Background

A large proportion of people living with HIV/AIDS in most European Union/European Economic Area (EU/EEA) countries are migrants, that is, people currently living outside their country of birth [1]. In 2013, around one third (33 %; 3 160) of those in EU/EEA who were reported to have contracted HIV through heterosexual contact were migrants from countries with generalised HIV epidemics (mainly sub Saharan Africa) [2]. While most of these diagnoses were made for the first time in Europe, HIV acquisition is predominantly assumed to have occurred in the home country [3].

This assumption is often based on reports from clinicians who make deductive inferences according to a patient’s country of birth, time of arrival in the new country of residence, CD4 cell counts and the natural history of HIV infection. Clinicians may also use algorithms that are biased towards determining that patients born in countries with a generalised epidemic contracted HIV prior to migrating to Europe [4]. Often these estimates do not take into account additional factors. For example, it has been reported that CD4 cell counts close to seroconversion are considerably lower among those living with non-B HIV-1 viral subtypes, which are most common among migrants from outside Europe, particularly in Sub-Saharan Africa. [3, 5, 6]. Consequently, a lower CD4 cell value at diagnosis in those with non-B sub-types may wrongly suggest that individuals have been living with HIV for longer than they have; leading to the conclusion that HIV was acquired before migration. The presence of non-B subtypes among Europe-born populations may also be interpreted as evidence of sexual mixing between migrants and non-migrants.

Funding for primary prevention among migrant communities may be reduced or redirected if surveillance data suggest that individuals do not have HIV prevention needs after they have left their home country or that significant numbers of migrants are coming to Europe as “HIV health tourists” [7, 8]. There have been no previous efforts to collate the information about the proportion of people from countries with a generalised epidemic that contract HIV post-migration. If programme managers and policy-makers underestimate the degree of HIV acquisition within the EU/EEA for migrants from countries with generalised epidemics this may undermine the potential for reducing HIV prevalence and incidence through targeted primary and secondary prevention programmes and policy.

This paper presents a review of the evidence of post-migration HIV acquisition among migrants from countries with a generalised epidemic living in Europe. We examine quantitative studies and surveillance reports based on data on populations from countries with a generalised epidemic which include outcomes that estimate the probable country of acquisition, incident infections, or evidence of sexual mixing. In addition this paper profiles the ability of EU Member States surveillance systems to provide accurate monitoring information about probable country of HIV acquisition. We discuss the implications of these results in HIV prevention programming and policy across the EU/EEA region.

Methods

Systematic review

Nine electronic databases (Allied and Complementary Medicine; Cochrane Database of Systematic Reviews; Cumulative Index to Nursing & Allied Health Literature; Database of Abstracts of Reviews of Effects; EMBASE; Health Management Information Consortium; Health Technology Assessment; Medline; PsychInfo) were searched during May 2012Footnote 1. A detailed search strategy was used which combined synonyms for “HIV”, “migrant”, “assortative sexual mixing”, “sexual transmission” with demonyms for all countries with a generalised HIV epidemic (for the complete list search terms, a full description the review protocol please see report [9]). Searches were limited to studies conducted between 01/0/1/2002 and 31/12/2014 to provide the most up-to-date estimates. Studies written in English, French, Italian, Portuguese and Spanish were included. Additional grey literature was retrieved from four websites (United Nations Department of Economic and Social Affairs Population Division; European Health for All database, World Health Organization Regional Office for Europe; European Centre for Disease Prevention and Control), and relevant data were requested from individuals participating in the Member States Survey (see below). The search process was documented by compiling the search strategies used to explore each resource.

Selection criteria

Only studies conducted in countries with programmatic or surveillance links within the European Centre for Disease Prevention and Control (ECDC) were included in this reviewFootnote 2. Studies were eligible for inclusion if: the study population included migrant men or women from countries with a generalised HIV-1 epidemic AND the study included sub-group analysis based on race/ethnicity or country/region of origin OR at least 80 % of the study populations were from countries with a generalised HIV epidemic. Studies were only included if they reported on any of the following outcomes: proportion of target population infected with HIV in country of origin; proportion of target population infected in country of migration; estimate of incident HIV infections (not diagnoses) in target population in country of migration; probable country of infection/HIV acquisition and evidence of sexual mixing. Studies that reported mode of transmission but made no reference to whether sexual transmission took place pre- or post-migration were excluded at full paper screening stage. Qualitative studies (using in-depth interviews, focus group discussions, and document analysis), conference communications, pilots or feasibility studies were excluded.

Quality assessment

Studies were selected using a two-stage screening approach. Reviewers devised a checklist to independently screen all retrieved titles and abstracts. Studies were given an overall quality score which incorporated a number of factors drawn from the PRISMA [10] and NICE guidelines [11] including risk of bias, internal and external validity (See Table 1). Papers were graded as having an overall quality score of “Low”, “Medium” or “High”. We were aware that within this review few cohort or intervention studies would be retrieved which may lead to a systemic bias in quality assessment. As a result studies were rated within the paradigm of their study type and studies based on surveillance or cross sectional data were able to achieve overall quality scores of “Medium” or “High”. Studies that received a “Low” score or for which no information to perform quality assessment was available were excluded from the final review. Inter-reviewer reliability scores (Cohen κ) were calculated using Kappa in Microsoft Excel: a kappa of 0.68 for full paper screening and 0.64 for quality appraisal, indicated a high level of agreement between reviewers.

Table 1 Criteria used to assess the quality of papers included in full paper review
Full size table

Data extraction and analysis

Data were extracted using forms detailing study objectives; thematic areas; data collection; methodology (design; setting; population; sample size; geographical scope); results and outcomes (n or %); author defined strengths and limitations and gender specific issues. After data extraction for each paper, studies were grouped according to outcomes of interest. Narrative summaries of each outcome of interest are presented.

Member states survey

During August 2012, a survey was conducted among 30 EU/EEA Member states using an online survey software package SelectSurveyNet (ClassApps). The questionnaire was developed to gather information from representatives of each member state regarding their knowledge and surveillance of HIV and HIV transmission among migrants from countries with a generalised HIV epidemic. Nationally nominated HIV surveillance contact points were invited to complete the short, 14-item questionnaire which contained mainly open questions, allowing respondents to provide detailed responses. Survey questions were tailored to each country based on information that had been recently submitted to ECDC as part of the Dublin Declaration reporting processFootnote 3. Participants were also able to upload documents to support their responses and these documents were added to the systematic literature review process described above.

Results

In total 8125 documents were retrieved from all sources. Twenty-seven peer-reviewed papers (representing 26 studies) were found to fulfil the inclusion and quality assessment criteria and were therefore included in the final review. Fig. 1 summarises the outcome of the paper selection process.

Fig. 1
figure 1

Summary of study selection process

Full size image

Papers were included from six EU countries: United Kingdom (9); Netherlands (4); France (4); Spain (2); Belgium (1) and Italy (3). Two papers were also included from Switzerland and two further papers covered the entire European region. Studies were grouped according to outcomes of interest within three categories: Probable country of HIV acquisition; Estimates of incident HIV infections; and Evidence of sexual mixing. Tables 2 and 3 summarise the 26 studies’ characteristics and includes the quality appraisal process results.

Table 2 Summary of included studies: population characteristics
Full size table
Table 3 Outcomes and limitations of included studies
Full size table

Probable country of HIV acquisition and estimates of incident HIV infection

Estimates of probable country of infection and/or estimates of incident infection were found in 18 of the 27 papers selected for systematic review. In most of the papers, the study population included people from countries not considered to have a generalised HIV epidemic, but data were disaggregated which allowed reviewers to perform data extraction and compare data across countries. The estimates varied both within countries and across Europe, and covered a range of subgroups, including men who have sex with men (MSM) as well as heterosexuals. Table 4 shows the proportion of infections acquired among sub Saharan Africans post–migration in France, Netherlands, Switzerland and the UK. Table 5 shows infections acquired post-migration among individuals from the Caribbean and Asia in Italy, Netherlands, Switzerland and the UK.

Table 4 Proportion of infections acquired in European countries among people born in Africa or with Black African ethnicity
Full size table
Table 5 Proportion of infections acquired in European countries among people born in Caribbean or Asia or with Black Caribbean ethnicity
Full size table

While the generilisability and validity of all these papers were medium or high, there are some limitations to the data presented (Table 3). Data estimating the proportion of post-migration HIV acquisition from cross sectional studies in France, Italy, Switzerland and the UK relied on samples that were either very small [12] or were reported to be possibly biased and unrepresentative of the wider population from which the samples were drawn [13–15]. Estimates from The Netherlands were derived from mathematical models. The authors provided very little information about the data used to source the models nor did they discuss the consequent limitations such data placed on the models [16, 17].

Using a new method to ascertain likely country of HIV infection, the UK recently published estimates of place of HIV acquisition among 10,612 heterosexual migrants (9065 of black-African ethnicity) diagnosed with HIV in the UK between 2004 and 2010 [18]. The authors used CD4 at diagnosis to estimate year of infection for each of the adults in their study population and taking into account the variable “Year of Migration”, assigned probable place of HIV acquisition. The CD4-cell based method estimated that 33 % of the study population (26 %-39 %) acquired HIV while living in the UK, three times higher than national estimates of HIV based on clinic reports (11 %). The study also found that the proportion of persons who had acquired their HIV while living in the UK had increased from 24 % (16-39 %) in 2004 to 46 % (31 %-50 % in 2010 (p < 0.01). Similar analysis performed at a later point by the same authors estimated the increase among black African heterosexuals to be from 9.1 % in 2002 to 37 % in 2011 (p < 0.01), however this analysis included data based on clinical reports [19].

The authors discuss two major limitations with this method. First, the method cannot take into account travel (including vacations back to the country of origin) post-migration. Second, the method relies on key variables (CD4 counts, date of arrival and country of birth) and therefore it cannot be applied if these data are missing.

Evidence of sexual mixing

Fourteen studies examined sexual activity of people after they migrated to Europe (Table 2). Most studies used molecular epidemiology to describe the distribution of non-B subtypes and HIV transmission networks, and in so doing provide estimates of sexual mixing. These studies show that non-B subtypes are prevalent among people born within and outside Europe: in Switzerland, the proportion of non-B subtype virus increased from 22 % in 1996 to 33 % in 2009 [20]; Holguin et al. (2007) [21] found that 53 % of migrants and 14 % of native Spaniards in their study in Grand Canary were infected with non-B strains; Snoeck et al. (2002) reported a small number of non-B subtypes originated in Belgium (16 %) [22]; in Italy Tramuto et al. (2013) estimate that less than one in ten Italian-born individuals in their study were living with non-B HIV-1 subtypes [23]. More detailed analysis from Lai et al. (2013) showed that country of origin was independently associated with the probability of patients being detected in epidemiological clusters, although patients from countries with a generalised epidemic were less likely to be detected than those from Italy or South America [24]. Rivas et al. (2013) provide some evidence of sexual mixing by reporting a small proportion of B subtypes (3.3 %) among migrants from sub Saharan Africa living in an area of Spain [25].

Elford et al. (2007) [26] state that 80 % of black African heterosexuals reported sexual partners of the same ethnicity and van Veen et al. (2009) [27] found that 41 % of their sample had partners with different ethnicity (15 % with Dutch partners). Similarly Marisciano et al. (2013) reported that 50 % of men and 47 % of women had partners from a different country although most were from another African country [28].

In a cross sectional survey examining sexual behaviour among Surinamese and Antillean migrants in the Netherlands [29], the authors describe travel patterns of migrants and provide estimates of sexual mixing for a population they classified as having the potential to provide a transmission “bridge” between the Caribbean and The Netherlands. Those having unprotected sex in both countries, the so-called “bridge population”, reported partners of a different ethnicity less often than those in moderate or low risk groups (42 %, 59 % and 66 % respectively) [29].

Member states survey

Twenty-four countries responded to the survey (response rate: 80 %). Five countries reported having data on sexual transmission of HIV among migrant communities. Denmark and Germany provided some contextual information about sexual transmission of HIV, but were unable to provide estimates of the proportion of migrants acquiring their HIV post-migration. The Netherlands provided grey literature and peer reviewed papers (subsequently included in the literature review) showing within-country sexual transmission among migrant communities. The most specific data came from Norway (14 % of 152 migrants diagnosed with HIV in 2011 were thought to have acquired their HIV infection post-migration) and the UK, (see Rice et al.[18]).

The remaining member state respondents did not provide data on, or estimates of, sexual transmission of HIV among migrant communities.

Ascertaining probable country of infection

Fifteen countries include a Probable Country of Infection (PCOI) data field in their new diagnoses database. Survey respondents reported that PCOI is established by direct interview of the case or indirectly through a clinician report. A range of data are used to assign PCOI – in particular a combination of clinical (such as CD4 count and viral load at diagnosis) as well patient demographic information (COB and Date of Arrival). Thirteen countries collect enough data to use the CD4-cell based method of calculating PCOI developed by Rice et al. (2012) (Belgium, Denmark, France, Greece, Italy, Lithuania, Luxembourg, Malta, Portugal, Romania, Slovak Republic, Sweden, UK).

Discussion

This study draws together evidence on the extent to which sexual acquisition of HIV is occurring among migrants from countries with a generalised HIV epidemic after they have moved to the EU/EEA. Despite this systematic review of the literature, survey of EU/EEA Member states, and data reported by countries as part of monitoring the Dublin Declaration which all provide some evidence, it remains difficult to gain an accurate picture for the EU/EEA. The published literature is relatively sparse: 27 papers were retrieved from just seven countries. Pan-European studies were also included, but these were low on detail and did not provide specific estimates.

The data do provide evidence of on-going post-migration HIV acquisition, most notably by providing estimates of probable country of acquisition or of incident infection. The methodology for calculating these estimates varied across studies, making it difficult to assess their reliability and comparability. Figures for HIV infections contracted post-migration ranged from as low as 2 % among sub Saharan Africans in Switzerland, to 62 % among black Caribbean MSM in the UK. As this example demonstrates, study populations varied, with some samples based on country of birth, others ethnicity and some using both definitions as surrogate markers for migration. Such heterogeneity in the retrieved data makes it difficult to construct a succinct and clear understanding of on-going sexual acquisition and transmission, both within host countries or across the EU/EEA. Nonetheless, all studies found that post-migration acquisition of HIV is occurring and this must be measured and understood in order to adequately meet the HIV prevention needs of migrant communities.

It is possible that limitations in our review methodology prevented us retrieving relevant information. We were unable to include papers published in languages outside our language skill set. No papers were retrieved from the eastern part of the EU, though this is not unexpected as Eastern Europe does not have a large population of migrants from countries with a generalised HIV epidemic. A survey of EU/EEA Member States combined with data captured through the monitoring of the Dublin Declaration on Partnership to Fight HIV/AIDS in Europe and Central Asia were expedient and cost-effective ways of gathering data about specific countries. Nonetheless, these often relied on the knowledge and experience of one representative from a national body who may not have had all relevant information at their disposal.

Implications for HIV surveillance

Understanding and quantifying HIV transmission in a given population is complex. Effective practice would be to monitor trends in HIV incidence, but these measures are not straightforward and are often beyond the scope of routine surveillance systems. Trends in new HIV diagnoses are sometimes used as a proxy for incident cases, but these are also subject to testing biases as they are dependent on high uptake of regular HIV testing in the population of interest. New tests of recent or incident infections may provide greater insights into real time transmission dynamics. Another challenge is specific to HIV surveillance among migrant communities. Migration as a process is not static, and many migrants travel backwards and forward between their country of origin and country of residence, making estimates of place of infection subject to measurement error. Studies addressing sexual mixing patterns and sexual activity of migrants from countries with generalised epidemics do not necessarily make the best proxies when attempting to find evidence about probable country of HIV acquisition. For example, the increasing proportion of non-B subtypes shown in these papers may reflect on-going migration rather than post-migration HIV acquisition or transmission.

This study found that many more member states collect data that could be used in an objective method for assigning Probable Country of Infection than is expected from the published data. The lack of published evidence could indicate that this is an area requiring technical support to EU/EEA Member States to carry out this work. Additionally, it is possible that some publication bias exists, with authors unable or unwilling to publish data on a sensitive topic that may have a stigmatising impact on migrant communities particularly within the context of Europe’s ongoing immigration debate [7, 30].

EU/EEA Member States might be encouraged to analyse and publish such data if standardised methods of calculating Probable Country of Infection were developed and implemented in surveillance systems across the region. To achieve this a minimum dataset of CD4 cell count at diagnosis, date or year of arrival and country of birth on all newly diagnosed persons are likely to be required. In some settings, sentinel surveillance or repeat cross sectional surveys, could play an important role in providing this necessary evidence. Member States might also be encouraged to publish data if they formed meaningful partnerships with migrant and/or minority ethnic community organisations who could provide support and guidance about how to disseminate data sensitively [31]. Data relevant to post-migration sexual acquisition could then be used to plan and monitor HIV programmes that meet the needs of migrant communities in Europe.

Implications for prevention programming

Given the evidence for post-migration HIV acquisition among migrant groups in some EU/EEA Member States, there is a need for increased awareness among policy-makers of the HIV prevention needs of migrants from countries with generalised epidemics. This awareness will require additional attention and resources to improve primary prevention programmes targeted to the specific (culturally appropriate) needs of various migrant communities.

Many migrants remain sexually active during transit and after they have reached their destination country. Papers examining sexual mixing present evidence that the number and proportion of non-B HIV-1 subtypes are increasing in non-migrants and established minority ethnic communities across Europe [20–22, 32]. The propensity for assortative sexual mixing means that people from countries with generalised epidemics may continue to live in a community with a generalised epidemic even after they have moved to Europe, but be unaware of their HIV prevention needs in their new home.

Migrant MSM appear at particular risk of HIV acquisition post-migration. Behavioural data suggests assortative sexual mixing according to country of origin or ethnicity is not a prevailing feature of sex between men [33] as reflected in the predominance of B subtypes in these communities [32]. Acquisition of viral clades not prevalent in home countries supports post-migration acquisition and (as with heterosexual men) this highlights the need for primary prevention programs targeting these communities. Effective HIV prevention interventions would need to recognise that many MSM from countries with a generalised epidemic may not self-identify as gay men or disclose their sexual identity [34]. Surveillance data does not routinely report on migrant MSM and as such are unable to inform prevention programmes for this population.

Implications for policy

Countries who identify migrants as an important part of their country’s HIV epidemic should consider developing an evidence-based, long-term policy to introduce prevention programmes that reduce HIV-acquisition in these groups. To reduce acquisition countries would need to include policies around structural, behavioural and biomedical prevention interventions that are targeted to all communities, including migrants from countries with generalised epidemics.

Approximately half of the EU/EEA countries surveyed report that they do not provide ART to irregular/undocumented migrants, that is, to persons that cannot legally reside in the country [35]. Treatment as a means of reducing sexual transmission of HIV now forms a key part of the prevention paradigm, like other conditions of paramount public health importance such as tuberculosis [36]. However, policy responses such as mandatory screening contravene the WHO policy framework for HIV testing in Europe which states mandatory HIV testing for migrants and asylum seekers upon arrival violates basic rights and ethical principles and cannot be justified on public health grounds [37]. Improving access to HIV treatment for all infected persons, regardless of their administrative and or immigration status, could positively impact on reducing incident infections both within and beyond migrant communities. This would necessitate addressing the already identified barriers to HIV prevention, testing and care that exist for migrant communities [31]. Failure to ensure access to HIV treatment for all persons in need could prove detrimental to efforts to ameliorate the HIV epidemic.

Conclusion

There is limited published evidence about the acquisition of HIV within Europe among migrants from countries with a generalised epidemic. Individuals are certainly contracting HIV through sexual contact after they have moved to Europe and MSM appear to be at particular risk of post-migration HIV acquisition, yet this is rarely acknowledged within the literature. Only a few countries collect and publish data to enable robust estimates to quantify or monitor the place of HIV infection, which may have a detrimental impact on HIV prevention interventions.

The majority of countries that identify migrants as an important at risk population for HIV infection have put in place measures to estimate the distribution of HIV in these groups. Despite the many areas of concordance and agreement in Member States surveillance systems there remain a number of gaps in the processing and availability of this data. This therefore limits the ability of policymakers and programme managers who wish to have an impact on HIV incidence at country and regional levels.