Effects of region of birth , educational level and age on late presentation among men who have sex with men newly diagnosed with HIV in a network of STI / HIV counselling and testing clinics in Spain

A Diaz (adiaz@isciii.es)1,2, J del Romero3, C Rodriguez3, I Alastrue4, J Belda5, F J Bru6, M M Cámara7, M L Junquera8, I Sanz9, L J Viloria10, L Gil11, E Martínez12, F Gual13, M C Landa14, I Pueyo15, J M Ureña16, B Martínez17, J A Varela18, A Polo19, M A Azpiri20, M Diez1,2,21, for the EPI-VIH Study Group22 1. Área de Vigilancia del VIH y comportamientos de riesgo, Centro Nacional de Epidemiología, Instituto de Salud Carlos III, Madrid, Spain 2. Centro de Investigación Biomédica en Red de Epidemiología y Salud Pública (CIBERESP), Madrid, Spain 3. Centro Sanitario Sandoval, Madrid, Spain 4. Unidad de prevención de VIH e infecciones de transmisión sexual (UPS e ITS), Valencia, Spain 5. Unidad de prevención de VIH e infecciones de transmisión sexual (UPS e ITS), Alicante, Spain 6. Prevención ITS/SIDA, Instituto de Salud Pública, Ayuntamiento de Madrid, Madrid, Spain 7. Consulta de ETS, Servicio de enfermedades infecciosas, Hospital de Basurto, Bilbao, Spain 8. Consulta de ETS, Hospital Monte Naranco, Oviedo, Spain 9. Consulta de ETS. Plan de sida/ITS del País Vasco, San Sebastián, Spain 10. Sección de Vigilancia Epidemiológica, D.G. de Salud Pública, Cantabria, Spain 11. Centro de atención a las ITS (CAITS), Palma de Mallorca, Spain 12. Sección de Vigilancia Epidemiológica y Control de Enfermedades Transmisibles, D.G de Salud Pública y Consumo, Consejería de Salud y Servicios Sociales, La Rioja, Spain 13. Servicio de Promoción y Educación para la Salud, DG Salud Pública, Murcia, Spain 14. COFES Iturrama, Navarra, Spain 15. Centro de ITS, Hospital Duque del Infantado, Sevilla, Spain 16. Centro de ITS y Orientación Sexual, Granada, Spain 17. Unidad de promoción y apoyo a la salud (UPAS), Malaga, Spain 18. Centro de ITS, Gijon, Spain 19. Unidad de prevención de VIH e infecciones de transmisión sexual (UPS e ITS), Castellón, Spain 20. Consulta de VIH, Ambulatorio Olaguibel, Comarca ArabaOsakidetza, Vitoria, Spain 21. Plan Nacional sobre el Sida, S.G. de Promoción de la Salud y Epidemiología, Ministerio de Sanidad, Servicios Sociales e Igualdad, Madrid, Spain 22. The members of the EPI-VIH Study Group are listed at the end of the article

This paper analyses late presentation (LP) of HIV infection, and its determinants, among men who have sex with men (MSM) in Spain, newly diagnosed with HIV (2003-2011) in 15 sexually transmitted infection/HIV counselling and testing clinics.LP was defined as <350 CD4 cells/µL or AIDS.In total, 3,081 MSM were included (2,499 having CD4/AIDS); overall LP was 25.3%.LP was higher in men older than 34 years, those not previously HIV-tested (adjusted odds ratio (aOR):3.1;95% confidence intervals (CI):2.3-4.2) , and those tested > 12 months before diagnosis (12-24 months (aOR:1.4;95% CI:1.0-2.0);> 24 months (aOR:2.2;95% CI:1.7-3.0)).LP was less likely in MSM reporting a known HIV-infected partner as infection source or symptoms compatible with acute retroviral syndrome.'Region of birth' interacted with 'educational level' and 'steady partner as infection source': only African and Latin-American MSM with low educational level were more likely to present late; Latin-American men

Introduction
Delayed diagnosis and treatment of HIV infection is a huge problem worldwide, with important individual and public health consequences.People presenting with an impaired immune system at diagnosis have higher rates of morbidity and mortality than those diagnosed earlier [1,2], and treating them is more costly [3].Moreover, HIV-infected people unaware of their status may inadvertently spread HIV [4].
Different definitions have been used for late presentation of new HIV diagnoses [5], most of them based on different CD4 count levels at HIV diagnosis and/or simultaneous or recent diagnosis with acquired immunodeficiency syndrome (AIDS) [6][7][8].To facilitate data comparisons, a consensus definition was proposed in Europe in 2010 and 2011 defining advanced disease (AD) as presenting a CD4 count below 200 cells/µL or AIDS at diagnosis, and late presentation (LP) as having less than 350/µL or AIDS [8,9].
In the European Union/European Economic Area (EU/ EEA) countries, nearly half (49.3%) of new HIV diagnoses reported in 2012 were late presenters (defined as fewer than 350 CD4 cells/µL at diagnosis), with a range of 35-66% across countries [10].The proportion of LP was higher among heterosexuals (59.1%) and people who inject drugs (PWID) (55.9%) than among men who have sex with men (MSM), (38.4%).In Spain, using the same definition, 48% of new HIV diagnoses reported to the national surveillance system that same year presented late; variations across exposure categories are similar, with MSM having the lowest proportion of LP in comparison to PWID and heterosexuals (39%, 59% and 65% respectively) [11].
In addition to the surveillance system, data on LP are available in Spain from the CoRIS cohort and the EPI-VIH Network.CoRIS is a Spanish cohort of treatmentnaïve HIV patients which collects epidemiological/ clinical data in a broad setting [12]; LP in this cohort was 48.6% in the period 2004-2006 [2].The EPI-VIH Network includes all sexually transmitted infections (STI) and HIV counselling and testing (HCT) clinics operating in the main Spanish cities; these are low threshold public facilities attending all key populations at higher risk for HIV [13,14].Between 2003 and 2010, the proportion of LP in new HIV diagnoses in this network was 27.6% [15], significantly lower than what was found in the comprehensive surveillance system [11].
Both in Spain (cohort and surveillance data) and elsewhere, several socio-demographic and epidemiological factors have been associated with LP, such as male sex, age, migration, low socio-economic status and HIV transmission mode [2,5,16¬-18].Also, all the Spanish studies have showed that LP is less common in MSM than in other transmission categories, even after adjusting for other variables.However, little is known about factors affecting LP within this group.The objective of this paper is to analyse LP of HIV infection and its determinants among MSM newly diagnosed with HIV from 2003 to 2011 in the EPI-VIH Network.LP was defined as having a CD4 count below 350 cells/ µL in the first determination after HIV diagnosis and/ or AIDS at diagnosis, following recent European recommendations [8,9].The analyses were dealt with at the level of 'country/region of birth'; to assign 'region of birth', the World Health Organization Regional Office for Europe's classification was used.Frequency distributions for each variable and prevalence of LP, overall and stratified by different variables, were calculated.To evaluate the association between categorical variables, chi-squared tests were used; the associations between LP and individual explanatory variables were considered statistically significant at a p value <0.05.Trends in LP over time were analysed using a joinpoint regression model.To identify factors associated with  LP, logistic regression models were fitted using a backward elimination procedure.Associations were measured using the odds ratio (OR) and its 95% confidence interval (CI).Data analyses were performed using the STATA statistical software package Version 13 (Stata Corporation, College Station, TX, US).

Methods
The EPI-VIH Network is an HIV sentinel surveillance system, and the database was registered in the Spanish Data Protection Agency (registry number 2080910068).
No personal identifiers were collected.

Results
A total of 3,081 MSM newly diagnosed with HIV were identified during the study period.Of these  Model adjusted by all variables shown in Table 2 plus clinic of diagnosis.

Interaction between region of birth and educational level
Interaction between region of birth and steady partner as probable source of infection the definition of LP (593 cases had fewer than 350 CD4 cells/µL, 5 presented AIDS at diagnosis and 33 had both).The proportion of LP increased with age, and was higher in men with a low educational level (28.5%), in MSM born in Africa (47.1%) or Latin America (countries of the American continent where Spanish or Portuguese is the main national language) (28.8%), and among those without a previous HIV-negative test (36.1%).Conversely, the prevalence of LP was lower in men attributing their infection to intercourse with a known HIV-infected partner (19%) and those reporting an acute retroviral syndrome (15.2%) (Table 1).No trend was found in the prevalence of LP during the study period (Figure 1).
In ).Factors inversely associated with LP were reporting sexual contact with a known HIV-infected partner as the source of infection (aOR: 0.7; 95% CI: 0.5-0.9)or symptoms compatible with an acute retroviral syndrome (aOR: 0.5; 95% CI:0.4-0.7).Region of birth presented interactions with 'educational level' and with 'probable source of infection: steady partner': MSM born in Africa or Latin America, with low educational level (but not those with high educational level) had higher odds of presenting late, although, for Africans, results were on the edge of significance and confidence intervals were very wide due to the small sample size.Latin-American MSM attributing their infection to their steady partner (but not any other subgroup) were also more likely to present late (Table 2, Figure 2).

Discussion
This paper presents data on LP among MSM newly diagnosed with HIV in the EPI-VIH Network in Spain.Among MSM diagnosed in this network, LP is less common than in those diagnosed elsewhere, but presenting late is not evenly distributed, and the effect of region of birth on LP varies depending on the levels of two other variables.
Reducing diagnostic delay is a policy priority in Spain [20], and HIV testing is free of charge in all public facilities.Since 2009, HIV testing at least once a year has been recommended for MSM, and HIV testing guidelines that reinforce the importance of timely testing in this population have been issued recently [21].In order to increase HIV testing availability and to facilitate anonymous testing, several regions have implemented testing in pharmacies [22] and others have made available rapid HIV tests in STI clinics and primary health care centres [23,24].Testing programmes implemented by nongovernmental organisations are also playing an important role [25,26].
The proportion of LP among MSM found in this study (25.3%) is lower than what has been reported in this group in other Spanish settings: in a study performed in Barcelona from 2001 to 2009 the proportion was 47.7% [17], and in another analysis of data from 11 autonomous regions during the period 2007-2011, the figure was 39.1% [18], although in this case the definition of LP did not include AIDS.The prevalence in our study was also lower than the 34% reported in the United Kingdom in 2012 [27].This finding is not surprising since the main purpose of the clinics belonging to the EPI-VIH Network is to be highly accessible to people with a high perceived risk for HIV, irrespective of their circumstances [15].Also, MSM attending these clinics are probably very much aware of the importance of frequent testing: almost 80% of the participants in this study reported previous testing, while in the Spanish sample of the European MSM Internet Survey (EMIS-Spain) the proportion of men ever tested was 74% [28]; furthermore, our clinics are located in the main cities, and EMIS-Spain showed that MSM living in big cities were more likely to have been tested for HIV [28].
Participants older than 34 years were found to be more likely to have LP, and the risk increased with increasing age.This finding is frequent in studies analysing LP [16,18,29], and is consistent with results from a study conducted in England, Wales, and Northern Ireland in 2007, where MSM over 50 years of age were almost three times more likely to have a CD4 count of less than 350 cells/µL at HIV diagnosis [30].
Low educational level [2] and migrant status [17,18] have been described as predictors of LP in Spain, and poor education and little knowledge about HIV were also associated with being untested in an online Norwegian sample of MSM [31].In our study, there was an interaction between education and migrant status, so that Latin-American and African MSM with low educational level (but not those from the same regions with high educational level or men from other regions) were more likely to present late.These results might reflect factors, such as lack of knowledge about HIV infection, lack of access to HIV diagnostics in their country of birth or lack of knowledge about HIV testing policies/ facilities in Spain, operating mainly in less educated migrants.In EMIS-Spain, MSM with low to middle educational level or lesser knowledge about HIV/STI, and those who were not confident about accessing HIV testing were more likely never to have been tested for HIV although, surprisingly, Spaniards were less likely than migrants to have been tested [28]; the likely explanation for this finding is that, in this particular study, most participating migrants were Latin-American with better educational level than their Spanish counterparts.
The fact that Latin-American MSM who attribute their HIV infection to their steady partners are at greater risk of LP warrants further investigation and suggests that emotional factors have to be considered when analysing LP.In any case, Latin-American MSM living in Spain appear to be highly vulnerable to HIV.They are over-represented in new HIV diagnoses [11] and showed higher levels of risky behaviours in EMIS-Spain, despite being highly educated and very knowledgeable about HIV [32].
It seems logical that men experiencing symptoms compatible with an acute retroviral syndrome and those reporting a known HIV-infected partner as their probable source of infection would have sought HIV testing quickly and therefore be less likely to present late.It is less obvious why not having a health card had no effect on LP, but this is not so surprising in our setting because this card is not required to be tested for HIV.LP was inversely associated with repeat HIV testing, a finding also reported in Danish MSM [33].Our results even show an upward gradient of risk for LP as the time lag between the previous negative HIV test and the diagnostic test increased, underlining the importance for MSM to follow the recommendation of testing at least yearly.
This study has some limitations.Most importantly, MSM attending the EPI-VIH Network are not representative of the Spanish MSM population, thus our results cannot be extrapolated to all MSM in the country; unfortunately, educational level and probable source of infection are not collected in the regular surveillance system, thus preventing replication of the same analysis with these data.Furthermore, many persons were involved in data collection, thus increasing the probability of introducing mistakes; nevertheless, the EPI-VIH Network has been operating for many years, participating clinicians are very experienced, and a standard questionnaire is used to collect data.Finally, some degree of misclassification might exist if someone newly infected with HIV and presenting a low CD4 count was classified as late presenter.
On the other hand, we believe that results from this study are important to define effective interventions to increase HIV testing in the MSM subgroup that is probably at highest risk of infection.The need to test for HIV at least yearly should be further disseminated among the MSM community, and efforts should be made to increase awareness about symptoms suggestive of an acute antiretroviral syndrome.In addition, measures aiming at improving early diagnosis in poorly educated Latin-American and African migrants are a priority; in these MSM subgroups better knowledge about their situation is also needed to determine the best way to assist them.

Figure 1
Figure 1Prevalence of late presentation among men who have sex with men newly diagnosed with HIV, by year of diagnosis, EPI-VIH Network,Spain, 2003Spain,  -2011 (n=2,499)    (n=2,499)

Table 1
Characteristics of men who have sex with men newly diagnosed with HIV, and prevalence of late and not late presentation in different subgroups, EPI-VIH Network,Spain, 2003Spain,  -2011 (n=2,499)    (n=2,499) MSM: men who have sex with men.

Table 2
Factors associated with late presentation among men who have sex with men newly diagnosed with HIV, univariate/ multivariate analysis, EPI-VIH Network,Spain, 2003Spain,  -2011 (n=2,499)    (n=2,499) majority were born in Spain (64.7%), were in the 25-34 year age group (49.3%), and had secondary/university education (77.4%).Almost one in six reported symptoms compatible with an acute retroviral syndrome, and 78.7% had been tested for HIV previously (Table1).Median CD4 at presentation was 497 (interquartile range: 349-690).Overall, 631 MSM (25.3%) met aOR: adjusted odds ratio; CI: confidence interval; MSM: men who have sex with men; n.a.: not applicable; OR: odds ratio.Reference categories in brackets.Model adjusted by clinic of diagnosis.aForconvenience, adjusted odds ratio for interactions between these variables are shown in Figure2.