Trends in HIV testing , prevalence among first-time testers , and incidence in most-at-risk populations in Spain : the EPI-VIH Study , 2000 to 2009

M Diez (mdiez@isciii.es)1,2,3, M J Bleda4, J A Varela5, J R Ordoñana6, M A Azpiri7, M Vall8, C Santos9, L J Viloria10, C de Armas11, J M Ureña12, J Trullén13, I Pueyo14, B Martínez15, T Puerta16, M Vera16, I Sanz17, M L Junquera18, M C Landa19, E Martínez20, M M Cámara21, J Belda22, F J Bru23, A Diaz2,3, for the EPI-VIH Study Group24 1. Plan Nacional sobre el sida, Ministerio de Sanidad, Servicios Sociales e Igualdad, Madrid, Spain 2. Área de Vigilancia Epidemiológica del VIH/sida y Comportamientos de riesgo, Centro Nacional de Epidemiología, Instituto de Salud Carlos III, Madrid, Spain 3. Centro de Investigación Biomédica en Red de Epidemiología y Salud Pública (CIBERESP) 4. Consejo Superior de Investigaciones Científicas (CSIC), Instituto de Química Avanzada de Cataluña (IQAC), Barcelona, Spain 5. Centro ETS, Gijón, Spain 6. Unidad de Prevención y Educación Sanitaria sobre Sida, Murcia, Spain 7. Ambulatorio Olaguibel, Comarca ArabaOsakidetza, Vitoria, Spain 8. Unidad de ITS . Hospital Universitario Vall d’Hebron, Barcelona, Spain 9. Centro de Información y Prevención del Sida, Valencia, Spain 10. Sección de Vigilancia Epidemiológica. Dirección General de Salud Pública, Consejería de Sanidad, Santander, Spain 11. Centro Dermatolológico, Tenerife, Spain 12. Centro de ETS y Orientación Sexual, Granada, Spain 13. Centro de Información y Prevención del Sida, Castellón, Spain 14. Centro ETS, Seville, Spain 15. Unidad de Promoción y Apoyo a la Salud, Málaga, Spain 16. Centro Sanitario Sandoval, Servicio Madrileño de Salud, Madrid, Spain 17. Plan del Sida del País Vasco, San Sebastián, Spain 18. Unidad de ETS, Hospital Monte Naranco, Oviedo, Spain 19. COFES Iturrama, Pamplona, Spain 20. Sección de Vigilancia Epidemiológica y Control de Enfermedades Transmisibles, Dirección General de Salud Pública y Consumo, Logroño, Spain 21. Unidad ETS, Enfermedades Infecciosas, Hospital de Basurto, Bilbao, Spain 22. Centro de Información y Prevención del Sida, Alicante, Spain 23. Programa de Prevención del Sida y ETS, Madrid, Spain 24. Members of the EPI-VIH Study Group are listed at the end of the article


Introduction
During the 1980s and 90s, Spain had the highest AIDS incidence in western Europe (with a peak of 184 cases per million population in 1994), and HIV transmission was attributed in most cases to the use of contaminated material (injection equipment and substance injected) by people who inject drugs (PWID) [1].Widespread dissemination of highly active antiretroviral therapy (HAART) in 1996 produced a sharp decrease in AIDS incidence and mortality [2,3], while harm reduction and other public health programmes resulted in less drug injection [4][5][6].As a result, the number of new HIV diagnoses in PWID decreased, and sexual transmission emerged as the most common transmission category in the country, in particular among men who have sex with men (MSM).Meanwhile, people born abroad, whose presence was barely registered in the Spanish HIV epidemic before 2000, came to play an important part in it [7], as a consequence of the influx of foreignborn persons, whose proportion in the Spanish population increased from 2.9% in 1998 to 14.3% in 2012 [8].Of all foreigners living in Spain in 2012, 40.8% had been born in other European countries, 36.4% in Latin America, 12.6% in northern Africa and 3.7% in Sub-Saharan Africa [8].
As the benefits of HAART became obvious, early diagnosis and treatment of HIV infection became a priority in Spain, reflected in the national multi-annual strategic plans on HIV/AIDS [9].Anyone can be tested for HIV free of charge in primary care centres and other public facilities, and private laboratories also perform the test; free-of-charge antenatal care, including HIV testing, is offered to all pregnant women.Nevertheless, actions aiming at reducing late presentation have increased, especially since 2005: these include information campaigns, opening of HIV counselling and testing (HCT) community centres, and the introduction of rapid tests in community programmes, health centres and pharmacies.
In the framework of second generation surveillance for HIV [10], the systematic collection of prevalence data in key populations is recommended for countries with concentrated epidemics, such as Spain.To this end, in 1999, the so-called EPI-VIH Study was initiated in 10 clinics (EPI-VIH Network) specialising in sexually transmitted infections (STI) or HCT.What distinguish these clinics from other public health facilities offering testing is that they are low-threshold, highly accessible centres that traditionally have attended most-at-risk populations for HIV and other STI.Analysis of retrospective and prospective data collected in the network documented a decrease in HIV prevalence from 1992 to 2002 in all most-at-risk populations, including PWID, MSM, female sex workers (FSW) and high-risk heterosexual individuals (i.e.those with a history of risk behaviours for HIV infection) [11,12].(iii) HIV incidence among all people attending the network clinics.Yearly publications describing both people tested and prevalence results [13], as well as some preliminary incidence analyses from the network [14], have provided useful insight into the HIV situation in Spain and have been used to define health policy.
The objective of this paper is to describe trends in HIV testing, HIV prevalence among first-time testers, and HIV incidence in most-at-risk populations attending the EPI-VIH Network clinics in a 10-year period, from 2000 to 2009.

Methods
The EPI-VIH Network

HIV testing practices
All HIV tests performed in the EPI-VIH Network during the study period were analysed.The testing included pre-and post-test counselling and was performed using an enzyme-linked immunosorbent assay (ELISA) followed by a western blot to confirm positive results.
For each test, people undergoing testing were classified according to whether this was the first time they had ever been tested in one of the network's clinics (first-time tester) or whether they had previously been tested in the same clinic (repeat tester).Anamnestic information about previous tests was not collected because for the incidence analysis (described below), the exact date of all tests performed had to be

Study population
All participants meeting the definition of first-time testers during the study period were included in the prevalence analysis (Figure 1A).
To estimate HIV incidence, an open cohort of people tested two or more times for HIV at one of the clinics belonging to the EPI-VIH Network between 1 January 2000 and 31 December 2009 was identified.People were eligible to enter the cohort if they met the following three criteria: (i) a documented negative result in the first HIV test performed during the study period at one of the network's clinics; (ii) at least one additional documented HIV test in the same clinic; and (iii) less than three years between their last two documented HIV tests (Figure 1B).The three-year time frame was chosen to identify true incident cases and to improve the estimates' precision.
New HIV diagnosis was defined according to the European case definition [15].All FSW in our study had sex only with men whereas male sex workers fell into two categories: those who had sex with men (MSM sex workers) and those who had intercourse only with women.Since the latter were very few (n=242) and their HIV prevalence did not differ from that of other heterosexual men, it was decided to classify them as heterosexual men for all analyses.

Statistical analyses
We calculated the number of tests performed for firsttime and repeat testers each year, stratified by type of most-at-risk populations, as well as the distribution of study participants by the variables of interest.Chi-squared tests were used to compare categorical variables.
HIV prevalence and its 95% confidence interval (CI) was calculated overall and stratified by different variables.A new HIV diagnosis for the prevalence analysis was classified as HIV prevalent when the person tested met the definition of first-time tester.
Only patients meeting the criteria for inclusion in the above-mentioned open cohort were included in the incidence analysis.Incidence rates (IR) and their 95% CIs were calculated overall and stratified by different variables, assuming a Poisson distribution.Although personal identifiers were not used in the study, linkage of different tests performed for the same person in the same clinic was possible through the use of a unique identifying number: this allowed the identification of seroconverters and the calculation of person-years (py) of exposure in the incidence analysis.A new HIV diagnosis for the incidence analysis was classified as HIV incident (seroconverter) if the person tested was eligible to enter the open cohort and their HIV test changed from negative to positive, i.e. they were positive in their last visit to one of the network´s clinics having been HIV negative in their previous visit to the same clinic.Seroconversion was assumed to have taken place at the midpoint between the first HIV-positive visit and the last previous HIV-negative visit.To calculate py of exposure, non-seroconverters contributed the time elapsing between their first and last HIV tests during the study period, while seroconverters contributed the time elapsing between their first HIV test and the estimated date of seroconversion.For annual estimates of HIV incidence, we allocated each person's py to the years that they contributed.
To evaluate trends in the number of tests, HIV prevalence and incidence, joinpoint regression models were fitted [16].

Results
From 2000 to 2009, a total of 236,939 HIV tests were carried out in the EPI-VIH Network for 165,745 individuals (Table 1).Of these, 117,465 were tested for the first time ever in one of the network's clinics during the study period (3,599 with a positive result) and never came back, i.e. they were tested only once; 27,872 were first-time testers at some time between 2000 and 2009: the result of the first test was negative and later they underwent testing again, thus becoming repeat testers; and 20,408 were tested for the first time before the study period with a negative result and were retested again during the study period (these persons were always considered repeat testers).People who were tested once were eligible only for the prevalence analysis; those tested before 2000 were eligible for the incidence analysis if they met the criteria to enter the open cohort.The remaining 27,872 were included in the HIV prevalence analysis the first time they were tested during the study period, and also in the HIV incidence analysis if, later on, they met the criteria to enter the open cohort (Figure 1A and 1B).
In total, 145,337 individuals met the definition of firsttime tester during the study period and were eligible for the prevalence analysis, of whom 3,599 tested positive (Figure 1A).1B).

HIV tests
The

HIV prevalence
In total, 145,337 persons, of whom 3,599 were HIVpositive, were tested for the first time ever in one of the network´s clinics during the study period (Figure 1A).Of the 3,599 found to be positive,  2).These differences in prevalence were statistically significant.
Very high HIV prevalence estimates throughout the study period were found among MSM sex workers (19.0%; range: 10.5-24.5)and PWID (17.0%; range: 13.3-21.2),although in both groups, yearly estimates were rather unstable because the number of MSM sex workers was small and the number of PWID decreased over the study period.Among MSM, HIV prevalence during the study was 7.6%, ranging from 6.4% in 2001 to 9.4% in 2009.Prevalence was much lower in heterosexual individuals (0.9%; range: 0.7-1.1)and FSW (0.8%; range: 0.5-1.2).There were no differences by sex in HIV prevalence among heterosexual individuals, therefore combined estimates are presented.Joinpoint models fit to evaluate trends in HIV prevalence by transmission category showed a decreasing trend in heterosexual individuals and an increasing trend in MSM (Figure 2).
Foreign-born participants had a higher HIV prevalence during the study period than did those who were Spanish born (2.9% (95% CI: 2.7 to 3.0) vs 2.2% (95% CI: 2.1 to 2.3), p<0.05) and the same was true for each region of birth except eastern Europe.An increasing trend in HIV prevalence was observed among Latin American-born participants (p<0.05)(Table 2).

HIV incidence
In The overall HIV IR for the study period was 1.0 seroconversions/100 py (95% CI: 0.9/100 to 1.1/100), with no statistically significant differences by type of clinic.The IR was highest in the country´s biggest cities, Madrid, Barcelona, Valencia and Seville (1.5/100 py (95% CI: 1.4 to 1.7), 1.1/100py (95% CI: 0.8 to 1.5), 1.2/100py (95% CI: 0.9 to 1.4) and 1.0/100 py (95% CI: 0.8 to 1.3) respectively), although Bilbao, which has about the same population as Valencia or Seville, had a much lower IR (0.4/100 py (95% CI: 0.3 to 0.7)).In general, centres located in northern Spain had lower seroconversion rates than those located in the eastern and southern parts of the country.Men and transgender females had higher IR than women (1.8/100 py (95% CI: 1.6 to 1.9), 1.2/100 py (95% CI: 0.5 to 2.8) and 0.1/100 py (95% CI: 0.09 to 0.2) respectively).With respect to age, the peak IR was found in people younger than 20 years, but the sample size was small, with a wide 95% CI.The next highest IR was in the 35-39 year age group (1.2/100 py; 95% CI: 1.0 to 1.5).Spanish-born participants as well as those born elsewhere in Europe had higher seroconversion rates than people born outside Europe (Table 3).
Results by transmission category showed that the highest IR was among MSM sex workers (3.0/100 py), although the 95% CI was quite wide (2.2 to 4.1), followed by MSM who were not sex workers and PWID (2.5/100 py (95% CI: 2.3 to 2.7) and 1.6/100 py (95% CI: 1.1 to 2.2) respectively).Since there were no differences in HIV incidence between male and female heterosexual individuals, combined estimates are presented (Table 3).
Joinpoint models fit to evaluate trends in incidence showed a statistically significant increase in overall HIV incidence over the study period, but stratification by transmission category showed that the increasing trend was present only among MSM (Figure 3).

Discussion
During the study period, the number of HIV tests increased in MSM, heterosexuals and MSM sex workers but not among PWID and FSW.HIV prevalence decreased in heterosexual individuals and increased in MSM, remaining stable in the other transmission categories.HIV incidence was highest among MSM sex workers and showed an increasing trend in MSM.
The study provides seroconversion estimates among migrants after their arrival in Spain.
Early HIV diagnosis and treatment is a policy priority in Spain [9] and the total number of HIV tests performed increased from 24.8 per 1,000 inhabitants in 2002 to 38.2 per 1,000 in 2009 [17].Our results suggest that the increase in testing is not uniform across most-atrisk populations and that testing might actually be decreasing in some groups.The reduced number of tests among PWID seems logical because the PWID population is rapidly decreasing [6], but the decrease observed among FSW warrants further investigation.
There are some data on the size of the FSW population in Spain [18] but none on its trend, although a decrease in the number of FSW that might explain the decrease in testing in this group seems unlikely.A change in the testing facilities preferred by FSW seems equally unlikely, because the clinics in the network are free of charge, highly accessible and have a long tradition of caring for this population.A more likely explanation is worsening access to these clinics in the network due to the increased proportion of foreign FSW -from 46% in 1998 [19] to 93% in 2009 (our study) -and the shift in the working environment from outdoors to indoors (i.e.apartments or clubs), a tendency noted throughout Europe, which makes sex workers less accessible to intervention measures [20].The lack of HIV prevalence data among FSW working only indoors in Spain warrants further investigation.
Decreasing trends in HIV prevalence among all most-atrisk populations were reported from 1992 to 2001 in the 10 centres that initially comprised the network [11,12].Data from the expanded network show that from 2000 to 2009, HIV prevalence remained stable among PWID and male and female sex workers, and increased in MSM, whereas it continued to decrease only in heterosexual individuals.This last group has increased the most in size; thus, declining HIV prevalence could also be a result of testing more people at lower risk.
HIV prevalence in transgender women was higher than that found previously in other Spanish studies [21,22].
In comparison with other countries, our figure (24.5%) is higher than the 21.7% reported in the United States and the same as the 24.5% found in Italy according to a meta-analysis of studies published from 2000 to 2011 [23].The situation of MSM sex workers is of particular concern since they present the highest HIV prevalence and incidence estimates, in clear contrast to the situation in FSW.As there is a dearth of recent studies on MSM sex workers in Spain [21,24], more information is needed on the reasons explaining this situation.The high proportion of foreigners, use of illegal drugs and the particular stigma associated with male sex work could all be important factors, together with the high background HIV prevalence in MSM.
HIV prevalence among MSM attending clinics in the network (7.6%) was higher than that found in MSM in a community centre in Barcelona (5.4%) between 2007 and 2012 [25] but lower than figures for MSM reported in studies carried out in Barcelona in 2002 (18.3%),Bangkok, Thailand, in 2007 (30.8%) or 21 cities in the United States in 2008 (19%) [26][27][28]; nevertheless the downward trend previously observed in MSM in Spain [12] has been reversed, which is not surprising given our findings that HIV incidence is also on the rise in MSM.The overall IR for MSM in our study (2.5/100 py) was lower than that found among American or Italian MSM in the 1990s (2.8/100 py and 4.6/100 py respectively) [29,30] or that estimated in Bangkok during 2003 to 2007 (from 4.1 to 7.7%) [27], but higher than figures reported in London, England, in 1997 to 1998 (1.8/100 py) or Catalonia, Spain, during 1995 to 2001 (1.92/100 py) [31,32].
HIV prevalence among PWID in this study was a third to a half of what it was in the 1980s and 90s in Spain [11,33], although the overall prevalence was very high (17%), only lower than that found among MSM sex workers.During the study period, these figures remained stable.Compared with other studies, the HIV prevalence was lower than the 26.9% self-reported HIV infection among PWID entering drug treatment in 2009 in Spain [34] and similar to the 17.2% found in   [35].Surveillance data show that new HIV diagnoses in PWID in Spain have decreased sharply since 2004 [36], as a result of the decrease in the size of the PWID population and decreased HIV transmission after implementation of harm reduction programmes [4,6]; however, our findings show that seroconversion among current injectors, though much less frequent than in the past [37,38], is still very common.
Latin-American migrants have always been present in Spain, but their numbers increased greatly since the mid-1990s, and since 2000 people from other areas have also been attracted to the country [8].Migration into Spain is therefore a relatively recent phenomenon, so the generally higher HIV prevalence among foreignborn people in our study probably reflects the HIV prevalence in their country of origin, as well as differences in the distribution of HIV transmission categories between those who were Spanish-born and the other groups.However, the diagnostic delay among those newly diagnosed with HIV in Spain is greater among people who are foreign-born [39], in spite of them being entitled to HIV testing and care free of charge; thus, it is possible that they are experiencing difficulties in accessing HIV testing and care due to discrimination or other barriers.
This study has some limitations.People attending the clinics in the EPI-VIH Network are by no means representative of the general Spanish population and the reasons for seeking care in the network are not independent of HIV infection: thus the results cannot be extrapolated to other settings.However, while the estimates presented probably reflect only the experience of the population at greater risk of HIV infection in each transmission category, the clinics have been operating on the same basis for many years, so if selection biases are present, they are unlikely to affect the results on trends.A change in the populations attending the clinics could influence trends, thus multivariate analyses need to be performed to have a better insight into the results.The questionnaires were administered by many individuals, making it difficult to control reproducibility.Testing patterns in the different most-at-risk populations might have evolved over the years in a different way.Finally, for some groups, e.g.transgender individuals, the numbers were very small.
The information obtained with this study shows the changing face of HIV epidemiology in Spain, is of great public health relevance and very useful for adjusting HIV policy to the real needs.In contrast with the 1980s and 90s, when most HIV infections occurred among PWID, MSM have emerged as the population at greatest risk for HIV in Spain and should be given priority in preventive efforts.Nevertheless, although injection of drugs is becoming rare, PWID should not be forgotten since HIV prevalence and incidence in this group remains very high and shows no signs of decreasing.
HIV infection remains a huge problem in transgender women and MSM sex workers that warrants further research in preventive interventions in these groups.With regard to FSW, efforts should be made to guarantee access to HIV testing and prevention, in particular for those working in places not easy to reach.Investigation of the situation of those who are hard to reach is also a priority.In addition, this study illustrates the increasing impact of migrants on the HIV epidemic in Spain; interventions tailored to their needs should therefore be urgently promoted.

Figure 1
Figure 1 Flowchart of inclusion of persons undergoing HIV testing in prevalence and incidence analyses, EPI-VIH Study, Spain, 2000-09

Figure 3
Figure 3 HIV incidence rate a by year of diagnosis and transmission category, EPI-VIH Study, Spain, 2000-09 (n=30,679) From 2000, 10 more clinics joined the EPI-VIH Network and the EPI-VIH Study was expanded.The aim of the new project was to prospectively collect and analyse data on the following: (i) the number of HIV tests carried out in the EPI-VIH Network and characteristics of people requesting testing; (ii) HIV prevalence in mostat-risk populations attending the network clinics; and
test during the study period and therefore were excluded from the open cohort.Of the remaining 35,027 HIV-negative individuals with two or more HIV tests during the study period, 4,348 were excluded because more than three years had passed between their last two tests, leaving 30,679 HIV-negative subjects eligible for the cohort.Over the study period, they totalled 64,104.2py of follow-up and 642 seroconverters were identified among them (Figure There were initially 48,280 HIV-negative repeat-testers potentially eligible to enter the open cohort: the 27,872 and 20,408 mentioned above.Of the latter, 13,253 had only one HIV