The American Public Health Association defines an outbreak as an “… increase above the expected number of persons with a communicable disease …”¹ One such outbreak of infectious syphilis in Sheffield (n = 516 000) (fig 1) has paralled the recent increase in the indentification of infectious syphilis in the British Isles.^2–5 and we have explored the utility of our traditional control method, contact-tracing or partner notification, in our control attempts. These increases have reflected the worldwide situation, where the re-emergence of the disease after many years of apparent control is associated with sociomedical factors such as urbanisation, poverty and major political transitions.⁶

We have discussed earlier how public health strategies have been engaged to control local, national and international outbreaks, from labour-intense and individualised contact-tracing (or partner notification), to social network analysis, cluster-testing, population and targeted screening; even mass treatment.⁶ Different strategies in local situations are generally unproven: one method may be as effective as any other: controlled trials are impossible. Our facility uses the traditional control method of contact-tracing, which aims to break the chain of transmission of infection by identifying, counselling and testing sexual partners of individuals with sexually transmitted infections (STI). All patients with STI are interviewed by a health adviser according to the national guidelines.⁷ Patients are offered the choice of informing partners themselves (patient referral), or supplying details for a health adviser to inform the partner, without their identity being disclosed (provider referral); patients intending to inform their own partners may agree to health adviser intervention if the partner has not presented within an agreed time (conditional referral). To prevent future infection, patients are offered risk-reduction counselling and free condoms. The recommended review periods used were 3 months for primary syphilis, 2 years for secondary syphilis and for early latent.⁸ Cooperation with partner notification is voluntary in the UK: there is no statutory requirement to name partners, or to attend for treatment if notified of exposure.

Partner notification has been put to much effective use in Sheffield in outbreaks of gonorrhoea and chlamydial infection,⁹ yet we have seen that many kinds of individual and situational dynamics can impair best contact-tracing efforts. We were not certain that this approach was optimal in infectious syphilis. Thus, syphilis outbreaks may demand new strategies of control. The traditional approach seems to work well at controlling outbreaks in persons whose sexual preference is heterosexual, but has not been as effective in controlling outbreaks in men who have sex with men (MSM).² Other approaches have been used in that situation, including offering screening in venues, such as saunas, clubs or gay “bath houses”, where sexual partners are met.^4,6 These venues may also be used to advertise the outbreak, and advise people where to go for testing. The internet may also be used.⁴

Until recently, the incidence of syphilis in Sheffield has been low (fig 1). Of late, however, we have seen what at first sight seemed to be an “outbreak”. Some authors have suggested that alternative strategies for control are best tailored to the phase of the local outbreak,¹⁰ for example, in the developing outbreak compared with the established one. In this light, the present work critically explores this putative outbreak and considers whether our strategy of contact-tracing was the most appropriate.

METHODS

Numbers and case reports were taken from computer-generated data held in the Department of genitourinary medicine (GU med) at the Royal Hallamshire Hospital, the only facility in the city that treats infectious syphilis. These data are normally generated to fulfil the statutory obligation of UK STI facilities (departments of GU med) to submit disease incidence figures to the Department of Health using the quarterly and annual “KC60” returns. We triangulated patient-referential incidence data by parallel computer-held records used by the contact tracers (health advisers; epidemiologists). These records reference demographic, sexual activity and sexual preference data, and may be cross-referenced to the numbers, and therefore contact-tracer records and clinical charts, of sexual contacts. We started to suspect an outbreak from our standard monthly incidence review and also the within-clinic sense that many of these cases were linked. At this point, it is our usual practice to start mapping cases as shown below. Case notes including partner notification records were reviewed for heterosexual cases of syphilis from the 12 months after the first case of locally acquired infection in October 2004. Homosexual cases were reviewed for 12 months from January 2004. There were no cases among bisexual men. Health advisers, who are responsible for partner notification, interviewed all patients with infectious syphilis during the study period: none refused to take part.

Data recorded for each partner (where available) included name, age/date of birth, address, telephone number, dates of first and last sexual contact, relationship type (regular, ex, casual) type of sex (oral, vaginal, anal), condom use, where met (if casual partner), plan agreed (patient, provider or conditional referral) at each interview and outcome (partner diagnosis/date; “untraced” or “informed but failed to attend”). All interviews and data recording were undertaken by health advisers having a mean of 10 years experience. Duration of interviews were not recorded, but are estimated to average 15 min.

Transmission networks constructed at the time from partner notification data were reviewed and updated. Partner notification processes and outcomes were analysed.

RESULTS

Outbreak statistics

Compared with the 7 years 1996–2002, during which a mean of 2.6 cases/year of infectious syphilis were seen, the 3 years 2003–5 saw 13, 15 and 16 cases/year respectively (Poisson CI (95%) 12.51–16.88 for the 3 years. p<0.05 for each year) (Simple Interactive Statistical Analysis (SISA), quantitative skills). During this time, there were no significant changes in the population served by the Sheffield department, nor in diagnostic and laboratory test methods.

Demographics

Data for heterosexual syphilis showed one large network of 15 connected individuals (fig 2), and two smaller ones each of four individuals. Of the 23 individuals, 10 had 11 cases of syphilis, six contacts tested negative and six were untraced. Of the 11 cases, (three women,), six were primary, four were secondary and one was early latent. Networks for MSM (fig 3 and fig 4) show 34 individuals, of whom ten had infectious syphilis (five primary, four secondary and one early latent), eight were uninfected, one was notified but failed to attend and 16 were untraced.

All three female heterosexual cases were white, British with an average age of 24 years. Two had concurrent STI (trichomonas and/or Chlamydia) and one was a former commercial sex worker (CSW). Male heterosexual cases showed a higher age (average 30 years; range 22–44 years) and greater ethnic diversity (a group of four white patients; three black patients; one Asian: all British). Four had concurrent STI (Chlamydia and/or primary herpes).

Of the MSM cases, average age was 34 years (range 21–55 years) and 9/10 were white British. Five had a concurrent STI diagnosed (2 gonorrhoea; 2 chlamydia; 1 non-specific urethritis) and one was previously known HIV positive.

All cases (heterosexual and MSM) were local residents and none gave a history of intravenous drug use. Most heterosexual cases (9/11) lived in areas of social deprivation compared with 2/7 MSM cases. Deprivation was determined by the proportion of postcodes below the 20th centile on the national Index of Multiple Deprivation, Super Output Area Levels (http://www.communities.gov.uk) (Sheffield has 29% population falling below the 20th centile).

Partner notification outcomes

Table 1 summarises case episodes using the traditional multiple counting system⁷. Episodes are counted rather than individuals: one individual may be counted several times as a case and contact(s) (appendix 1, table A; appendix 2, table B) The large heterosexual network (fig 2) shows the key role played in transmission by case 3, who is directly or indirectly linked to the six other cases. Most of her contacts were one-night stands—including one client from occasional escort work. The other three partnerships involved regular sex over a 2-month or 3-month period. There was a high level of concurrency: during September 2004, she was sexually active with two regular and three casual partners.

Transmission dynamics for MSM are less clear because most contacts (16/25; 64%) remained untraced. Of these, 88% (14/16) were from outside Sheffield–eight from elsewhere in the UK and seven from abroad. Most UK contacts were met over the internet, in telephone chat rooms, or in gay venues in cities known to have high rates of syphilis among MSM.

Partner notification interviews

A review of heterosexual case 3’s partner notification records shows the painstaking process required to uncover and trace those at risk. On her first interview she named two people (contacts 5 and 6), both of whom she said she would inform herself. During a second interview 3 days later, she had not been able to notify them, so agreed to provider referral; she supplied details for contact 5, and promised to bring a phone number for contact 6; she also named a third partner (contact 7), who was untraceable. During a third interview 4 days later with a different health adviser, she named a fourth partner (contact 8) who she said she would inform herself. Attempts to interview her again were unsuccessful. She was later named by two patients with syphilis (cases 4 and 7), whom she had not named or informed. We were not able to link them to her straight away because they only knew her by an alias; eventually contact 5 confirmed she used both names. She is also probably linked to case 1, who described an unnamed casual partner he met in the club where she met other contacts; however, the description he gave did not match, so it is possible there was another untraced infected female.

Patient or provider referral?

Provider referral was preferred to patient referral for the majority (8/11) of sought heterosexual contacts (ie, those who had not already attended and were believed to be traceable). Of the eight tested contacts who attended following notification, (contacts 4, 5, 12, 14, 15, 16, 19, 26), seven had been informed by a health adviser. Four of the eight provider referrals were agreed at follow-up interview. Time between index and contact test ranged from 8–72 days (median 30).

Patient referral was preferred by MSM for 11/15 (73%) of sought contacts, although provider referral was more likely to lead to contact attendance: 6/11 (55%) of patient referrals attended, compared with 3/4 (75%) provider referrals.

DISCUSSION

While we have not seen as extensive an outbreak as seen recently in Dublin, for example, the increase in incidence which we saw in 18 months was unusual. Our “outbreak” seems to have been of a pattern intermediate between the large MSM outbreaks in Brighton and Manchester and the purely heterosexual one seen in the Fens.^2–5

The heterosexual outbreak in Sheffield seems to have been a “spread network”.¹⁰ A majority of the contacts were traced and tested, and since the end of 2004, no new infected patients were added to the network. Thus, a good measure of control was effected by partner notification. The MSM outbreak was harder to classify, since it was mostly comprised of sporadic cases probably acquired outside of Sheffield. Some of our cases showed a “starbust” pattern.¹¹ It may be that these cases form part of a larger maintainance, or dendritic,¹¹ network which extends over other communities, including the internet, where MSM from Sheffield meet partners. That infection persists in such networks is in part due to the inadequacy of partner notification when there are large numbers of untraceable contacts.

The higher proportion of linked cases in the heterosexual outbreak shows greater success with contact-tracing for this population than for MSM, who more frequently named casual untraceable partners from outside Sheffield or abroad, and were less receptive to provider referral. Standard contact-tracing methods alone may not be effective for such a situation. Having said that, the identification, diagnosis, treatment and therefore isolation of one infected individual removes them from the infection pool.

A review of partner notification records of heterosexual case 3 (the “night-club” patient) reveals the painstaking process required to uncover and trace those at risk. Multiple interviews were necessary to acquire data of several contacts and illustrates the value of repeating partner notification interviews to uncover contacts not disclosed initially: overall 25 interviews were conducted with the ten heterosexual cases, (average 2.5; range 1–4). The importance of cross-referencing cases to build a more complete partner/risk history is also highlighted.

Repeating partner notification interviews increases the proportion of provider referral agreements, because it gives patients who have initially opted for patient referral a chance to change their mind, and/or to bring contact information for provider referral. This tendency to switch from patient to provider referral, noted elsewhere,¹² illustrates the difficulty patients may have in notifying partners themselves, and the importance of health adviser support through repeat interviews.

Strategies other than contact-tracing have been developed in areas where the incidence of syphilis has dramatically increased.^10,11 An outbreak in Dublin between 1999 and 2001 has led to an active educational campaign promoting sexual health screening.⁴ Information on syphilis and on-site serological testing was made available in venues where people met their sexual partners. Advertisements were placed on popular internet sites and in gay mens’ magazines. Other areas with high rates of syphilis have used strategies similar to these. In British Columbia, an area which sees a high incidence of syphilis associated with commercial sex workers, a social network approach was used in addition to standard contact-tracing.¹¹ “Street nurses” were sent to popular venues where people met their partners, to conduct interviews and offer testing. They also arranged follow-up and treatment outside of the clinics. This resulted in a significantly increased proportion of cases identified by the street nurses, and increased the percentage of cases linked to a previous case.

Some of these strategies could have been put into action in Sheffield to help reduce the number of untraceable patients in the heterosexual “outbreak”. As most of the patients who were involved met their partners in a particular nightclub, information about syphilis and details of where to attend for testing could have been made available there. However, the outbreak was successfully controlled through partner notification so these strategies and other possible ones (eg, cluster or social network testing¹³) did not prove necessary. It was difficult to locate the MSM outbreak in Sheffield to any particular site and a large proportion of the contacts were from other areas of the UK and from abroad. In the event of a sustained outbreak, a multi-disciplinary approach is called for involving collaboration between public health, genitourinary medicine (GUM), microbiology and infectious diseases teams. It is also helpful to understand the phase of the outbreak, to allow tailored interventions to be formulated.¹⁰

Generally, we have been reassured by our analysis that our standard approach of partner notification has paid dividends. We managed to effect a measure of control in the heterosexual outbreak, and the corresponding one in MSM continues to be sporadic. However, with location-specific epidemiology and rapidly rising incidences of syphilis, local clinics should monitor their outbreaks and should be aware of new and innovative interventions, which might be adapted and implemented.^13,14 These must be appropriate, accessible and acceptable to those at greatest risk. They may include the development of rapid, sensitive and specific diagnostic tests for syphilis, including a salivary assay for use in the field: significant numbers of our general clinic population decline venipuncture. This would allow screening to be more easily extended to venues such as saunas and clubs. Treatment regimes might be given as a single dose, although we have found that daily treatment with procaine penicillin—for example, is fruitful in adopting an iterative approach to contact tracing.

CONCLUSION

We have confirmed the experience of others that, in the face of an apparent outbreak, clinicians locally must explore the nature and parameters of the epidemic, and engage a mixture of control methods to interrupt infection spread by optimising case-finding and isolation by treatment, and by appropriate outreach initiatives.