Schumacher, Christina M. MHS*; Ellen, Jonathan MD†; Rompalo, Anne M. MD, ScM‡
AFTER FALLING PRECIPITOUSLY THROUGHOUT the 1990s and reaching an historic low in 2000, syphilis incidence in the United States began to rise. Although the epidemics in the late 1980s occurred among poor inner city African Americans and were attributed mostly to increased cocaine use,1 the overall increase in syphilis incidence since 2000 has been attributed to increases among men having sex with men (MSM).2
These recent epidemics represent disease re-emergence with reversal of decreasing incidence—the epidemic has shifted from a low-level endemic or decline phase to an increasing incidence or growth phase. In some cities, the disease has progressed to a hyperendemic phase of stable but high incidence.3 In theory, the transition to growth and hyperendemic phases is related to changes in population dynamics such that transmission becomes less dependent on a small, high-risk sexual networks and more dependent on lower-risk sexual networks.4 Because syphilis is generally maintained among relatively small, high-risk sexual networks, new epidemics occur either when the disease is introduced into a different high-risk sexual network or when general behavior changes allow mixing between high-risk and low-risk sexual networks.
Given the interplay between epidemic patterns and changes within sexual networks, certain interventions may be appropriate only during certain epidemic phases. For example, screening programs and mass media campaigns would be appropriate during a hyperendemic phase, when the sexual transmission networks are diffuse and the pathogen is spreading among relatively large low-risk populations. In contrast, interventions during growth, decline or endemic phases should be focused on reducing the disease among high-risk groups through targeted screening programs and community outreach strategies.3,5–7 However, little empiric data are available that identify epidemic phases and illustrate the relationship between epidemic phases and population characteristics.
In the mid 1990s, Baltimore, MD, experienced a large syphilis epidemic, which peaked in 1997.8 Like the epidemics in other US cities that occurred in the late 1980s, the large epidemic was mostly heterosexual, among African Americans and attributed to increases in crack cocaine use, sex for drugs, and prostitution.9 Syphilis incidence in Baltimore had declined significantly between 1997 and 2002, when surveillance activities indicated that syphilis was increasing among MSM and young women. (Baltimore City Health Department surveillance data, unpublished 2003).
Using data from Baltimore, we sought to determine whether shifts in syphilis epidemic phases are associated with changes between or within high-risk populations. We hypothesized that demographics and reported risk factors (epidemiologic profile) changed as the infection progressed through epidemic phases, and that the epidemiologic profiles of syphilis cases in each phase are different from one another.
Materials and Methods
The Baltimore City Health Department (BCHD) maintains an electronic registry of all syphilis cases, and initiates follow-up interviews and case management services upon receipt of a laboratory report of positive syphilis serology, treatment for syphilis reported by a health care provider, or a person named as a sexual contact of an early syphilis (primary, secondary, or early latent) case. Case management is assigned to disease intervention specialists who conduct field investigations and attempt to interview each case and his or her identified sexual partners. Information routinely collected during the interview includes demographics, sexual risk history, substance abuse, and the names and contact information of potentially exposed partners. Standard Centers for Disease Control and Prevention forms are used to record the information from interviews; all records are then entered and maintained in the Centers for Disease Control and Prevention Sexually Transmitted Disease*Management Information System Registry.
Demographic and behavioral risk information for all early syphilis cases reported to the BCHD between January 1994 and December 2004 were abstracted from Sexually Transmitted Disease* Management Information System Registry. Because risk factors for syphilis often differ by gender, we stratified all analyses by gender and excluded all cases with missing gender data. We also excluded those cases with missing data on report and diagnosis dates because this information was needed to assign epidemic phases.
Epidemic phases were defined through visual inspection of the infection date-based epidemic curve in order to eliminate lag-time bias when inspecting trends over time. Infection dates were estimated by subtracting the stage specific median incubation period from the diagnosis date or from the report date if the diagnosis date was missing.10 Because the Centers for Disease Control and Prevention Program Operations recommend quarterly analysis of disease trends,11 the epidemic curve was constructed by parceling infections by quarter. The epidemic curve and the designation of epidemic phases are shown in Figure 1. We defined epidemic phases using the framework described by Wasserheit and Aral.3 Because the information before 1994 was unavailable, we could not accurately assess whether infections in 1994 and early 1995 represented an increase in incidence or the endemic level of infection, and we excluded cases that occurred before the second quarter of 1995. We defined the period of high incidence between the second quarter of 1995 and the second quarter of 1997 (Q2 1995 to Q2 1997) as a hyperendemic phase, and the period of declining and stable incidence from Q3 1997 to Q4 2001 as a decline phase. As infections increased slightly between Q1 2002 and Q4 2004, we considered this a growth phase because Baltimore City is considered a high-morbidity area, and any increases in incidence may trigger an outbreak investigation.
All demographic and behavioral risk data were investigated as predictors of being infected during a particular epidemic phase. We dichotomized all variables. In Baltimore, only a small proportion of syphilis cases (2.3%) are not classified as African American or Caucasian, so we collapsed racial categories to African American and non-African American. Reporting providers were categorized as being under the jurisdiction of the BCHD (BCHD clinics) and others. For contact tracing purposes, the number of sexual partnerships is assessed over the maximum limit of the stage-specific incubation period (90 days for primary, 6 months for secondary, 1 year for early latent).12 For behavioral variables, disease intervention specialists document only affirmative interviewee answers; therefore, the reference point for these variables is No/Refused rather than No. BCHD expanded the risk portion of the standard interview in 1999, and in order to compare changes in risk factors across epidemic phases, we restricted our investigation to factors that were assessed throughout the entire analysis period. For this reason, men are considered to be MSM if they reported having sex with a male partner or if they self-identified as MSM; drug use was restricted to intravenous drug use only.
Statistical analyses included simple logistic regression. In the logistic regression models, we used the epidemic phase as the outcome and the decline phase as the reference group. Because we performed multiple comparisons, we used a Bonferroni correction when assessing statistical significance. Because there are 3 outcomes and 3 possible comparisons, we considered results statistically significant for P values = 0.016 (0.05/3). All analyses were conducted using Stata Version 9.0. (College Station, TX).13
Surveillance data indicate that since 2002 (growth phase), increases in syphilis incidence have been among MSM and young (≤25 years) noninjecting drug using women. We repeated our analysis in each of these subgroups to investigate whether demographics or behaviors changed within these high-risk groups in relation to epidemic phase shifts.
Between January 1994 and December 2004, the BCHD received 8412 reports of early syphilis. Of these, 16 (0.2%) were excluded due to missing data on gender. For 160 cases (2%), we used the report date as a proxy for the diagnosis date to calculate infection dates for the epidemic curve; no cases were missing data on both diagnosis and report dates. A total of 1005 (12%) cases were excluded for occurring before the second quarter of 1995. For the final analysis, 3354 infections occurred during the hyperendemic phase (1788 men and 1566 women), 3145 during the decline phase (1581 men and 1554 women), and 892 during the growth phase (566 men and 326 women). See Figure 1.
Changes in Profiles Among All Men and Women
Demographic and reported risk behaviors across epidemic phases are shown in Tables 1 and 2. In general, the demographics of both men and women syphilis cases were similar in both the hyperendemic phase and the decline phase. Compared to the decline phase, both men and women infected with syphilis in the growth phase were more likely to be of races other than African American [men: odds ratio (OR) 6.21, P <0.001; women: 7.76, P <0.001], and were more likely to be of Hispanic ethnicity (men: OR: 7.03, P <0.001; women: 6.14, P <0.001). In addition, both men and women infected with syphilis in the growth phase were 40% less likely to be older than 25.
Changes in reported risk behaviors over time differed between men and women. The most dramatic change in reported risk factors among men was the increase in those who reported having male sex partners; men were about 5 times more likely to report sex with other men in the growth phase than during the decline phase (P <0.001). The number of reported sex partners and reporting injection drug use were similar across all 3 phases. Men infected in both the hyperendemic phase and the growth phase were less likely to report engaging in commercial sex than in the decline phase.
Reported risk behaviors among women were similar in the hyperendemic and decline phases; however, women infected during the growth phase were almost twice as likely to report injection drug use (OR 1.85, P <0.001) and nearly 3 times more likely to report more than 5 sex partners (OR 2.72, P <0.001) when compared to the decline phase. Although women infected during the growth phase were more likely to report commercial sex work than in the decline phase, this result was not statistically significant.
Changes in Profiles Among MSM and Noninjection Drug-Using Young Women
Four hundred fifty-three MSM were infected between April 1995 and December 2004: 131 during the hyperendemic phase 1, 135 during the decline phase, and 187 during the growth phase. Demographic and behavioral risk profiles of MSM were similar between the hyperendemic phase and the decline phase. (Table 3). However, in the growth phase, MSM were 3.5 times more likely to be of races other than African American (P <0.01); 70% less likely to be over 25 years of age (P <0.001); 75% less likely to report injection drug use (P <0.001); and 76% less likely to report engaging in commercial sex (P <0.01). When compared to the hyperendemic phase, MSM in the growth phase were 6 times more likely to be to be non-African American (P <0.001), 66% less likely to be over 25 years old (P <0.01), 74% less likely to report injection drug use (P <0.01), and 64% less likely to report engaging in commercial sex (P <0.01).
Among the 955 women under 26 years old with early syphilis not reporting injection drug use, (422 in the hyperendemic phase, 425 in the decline phase, 108 in the growth phase) those diagnosed in the hyperendemic phase and the decline phase were similar with regard to demographic and risk profiles (Table 4); however those in the growth phase were 6 times more likely be of races other than African American (P <0.001), 4.3 times more likely to report more than 5 sex partners (P <0.01) and about twice as likely to report commercial sex work (P = 0.04) than those diagnosed during the decline phase. When compared to the hyperendemic phase, young women infected with syphilis in the growth phase were 8 times more likely to be non-African American (P <0.001) and about 3 times more likely to report more than 5 sex partners (P <0.01). Other demographics and behaviors were similar between the 2 epidemic phases.
The epidemiologic profile of persons infected with syphilis during the hyperendemic and decline phases were similar; however, the profile of those infected during the growth phase differed from those of both the hyperendemic and decline phases. When compared to the decline phase, persons infected during the growth phase were more likely to be MSM and women under 26 years of age. MSM infected during the growth phase were more likely to be younger and less likely to be African American, to report injection drug use, and to report engaging in commercial sex than MSM infected during the hyperendemic or decline phases. Young women were more likely to be non-African American and more likely to report commercial sex work in the growth phase than in the other 2 phases.
Inspection of the overall incidence of early syphilis (Fig. 1) shows a marked decline after 1997 until 2002, and our epidemiologic profile analysis shows that during the hyperendmic and decline phases, transmission was mostly heterosexual and among African American adults. After 2002, although the overall disease rates increased only slightly, our results indicate that the relative proportion of cases increased among 2 different subpopulations: MSM and young women, a potential inflection point in the epidemic curve. Continued transmission of syphilis within these subpopulations sparked a new epidemic in Baltimore,14 despite increased surveillance and control efforts that were implemented by the BCHD in the late 1990s.
The re-emergence of syphilis among MSM in the United States and Europe that began in the late 1990s has been well documented.15,16 Reported risk factors for syphilis among MSM have included concurrent human immunodeficiency virus infection and use of Viagra and methamphetamines, as well as high numbers of concurrent sexual partnerships, anonymous sex partners, high rates of unprotected sex, and meeting sex partners on the Internet.17–19 Our data indicate that in Baltimore, the prevalence of behavioral risk among MSM has decreased or stayed the same over time, whereas the rate of syphilis among MSM has increased. In particular, the number of reported sexual partners has not changed over time, and only a small percentage of men are reporting more than 5 sex partners in all 3 epidemic phases. We were unable to assess any changes in drug use patterns other than injection drug use over time because the BCHD did not begin collecting this data until 2002. However, very few MSM in the growth phase reported using methamphetamines or Viagra.
Along with MSM, young, noninjecting drug-using women who report more than 5 sex partners or commercial sex work represent 2 documented core groups for syphilis transmission. It is possible that the increased likelihood of infection among these subpopulations during the growth phase is because of the marked decline of syphilis among African Americans throughout the late 1990s. However, when we inspected incidence of syphilis from 1995 to 2004 stratified by race and gender, the incidence of both non-African American men and women began to increase after 2002, whereas rates of African American men remained steady and rates of African American women continued to decrease (data not shown). These results indicate that the transition from the decline to growth phases was associated with a shift in the demographic groups most affected by syphilis rather than changes in risk behaviors in the affected subpopulations over time.
Another explanation for the increase in MSM and young women infected with syphilis in the growth period is that the MSM and young, women commercial sex worker populations have increased over time in Baltimore and that our results are an artifact of changes in the composition of Baltimore's population. However, reliable estimates on MSM and commercial sex worker populations are not available, and we are unable to assess whether the changes in the incidence of syphilis within these groups are due to increases in these subpopulations.
Although the risk profiles of syphilis cases have changed over time, it is important to note that these risk profiles are based on self-reported behaviors. During the interview process, cases may refuse to answer any question, and the interview sheet is marked only if the response is “yes.” Because of the data collection process, there is no way of ascertaining whether a case responded “no” to a particular question or whether the case refused to answer; therefore, the absence of reported behaviors in this analysis does not necessarily correlate to the absence of the behavior itself. However, it is likely that the rate of refusing to answer particular questions is nondifferential across the epidemic phases, and any biases in our results would tend toward the null. Another limitation is the difficulty in finding cases that exhibit the riskiest behaviors. It is also unlikely that the health department's ability to locate and interview these cases changed across the epidemic phases so any bias would lead towards an underestimation of the true change in risk behaviors between epidemic phases.
Additionally, to our knowledge, no standard guidelines for determining when epidemic phase shifts occur exist. Blanchard describes a general classification scheme for identifying epidemic phases: the epidemiologic perspective of analyzing incidence/prevalence or the population perspective of analyzing geographic concentration, network structures, or transmission chains.4 We defined epidemic phases based on visual inspection of the infection-date based epidemic curve using the definitions described by Wasserheit and Aral3 and our assessment of the time trends in syphilis incidence in Baltimore. It is possible that some cases may have been misclassified about epidemic phase. Any misclassification would be nondifferential, and our results would underestimate the true association.
As noted previously, we were unable to assess how syphilis transmission and drug use other than injection drug use changed over time in this analysis because of the data collection process used by the BCHD before 2002, when a more comprehensive risk factor interview was incorporated.
The broad shifts in demographic and behavioral risk factors in this analysis related to changes in the epidemic phase are consistent with other findings. Broad behavioral changes among MSM in Britain preceded a shift in from an endemic phase to a growth phase,20 and changes in the sexual network structures were found to be indicators of epidemic phases for chlamydia in Colorado Springs.21 Among syphilis networks in Baltimore, changes in the network structure were found between the latter half of hyperendemic phase and the decline phase (through 2000); the sexual network during the hyperendemic phase was more central, had more components, and was less dense than the network in the decline phase.22 Additionally, throughout the hyperendemic phase, geospatial clustering of syphilis cases became more diffuse and then receded to a small areas of the city during the decline phase.23 These results are consistent with our hypothesis that syphilis receded from the general African American population to smaller, high-risk groups and then spread among MSM and young, non-African American women.
In the era of syphilis elimination, any inflection point in an epidemic should be investigated to determine whether the trend represents a shift in the subpopulation most affected by syphilis or a reversal of declining incidence within the same subpopulation. Ascertaining why the disease is transitioning from a decline or endemic phase to a growth phase would allow health departments to properly focus intervention strategies, which in turn could prevent future epidemic transmission. We have retrospectively shown that in Baltimore, the slight increase in incident syphilis between 2002 and 2004 can be attributed to increases in transmission among non-African American MSM and young, non-IDU women involved in commercial sex work. The BCHD has been conducting syphilis control activities among MSM and young women since 2003, when surveillance activities detected an increase of syphilis among these groups. However, our phase-based analyses indicate that the shift in syphilis epidemiology began in 2002, underscoring the utility of such an analysis in real time, so that interventions can be properly focused at the beginning of a potential growth phase.
1. Rolfs RT, Nakashima AK. Epidemiology of primary and secondary syphilis in the United States, 1981 through 1989. JAMA 1990; 264:1432–1437.
2. Centers for Disease Control and Prevention. Sexually Transmitted Disease Surveillance, 2005. Atlanta, GA: US Department of Health and Human Services, 2006.
3. Wasserheit JN, Aral SO. The dynamic topology of sexually transmitted disease epidemics: Implications for prevention strategies. J Infect Dis 1996; 174 (suppl 2):S201–S213.
4. Blanchard JF. Populations, pathogens, and epidemic phases: Closing the gap between theory and practice in the prevention of sexually transmitted diseases. Sex Transm Infect 2002; 78 (suppl 1):i183–i188.
5. Low N. Phase specific strategies for the prevention, control, and elimination of sexually transmitted infections: Case study in Lambeth, Southwark, and Lewisham, London, UK. Sex Transm Infect 2002; 78 (suppl 1):i133–i138.
6. Hodgins S, Peeling RW, Dery S, et al. The value of mass screening for chlamydia control in high prevalence communities. Sex Transm Infect 2002; 78 (suppl 1):i64–i68.
7. Patrick DM, Rekart ML, Jolly A, et al. Heterosexual outbreak of infectious syphilis: Epidemiological and ethnographic analysis and implications for control. Sex Transm Infect 2002; 78 (suppl 1):i164–i169.
8. Division of STD Prevention. Sexually Transmitted Disease Surveillance, 1998. Atlanta, GA: Department of Health and Human Services, Centers for Disease Control and Prevention (CDC), 1999.
9. Outbreak of primary and secondary syphilis—Baltimore City, Maryland, 1995. MMWR Morb Mortal Wkly Rep 1996; 45:166–169.
10. Schumacher CM, Bernstein KT, Zenilman JM, et al. Reassessing a large-scale syphilis epidemic using an estimated infection date. Sex Transm Dis 2005; 32:659–664.
12. Sexually transmitted diseases treatment guidelines 2002. Centers for Disease Control and Prevention. MMWR Recomm Rep 2002; 51:1–78.
13. Stata Statistical Software: Release 9 [computer program]. College Station, TX: StataCorp LP, 2005.
14. Division of Sexually Transmitted Disease. Primary and Secondary Syphilis Case Rates, Maryland Counties and Baltimore City, 1997–2006. Baltimore, MD: Maryland Department of Health and Mental Hygiene, 2007.
15. Peterman TA, Heffelfinger JD, Swint EB, et al. The changing epidemiology of syphilis. Sex Transm Dis 2005; 32:S4–S10.
16. Fenton KA, Imrie J. Increasing rates of sexually transmitted diseases in homosexual men in Western Europe and the United States: Why? Infect Dis Clin North Am 2005; 19:311–331.
17. Resurgent bacterial sexually transmitted disease among men who have sex with men—King County, Washington, 1997–1999. MMWR Morb Mortal Wkly Rep 1999; 48:773–777.
18. Wong W, Chaw JK, Kent CK, et al. Risk factors for early syphilis among gay and bisexual men seen in an STD clinic: San Francisco, 2002–2003. Sex Transm Dis 2005; 32:458–463.
19. Taylor M, Aynalem G, Smith L, et al. Correlates of internet use to meet sex partners among men who have sex with men diagnosed with early syphilis in Los Angeles County. Sex Transm Dis 2004; 31:552–556.
20. Simms I, Fenton KA, Ashton M, et al. The re-emergence of syphilis in the United Kingdom: The new epidemic phases. Sex Transm Dis 2005; 32:220–226.
21. Potterat JJ, Muth SQ, Rothenberg RB, et al. Sexual network structure as an indicator of epidemic phase. Sex Transm Infect 2002; 78 (suppl 1):i152–i158.
22. Cunningham SD, Michaud JM, Johnson SM, et al. Phase-specific network differences associated with the syphilis epidemic in Baltimore city, 1996–2000. Sex Transm Dis 2004; 31:611–615.
23. Gesink L, Bernstein KT, Serre ML, et al. Modeling a syphilis outbreak through space and time using the Bayesian maximum entropy approach. Ann Epidemiol 2006; 16:797–804.