Assessment of an outreach street-based HIV rapid testing programme as a strategy to promote early diagnosis : a comparison with two surveillance systems in Spain , 2008 – 2011

M J Belza1,2, J Hoyos (jhoyos@isciii.es)2,3, S Fernández-Balbuena2, A Diaz2,3,4, M J Bravo2,3, L de la Fuente2,3, the Madrid HIV rapid testing group5 1. Escuela Nacional de Sanidad, Instituto de Salud Carlos III, Madrid, Spain 2. CIBER de Epidemiología y Salud Publica, Madrid, Spain 3. Centro Nacional de Epidemiologia, Instituto de Salud Carlos III, Madrid, Spain 4. Ministerio de Sanidad, Servicios Sociales e Igualdad, Spain 5. Members of the group are listed at the end of the article


Introduction
In European countries with HIV epidemics similar to Spain's [1,2], between 20 and 35% of the HIV-infected population remains undiagnosed [3].Of the new diagnoses reported in Spain in 2012, 47.6% had a CD4 count under 350 cells/µL [4].Late presenters have both higher morbidity and mortality [5,6] and higher rates of transmission than those who present early [7], therefore, promoting earlier diagnosis is a top priority to fight the epidemic [8].
In Spain, HIV testing is a non-routine procedure and, until recently, has been performed at all levels of the public health system, confidentially and free of charge, when requested by the patient or when considered necessary by the health provider.However, recent regulatory changes limit its access to migrants with illegal administrative status [9].In some cities, testing is also offered in HIV-sexually transmitted infection (STI) clinics where it is also performed confidentially and free of charge.
Additionally, programmes offering rapid testing in non-clinical settings have proliferated in recent years.They are very heterogeneous in terms of target population, appointment requirement, duration of the counselling provided and type of rapid test used, but most are carried out at the premises of the communitybased organisations (CBO) that run them.Despite their expansion in recent years, their effectiveness in terms of seropositivity rates, linkage to care and capacity for promoting early diagnosis have rarely been externally evaluated, and their outcomes have rarely been compared with clinical settings.
Given the fixed nature of these programmes' venues, they reach people who necessarily perceive themselves at risk of being infected or who have interiorised the routine of testing periodically.Conversely, they might miss people with low risk perception, who do not feel the need to be tested.Offering rapid testing in highly visible locations could promote diagnosis in populations that do not actively seek testing in other venues.
Testing is offered free of charge in a wide range of settings in Spain, and this paper aims to appraise the added value of a multi-site outreach programme offering rapid HIV testing in the street.To do this, we first analyse the characteristics of the population tested and the programme's capacity to reach people and link them to care early in the course of HIV infection and compare results against existing services.Second, we determine whether those diagnosed may constitute a population that is either not seeking HIV testing or testing too infrequently, and discuss to what extent the programme reduces time from infection to diagnosis.

Setting and study period
The programme was run by Madrid Positivo, a nongovernmental organisation and it was conducted during three periods: May 2008-December 2008 (season 1, 62 days in total), July 2009-July 2010 (season 2, 65 days) and November 2010-December 2011 (season 3, 35 days).In all three seasons, a mobile unit was located in Chueca, a busy commercial city-centre neighbourhood of Madrid frequented by young people, with a high proportion of gay residents and a high number of gay businesses (hereafter referred to as 'the gay neighbourhood').It was also deployed in a Madrid neighbourhood with high migrant concentration (hereafter referred to as 'the migrant neighbourhood') (season 3) and in locations outside the city of Madrid with no relation to the gay scene (season 2-3) (additional data available upon request).The programme usually operated in the afternoon, and on certain days throughout the day.The regularity and the days on which the programme was implemented depended on permissions granted by local authorities to deploy the mobile unit in public spaces.

Data collection, rapid test results and linkage to care procedures
Individuals signed an informed consent and entered the mobile unit, where a nurse or doctor completed a brief pre-counselling session, and performed the test (Determine HIV-1/2 test).While waiting for the result, individuals completed an anonymous self-administered paper-based questionnaire code linked to their test (sections used available from authors upon request).The core survey was the same throughout the three seasons and included sociodemographic and behavioural questions (number of sexual partners, condom use, STI history and injecting drug use) and also on HIV-testing history (previous testing experience and time since last test).The questions assessing involvement in gay culture, self-identified sexual orientation, last testing location, main reason for testing today, reason for testing in the programme and future testing intentions were only included during certain periods that will be specified as table footnotes.For those with limited proficiency in Spanish, a form was designed to collect basic socio-demographic, behavioural and HIVtesting history data, in English and French.
In season 1, those with a reactive rapid test were referred to a collaborating STI/HIV diagnostic centre or advised to see their general practitioners.They were asked to give a telephone number in order to obtain their confirmation result and to keep in contact for support during linkage to care.Confirmation results and CD4 count were obtained through direct contact with either the individual or the collaborating diagnostic centre.To shorten the diagnostic process and facilitate linkage to care, from season 2 onwards blood was extracted at the mobile unit and immediately sent to a collaborating STI/HIV diagnostic centre for confirmation.Subjects were then contacted and an appointment set for the result communication and, if positive, the collaborating centres performed a clinical and immunological evaluation for antiretroviral therapy (ART) initiation.We considered as 'linked to care' all the individuals who visited a health centre (mainly the collaborating STI/HIV diagnostic centres) to receive the confirmation result or to ask for a confirmation test.Early diagnosis was defined as having a CD4 count of ≥350 cells/µL.Those who revealed during post-test counselling that they had previously tested positive for HIV were excluded from the analysis.

Data analysis
A descriptive analysis was carried out by stratifying the sample into three groups: women, men who have sex exclusively with women (MSW), and men who have sex with men (MSM).Men included in the latter group were those who reported ever having had sex with men.Using the same stratification, we analysed testing history and other testing-related variables.In the MSM group, a further stratification by serostatus was conducted.Differences between the three groups were assessed using the chi-squared test.
In the analysis of HIV positive individuals, we considered new diagnoses (n=133), those rapid test results with a positive confirmation (n=114) and those for which the confirmation result remains unknown (n=19) (Figure 1).To evaluate the programme's capacity to detect previously undiagnosed HIV infections, we present the distribution of persons tested and the prevalence of infection with its 95% confidence interval (CI) by programme location.In each location we performed the same analysis by transmission category and in the MSM by place of birth.Due to limited sample sizes this analysis could not be conducted in heterosexuals or injecting drug users (IDU).We also estimated the prevalence of infection by transmission category (regardless of programme location) and in MSM, by age, country of birth and educational level.Within HIV-positive individuals we estimated the percentage linked to care, and the proportion diagnosed at a late stage of infection, both globally and among MSM.MSM: Men who have sex with men; MSW: Men who have sex with women.a n=70 were not included because they had never had sexual relations or had ever injected drugs ;n=173 men were excluded because they could not be classified as MSM or MSW.b Did not attend a collaborating centre and could not be contacted directly to obtain confirmation results, however they were considered new diagnoses in the analysis.
and large Spanish cities.They offer voluntary, confidential and free-of-charge HIV testing, and some also offer anonymous testing.No legal documents were required during the study period (2008)(2009)(2010)(2011).This system collects some limited information about people tested including the test result [10].SNHSS collects new HIV diagnoses data reported by 17 of the 19 Spanish autonomous regions (71% of the Spanish population).It is the best approximation of the characteristics of the national epidemic and its evolution [4].Both EPI-VIH and SNHSS information is collected using data collection forms completed by a healthcare professional.When comparing our results with those of both surveillance systems, we took into account available data from the years when the programme was implemented: 2008-2010 in the case of EPI-VIH and 2008-2011 in SNHSS.Given the low number of HIV-positive women and MSW in the programme, the characterisation of HIV-infected individuals was restricted to MSM.The capacity of the programme to reduce periods of undiagnosed infections is gauged by i) analysing whether its good visibility prompted testing in individuals who had not though about it and ii) by analysing testing history and future testing intention of HIV-infected participants.Again, due to sample size limitations these analyses were limited to HIV-positive MSM.The study was approved by the Instituto de Salud Carlos III's institutional review board.

Sociodemographic, behavioural characteristics and sexual orientation
Of the 9,166 people tested we excluded from the analysis 70 individuals who had never had sexual relations or injected drugs and 173 men who did not answer the question on sexual behaviour and could not be classified in either of the two subgroups (Figure 1).Of the   8,923 analysed (Table 1), 34.6% were women, 31.7%MSW and 33.7% MSM.Some 60.7% were tested in Madrid´s gay neighbourhood, 8.9% in Madrid´s migrant neighbourhood and 30.4% outside of Madrid (Table 1).The proportion of MSM was notably higher in the gay neighbourhood than in the other two areas (Figure 2).Fifty two percent were under 30 years of age, 21.4% were born in Latin America (defined as people born in countries of the American continent where Spanish or Portuguese is the main national language), 10.8% in other countries and 47.7% had a college degree.During the previous 12 months, 50.6% of the women and 51% of the MSW had had unprotected sex with heterosexual occasional partners.Unprotected sex with homosexual occasional partners was reported by 36.5% of the MSM.Three per cent reported having ever injected drugs (Table 1).

Testing related information
Twenty six per cent of the MSM had never been tested before (18.4% among new diagnoses); this percentage was higher in women (63.8%) and MSW (61.5%).MSM had the shortest time between previous and current testing: 40.2% of MSM had tested in the previous 12 months (although 26.3% of HIV-positive MSM had last been tested two or more years previously).The most common location for most recent test was primary care (26.8%) and having had unprotected sex with occasional partners (34.1%) was the most common reason for testing that day.Testing as a part of a periodical routine was the second most cited reason (12.8%), mainly because 24% of the MSM reported it.This percentage was 4 times lower in the HIV-positive MSM (6.3%) (Table 2).Some 57.5% of all attendees got tested in the programme because they passed by, saw it and decided to take it.Regarding testing intentions, 22.2% said they probably or certainly would not have been tested in the next 12 months if they had not done so that day.This percentage was lower in MSM (10.4%), and even lower in MSM diagnosed with HIV (3.1%) (Table 2).

Analysis of newly diagnosed individuals: comparison with EPI-VIH and SNHSS
The overall HIV prevalence was  3).

Discussion
The programme described in this paper reached a diverse and under-tested population.However, it diagnosed MSM almost exclusively and presented very low positivity rates when conducted outside of Madrid.We found that the prevalence of infection was half than that of EPI-VIH and that HIV-positive MSM were younger, more frequently from Latin America and had a higher level of education than in EPI-VIH.Compared with SNHSS, MSM diagnosed in the programme were also younger and more frequently from Latin America.The good visibility of the mobile unit led to reduced periods of undiagnosed infection in six out of ten MSM who happened to see it and decided to get tested.The programme was able to reach and diagnose a group of MSM who did not test for HIV with the recommended frequency.Eight out of ten were linked to care and the percentage of late diagnoses was similar to EPI-VIH but half of what was reported in SNHSS.
This study contributes substantially to the scarce body of European literature that analyses community-based testing.It evaluates a programme that is different in terms of its setting and the population it served.
Comparing the programme with other healthcare settings puts the results into perspective, which has rarely been done before.Finally, we evaluate how this programme's good visibility could contribute to the control of the epidemic.
Most of the European-based published studies have evaluated programmes where MSM are the target population [11][12][13][14][15].This is one of the few not specifically focused on them.The good visibility of the programme, and its deployment in settings not related to the gay community may have prompted testing in lower-risk individuals who otherwise would not have thought of it.In fact, two thirds of those who took the test in the mobile unit were either MSW or women, and within these two groups, around six out of 10 had never tested for HIV before.This capacity of the programme to promote HIV testing in populations with no previous testing history has been described elsewhere [16].
The overall prevalence is similar to that found by the handful of studies that evaluate programmes outside clinical settings which do not target vulnerable populations [17][18][19][20].It varied by location and was higher in the city of Madrid.The capacity of the programme to reach undiagnosed individuals was substantially lower than in EPI-VIH, which is consistent with the only study comparing a CBO with diagnostic clinics in the past [11].Almost all of the new diagnoses were in the MSM group which is unsurprising in a country where they are the most vulnerable population for HIV infection.However, their weight among the new diagnoses is much higher than in the national figures represented by SNHSS, and also higher than in EPI-VIH.
Prevalence in heterosexuals was three times lower than in EPI-VIH and very similar to the prevalence found by another study conducted in primary care in Madrid [21].This raises the question: is it appropriate to carry out programmes of this nature in locations frequented by lower risk groups?The question is even more pertinent if we consider that, while cost per diagnosis was not assessed, the only example we found concluded that it was considerably higher in outreach settings than in STI/HIV clinics [22].
To evaluate the programme's potential to prevent onward transmission, we must consider factors other than prevalence of infection.Reducing the time an infection remains undiagnosed as a result of the implementation of the programme is a key factor: the sooner an infection is detected, the sooner the community will benefit from viral load reductions and behavioural change.In this sense, good visibility translated into an earlier diagnosis in more than half of the HIVpositive MSM who were not actively seeking to be tested that day: they happened to see the mobile unit and decided to test.According to testing intentions, the time of undiagnosed infection would be reduced by up to one year: almost all of the HIV-infected individuals reported that they would have sought testing in the next 12 months.According to their testing history, however, the gain would be higher: two out of 10 had never tested before and an additional 25% had tested more than two years ago, which is much longer than recommended for this group [8].It is noteworthy that very few of the HIV-positive MSM reported having tested as a part of a routine check-up.Reaching out to high-risk populations who have not internalised testing as a part of a routine could shorten the time from diagnosis to infection and therefore reduce onward transmission.
The programme also showed a capacity to promote early diagnosis in MSM: only two of 10 had CD4 <350 cells/µL, which represents a remarkable improvement when compared with SNHSS.Likewise, programme attendees were also diagnosed earlier than EPI-VIH patients but differences were much smaller.Evidently, as the percentage of delayed diagnosis drops there is less room for improvement.It is also true that HIVpositive MSM were younger and had a higher level of education than in EPI-VIH and SNHSS.The association of delayed diagnosis with increasing age and, in men, with low educational level has been described previously [23][24][25]26], which means that the programme is reaching subpopulations that are a priori at lower risk of being diagnosed late.On the other hand, in comparison with EPI-VIH and SNHSS, the programme was good at reaching Latin Americans.In Spain, unlike other European countries, they are the largest migrant group, but it is also true that delayed diagnosis in this group is similar to that of the Spanish-born population [4].
Obviously, diagnosis is useless if it is not followed by linkage to the health system for ART eligibility.The linked-to-care percentage was similar to that of the few European studies reporting this outcome [11,12,15] but the definitions used in those studies are not clearly stated and precautions should be taken when comparing results.
The present study has several limitations.First, there is the possibility that some of those with unavailable data on confirmation result did not attend a health centre for this purpose.However, given that access to testing in Spain was universal and performed confidentially in a wide range of settings, we believe that this possibility is minimal.Second, metrics used in 'linkage to care' definitions are heterogeneous.Some are based on clinic visits (as for this study) and some on laboratory monitoring tests.Nevertheless, they always include a time period since diagnosis to either first clinic visit or first laboratory monitoring test [27].
Unfortunately, this parameter was not assessed in the present study and, if included, our 'linkage to care' percentage would probably be lower.Third, when interpreting late diagnosis figures, we should keep in mind that they are calculated by factoring in only those individuals for whom we have a CD4 count.It is not known whether those with no data available are more affected Percentages calculated on those with data available on CD4 count Percentage by late diagnosis.However, the effect of this limitation in the comparison with EPI-VIH is limited, as the proportion of individuals with an unknown CD4 count (33.4%) is similar.Conversely, it might have a greater influence when comparing our study with SNHSS, since the percentage is notably lower (13%).Finally, our data is based on self-reports and could be affected by social desirability bias.However, the use of an anonymous and self-administered questionnaire may have helped to obtain franker answers in the sensitive aspects of the survey.
By giving individuals the chance to rapidly check their serological status, this highly visible programme helped to diagnose a high proportion of individuals who were not actively seeking to get tested.Thus, it diminished the period during which the infection remains undiagnosed and therefore has the potential to reduce onward transmission in a population with high levels of sexual risk behaviours who are not testing with the recommended frequency.All this translated into a substantial contribution to early diagnosis in the MSM group in which late presentation at the population level -despite being lower than in other groupsis still too high.However, the HIV prevalence is notably lower if we compare it with a clinical setting serving atrisk populations.In order to improve its performance from a cost perspective, this programme should concentrate in locations highly transited by MSM.
-based organisation; MSM: men who have sex with men; MSW: men who have sex exclusively with women; n.a.; not applicable; STI: sexually transmitted infection.The programme was conducted during three periods: May 2008-December 2008 (season 1, 62 days in total), July 2009-July 2010 (season 2, 65 days) and November 2010-December 2011 (season 3, 35 days).a p value refers to chi-squared test between MSM, Women, MSW.b Included in season 2 onwards.c Included in the first season, but starting October 2008.

Figure 3
Figure 3 Percentage of newly diagnosed HIV infections with CD4 count <350 cells/µL in a street-based HIV rapid-testing programme, Spain (2008-2011) compared with EPI-VIH (2008-2010) and the Spanish National HIV Surveillance System (2008-2011) Rapid test and confirmation results, availability of data on linkage to care and CD4 count for people who underwent testing in a street-based HIV rapid-testing programme,Spain, 2008Spain,  -2011 Number of people tested and prevalence of HIV infection by programme location, transmission group and place of birth, in a street-based HIV rapid-testing programme, Spain, 2008-2011 IDU: injecting drug users; MSM: men who have sex with men; MSW: men who have sex exclusively with women.

Table 1
Sociodemographic, sexual identity, gay community involvement and behavioural risk variables of people receiving rapid HIV testing in a street-based programme,Spain, 2008Spain,  -2011 (n=8,923)    (n=8,923)

Table 2
Testing history, reasons for testing and future testing intentions of people who underwent testing in a street-based HIV rapid-testing programme,Spain, 2008Spain,  -2011 (n=8,923)    (n=8,923) a p value referred to chi-squared test between MSM, women, MSW.b Variable not included in season 3. c Included in season 2 onward.d Included in season 1 and 2. In season 3 it was included starting from June 2011.

Table 3
Comparison of people newly diagnosed with HIV by a street-based HIV rapid-testing programme, Spain, 2008-2011, with those from EPI-VIH (2008-2010) and the Spanish National HIV Surveillance System (2008-2011) (SNHSS) CI: confidence interval; EPI-VIH: network of 20 Spanish STI/HIV diagnostic clinics; n.av.: data not available; SNHSS: Spanish National HIV Surveillance System.