Article Text

PDF

Variations in the epidemiology of primary, secondary and early latent syphilis, England and Wales: 1999 to 2008
  1. Heather Jebbari1,
  2. Ian Simms1,
  3. Stefano Conti2,
  4. Andrea Marongiu1,
  5. Gwenda Hughes1,
  6. Helen Ward1,
  7. Cassandra Powers1,
  8. Daniel Rh Thomas3,
  9. Barry Evans1
  1. 1HIV and STI Department, Health Protection Services, Colindale, Health Protection Agency, London, UK
  2. 2Statistics, Modelling & Bioinformatics Unit, Health Protection Agency Centre for Infections, London, UK
  3. 3NPHS Communicable Disease Surveillance Centre, Temple of Peace and Health, Cardiff, UK
  1. Correspondence to Dr Ian Simms, Epidemiologist, HIV and STI Department, Health Protection Services, Colindale, Health Protection Agency, 61 Colindale Avenue, London NW9 5EQ, UK; ian.simms{at}hpa.org.uk

Abstract

Objective To investigate factors associated with variations in diagnoses of primary, secondary and early latent syphilis in England and Wales.

Methods Data were derived from two sources: diagnoses made in genitourinary medicine clinics reported on form KC60, and information collected through National Enhanced Syphilis Surveillance (NESS). Multinomial regression modelling was used for data analysis.

Results Between 1999 and 2008, 12 021 NESS reports were received, 54% of KC60 reports. The dominant profile of the epidemic was one of white men who have sex with men aged 35–44, often co-infected with HIV, centred in larger cities. During this period, the proportion of primary cases increased over time, while the proportion of secondary cases fell. Primary cases exceeded secondary cases by 2004. The proportion of early latent cases remained relatively stable over time and tended to be lower than that of primary and secondary infection. Patients who attended because they had symptoms of infection, had been identified through partner notification, were HIV positive, and were UK born were more likely to present with primary or secondary infection than with early latent infection. A higher proportion of early latent cases were seen among patients who were Asian, had contacted sexual partners through saunas, bars and the internet, had untraceable partners, and had acquired infection in Manchester.

Conclusions The continuing syphilis epidemic indicates that control has only been partially effective, with ongoing transmission being sustained. Intensive and targeted efforts delivered locally are required to interrupt further transmission.

  • Infectious syphilis
  • epidemiology
  • stage of infection
  • England & Wales
  • syphilis

Statistics from Altmetric.com

Introduction

In England and Wales the resurgence of infectious syphilis started in 1997 with an outbreak among heterosexual men and women in Bristol. Since then, diagnoses made at genitourinary medicine (GUM) clinics have risen from 285 in 1997 to 3265 cases in 2008.1 2 This dramatic change in incidence has been focused on major urban areas, particularly London and Manchester, but outbreaks have been seen throughout England and Wales, and there are striking similarities within outbreaks seen in Europe, North America and Australia.2 The epidemic has been influenced by developments in the HIV epidemic and behavioural change in men who have sex with men (MSM) and has been characterised by a rapid increase in diagnoses made in MSM, and a high proportion of HIV co-infections. An outbreak among heterosexuals, including young people, has developed alongside the larger MSM epidemic.

The natural history of syphilis suggests that an early epidemic dominated by primary and secondary infection would be expected to develop into an epidemic predominantly composed of early latent infections.3 4 Each phase presents a different public health challenge. Here we investigate the factors that are associated with variations in the stage of diagnosed infections over the past 10 years and consider the implications of these findings in terms of public health intervention.

Methods

Data sources

Two data sources were used: diagnoses made in GUM clinics reported on form KC60, and information collected through National Enhanced Syphilis Surveillance (NESS) which was also undertaken only within GUM clinics. Both methodologies have been previously described and discussed in detail.5 In brief the KC60 was an aggregate mandatory return, the number of cases being recorded by gender and male sexual orientation for primary, secondary and early latent syphilis.i Age group was also recorded but only for primary and secondary infection. NESS is a voluntary reporting system based on the collection of disaggregate (patient level) data including: gender, age, ethnic background, sexual orientation, stage of infection, HIV status, geographic area where the infection was likely to have been acquired, and connections with social and sexual networks.

Definitions

Definitions of early latent syphilis vary between countries. The definitions used for primary, secondary and early latent infection within the UK are given in the guidelines published by the British Association for Sexual Health and HIV (BASHH; http://www.bashh.org). The BASHH guidelines define early latent syphilis as ‘Treponema pallidum infection diagnosed on serological testing with no symptoms or signs within the first 2 years of infection: this is early latent syphilis and beyond that late latent syphilis’. In the KC60 return, primary and secondary infections were recorded under a single combined category (A1, A2) and early latent (A3) was recorded separately. In contrast, under the NESS collection system, primary (A1), secondary (A2) and early latent (A3) are recorded separately.

Data completeness and representativeness

The NESS dataset is a subset of the KC60 dataset. KC60 data are shown as a reference point for the NESS data, although only the patient level NESS dataset was used in the subsequent detailed analysis. Completeness of reporting varies over time and by Strategic Health Authority (SHA): reporting from the North West and North East SHAs was close to 100%. NESS began in 1999 in the North West, while Wales and most other English regions followed from 2003, the East of England and East Midlands regions submitting data for the first time in 2007. Coding accuracy varies between the two datasets.6 For example, the patient management software used in some clinics automatically defaults male sexuality to ‘heterosexual’, leading to the incorrect assignment of some patients to this category. Regional variation in diagnoses reported through NESS was the area of most concern in terms of selection bias. To investigate this in more detail, the SHA of the diagnosing GUM clinic was included in the statistical model.

Statistical analysis

Statistical analysis was only undertaken on NESS data. The relationship between stage of infection and selected explanatory variables was investigated by using multinomial (also known as multi-categorical logistic) regression modelling (details given in online appendix).7 The probability of presenting with syphilis at each stage of infection was estimated by year and for different patient characteristics from the model using maximum likelihood. Predicted probabilities were plotted to investigate infection dynamics over time. These figures are shown in the main text and the online appendix.

Results

Between 1999 and 2008, annual cases of infectious syphilis reported through the KC60 system rose from 415 to 3265, an increase in incidence from 0.79 per 100 000 (denominator=total population) to 5.99/100 000. A cumulative total of 22 156 diagnoses were made over the period. For the same period, NESS reports were received for 12 021 cases, 54% of the KC60 total. The NESS and KC60 datasets were compared in terms of the outcome variable—stage of infection—and an overview of the proportions recorded in each dataset indicates that there was a good general level of qualitative agreement (figure 1). This suggests that, overall, NESS gave a consistent representation of the distribution of diagnoses over time. The analysis was undertaken on those patients for whom complete information was available. After exclusion of cases with missing values, 11 838 were used for the statistical analysis (table 1). The majority (73%, 8656/11 838) of cases were seen in MSM compared with 16% (1938/11 838) in heterosexual men and 10% in heterosexual women (1244/11 838). Most cases were seen in patients of white ethnic background (9491/11 594 or 82%), whereas black and Asian groups accounted for 10% (1126/11 594) and 6% (656/11 594), respectively. Primary syphilis was reported in 43% (4698/10 997) of cases, secondary in 33% (3677/10 997), and early latent in 24% (2622/10 997). Of the 233 patients identified through antenatal screening, 97% were women and 141 (61%) were diagnosed with early latent syphilis. MSM reported a median of two (IQR one to five) sexual partners in the 3 months before diagnosis, whereas the median number reported by heterosexual men and women was one (IQR one to two). Just over a third of syphilis patients (3087/7870 or 39%) thought that they had been infected through oral sex, 92% of whom (2847/3087) were MSM.

Figure 1

Comparison of proportion of primary, secondary and early latent infection reported to the KC60 and National Enhanced Syphilis Surveillance (NESS): 1999 to 2008.

Table 1

Absolute (relative) frequencies of syphilis diagnoses reported through National Enhanced Syphilis Surveillance (NESS) by gender and male sexual orientation: 1999 to 2008

Results of the multinomial regression model are summarised in tables 2 and 3. Each estimated regression coefficient provides a measure of increased (if positive) or decreased (if negative) likelihood of presenting with either primary or secondary syphilis, relative to early latent. For example, an HIV-positive individual would be expected to have a 1.62=exp{0.48} higher OR of presenting with primary syphilis infection (as opposed to early latent) compared with an HIV-negative patient.

Table 2

Comparison of primary with early latent infection: estimates and CIs for multinomial regression modelling*

Table 3

Comparison of secondary with early latent infection: estimates and CIs for multinomial regression modelling*

Patients who attended because they had symptoms of infection were significantly more likely to present with primary infection than with early latent infection (p<0.001) (table 2). Patients brought in through partner notification (p=0.051), who were HIV positive (p=0.052) and who were UK born (p=0.064) were also more likely to present with primary infection, although this was of borderline significance. In contrast, patients who were Asian (p=0.039), or those who had contacted sexual partners through saunas (p=0.001), bars (p<0.01) and the internet (p=0.01), had untraceable partners (p=0.029) and had acquired infection in Manchester (p=0.043) were significantly less likely to present with primary infection than with early latent infection.

The pattern was similar when secondary and early latent infection was compared (table 3). Patients who attended because they had symptoms of infection (p<0.001), had been identified through partner notification (p=0.022), were HIV positive (p<0.001) and were UK born (p=0.045) were significantly more likely to present with secondary infection than with early latent infection (table 3). Patients who had contacted sexual partners through saunas (p=0.005), bars (p=0.002) and the internet (p=0.005) were significantly less likely to present with secondary infection than with early latent infection.

There was no significant association between stage of infection at presentation and the health region reporting the data, suggesting that inconsistent regional reporting is unlikely to have influenced the main findings (tables 2 and 3).

In general, the proportion of diagnosed cases of primary infection increased over time, whereas the proportion of secondary cases fell (figure 2 and figures A1–A5 online). For all the variables studied, the proportion of primary cases had exceeded that of secondary cases by 2004. The proportion of early latent cases remained relatively stable over time and was lower than that of primary and secondary infection. MSM were more likely to present with primary infection, whereas heterosexual men were more likely to present with secondary infection (figure 2). Women were most likely to present with primary infection, but a relatively high and increasing proportion presented with early latent infection (figure 2). The relative proportion of the stages of infection varied substantially depending on the reasons for attending clinical care, a higher proportion of early latent cases being seen in those attending for routine clinical care (figure A4). The proportion of primary cases seen in patients who acquired their infection in Brighton exceeded that of secondary cases at least 3 years before this was observed for those infected in London, Manchester and other parts of the UK (figure A5).

Figure 2

Probability of infection with syphilis by stage of infection, gender and male sexual orientation (model estimates): 1999 to 2008.

Discussion

Over the last 10 years the epidemic of infectious syphilis has remained ‘young’, with a high proportion of primary and secondary cases. The decline in diagnoses of secondary infection and the increase in diagnoses of primary infection indicate that infection is being detected and managed at an earlier stage of infection through increased disease awareness, better case ascertainment, and increased access to GUM services. However, in the early 2000s, waiting times to access GUM services increased substantially, caused by large increases in attendance at already overstretched clinical services, and this is likely to have contributed to an increase in the duration of infectiousness.2 The problem was recognised by the Department of Health (England), and the proportion of people contacting GUM services seen within 48 h now approaches 100%.

The consistent level of early latent cases, although low, indicates that primary and secondary infections are going undiagnosed, but it is not possible to predict the number of undetected cases present in the population. Undiagnosed cases have a number of public health implications. Some infections will be resolved through indirect antimicrobial treatment, but, of those that progress to tertiary syphilis, around a third are likely to develop clinical manifestations of late syphilis, which may be seen in clinical settings such as neurology and cardiology over future decades. The high proportion of syphilis diagnoses in MSM who are co-infected with HIV reflects the close relationship between the epidemics. Syphilis infection is known to facilitate HIV transmission and consequently may be contributing to increased HIV incidence. This highlights recommendations that MSM should have an annual sexual health screen, including testing for HIV where not already diagnosed, supported by improved laboratory turnaround times for the diagnosis of syphilis.8 9 MSM need to be aware that syphilis can be transmitted through oral sex. There is also an increased risk of congenital syphilis in women of reproductive age. Congenital syphilis can be prevented through first-trimester screening supported by treatment and partner notification. This control method is cost-effective but highly dependent on well-structured healthcare pathways. In 2007, 95% of pregnant women were screened for syphilis in England, although uptake varied from 82% to 98% between regions.10 This level of screening is slightly higher than that for HIV screening (93% nationally, range 85–97%) despite there being no national target for syphilis screening in pregnancy. Over the period studied, this important public health intervention identified 22% of infections seen among women. The cases of congenital syphilis that have emerged reflect failures of both the antenatal screening and adult syphilis intervention strategies.11 12

The dominant profile of the epidemic is one of white MSM aged 25–34, many of whom are co-infected with HIV, and over a third believed that they had acquired infection through oral sex. This risk profile has been observed consistently since the re-emergence of infectious syphilis over a decade ago.1 The associations between primary and secondary infection and social networks within saunas, bars and the internet reflect risk taking among MSM accessing both traditional ‘sexual marketplaces’ and internet chat rooms. The easy acquisition of sexual partners via the internet has joined previously isolated networks and reduced the time taken for epidemics to evolve.

Infections among heterosexuals have also increased, with the East Midlands region seeing a high proportion (54%) of heterosexual cases. The profile of these cases is more diverse than those seen in MSM. In the early stages of the epidemic, heterosexual outbreaks were generally isolated and linked, for example, to travel abroad.1 More recently, heterosexual infections have been acquired through local sexual networks within the UK, sex work and, in some cases, among students and young people aged less than 20.13 14 In England and Wales, GUM clinics offer free, open access services that are widely advertised, but some of those at risk of infection find accessing services difficult. For example, outbreak investigations showed that some young people diagnosed with syphilis were marginalised in society in terms of socioeconomic circumstances.15 They were not registered with health services and did not attend services when they experienced issues with their health. Increased risk of infection among patients of Asian ethnicity may reflect infection acquired abroad, but it may also be caused by barriers to accessing sexual health services that deter patients from seeking clinical advice.16 These observations highlight the importance that all young people should be able to access comprehensive sexual health services for testing, treatment and management.

The main potential limitation of the study is the use of voluntary NESS dataset. The NESS dataset includes around 60% of the diagnoses recorded in the KC60 dataset and consequently it may not be representative of the overall pattern of diagnoses. High regional variation is a feature of the epidemic, but analysis of the NESS dataset showed that the presence of region as a factor in the analysis did not have a significant effect on the model, indicating that the enhanced dataset was not geographically biased.

Over the past decade, diagnoses of infectious syphilis have increased in Western Europe and the USA, but few countries collect detailed information by stage of infection. In the USA in 2008, primary and secondary infection accounted for 52% of diagnoses of early syphilis, whereas early latent infection accounted for 48%.17 In contrast, 24% of diagnoses of early syphilis made in England and Wales were early latent. However, comparisons are difficult to make given variations in the quality of surveillance data, sexual behaviour, access to clinical services and treatment, and the different case definitions used for early latent syphilis. For example, the definition used in the USA includes latent cases seen within the previous 12 months, a shorter period than the 24 months used in the UK definition.

The re-establishment of syphilis as an endemic infection reflects a failure of control strategies. Partner notification, which is central to the detection of the infection, has been of limited use to control efforts because of the high proportion of anonymous partners seen in MSM. Other forms of control include a combination of interventions such as the promotion of early diagnosis and treatment through measures such as facilitating access to sexual health services, antenatal screening, diagnosis and treatment, and promoting behavioural changes, such as increased condom use and reducing the number of sexual partners. As the syphilis epidemic continues to develop, sustained, intensive and targeted efforts to interrupt further transmission need to be maintained and intensified. Locally based interventions that penetrate sexual networks identified through local partner notification and surveillance initiatives will probably be the most effective method of controlling infection.

Key messages

  • Over the last 10 years, the infectious syphilis epidemic has remained ‘young’, with a high proportion of primary and secondary cases.

  • The increase in primary syphilis as a proportion of early syphilis indicates that infection is detected and managed at an earlier stage of infection.

  • Sustained, intensive, targeted efforts to interrupt further transmission need to be maintained and intensified.

  • Locally based interventions that penetrate sexual networks identified through partner notification and surveillance initiatives will probably be the most effective method of controlling infection.

Acknowledgments

We thank GUM clinicians for reporting cases of infectious syphilis through the enhanced surveillance systems operated by the HPA Regional Epidemiology Units and HPA Centre for Infections, and Dr Marion Lyons for providing surveillance data for Wales.

References

View Abstract

Supplementary materials

Footnotes

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • i The KC60 statistical return was replaced by the Genitourinary Medicine Clinic Activity Dataset as at 1 April 2009.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.