 1999 and 2015, and noted that participants ages 18–30 were the age group with the highest risk of syphilis infection.⁵ Population based data from New York City captured syphilis incidence specifically among adolescents living with HIV in the years 2001–2002,⁶ but there are few other published reports describing the current burden of syphilis in this key population.

Additionally, little research has been conducted exploring what risk factors may be driving increasing syphilis rates among young people. Studies of young people in Brazil found that condom use, age of sexual debut, and number of sexual partners were associated with risk for syphilis infection.^7–9 While behavioral factors such as condom use and limiting the number of sexual partnerships are key elements of syphilis prevention strategies, these behaviors are embedded in a larger socio-ecological context that influences how behaviors translate to vulnerability to infection.¹⁰ Moreover, exploring factors at the neighborhood and community levels may better guide efforts to address the excess burden of syphilis that is shouldered by racial/ethnic minorities and people living with HIV, as research has consistently confirmed that behavioral risk factors are not the primary drivers of disparities in STIs in vulnerable populations.^11,12

Considering the rising rates of syphilis, particularly among youth, people living with HIV, and MSM and transgender women, health systems serving these key populations will need the resources to provide comprehensive syphilis-related care. This package of services requires not only routine screening and prompt treatment for infection, but also clinician education to recognize and presumptively treat cases where suspicion of syphilis is high, partner treatment of known syphilis-positive individuals, and primary and secondary prevention counseling to reduce the risk of initial and recurrent infections.¹³ Therefore, an effective syphilis prevention system also needs to effectively encompass and care for patients’ sexual networks.

Given these needs, the present study aims to (1) describe patterns of syphilis infection and syphilis-related care utilization among AYALH in Philadelphia, and (2) identify individual and neighborhood level factors that are associated with syphilis infection and care utilization in AYALH.

METHODS

Study design and study sample

We conducted a retrospective cohort study of all AYALH receiving care at the Adolescent Initiative at the Children’s Hospital of Philadelphia (CHOP) who received a syphilis test and/or who received benzathine penicillin for syphilis treatment between July 1^st, 2011 December 31^st, 2018, extracting all medical and laboratory patient encounters for the included patients during this time period. The Adolescent Initiative is a Ryan White-funded outpatient clinic focused on providing comprehensive care to adolescents and young adults living with HIV. A participant’s baseline visit was the first visit during the observation period, and observations were censored at their last visit in the observation period. Study procedures were reviewed and approved by the CHOP Institutional Review Board.

Assessment and Measures

Data were extracted from the electronic health record (EHR-EPIC, Verona, WI) and were cleaned and restructured by members of the research team. Syphilis-related care utilization was defined as documented administration of benzathine penicillin in the EHR associated with a rapid plasma reagin (RPR) test for syphilis. Visits in which a participant received syphilis-related care were further categorized as one of the following: confirmed syphilis case, suspected syphilis case, or partner treatment. Confirmed cases were defined as cases in which syphilis treatment was associated with a positive RPR result and a new positive treponemal test and/or a four-fold rise in RPR titer.¹³ Suspected cases were defined as cases where syphilis treatment was associated with symptoms of syphilis infection, but where results from laboratory testing did not confirm syphilis infection. Partner treatment cases were defined as cases where syphilis treatment was administered to the participant after reported recent exposure to syphilis through a sexual partner, but where laboratory testing did not confirm infection. All participant charts underwent manual chart review by the senior author (S.W.) to classify cases according to these definitions. For participants whose case definition could not be clearly determined, a second author (H.K.) conducted an independent review of the chart to determine case definition. Conflicting case determinations were resolved by discussion and consensus between the two authors.

Participant demographic data were also extracted from the medical records. Race and ethnicity were collapsed into two categories (racial/ethnic minorities and non-racial/ethnic minorities) due to lack of variability given that the vast majority of the clinic population identify as Black and/or Latinx. Insurance coverage was measured using the latest known insurance payor on record for the participant, and was labeled as either private insurance, public insurance, or uninsured/self-pay (i.e., Ryan White-funded care).

Census tract-level neighborhood factors were extracted from the American Community Survey (ACS) 5-year estimates for 2017. ¹⁴ A social disadvantage index (SDI) was created from seven items from the ACS which measure neighborhood level social disadvantage (i.e., unemployment rate, percent of households receiving public assistance, percent very low-income persons, percent low-income persons, percent with less than high school education, percent female-headed households with children <18 years, percent renter-occupied houses, and percent changing residences within the last three years; alpha =.84).¹⁵ The measures were each standardized to their national means and standard deviations and then summed, creating a composite z-score where zero indicates being at the national average for all seven measures. Distance to clinic was measured using Esri ArcGIS Network Analyst (Redlands, CA). The total driving distance was measured in miles from subject’s home address to the Adolescent Initiative clinic. A Diversity Index was computed that measured the probability that two individuals chosen at random from a given census tract would belong to different racial/ethnic groups, and could range from 0–100 with lower numbers representing less diversity and high numbers representing greater diversity. For reference, the average Diversity Index in the United States is 65.¹⁶

Statistical Analysis

To characterize the study sample, we described baseline demographic characteristics for study participants including age, race/ethnicity, sex assigned at birth, and insurance status. We also summarized the total observed time on study and the number of clinic visits for each participant and provide a summary of confirmed syphilis cases and syphilis-related care utilization.

We calculated syphilis incidence rates for the overall study cohort, by key baseline demographic characteristics, and by calendar year across the study period. A chi-square test for trends was used to test for a trend in incidence by calendar year. The numerator for these incidence rates included all confirmed syphilis cases (including recurrent cases for participants with multiple confirmed cases) and the denominator was the follow-up time in years.

In order to explore how demographic and neighborhood level factors were associated with incident syphilis and syphilis-related care utilization, we fit two Prentice-Williams-Peterson Gap Time models (PWP-GT).¹⁷ The PWP-GT model applies concurrent Cox regression models which are stratified based on the number of prior events an individual has experienced, where individuals only enter the next strata contingent on having an event in the previous strata. This modeling strategy allows for the occurrence of multiple events within a single individual to be incorporated into a survival analysis. Analyses were performed in R 3.5.1 ¹⁸ using the psych package ¹⁹ and the survival package.²⁰

RESULTS

Our study cohort included 335 unique patients with a total of 6,047 participant encounters. Participant characteristics at their baseline visit and a summary of syphilis diagnoses and treatments for this cohort are presented in Table 1. The cohort was largely non-Latinx Black/African American (83%) and assigned male sex at birth (88%). The median age at first observation in the dataset was 21 (IQR: 18–22), and 26% were 18 years of age or younger at first observation. Most participants resided in Philadelphia county (85.7%), lived in neighborhoods with a low diversity index (median: 34.35, IQR: 19.0– 60.0), and scored higher than the national average for social disadvantage (mean SDI: 6.86, SD: 5.75).

Table 1:

Baseline participant characteristics and summary of confirmed syphilis cases and syphilis-related care utilization

	Overall (n=335)
Age, years (median, IQR)	21 [18–22]
Male sex assigned at birth (n,%)	295 (88)
Race/ethnicity (n,%)
Non-Hispanic Black/African American	279(83)
Hispanic/Latinx	18 (5)
Non-Hispanic White	13 (4)
Multiracial	2 (1)
Asian	1(<1)
Other	12(4)
Refused	10 (3)
Insurance status (n,%)
Public insurance	139 (41.5)
Private insurance	46 (13.7)
Uninsured	135 (40.3)
Missing	15 (4.5)
Diversity Index a (median, IQR)	34.35 (19.0– 60.0)
Distance to clinic, miles (median, IQR)	4.45 (2.91–8.09)
Social disadvantage index b (mean, SD)	6.86 (5.75)
Resides in Philadelphia County (n, %)	287 (85.7)
Number of clinic visits (median, IQR)	10 (5–21)
Follow up time, months c (median, IQR)	19.2 (7.1–40.7)
Syphilis confirmed cases (n, %)
No cases	252 (75.2)
One case	61 (18.2)
Two cases	18 (5.3)
Three cases	4 (1.2)
Syphilis-related care utilization (n, %)
Never treated	247 (73.7)
One treatment	52 (15.5)
Two treatments	24 (7.2)
Three or more treatments	12 (3.6)

^aProbability that two individuals chosen at random from a given census tract would belong to different racial/ethnic groups.

^bComposite z-score of social disadvantage index, where zero indicates being at the national average for all seven measures.

^cFollow-up time was calculated as the difference between the first and last visit in the study period.

Syphilis-related care was provided 145 times in this data set to 88 unique participants. Of these treatment administrations, 109 were for confirmed syphilis cases, 25 were for known exposures to a partner with syphilis that did not later yield a confirmed syphilis diagnosis, and 11 were for suspected syphilis infection based on symptoms that did not later yield a confirmed syphilis diagnosis. Among the confirmed cases, 74 (68%) were asymptomatic, 11 (10%) presented with a rectal, genital, or orofacial ulcerative lesion consistent with primary syphilis, 16 (15%) presented with symptoms consistent with secondary syphilis (n=15 with rash, n=1 with condyloma lata), one (1%) presented with symptoms consistent with neurosyphilis, and seven (6%) presented with generalized symptoms that did not correspond to a stage of syphilis. Among the partner treatment cases, 24 (96%) were asymptomatic and one (4%) had generalized symptoms that did not correspond to a stage of syphilis. Among the suspected syphilis cases, eight (73%) presented with symptoms consistent with primary syphilis and three (27%) presented with a rash consistent with secondary syphilis.

The 109 confirmed syphilis cases were attributed to 83 unique participants. Of those with at least one confirmed case, 22 participants (27%) had multiple episodes of syphilis; among these, 18 (82%) participants had two episodes of syphilis and four participants (18%) had three episodes. The median time between recurrent syphilis cases was 14 months (IQR: 9–20 months).

The overall incidence rate of confirmed syphilis infection was 13.50 (95% CI: 10.9–16.5) cases per hundred person-years (Figure 1). Participants who were assigned male sex at birth had greater incidence rate of syphilis (15.0 per 100 person-years, 95% CI: 12.1–18.4) than participants assigned female sex at birth (2.4 per 100 person-years, 95% CI: 0.3–8.6). Participants who were between 15–18 years of age at their baseline visit had greater incidence rate (15.8 per 100 person-years, 95% CI: 11.3–21.5) than participants aged 19–24 at baseline (12.2 per 100 person-years, 95% CI: 9.2–15.9). The syphilis incidence rate showed an increasing trend over the course of the study (χ²=17.18, df=1, p<0.001), from 2.6 (95% CI: 0.1–14.7) cases per hundred person-years in 2011 to 25.1 (95% CI: 14.3–40.7) cases per hundred person-years in 2018 (Figure 2).

Figure 1:

Syphilis incidence per 100 person-years among full study cohort and by demographic variables.

Figure 2:

Syphilis incidence per 100 person-years among study sample by year.

In multivariate PWP-GT models (Table 2), participants with male sex assigned at birth showed significantly higher hazard of confirmed syphilis infection (HR: 6.12, 95% CI: 1.53–24.48) and of utilizing syphilis-related care (HR: 7.11, 95% CI: 1.93–26.17). Greater age was associated with lower hazard of confirmed syphilis infection and of utilizing syphilis-related care; each year of age was associated with 0.64 times lower hazard of confirmed syphilis (95% CI: 0.58–0.72) and 0.69 times lower hazard of utilizing syphilis-related care (95% CI: 0.62–0.76). Additionally, each addition mile of distance between a participant’s residence and the clinic was associated with 0.97 times lower hazard of confirmed syphilis infection (95% CI: 0.94–1.00) and 0.96 times lower hazard of utilizing syphilis-related care (95% CI: 0.93–0.99). Race, insurance status, neighborhood diversity index, neighborhood social disadvantage index, residence in Philadelphia County, baseline visit year, and number of clinic visits were not found to have significant relationships with the hazard of confirmed syphilis infection or of utilizing syphilis-related care.

Table 2:

Recurrent survival regression models (PWP-GT) of demographic and neighborhood factors on confirmed syphilis cases and syphilis-related care utilization

	Adjusted Hazard Ratio, Confirmed Syphilis Cases (aHR, 95% CI)	Adjusted Hazard Ratio, Syphilis-Related Care Utilization (aHR, 95% CI)
Age, years	0.64 (0.58, 0.72)***	0.69 (0.62, 0.76)***
Male sex assigned at birth	6.12 (1.53, 24.48)*	7.11 (1.93, 26.17)**
Racial/ethnic minority	1.09 (0.40, 2.99)	0.85 (0.34, 2.12)
Insurance
Private insurance	Ref	Ref
Public insurance	0.82 (0.44, 1.54)	0.73 (0.47, 1.47)
Uninsured	0.90 (0.44, 1.85)	0.79 (0.41, 1.51)
Diversity indexa	1.01 (1.00, 1.01)	1.01 (1.00, 1.01)
Distance to clinic, miles	0.97 (0.94, 1.00)*	0.96 (0.93, 0.99)*
Social disadvantage indexb	0.99 (0.95, 1.04)	1.00 (0.96, 1.04)
Resides in Philadelphia County	1.35 (0.46, 3.96)	0.64 (0.27, 1.53)
Year of baseline visit	0.93 (0.78, 1.11)	0.90 (0.78, 1.04)
Number of visits during study period	0.99 (0.98, 1.01)	1.00 (0.99, 1.02)

^aProbability that two individuals chosen at random from a given census tract would belong to different racial/ethnic groups

^bComposite z-score of social disadvantage index, where zero indicates being at the national average for all seven measures.

^*p<0.05

^**p<0.01

^***p<0.001

DISCUSSION

Our study found a high incidence rate of syphilis infection among AYALH in Philadelphia, and a substantial number of participants who experienced recurrent syphilis. Additionally, a trend towards increasing syphilis incidence was seen over the course of the study. The incidence rate of syphilis found in this study population was higher than those reported in similar settings in recent years.^5,21 Our data also revealed 36 cases where penicillin was administered for partner treatment or presumptive treatment based on symptoms where RPR testing did not confirm a syphilis infection. These cases can be seen as clinical indicators of heightened syphilis risk within sexual networks and represent a key opportunity to delivery enhanced STI prevention resources to network members. Considering recent trends of increasing syphilis rates across the country, these results highlight a critical need to direct robust and comprehensive syphilis prevention resources to AYALH. Substance use and mental health disorders have been found to be associated with a range of adverse outcomes among AYALH ^22,23 and contribute to increased risk for syphilis and other STIs. Providing AYALH with comprehensive mental health and substance use treatment resources, in addition to frequent syphilis screening and prompt treatment, could help to address the drivers of syphilis infection among this vulnerable population.

Infectious disease experts have suggested that using doxycycline as either pre-exposure or post-exposure prophylaxis against syphilis could be a promising primary prevention intervention to decrease syphilis incidence among high-risk populations.²⁴ Preliminary studies have found that both daily doxycycline as pre-exposure prophylaxis ²⁵ and event-driven doxycycline as post-exposure prophylaxis ²⁶ resulted in decreased incidence of syphilis. In our study, while a high proportion (27%) of AYALH had repeat syphilis infections, the median time between recurrent syphilis cases was 14 months (IQR: 9–20 months), which may not justify daily doxycycline use for syphilis prevention. Given this time frame, clinicians should consider the risks of prolonged doxycycline exposure including antimicrobial resistance, potential alteration of the microbiome, and common adverse events such as photosensitivity and gastroenterological upset. More data are needed to determine whether event-driven prophylaxis may be a more feasible strategy.

In multivariate survival models, younger participants and those assigned male sex at birth showed greater risk for syphilis infection, a finding which is consistent with previous studies,^5,6 yet few other demographic or neighborhood characteristics were associated with risk for infection. This result was contrary to our original hypothesis that neighborhood level factors would be associated with syphilis risk, given that prior research has found that neighborhood level factors, including measures of poverty and segregation, were associated with risk for STIs.^27–29 However, given that the sample in this study represents a population that faces multiple social and structural barriers to optimal health (e.g. encountering HIV-related stigma, residing in racially segregated and economically under-resourced neighborhoods) and were somewhat homogenous in terms of neighborhood social disadvantage (88% of participants lived in neighborhoods with higher social disadvantage scores that the national average) and in terms of racial residence segregation (79% of participants lived in neighborhoods with lower diversity index than the national average), it is possible that there was simply not enough variation in neighborhood conditions to detect meaningful associations with syphilis risk. These results suggest that similar populations of socially marginalized youth are likely to also be highly affected by syphilis. Interventions should focus on scaling up primary syphilis prevention resources for all young people living with HIV, and should begin with entry into HIV care, given that the youngest members of our cohort were seen to be at greatest risk for acquiring syphilis. Contrary to our hypothesis that distance to clinic would present a structural barrier to STI prevention resources, we found that participants living further from the clinic were at lower risk of acquiring syphilis. This could reflect that participants living far from the clinic were more likely to seek testing and treatment for STIs in other clinical setting. Additionally, it is possible that those living far from the clinic were members of sexual networks with lower syphilis prevalence, given that Philadelphia has experienced higher rates of syphilis than the surrounding areas in recent years.¹ Additionally, we were surprised to find that insurance status was unrelated to syphilis risk, as previous studies of people living with HIV have found that patients without health insurance had an increased risk of syphilis infection.⁵ Considering that the cost of care for all patients without insurance at this clinic was covered by Ryan White funding, this critical source of safety net funding may have dampened the impact of lacking insurance on risk for syphilis.³⁰

Limitations of the study include generalizability of the results, as our data came from a single clinic in a large urban center, and the findings here may not be applicable to other clinical settings or to rural areas. Additionally, our sample was homogenous in terms of race and sex, with most participants being assigned male at birth and identifying as Black or African American. This is partially a reflection of the epidemiology of the HIV epidemic among youth in the U.S., which disproportionately affects racial/ethnic minority men. Future research should include a more diverse cohort of young people both to improve generalizability and to increase the power to capture effects of racial segregation and other structural drivers of health disparities. We were also unable to measure the gender identity of participants and only captured the sex that participants were assigned at birth, which limits our ability to study gender-specific health outcomes in this study, especially for transgender youth. Additionally, our study may underestimate the number of recurrent syphilis cases experienced by the study cohort. Some participants included in this cohort may have been treated for syphilis prior to the study period either in clinic or in the community, and thus were not actually syphilis naïve, although we were unable to ascertain what proportion of the sample met this condition. Therefore, to avoid introducing misclassification bias, we labeled all participants with a baseline case index of zero at the beginning of our observation period.

CONCLUSION

Our study found high incidence of syphilis among a sample of AYALH. The data highlight the importance of increasing syphilis prevention resources available to this population. Integrating comprehensive STI prevention services into HIV care and improving syphilis prevention services in communities with high incidence and recurrence rates should be a priority in future intervention work.