Skip to main content

Table 3 Outcomes and limitations of included studies

From: A systematic review of post-migration acquisition of HIV among migrants from countries with generalised HIV epidemics living in Europe: mplications for effectively managing HIV prevention programmes and policy

Study reference Results/Outcomes Quality scores Limitations
Aggarwal et al. (2006) [32] Distribution of non-B subtypes: Black African 149/154; Black Caribbeans13/42 • Country of infection B subtypes: Black African 3 UK (n = 5); Black Caribbean 13 UK, 5 Caribbean, 11 undetermined (n = 29) • Country of infection non-B subtypes: Black African 98 before migration, 14 UK, 34 undetermined (n = 149); Black Caribbean: 13/13 UK •Overall infected in the UK: Black African 17/154 Black Caribbean: 26/42. SD: +++ V: +++ G: +++ OS: +++ Poor test specificity differentiating between subtypes B and D likely to be a significant factor in limiting the use of serotyping among black Africans. No standardized prospective data collection. Designation of likely country of infection based, on poorly documented variables from medical records (possible reporting bias). Findings may represent an underestimate of number of infections acquired through overseas travel.
Burns et al. (2009) [38] Country of Acquisition: 61(23.2 %) “Definitely acquired HIV abroad”; 44 (16.7 %) “Probably abroad”; 16 (6.1 %), “Definitely acquired in the UK”; 142 (54.4 %) ‘Indeterminate cases’. All cases (determinate and indeterminate): UK acquired: 25.1 % - 35.4 %, Acquired abroad 60.8 % - 67.3 %. SD: +++ V: +++ G: +++ OS: +++ Acquisition of HIV in UK negatively associated with late presentation therefore findings may underestimate infection acquired in UK. Sample only includes Africans living in London, therefore may not be representative to all HIV positive Africans in the UK.
del Amo et al. (2011) [33] Most studies among sub-Saharan African migrants report infections acquired in the country of origin; includes studies in Denmark, Spain, UK & Canada. Some evidence of post-migration HIV acquisition in EU countries (includes Latin American MSM & other migrant populations); evidence of acquisition during return visits to country of origin. SD: +++ V: +++ G: +++ OS: +++ The search strategy includes only articles in English; research in other languages, the grey literature and conference abstracts not included.
Dougan et al. (2005) [39] New diagnosis: probable country of infection reported for 38 % of BME MSM. Born outside and infected in the UK: 38 % of black African (BA), 27 % of black Caribbean (BC) • Born and infected outside the UK: 50 % of BA, 37 % of BC • Undiagnosed prevalence % (CI): All BME 4.3 (4.1-4.5), BC 4.6 (3.3-6.2), BC 15.8 (11.7-20.8). SD: +++ V: +++ G: +++ OS: ++ Incomplete data in some variables (country of birth and infection in the new diagnosis study). Heterogeneous population compared for each of the outcomes.
Dougan et al. (2004) [40] Heterosexual men infected in UK: 37 (country of birth unknown: 24) • Heterosexual women infected in UK 66 (country of birth unknown: 31) • MSM infected in the UK 48 (Country of birth unknown: 24). SD: +++ V: +++ G: +++ OS: +++ Heterosexual transmission of HIV among Caribbeans within UK likely to be underestimated. If exposure to HIV has occurred in more than one country, the country with the highest prevalence will be assigned the likely country of infection. Missing data about country of birth may have had an impact on the review.
Dougan et al. (2005) [15] Probable country of infection MSM born in Africa: Infected in Africa =46.4 %; UK = 45.5 %; Other = 8.2 % • MSM born in Caribbean: Infected in Caribbean = 50.0 %; UK = 42.6 %; Other: 7.4 % • MSM born in Asia: Infected in Asia = 30.6 %; UK = 61.2 %;Other = 8.2 %. SD: +++ V: ++ G: +++ OS: ++ Country of Birth unknown for almost 50 % of sample; results may underestimate the number of diagnosis among MSM born abroad & proportion of MSM infected abroad because of clinician and patient reporting bias. Unclear whether permanent migrants or visitors.
Hamers & Downs (2004) [41] Most HIV infections diagnosed in migrants probably acquired in country of origin. UK: 75 % of heterosexual infections diagnosed in 2002 probably acquired in Africa • Germany: new diagnoses increased in 2002 among heterosexuals from countries with generalised epidemics, majority infected in their countries of origin • Sweden: > 80 % of HIV infections acquired through heterosexual contact were probably acquired abroad • Denmark: 37 % of all diagnoses in 2002 were among migrants; 59 % infected through heterosexual contact, most infected abroad. • Belgium: 4016/5515 of infections ever diagnosed in heterosexuals were in non-Belgians—mostly Africans. SD: +++ V: ++ G: +++ OS: +++ Based on secondary data. No clear that reviewed literature was quality assessed.
Lot et al. (2004) [14] Patients infected through heterosexual transmission: 690 patients (47 % from SSA). No data on the nationality or ethnicity of MSM • Proportion of recent infections among heterosexuals: SSA 26 % vs France 44 % (p = 0.0001). SD: +++ V: +++ G: +++ OS: +++ Based on preliminary data. Late reporting and longer follow-up periods could show larger differences in recent infections. Additionally, the authors do not report on the ethnicity of patients just country of origin. No data on the nationality of MSM or IDUs.
Pezzoli et al. (2009) [12] HIV-1 detected in 0.97 of participants (95 % CI 0.90–1.2) • Avidity Testing (n = 27) Six (22.2 %) probably acquired in Italy by migrants from: SSA (n = 3), eastern Europe (n = 2), and Latin America (n = 1). All 4 (14.8 %) who acquired infection before migration were migrant SSA. SD: +++ V: +++ G: +++ OS: ++ Recruitment was not evenly balanced between centres; the study acceptance rate was 73.6 %. Place of infection could not be determined for 17 (63.0 %) of 27 persons (this is presumably due to recall bias). Very small sample size for avidity testing.
Rice et al. (2012) [18] Probable place of infection: 33 % (26 %-39 %) acquired HIV while living in the UK • Percentage increased from 24 % (16 %-39 %) in 2004 to 46 % (31 %-50 %) in 2010 (p < 0.01). SD: +++ V: +++ G: +++ OS: +++ CD4 cell method may over estimate UK as place of infection since the longer a person is in the UK, the more likely they are to have been assigned UK as place of infection, despite travel habits and behaviour. Missing data for approximately 40 % of eligible adults. Unclear robustness of model used to calculate date of infection.
Rice et al. (2014) [19] Probable country of infection: The percentage of BA heterosexual adults probably acquiring HIV in the UK increased from 9.1 % (276/3019) in 2002 to 37 % (444/1202) in 2011 (P < 0.01). SD: +++ V: +++ G: +++ OS: +++ Definition of heterosexual is based on probable route of infection as reported by clinics, and there is potential for misclassification. (See above for limitations in assigning country of infection).
Semaille et al. (2008) [13] Proportion recent infections: French heterosexuals 27 %, SSA heterosexuals 8.4 %. OR for French (Ref.SSA): 3.95 (3.36-4.64). SD: +++ V: +++ G: +++ OS: ++ Difficult of interpret available data.
Sinka K et al. (2003) [4] Probable Country of Infection for black African and black other (n = 7741): UK or Rest of Europe 523/ (6.76 %); Africa: 6163 (79.6 %). SD: +++ V: ++ G: +++ OS: ++ Limitations of surveillance data due to missing variables, particularly country of birth, ethnicity and country of acquisition. Heterosexual transmission underestimated due to how this data is recorded.
Staehelin et al. (2004) [42] Infection pre- migration (“with great certainty “or “presumably”) SSA: 78 (86.5 %) SEA: 13 (50.5 %) • Infection post-migration SSA: 2 (2.2 %) SEA: 6 (25 %). SD: +++ V: +++ G: +++ OS: +++ Source of infection not analysed because of poor availability of data. Sample size of SSA: only 92 patients. The robustness of the methodology for “Time of HIV-infection and migration” was not discussed directly; authors cite evidence there is no difference in the natural history of HIV infection in patients of differing ethnicity.
Valin, et al. (2004) [43] Probable country of infection: 44 % SSA, 29 % France, 27 % unknown country. SD: ++ V: ++ G: +++ OS: ++ Proportion of patients who arrived in France after 1999 (34 %) overestimated. Study population includes naturalized citizens; is not representative of the entire the HIV-positive population originating from sub-Saharan Africa and living in France. Some questionnaire items may be subject to reporting bias.
von Wyl V et al. (2011)* [20] Proportion of non-B subtype viruses: Increased from 22 % in 1996 to 33 % in 2009 • Over 80 % of all non-B infections among Africans may have originated outside of Switzerland: 20 % of all sequences from this group were contained within Swiss-specific clusters. SD: +++ V: +++ G: +++ OS: +++ Sampling bias (substantially alleviated by the high representativeness of the SHCS), linkage between individuals can never be established with absolute certainty.
Xiridou et al. (2010) [16] “New” Infections (Estimated 1.50 new infections/100,000 people/year): 53 % of new infections among migrant Africans (32 % acquired in The Netherlands), 26 % among Caribbean Migrants (18 % acquired in the Netherlands). SD: +++ V: +++ G: +++ OS:++ Data used in model taken from different studies, therefore difficulty to assess research quality.
Xiridou et al. (2011) [17] Incidence of HIV among heterosexuals:1.50 new infections per 100,000 individuals per year in 2010 (infections occurring as a result of sexual contacts in The Netherlands or during trips of migrants to their home country).• Sub-group analysis 67.18 new infections/100,000 African migrants, 12.12 /100,000 Caribbean migrants, 0.47/100,000 Dutch local. SD: +++ V: +++ G:N/A OS: ++ Model does not take into account differences between 1st and 2nd generation migrants.
Elford et al. (2007) [26] Assortative Mixing: 80 % of BA heterosexual men and women reported sexual partners were also BA. SD: +++ V: +++ G: +++ OS: +++ High-risk sexual behaviours may be underreported because of social desirability bias or because of the associated stigma. Selection bias from response rate; not broadly representative of those living with HIV as sample exclusively from London.
Holguin et al. (2007) [21] Prevalence of HIV-1 non-B subtypes and recombinants 40 (28.8 %) samples, Migrants: 28 (53 % of all migrants in study - 75 % acquired their infection through sexual contact with people born in African) Native Spaniards: 12 (13.7 % of all native Spaniards in the study - 4 most likely acquired HIV-1 through unprotected sex in sub Saharan Africa; 3 with Africans residing in Spain; 2 with partners from Spain; 2 sexual contact with sex workers and 1 MSM with multiple partners). SD: + V: ++ G: + OS: ++ Number of non-B subtypes among newly diagnosed native individuals is biased and could be underestimated. Subtyping of a large number of samples would be required to determine if the incidence of HIV- 1 non-B variants is increasing over time in the newly diagnosed native population.
Kramer (2008) [29] Sexual mixing (sexual partner with differing ethnicity) High risk = 42 % (84 % unprotected), Moderate risk = 59 % (no data), Low risk = 66 % (no data). SD: +++ V: ++ G: ++ OS: ++ Convenience sample and social desirability bias Includes both first generation and second generation migrant with no distinction drawn between them in analysis.
Lai (2013) [24] Sexual mixing: 50 % of men and 47 % of women reported partners born in different countries. Most partners from a different African country (men 19 %; women 20 %). SD: + V: ++ G: + OS: ++ Convenience sample; low response rate (14 %); desirability bias; data does not support some conclusions reached in the discussion.
Marsicano et al. (2013) [28] Factors associated with epidemiological networks: Country of origin independently associated with the probability of isolates being detected in clusters OR for Italian vs. African origin: 5.3, 95 % CI: 2.2–12.9, P < 0.001; South American vs. African origin: 25.6, 95 % CI: 2.0–162.0, P < 0.001. SD: +++ V: ++ G: ++ OS: +++ ARCA database has relative lack of country of origin and risk factor information for some patients which could have weakened the strength of the detected associations. Clusters were probably underestimated and incomplete due to missing data.
Rivas (2013) [25] Proportion of B subtypes: Total 4 (3.3 %); Migrants from Equatorial Guinea 2(2.9 %); sub Saharan Africa; 1 (5.6 %) P = 0.47. SD: ++ V:++ G:++ OS:++ Sample disproportionately represented by women and elderly people so might not reflect wider Equatorial Guinea community. Poor justification for some conclusions e.g. low CD4 cell counts = imported infections.
Snoeck et al. (2002) [22] Country of Infection: 45 % Africa, 2 % South-America, 6 % rest of Europe or USA. • Origin of the virus (P = 0.0004): Belgium (19) Subtype B = 16; Non-B = 3; Other (22) Subtype B = 3; Non-B = 19 • No association between nationality and subtype (P = 0.06). SD: +++ V: ++ G: ++ OS: ++ Small sample size. Disproportionate numbers of female non-Belgians than male non-Belgians in the study population may have introduced a bias.
Tramutoet al (2013) [23] Proportion of non-B subtypes:, 107 (69.0 %) were infected with B strains, whereas non-B subtypes were detected in 48 subjects (31.0 %).Only 9.7 % (n = 11/113) of Italian-born subjects were infected with non-B HIV-1 variants. 3 (7.9 %) Africans were infected with B subtypes. SD: +++ V: +++ G: ++ OS: +++ Data does not support some of the conclusions. Authors do not acknowledge limitations of surveillance data.
van Veen et al. (2009) [27] Sexual mixing: Partners from the same ethnicity 59 %, Partners with differing ethnicity 41 % (15 % with Dutch partners; 21 % with partners of “Other” ethnicity; 5 % with both Dutch and “Other”). SD: +++ V: +++ G: ++ OS: +++ Desirability bias; convenience sample; auto-selection bias.
  1. BA = Black African, BC = Black Caribbean, SSA = Sub Saharan African, SEP = Socioeconomic Position, MSM = Men who have sex with men SD = Study Design, V = Validity, G = Generalisability, OS = Overall Score, N/A = Not applicable