Pathways to, and use of, sexual healthcare among Black Caribbean sexual health clinic attendees in England: evidence from cross-sectional bio-behavioural surveys

Background In England, people of Black Caribbean (BC) ethnicity are disproportionately affected by sexually transmitted infections (STI). We examined whether differences in sexual healthcare behaviours contribute to these inequalities. Methods We purposively selected 16 sexual health clinics across England with high proportions of attendees of BC ethnicity. During May–September 2016, attendees at these clinics (of all ethnicities) completed an online survey that collected data on health service use and sexual behaviour. We individually linked these data to routinely-collected surveillance data. We then used multivariable logistic regression to compare reported behaviours among BC and White British/Irish (WBI) attendees (n = 627, n = 1411 respectively) separately for women and men, and to make comparisons by gender within these ethnic groups. Results BC women’s sexual health clinic attendances were more commonly related to recent bacterial STI diagnoses, compared to WBI women’s attendances (adjusted odds ratio, AOR 3.54, 95% CI 1.45–8.64, p = 0.009; no gender difference among BC attendees), while BC men were more likely than WBI men (and BC women) to report attending because of a partner’s symptoms or diagnosis (AOR 1.82, 95% CI 1.14–2.90; AOR BC men compared with BC women: 4.36, 95% CI 1.42–13.34, p = 0.014). Among symptomatic attendees, BC women were less likely than WBI women to report care-seeking elsewhere before attending the sexual health clinic (AOR 0.60, 95% CI 0.38–0.97, p = 0.039). No ethnic differences, or gender differences among BC attendees, were observed in symptom duration, or reporting sex whilst symptomatic. Among those reporting previous diagnoses with or treatment for bacterial STI, no differences were observed in partner notification. Conclusions Differences in STI diagnosis rates observed between BC and WBI ethnic groups were not explained by the few ethnic differences which we identified in sexual healthcare-seeking and use. As changes take place in service delivery, prompt clinic access must be maintained – and indeed facilitated – for those at greatest risk of STI, regardless of ethnicity. Electronic supplementary material The online version of this article (10.1186/s12913-019-4396-3) contains supplementary material, which is available to authorized users.

The sexual healthcare behaviour of people of BC ethnicity in the UK is under-researched: a recent systematic review found few studies which focus on this topic [1]. Nationally-representative data from Britain's general population (collected 2010-12) show that BC men and women were more likely to report SHC attendance within the previous 5 years than White British men and women, after adjusting for confounders [3]. In a 2004-5 survey in SHCs across England, symptomatic BC men (but not women) experienced less provider-delay in accessing clinics than symptomatic White men did, but no other statistically-significant differences were observed in either care-seeking, or in sexual behaviour since seeking care [20]. Access to SHCs has changed since then: patients' pathways to clinic shortened between 2004 and 5 and 2009 [17], but subsequently may have lengthened [21]. We conjectured that ethnic differences in pathways to, and use of, SHCs, might exist and contribute to explaining the high STI diagnosis rates among BC SHC attendees. To explore this, we compared sexual healthcare seeking and use (hereafter, 'sexual healthcare behaviours') among BC and White British/Irish (WBI) attendees (the ethnic majority). We also compared sexual healthcare behaviours by gender, within these ethnic groups.

Study design, study population and sampling
We developed a Bio-Behavioural Enhanced Surveillance Tool (BBEST) to explore factors influencing STI among key risk groups [22]. Development of the BBEST included formative qualitative research with BC people [23], stakeholder engagement (which continued throughout the study) and piloting [22]. We purposively-selected 16 SHCs across England with high proportions of BC attendees. Between May and September 2016, people attending these clinics (of all ethnicities) were invited to complete the BBEST online survey (Additional file 6), which they accessed on tablets provided at the clinics, or their own devices. Screening questions routed eligible attendees (aged ≥15 years old and sexually-active in the past year) to the full survey, which included questions about sexual health service use, symptoms, sexual behaviour whilst symptomatic, and partner notification. 1 Survey data were individually linked, with participants' consent, to an extract of data routinely collected by SHCs for national STI surveillance (GUMCAD STI Surveillance System), including STI diagnoses. We restricted our analyses to the 627 BC and 1411 WBI attendees (99.4%) who gave their gender as male or female. (We present findings for other ethnic groups in Additional files 1, 3, 4 and 5). These sample sizes gave us adequate statistical power (80%) to detect as statistically significant (at the 5% level) differences by ethnic group (BC vs. WBI) of, for example, 5% vs. 12% among men and 5% vs. 11% among women (i.e. for behaviours with low prevalence), and differences by ethnic group (BC vs. WBI) of, for example, 40% vs. 53% among men and 40% vs. 52% among women (i.e. for more prevalent behaviours). These power calculations also allow for a design effect of 1.2, reflecting how participants were clustered by clinic.

Data analysis
Data were analysed using STATA v14, using survey commands to take account of the clustering of participants by clinic. We stratified the data by gender, and used logistic regression to obtain crude odds ratios (ORs) comparing BC and WBI participants. We then stratified by these two ethnic groups, and used logistic regression to compare participants by gender. We used multivariable logistic regression to account for possible confounders, separately for each comparison, using saturated models. Statistical significance was considered as p < 0.05 for all analyses.
In our analysis of reasons for SHC attendance, we examined reasons for attendance that participants reported in the survey, and used clinical data to identify participants with recent STI diagnoses (within the 6 weeks prior to their attendance at which the survey was completed) which we considered very probably related to clinic attendance. 2

Ethics
This study was approved by the NRES Committee South Central -Oxford C, ref.: 15/SC/0223. Table 1 presents participants' characteristics, and comparisons by ethnic group and gender (findings for a wider range of ethnic groups are presented in Additional file 1). In this section and throughout the text of the Results, we first comment on ethnic differences among women, then among men, and finally, gender differences among BC participants.

Sample characteristics
BC and WBI women were similar in age (medians 26, 25 years respectively). 23.0% of BC and 8.8% of WBI women were born outside the UK. BC women were less likely than WBI women to be educated beyond GCSEs (77.6, 84.9%, respectively), but equally likely to be in work (just over 70%). Slightly more BC women than WBI women self-identified as heterosexual (94.3, 92.4% respectively). BC women reported fewer partners than WBI women: e.g. 8.5% BC women, compared with 17.3% WBI women, reported 5 or more partners within the past year. BC women were also less likely to report new partners in this timeframe. However, there were no ethnic differences in women's current partnership type(s). There were also no ethnic differences in the proportions of women: reporting condomless last sex (around 70%), or considering themselves at risk of STI (around 44%).
Over a third (35.3%) of BC men were aged under 25, compared with 26.7% WBI men. 20.6% of BC men were born outside the UK, compared with 9.5% WBI men. A lower proportion of BC men than WBI men were educated beyond GCSEs (68.8, 83.8% respectively), and a lower proportion were in work (73.9, 83.1%). In our sample, 87.7% BC men self-defined as heterosexual, compared with 75.4% WBI men. 3 No ethnic differences were observed in the number of partners men had in the past year; around 39% reported 5 or more partners, and almost 83% reported new partners. While there were no ethnic differences in the proportions of men reporting no current partnership, current steady or casual partnerships, 35.2% BC men reported uncommitted regular partnerships, compared with 21.5% WBI men. There were no ethnic differences in the proportions of men reporting condomless last sex (around 65%), nor in the proportions considering themselves at risk of STI (just over 70%).
BC men were somewhat older than BC women (OR for being age 25 or older: 1.52, 95% CI: 1.02-2.25). While there were no gender differences observed in the proportions of BC attendees born outside the UK, nor the proportions in employment, a lower proportion of BC men than BC women were educated beyond GCSEs (OR 0.64, 95% CI:0.46-0.89). Despite our survey slightly under-sampling MSM, 2 BC men were more likely than BC women to self-identify as non-heterosexual (OR 2.29, 95%CI:1.04-5.03). BC men reported more partners in the past 12 months than BC women (OR for > 1 partner, vs. 1 partner: 5.08, 95%CI:2.88-8.97; OR for ≥5 partners, vs. < 5 partners: 6.75, 95% CI:4.08-11.17), and were more 1 Throughout this paper we use the terms partner/partnership to refer to sexual partners/partnerships, regardless of their duration, or romantic or social significance. 2 The survey question on reasons for clinic attendance did not include a response option for recent diagnosis or attendance for treatment. 3 See footnote b to Table 1.

Reasons for attending SHCs
The two main reasons for attending clinic were having (had) symptoms, and wanting an asymptomatic check-up, together reported by almost three-quarters of attendees. After adjusting for variables which were statistically significant in Table 1, BC women's attendance was more likely to be related to recent STI diagnosis/es, than WBI women's, and specifically bacterial STI diagnosis/es (AOR 2.98, 95% CI:1.43-6.23; 4.1% BC and 1.3% WBI women had recent bacterial STI diagnosis/es). BC women were more likely than WBI women to report attending because they were contacted by the clinic (AOR 3.64, 95% CI: 1.41-9.38), but less likely to report attending for a contraceptive or reproductive health reason (AOR 0.64, 95% CI: 0.45-0.92). No other ethnic differences were observed in women's reasons for attendance ( Fig. 1, data presented in Additional file 2). (Additional file 3 presents findings for a wider range of ethnic groups). Among men, no statistically-significant ethnic differences were observed in the proportions with recent STI diagnosis/es, nor in reported reasons for attendance, except that BC men were more likely to report attending because of a partner's symptoms or STI diagnosis (AOR 1.81, 95%CI:1.18-2.79).
Among BC attendees, no gender differences were observed in recent STI diagnoses. The only reasons for attendance which differed by gender were: partner's symptoms/diagnosis (more commonly-reported by BC men than BC women, AOR: 4.36, 95% CI:1.42-13.34); and contraceptive/reproductive health reasons (unsurprisingly much less commonly-reported by BC men than BC women, AOR 0.04, 95% CI:< 0.01-0.70).

Symptomatic attendees' pathways to clinic
We now focus on the pathways to clinic of participants who reported being symptomatic as a reason for attendance, around four-in-ten of the sample (Table 2; Additional file 4 presents findings for a wider range of ethnic groups).
Around a third of symptomatic women reported symptom onset within the last 7 days, and around half reported having had sex since symptoms began, with no ethnic differences observed. Symptomatic BC women were less likely than their WBI counterparts to report having tried to get treatment/advice elsewhere before attending clinic (AOR: 0.60, 95% CI: 0.38-0.97). No ethnic differences were observed among symptomatic men, in reporting: duration of symptoms, seeking treatment/advice elsewhere before attending clinic, or sex since symptoms began. No gender differences in these variables were observed among symptomatic BC attendees.

Attendees' previous experiences of STI and partner notification
We now focus on the subsample of attendees reporting previous STI diagnosis or treatment (hereafter 'previous STI'), exploring their most recent episode (Table 3; Additional file 5 presents findings for a wider range of ethnic groups). Previous STI was more commonly reported by BC women than WBI women (61.2% compared with 40.0%), with a smaller difference between BC and WBI men (60.9, 50.6% respectively), and no gender difference among BC attendees.
Almost half of attendees reporting previous STI were last diagnosed or treated within the past year, with no differences by gender or ethnic group. BC women and BC men were more likely than their WBI counterparts to report that their most recent episode included bacterial STI(s) and/or TV (AORs: women: 1.97, 95% CI:1.52-2.55; men: 3.37, 95% CI:2. 17-5.24). No statistically-significant gender differences were observed among BC attendees.
Of those reporting that their most recent STI was bacterial or TV, four-fifths recalled that clinic staff advised partner notification (PN), with no ethnic or gender differences. No ethnic differences were observed in the proportion of women reporting notifying all their partners (around four-fifths). However, BC men were more likely than WBI men to report notifying all of their partners (AOR for not notifying all partners, or cannot remember: 0.30, 95%CI:0.14-0.67). No gender differences in reporting this were observed among BC attendees.
Among those reporting notifying any partner(s) at this time, two-thirds reported notifying partner(s) in person, and half reporting doing so by telephone, text message, email, or social media (combined). Less than 4% had clinic staff notify partners for them. No ethnic or gender differences in notification methods were observed.
The three most commonly-reported reasons for not notifying (all) partners were similar between ethnic groups (however denominators were small). BC attendees of both genders reported being scared to tell partners (and BC men, being worried that their partner would leave them), which did not feature in WBI attendees' top-3 reasons.

Main findings
Our study showed that differences in STI diagnosis rates observed between BC and WBI ethnic groups were not explained by the few ethnic differences that were identified in sexual healthcare-seeking. These differences were that BC women were more likely than WBI women to have had recent bacterial STI diagnosis/es, or to attend because they were contacted by the clinic, while BC men were more likely than WBI men to report attendance due to a partner's diagnosis. Symptom duration was similar between ethnic groups, but symptomatic BC women were less likely than symptomatic WBI women to report first seeking care elsewhere. We observed no ethnic differences in symptomatic attendees' likelihood of reporting sex since symptom onset. Among those who had previously been diagnosed/treated for bacterial STI(s) or TV, there were no ethnic differences in reporting having been advised to notify partners, but BC men were more likely than WBI men to report notifying all of their partners at this time.

Strengths and limitations
Our research follows recommendations to assess needs and inequalities by ethnicity, to guide practical action [25], and contributes findings about an epidemiologicallyimportant population. Purposively-selecting clinics with high proportions of BC attendees enabled us to recruit a sample that included a relatively large number of this minority group. This enabled us to perform adjusted analyses (which were not done in a previous, similar study [20]) to control for observed differences in sociodemographic and sexual behavioural factors, which could otherwise obscure associations with health (care) behaviours. As others have done, we made gender-stratified comparisons by ethnicity [20], but additionally conducted ethnicity-stratified comparisons by gender, to explore the 'effect' of gender within ethnic groups.
Our findings are derived from detailed data collected at SHCs across England. Linkage to clinical data (achieved for the majority of survey-completers [22]) enabled inclusion of clinic-verified STI diagnoses. This has been done in previous studies of patients' pathways to clinic [17,26,27], but we were able to link to longitudinal clinical data, to include STI diagnoses associated with earlier or later attendances at the same clinic. While it is not possible to link diagnoses made in different clinics  Table 1. a For the ethnic comparison among women: the following binary variables: born in UK, education, sexual orientation, > 5/5+ partners in past 12 months (other sexual partner number variables were omitted due to likely covariance). b For the ethnic comparison among men: age as a continuous variable, and the following binary variables: born in UK, education, employment, sexual orientation, reporting regular but uncommitted partner(s). Additional file 2 provides the data used in Fig. 1 including within ethnic group comparisons by gender, and Additional file 3 provides detailed data on other ethnic groups   For the gender comparison among BC attendees: age as a continuous variable, and the following binary variables: education, sexual orientation, > 5/5+ partners in past 12 months (other sexual partner number variables were omitted due to likely covariance), reporting any steady partner(s), reporting any regular but uncommitted partner(s), reporting any casual partner(s), and self-perceived STI risk i For the gender comparison among White British/Irish attendees: age as a continuous variable, and the following binary variables: employment, sexual orientation, > 5/5+ partners in past 12 months (other sexual partner number variables were omitted due to likely covariance), reporting any steady partner(s), reporting any casual partner(s), reporting condom use at last sex, and self-perceived STI risk (because patient identifiers are clinic-specific), our use of a narrow (+/− 6 weeks) 'window' for diagnoses data probably minimises the impact of this issue. By surveying SHC attenders, we focus on a high STI risk population [4], with a key role in STI transmission and control. However, we were unable to collect data on those who did not use SHCs, and so do not know about non-attenders care-seeking or use of other services. In nationally-representative survey data, over 85% people who did not attend SHC in the past year, reported unsafe sex during this period [28]. Non-attenders also reported fewer markers of STI risk than attendees (after age-adjustment) [28], but nevertheless include people who are underserved, and thus epidemiologically-important.
Our analyses of sub-samples (e.g. Table 3) may have lacked the statistical power to detect some differences as significant, reflecting how we sought to examine ethnic differences within gender, and gender differences within ethnic group, which our original power calculations did not factor in. Our sampling strategy was designed to obtain sufficient numbers of SHC attendees of BC ethnicity, a relatively under-researched group. It was not designed to give us sufficient statistical power to additionally stratify our analyses by sexual orientation, and so we were unable to explore reasons why differences exist by sexual orientation, although these have been the focus of other studies [15,28,29]. Survey response rates were generally high (averaging 62.2%), but varied by clinic, related to clinics' ability to support survey administration [22]. Among survey-completers, consent to data linkage was slightly lower among BC than White participants (87.1% vs. 93.8%, p < 0.01) [22]. This is still relatively high, which minimises any differential influence on representativeness by ethnicity. We found some evidence that we may have over-sampled those with higher risk-profiles than SHC attendees in general [22]. The effect on our findings is unknown, but the highest-risk individuals are those of greatest concern in terms of individual and public health need.
All self-reported data is potentially subject to social desirability bias, but confidential, electronic self-completion may reduce this [30,31]. Sexual health knowledge may differ by ethnicity (suggested by large survey of London young people [32]), which could influence actual and reported behaviours, and ethnic differences therein.

Discussion of findings in relation to other research
Our sample of BC SHC attendees is likely to be at higher STI risk than the BC population as a whole [33], or the general population [4]. We found ethnic differences in reasons for attendance, whereas no differences were observed in a 2004-05 clinic survey [20], possibly related to statistical power, and plausibly influenced by SHCs' accessibility at the time. Nevertheless, our findings are similar to this earlier survey in that we found few ethnic differences in sexual healthcare-seeking behaviours, despite observing differences in STI diagnosis [20].

Meaning and implications
Differences in STI diagnosis rates observed between BC and WBI ethnic groups [14] are unlikely to be explained by the few differences identified in sexual healthcareseeking and use, according to our study of SHC attendees. In the context of persistent elevated STI risk among England's (and Britain's) BC population [3,14], we need to ensure their access to sexual healthcare is, at the very least, maintained, if not improved. In terms of STI control, it is encouraging that we found that symptomatic BC women attendees were less likely than WBI women to seek care elsewhere before attending a SHC, because using other services may lengthen care-seeking [17,27]. It is also encouraging that among those with previous bacterial STI/TV, a higher proportion of BC men than WBI men reported notifying all partners. Yet there is scope to improve such behaviours among all ethnicities, through clinic-based and broader health promotion and structural interventions. It is concerning that BC and WBI women attendees were equally likely to perceive themselves at risk of STI, given BC women attendees' greater likelihood of diagnosis. In a separate analysis of our dataset, heterosexual BC women reported lower recent partner numbers than their WBI counterparts [9], and this lower individual risk behaviour may affect risk perception. However, the picture is complex, with survey data from the general population suggesting little relationship between STI risk perception and both reported unsafe sexual behaviour, and engagement with sexual healthcare [34]. Greater engagement with asymptomatic check-ups is needed, among sexually-active people who are at elevated risk of STI who do not test for STIs, in line with Public Health England's advice [14].
Since data collection, funding cuts [35] and SHC closures have taken place in England, including some clinics which participated in this study. STI testing across London became accessible online, with an expectation that a significant proportion of patients could test using home self-sampling packs [36]. The effect of these changes on the 4.2% of Londoners of BC ethnicity [10] is unknown; generally, evidence about e-health use by ethnicity is lacking [37]. Evaluation of sexual health service reconfigurations must address impacts on access by ethnicity, to ensure that service changes lessen, or at the very least do not exacerbate, existing health inequalities.