Understanding the burden of bacterial sexually transmitted infections and Trichomonas vaginalis among black Caribbeans in the United Kingdom: Findings from a systematic review

Background In the UK, people of black Caribbean (BC) ethnicity continue to be disproportionately affected by bacterial sexually transmitted infections (STIs) and Trichomonas vaginalis (TV). We systematically reviewed evidence on the association between bacterial STIs/TV and ethnicity (BC compared to white/white British (WB)) accounting for other risk factors; and differences between these two ethnic groups in the prevalence of risk factors associated with these STIs, sexual healthcare seeking behaviours, and contextual factors influencing STI risk. Methods Studies presenting relevant evidence for participants aged ≥14 years and living in the UK were eligible for inclusion. A pre-defined search strategy informed by the inclusion criteria was developed. Eleven electronic databases were searched from the start date to September-October 2016. Two researchers independently screened articles, extracted data using a standardised proforma and resolved discrepancies in discussion with a third researcher. Descriptive summaries of evidence are presented. Meta-analyses were not conducted due to variation in study designs. Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines were followed. Results Of 3815 abstracts identified, 15 articles reporting quantitative data were eligible and included in the review. No qualitative studies examining contextual drivers of STI risk among people of BC ethnicity were identified. Compared to the white/WB ethnic group, the greater STI/TV risk among BCs was partially explained by variations in socio-demographic factors, sexual behaviours, and recreational drug use. The prevalence of reporting early sexual debut (<16 years), concurrency, and multiple partners was higher among BC men compared to white/WB men; however, no such differences were observed for women. People of BC ethnicity were more likely to access sexual health services than those of white/WB ethnicity. Conclusions Further research is needed to explore other drivers of the sustained higher STI/TV prevalence among people of BC ethnicity. Developing holistic, tailored interventions that address STI risk and target people of BC ethnicity, especially men, could enhance STI prevention.


Introduction
In the United Kingdom (UK), sexually transmitted infections (STIs) continue to be a public health concern [1,2]. Studies since the late-1980s have shown disproportionately high rates of bacterial STI diagnoses among black compared to white ethnic groups [3,4]. Since 2000, clinic-based studies [5,6], national probability surveys [7,8], and surveillance data [9,10] have distinguished between 'black Caribbean' (BC), 'black African' (BA) and 'black other' (BO) ethnicities, and have consistently shown that people of BC ethnicity in particular experience the highest rates of infection with Trichomonas vaginalis (TV) and bacterial STIs. Rates of gonorrhoea are 8-12 times higher among people of BC ethnicity compared to people of white ethnicity [9,11]. Age and sex related variations exist in rates of gonorrhoea and chlamydia diagnoses across and within ethnic groups, including among people of BC ethnicity [11]. The risks of coinfection and reinfection with bacterial STIs are also higher among BC compared to other ethnic groups [12][13][14].
Understanding the social patterning and determinants of poor sexual health is key to promoting equitable health and informing evidence-based public health policy and practice [8]. Given the high burden of bacterial STIs and TV among people of BC ethnicity in the UK, we undertook a systematic literature review to examine if ethnic variations in factors known to be associated with STIs/TV (for example: age) explain ethnic variations in STIs/TV at a population level. Specifically, we examined the evidence on the association between bacterial STI/TV and ethnicity (BC compared to white/white British (WB)) because the latter is the predominant ethnic group in the UK corresponding to approximately 90% of the population [15]. We also examined variations between these two ethnic groups in the prevalence of behavioural risk factors associated with STIs, sexual healthcare seeking behaviour, and contextual factors influencing STI risk.

Inclusion criteria
Studies examining the association between ethnicity (BC compared to white/WB) and bacterial STIs/TV were eligible for inclusion. We also included studies if the outcome variable considered bacterial and viral STIs together but predominantly comprised of bacterial STIs. Studies conducted only among people living in the UK were included because our focus was to understand the factors driving the sustained disproportionate burden of these infections among BC people in the UK. Studies examining variations in the prevalence of risk behaviours associated with these infections and in sexual healthcare seeking behaviours between these two ethnic groups, and studies exploring contextual drivers of STI risk among BC people were also eligible for inclusion. Studies conducted among persons aged �14 were eligible for inclusion because the rates of bacterial STI diagnoses increase substantially from the age of 14 [2].

Exclusion criteria
Studies that did not differentiate between different 'black' ethnic groups were excluded due to variations in STI prevalence among different black communities, and because 'black' is a heterogeneous category, including for example, variations in history of migration, risk behaviours, and background STI prevalence in home countries among migrant populations [7]. Additionally, we excluded studies that met the inclusion criteria but provided scant data [16,17], determined via discussion between two researchers. We also excluded studies not written in English.

Search strategy
The following electronic databases were searched from the start up to September-October, 2016 (S1 Table) for empirical studies using a pre-defined search strategy informed by the review inclusion/exclusion criteria (S1 File): Medline, Embase, Cinahl, Psychinfo, Scopus, Web of Science, British Humanities Index, Applied Social Science Index and Abstracts, International Bibliography of the Social Sciences, Sociological Abstracts, and the Cochrane database of systematic reviews. Search terms were adapted to meet the requirements of different databases and search results imported into Endnote software. We also contacted two researchers in the field for unpublished papers/reports.

Screening and data extraction
Following merging and deduplication of search results, two researchers independently screened 5% of all titles and abstracts to develop consensus for inclusion of studies, using prespecified screening questions (S2 Table) which were informed by the inclusion/exclusion criteria. Subsequently a researcher screened the remaining titles and abstracts. Reference lists of studies included in the review were also screened. Once the eligible papers were identified, data were extracted and quality appraisal was conducted independently by two researchers for five of the included studies to pilot the data extraction pro-forma (S2 File) and quality appraisal pro-forma (S3 Table) and checked for concordance. The quality appraisal pro-forma was adapted from the NICE guidance for 'quality appraisal checklist for quantitative studies reporting correlations and associations' [18]. This checklist seeks to assess the key population criteria for determining the study's external validity, i.e. the extent to which the findings of the study are generalizable to the study's source population. It also seeks to assess the internal validity of the study using various criteria, i.e., that the study has been carried out carefully and the identified associations are valid. Disagreements regarding study inclusion and discrepancies in data extraction and quality appraisal were resolved through discussion with a third researcher. Subsequently data extraction and quality appraisal of all the remaining papers was conducted by a researcher and was checked by another researcher.

Data synthesis
Quantitative studies included in the review varied in study design, methods, definitionsand measurement of outcomes and explanatory variables. Thus a meta-analysis was not conducted to avoid the risk of deriving misleading conclusions [19], and instead we conducted a narrative synthesis of evidence for each of our research questions. Studies were analysed according to the type of STIs, and study design. Descriptive summaries are presented on the reported evidence relating to (i) the association between ethnicity (BC vs white/WB) and STIs/TV after adjusting for other factors, and (ii) variations in the prevalence of risk behaviours associated with STIs/TV, and (iii) variations in sexual healthcare seeking behaviours between these two ethnic groups.
We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines (S4 Table) [20].

Results
We identified 3815 records (Fig 1). Fifteen quantitative studies were eligible and included in the review. Of these 13 studies were identified during electronic searches, one through screening of the reference list of an included paper, and one was a paper published by our research team which was under review when we conducted electronic database searches and was subsequently published in 2017 (Table 1). No qualitative studies that examined contextual drivers of STI risk only among people of BC ethnicity were identified.

Characteristics of included studies and participants
Of the 15 studies included in the review (Table 1), ten examined risk factors either for single or multiple STIs accounting for ethnicity [6][7][8][9][10][11][21][22][23][24]. Of these, most studies examined the risk factors for chlamydia or gonorrhoea, but a few examined risk factors for TV [10], syphilis [9], or 'any bacterial STIs' [7,8]. Furthermore, one study examined factors associated with the risk of gonorrhoea and chlamydia co-infection [12]. Another study examined the risk of re-infection with gonorrhoea [14], and a study examined the risk of acute STIs among patients reattending a sexual health clinic within one year [13]. Differences between BC and white/WB ethnic groups in the prevalence of behavioural risk factors associated with bacterial STIs/TV were examined by three studies [7,8,25], and in sexual healthcare seeking behaviours by three studies [7,25,26].
Study design. Two studies used data from national probability surveys in Britain (England, Scotland and Wales) [7,8], and three used data from sexual health clinics in England only, comprising a cross-sectional survey [26], a case-control study [22], a retrospective case note review [12]. Six studies used routine or sentinel surveillance data on clinic attendees in England [6, 9-11, 13, 14]. Three studies used data from England's National Chlamydia Screening Programme (NCSP) which targets sexually-active young people aged 14-24 years [21,23,24]. One study reported data from a cross-sectional survey in secondary schools in London among students aged 11-16 years [25].

Quality assessment of included studies
As shown in S4 Table, of the 15 studies included in the review, the majority had reported data that enabled assessment of internal and external validity of the study results. As shown in Table 1 and S3 Table, the scores for internal and external validity of study results of four studies (two national probability surveys and two studies using surveillance data from all sexual health clinics in England) [7][8][9][10] were higher than for the other studies. For the other eleven studies, either the generalisability was limited (for example, due to sample selection or recruitment bias [21,23,24], or their internal validity was limited (for example, due to low participant response rate [22].

Associations between ethnicity and STIs adjusting for other factors
Studies examined the association between ethnicity (BC compared to white/WB) and different types of bacterial STIs/TV diagnoses adjusting for a diverse range of factors ( Table 2). Studies that used surveillance data showed that the association between ethnicity and diagnoses of  gonorrhoea [6,9,11], chlamydia [6,11], syphilis [9], and TV [10], was only partially explained by differences between BC and white/WB ethnic groups in their sociodemographic characteristics, including age, area-level deprivation, sexual orientation, and in previous STI diagnosis. Similarly, among participants screened for chlamydia for the English NCSP, the risk of chlamydia positivity among BC compared to white/WB young people continued to be higher despite accounting for socio-demographic factors and for screening venue [21,23,24]. This finding was also observed in a case-control study [22] and in studies that used national probability survey data which in addition to sociodemographic factors had adjusted for individuallevel economic status and sexual behaviours [7,8], and recreational drug use [8]. Taken together, these studies indicate that variations between BC and white/WB ethnic groups in a number of socio-demographic, economic and behavioural factors do not fully explain differences in STI prevalence between these ethnic groups. The greater risk of reinfection with acute STIs among BCs re-attending sexual health clinics compared to those of white/WB ethnicity was also not fully explained by differences in these two ethnic groups in the socio-demographic factors, behavioural risk factors, and history of previous STI diagnosis [13]. Another study conducted among sexual health clinic attendees showed that adjusting for sexual orientation, area-level deprivation, and history of gonorrhoea diagnosis explained the greater risk of repeat gonorrhoea infections observed among those of BC ethnicity compared to those of white/WB ethnicity [14]. With regards to the risk of co-infection with chlamydia and gonorrhoea, one study found that ethnic differences in age and sex (being <20 years and female) explained the higher risk of co-infection among BC people [12].

Ethnic differences in the prevalence of behavioural factors associated with STIs
As shown in Table 2, regardless of study design and ethnicity, factors such as age, multiple partners, and condomless sex were associated with STI diagnosis among men and women LGV/Donovanosis, genital warts (1st episode), genital herpes (1st episode), molluscum contagiosum, or scabies/pediculosis; '++' Indicates that for the stated external or internal validity checklist question, the study has been designed or conducted in such a way as to minimise the risk of bias; '+' Indicates that either the answer to the checklist question is not clear from the way the study is reported, or that the study has not addressed all potential sources of bias for that particular aspect of study design. https://doi.org/10.1371/journal.pone.0208315.t001 Sexually transmitted infections among black Caribbeans in the UK  [6-8, 21, 22]. Individual-level economic status [8], sexual orientation [6], recreational drug use [8,22] and paying for sex [7,8] were found to be associated with STI diagnosis only among men, whereas marital status [8,22], smoking [22], having new sex partners from another country [7], concurrent partnerships, sexual debut <16 years, and lack of sexual competence at sexual debut [8] (sexual debut was considered as sexually competent if the study participants reported: an absence of duress and regret about timing; autonomy of decision; and that a reliable form of contraception was used) were associated with STI diagnosis only among women. Data on factors that influence STI risk behaviours were available only from one longitudinal school-based survey which showed that depressive symptoms, low family support, and substance use were the strongest predictors of reporting sexual debut <16 years, having more than one sex partner, or condomless sex [25]. Year of first diagnosis � , age group among heterosexuals, age group among MSM#, sexual orientation � , resident in same town as clinic, area-level deprivation (IMD) � , reported history of gonorrhoea � , reported history of any STI, number of partners in last 3 months � , sex with a high risk sexual partner (past 12 months), currently a sex worker, non-completion of clinic's behavioural pro-forma � , ever injected drugs � Association with the STI examined; # Associated with the STI examined only among men; ¶ Associated with the STI examined only among women; † Associated only with chlamydia in women and with gonorrhoea and chlamydia in men; Variables which are not marked with any of the symbols �# ¶ were either removed from multivariable logistic regression models because of lack of association in earlier iterations of the model, or were not statistically significant in the model which produced the adjusted odds ratios are presented in this The following section describes how behavioural factors associated with STIs varied for BCs relative to white/WB ethnic groups (Table 3).
Age and sexual competence at sexual debut. The national probability surveys show that compared to white/WB men, the proportion of BC men reporting early sexual debut, i.e., <16 years, was double (56.3%-60.6% vs. 26.7%-27.9%) [7,8] and similarly, a London school-based survey estimated sexual debut �13 years to be 35.0% and 10.0% among BC and white/WB young men respectively [25]. The national survey also showed that the proportion of BC men reporting that they were sexually competent at sexual debut was lower (32.9%) than among white/WB men (47.4%) [8]. In contrast, among BC and white/WB women, the prevalence of reporting sexual debut <16 years was similar in the national probability surveys (~20.0%) [7,8] as was the prevalence of sexual debut �13 years in the London-school based survey (~5.0%) [25]. The national survey showed that the reporting of sexual competence at sexual debut among BC women was lower (40.9%) than among WB women (47.9%) [8].
Partner numbers. In national probability surveys, the proportion of BC men reporting five or more partners in the last five years was higher (range: 27.1%-35.7%) than for white/WB men (range: 13.9%-21.0%) [7,8], whereas the proportions were similar among BC and white/ WB women (range: 7.7%-11.9%) [7,8]. Similarly, among 13-16-year-olds in the London school-based survey, the proportion of BC men reporting two or more partners ever was higher (57.0%) than for white/WB young men (17%) but proportions were similar among BC and white/WB young women (11.0%, 7.0% respectively) [25]. Among these 13-16 year olds, the proportion of BC men (49.0%) and women (22.0%) reporting 'ever' having sex was higher than for white/WB men (31.0%) and women (16.0%).
Sexual behaviours. In national probability surveys, the proportion of BC men reporting concurrency in the last year among sexually active participants [7] or in the last five years [8] was higher (~26.0%) than for white/WB men (~14.0%); however, this difference was explained by differences in age between these two ethnic groups [8]. The proportions of BC and white/WB women reporting concurrency (8.0%-11.5%) did not vary by ethnicity. The proportions of BC and white/WB men reporting paying for sex in the last five years were and 5.1% and 3.1% respectively, and among women no BC women reported this behaviour and it was reported by 0.03% white/WB women [8]. There were minor differences in the proportions of BC and white/ WB men reporting one or more new same-sex partnerships in the last year (2.1% and 1.4% respectively) and among women it was 1.0% and 0.9% respectively [7]. The proportions of BC men and women reporting 'ever' having genital contact with same-sex partners was lower (1.1% and 2.2% respectively) than for white/WB men and women (5.7% and 6.6% respectively) [8].
The national probability survey showed that the proportion of BC men reporting condomless sex with two or more partners in the last year was higher (11.6%) compared to white/WB men (7.4%) whereas the proportions of BC and WB women reporting this behaviour were similar (~6.0%) [8]. In the London school survey of 13-16 year olds [25], the proportions of men 'ever' having condomless sex were similar among BC and white/WB men (~12%) and women (8.0%). However, the proportion reporting condom use at last sex was higher among BC men and women (93.0% and 74.0%, respectively) than among white/WB men and women (76.0% and 68.0%, respectively) [25]. Likewise, the proportions reporting new sex partners from outside the UK in the past five years was higher among BC men and women (20.4% and 18.1%, respectively) than among white/WB men and women (13.2% and 6.3%, respectively) [7].

Substance use
With regards to substance use, the national probability survey shows that similar proportions of BC and white/WB men and women reported recreational drug use in the past year (12.6% and 15.6% respectively in men and 11.2% and 7.0% respectively in women) [8]. Whereas BC men and women were less likely to report current smoking (19.6% and 21.4%) than white/WB men and women (26.5% and 25.5% respectively) [8].

Ethnic differences in sexual healthcare seeking behaviour
Three studies (Table 4) reported data on variations in sexual healthcare seeking behaviours among BC and white/WB ethnic groups [7,8,26]. Both nationally representative surveys reported that a higher proportion of BC people attend sexual health clinics than white/WB people [7,8]. Additionally, a clinic-based survey reported that symptomatic BC men were less likely to delay seeking care (i.e., waited more than seven days after symptoms started before seeking care) than white/WB men; however, the former were more likely to have multiple sex partners when symptomatic [26]. Among women, similar proportions of BC and white/WB women reported a delay in seeking care (44.4% and 48.1% respectively) or to have sex after symptoms started (49.2% and 54.2% respectively).

Discussion
This is the first systematic review to examine factors influencing the sustained disproportionate burden of bacterial STIs and TV among people of BC ethnicity in the UK. Our findings highlight that in most studies, the higher risk of STI acquisition among BCs compared to white/WB ethnic group persisted after adjusting for various socio-demographic factors, behavioural risk factors, and substance use. Importantly, however, our review suggests that the higher prevalence of sexual risk behaviours, including early sexual debut, concurrency and larger partner numbers reported among BC men compared to white/WB men potentially contributes to their disproportionately high STIs rates. In contrast, as shown in Tables 2 and 3, the higher STI burden among BC women compared to white/WB women exists despite adjusting for known risk factors for STIs, and there are similarities in the reporting of the prevalence of risk behaviours among women from these two ethnic groups. These findings suggest that STI prevention efforts targeting behaviour change among BC men might be more effective in this population. Encouragingly, our review also suggests that people of BC ethnicity are more likely to access sexual health clinics than people of white/WB ethnicity, which has implications for delivering STI prevention interventions through sexual health clinics to reach this population. In terms of clinical practice, the limited existing evidence on co-infection and reinfection with bacterial STIs among BC people suggests that retesting following treatment and enhancing partner notification, especially among young women, could be beneficial for improving sexual health outcomes for this population group.

Strengths & limitations
We included studies from the start of the electronic databases searched up to October 2016. The earliest study that met our inclusion criteria was published in 2000 [6], thereby reducing heterogeneity between studies in study populations over time. Sexual behaviour and its associated influences have changed little at a population-level since 2000 [27][28], but changes in the delivery of sexual healthcare in recent years are likely to have influenced sexual healthcare seeking [29]. The introduction of more sensitive STI testing techniques over time may also have influenced STI diagnosis rates, but their application is unlikely to have varied by ethnic group. Our findings should be interpreted with caution given the variation in study methodologies which meant that we were unable to conduct a meta-analysis. For example, some studies used a generic 'white ethnicity' as the reference group while others specifically used WB ethnicity. Such conflation of people of white/WB ethnicity could have introduced bias especially in studies following the accession of ten central and eastern European countries to the European Union in 2004, resulting in increased numbers of people identifying as 'white other' [8,30]. Differences in sexual behaviour between 'white other' and WB populations have also been reported [31]. The majority of the studies included in the review had limited internal and/or external validity, therefore the review results should be interpreted with caution, especially as most of the studies used data collected from sexual health clinic attendees in England. While we excluded studies that did not differentiate between people of BC ethnicity from other black ethnicities, we acknowledge that considerable heterogeneity exists within the BC ethnic group [32].

Implications for future research and practice
Addressing inequalities is one of the priorities of the sexual health improvement framework in England [33]. Previous national and local policies have tended to prioritise HIV prevention to the exclusion of other STIs [34]. High rates of STI diagnoses in people of BC ethnicity have been highlighted since the early 2000 [11] yet there has been a relative dearth of studies addressing this issue. Our systematic review strengthens the evidence-base by enhancing understanding of the factors influencing ethnic differences in STIs which is vital for understanding research gaps and improving STI prevention interventions. It has been argued that STI prevention should focus on young people irrespective of ethnicity and account for gender differences [35][36][37] because patterning of risky and protective behaviours is mediated by 'youth' [25,37]. However, this argument overlooks the role that ethnic identity may play in influencing STI risk.
The higher burden of STI diagnosis observed among BC women relative to white/WB women in the absence of behavioural differences highlight the need to conduct partnershiplevel studies of STI risk to inform STI prevention efforts. We did not find any qualitative studies conducted specifically among people of BC ethnicity. However qualitative studies of young people in the UK from major ethnic groups or of black ethnicity have shown that the broader social context, including religion [37], gender norms related to sex and sexuality influence partnership types and hamper condom use, exacerbating STI risk, especially among young women [34,36,37]. These studies have also highlighted a preference for same-ethnicity longterm partnerships but greater likelihood of disassortative mixing in casual partnerships [36,37]. However, none of the studies included in our review examined the impact of sexual mixing patterns by ethnicity on STI risk. Better evidence is also needed on the impact of ethnicity related stigma and discrimination on STI related risk and healthcare seeking behaviours [35]. Mixed-methods research could improve our understanding of, for example, the role of partnership dynamics on STI risk among people of BC ethnicity and is currently underway [38].
Supporting information S1