HIV testing strategies employed in health care settings in the European Union/European Economic Area (EU/EEA): evidence from a systematic review.

OBJECTIVES: Despite the availability of HIV testing guidelines to facilitate prompt diagnosis, late HIV diagnosis remains high across Europe. The study synthesizes recent evidence on HIV testing strategies adopted in health care settings in the European Union/European Economic Area (EU/EEA).
METHODS: Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed and systematic searches were run in five databases (2010-2017) to identify studies describing HIV testing interventions in health care settings in the EU/EEA. The grey literature was searched for unpublished studies (2014-2017). Two reviewers independently performed study selection, data extraction and critical appraisal.
RESULTS: One hundred and thirty intervention and/or feasibility studies on HIV testing in health care settings were identified. Interventions included testing provision (n = 94), campaigns (n = 14) and education and training for staff and patients (n = 20). HIV test coverage achieved through testing provision varied: 2.9-94% in primary care compared to 3.9-66% in emergency departments. HIV test positivity was lower in emergency departments (0-1.3%) and antenatal services (0-0.05%) than in other hospital departments (e.g. inpatients: 0-5.3%). Indicator condition testing programmes increased HIV test coverage from 3.9-72% before to 12-85% after their implementation, with most studies reporting a 10-20% increase. There were 51 feasibility and/or acceptability studies that demonstrated that HIV testing interventions were generally acceptable to patients and providers in health care settings (e.g. general practitioner testing acceptable: 77-93%).
CONCLUSIONS: This review has identified several strategies that could be adopted to achieve high HIV testing coverage across a variety of health care settings and populations in the EU/EEA. Very few studies compared the intervention under investigation to a baseline, but, where this was assessed, data suggested increases in testing.

Introduction

In 2017, 49% of people diagnosed with HIV infection were first identified at a late stage of infection (CD4 count < 350 cells/µL) in Europe 1. Late diagnosis is associated with increased risk of morbidity and mortality 2, 3 as well as increased risk of onward transmission of HIV as a consequence of delayed initiation of treatment 4. The Joint United Nations Programme on HIV/AIDS (UNAIDS) set the global 90‐90‐90 target where 90% of all people with HIV infection should be diagnosed, 90% of those diagnosed should receive HIV treatment and 90% of those on treatment should have a suppressed viral load by 2020 5. HIV testing is therefore a vital first step in the HIV care continuum and in Europe it has historically been offered in traditional health care settings, such as sexual health clinics, antenatal services and voluntary counselling and testing sites. Testing guidance for sexually transmitted infection (STI)/genitourinary/dermato‐venereology clinics exists at national, European and international levels promoting universal testing offer 6, 7, 8, 9, 10. However, other health care settings that are nonspecialist for HIV and where patients are presenting for the management of other conditions present opportunities to increase HIV testing, thereby reducing undiagnosed infections. In 2016, an estimated 101 400 people were living with undiagnosed HIV infection in the European Union/European Economic Area (EU/EEA), and, although this represents a decline in the number since 2012, it highlights the continued need for effective HIV testing programmes to improve HIV test coverage 11.

The World Health Organization (WHO) consolidated guidelines on HIV testing services, recommending that HIV testing services should be integrated with other relevant clinical services such as those for tuberculosis (TB), maternal health, sexual and reproductive health and harm reduction programmes, especially as these services attract populations considered to be at higher risk for HIV infection 9. The guidelines endorse the use of provider‐initiated testing and counselling when the epidemic is generalized and the routine offer of testing for all clients in all health facilities (including primary care, inpatient and outpatient services and all services for key populations) is recommended as an effective way to identify people with HIV infection.

Although guidance is available from international organizations and national public health bodies to inform service provision for HIV testing 12, health care providers within European countries need to be able to operationalize these into clinical practice so as to diagnose HIV infection at an earlier stage of infection. An evidence synthesis published in 2010 by the European Centre for Disease Prevention and Control (ECDC) showed that a number of strategies could reduce missed opportunities for HIV testing, including indicator condition (IC)‐guided testing, which involves offering testing to all patients presenting to care with an AIDS‐defining illness or with an HIV ‘indicator’ condition (IC) 8. An HIV IC is a condition associated with an undiagnosed HIV prevalence of at least 1 per 1000 13. Other strategies were routine HIV testing implemented as part of routine care in health care settings and the use of rapid tests that offer immediate results.

A recent evaluation of the 2010 ECDC guidance found that, although the document was considered important for policy and guideline development, an update to the guidance was necessary to incorporate new approaches and technologies that have been adopted to increase testing offer and coverage in recent years 14. The purpose of this paper is to present the recent body of evidence on HIV testing strategies employed in health care settings in Europe. Additionally, the paper reviews the evidence on testing provision strategies that increase HIV testing coverage and on the feasibility and acceptability of HIV testing strategies. The systematic review described here was conducted as part of a wider review of the evidence on HIV testing in the EU/EEA and barriers to testing to update the 2010 ECDC HIV testing guidance.

Methods

This review followed the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) statement for the reporting of systematic reviews 15. A full description of the methodology is described elsewhere 16. Briefly, five electronic databases (OVID Medline, Embase, PsycINFO, Scopus and the Cochrane Library of Systematic Reviews) (January 2010 to March 2017) and the proceedings of six conferences (2014–2017) were searched using key search terms covering concepts including ‘HIV’, ‘HIV testing’ and ‘Europe’ (Tables S1–S5). Only studies pertaining to adults and set in the 30 EU/EEA countries (Table S6) and outside occupational settings were included in the review. No language restrictions were applied.

Two reviewers independently undertook title review, full‐text review and data extraction. ECDC completed all reviews for non‐English studies, with data extraction in English. Data on qualitative and quantitative outcome indicators were extracted, including information on HIV coverage, test positivity and intervention feasibility and acceptability. Two reviewers carried out quality assessment and risk of bias assignment for published studies based on National Institute for Health and Clinical Excellence (NICE) checklists and the AXIS quality assessment tool (Table S7). Studies were rated as being high, medium or low quality and having high, medium or low bias. Conference proceedings were not appraised for quality and bias. Critical appraisal results can be found in Table S8.

This paper focuses on studies of HIV testing in health care settings including STI clinics, primary care, hospitals, pharmacies, prisons, drug services and TB services. HIV testing strategies involved interventions categorized as testing provision, education programmes, campaigns, use of communication technologies, use of clinical decision‐making tools and other interventions. Data from HIV testing provision studies were examined for the impact of testing on HIV testing coverage. Studies documenting the feasibility and/or acceptability of HIV testing interventions studies were also included. The remaining studies covering barriers to testing, economic evaluations, audits and non‐health care settings are included in the wider systematic review findings that informed the guidance (n = 238) 17. European regions referred to are based on the United Nations geoscheme for Europe.

Results

Study identification and overview

The searches yielded 15 004 records after de‐duplication; after full‐text review, 368 studies were included in the overall systematic review (Figure 1). Of the 368 studies, 130 are described in this paper, exploring interventions and feasibility of HIV testing in health care settings in Europe, including 84 peer‐reviewed articles and 46 conference proceedings (Table S8) 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147.

Figure 1

Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) flow diagram. EU/EEA, European Union/European Economic Area.

Studies were from 13 of the 30 EU/EEA countries. Most studies were from Northern Europe (n = 78; 64%), followed by Western (n = 23) and Southern Europe (n = 24). There was only one study from Eastern Europe 109 and four studies set across multiple European countries 110, 114, 127, 128. The majority of the studies from Northern Europe were from the UK (92%). Other than the UK, there were two countries with more than five studies (Spain and France).

Studies were set in a range of health care facilities, the most common being primary care (n = 45) followed by inpatient services (n = 25), STI clinics (n = 24) and emergency departments (n = 23). Other health care testing sites for HIV included outpatient services (n = 16), prisons (n = 4) and pharmacies (n = 4). Almost a fifth of studies were conducted in more than one setting type.

There were a number of interventions implemented to increase HIV testing in health care settings, including innovative/improved testing provision (n = 94), use of testing campaigns (n = 14), use of communication technologies (n = 4), education and training for staff and patients (n = 20), use of tools to aid clinical decision‐making (n = 10) and relocation of a clinic to a higher men who have sex with men (MSM) density area (n = 1). Twenty‐two studies applied strategies with multiple interventions to increase testing.

The quality of the peer‐reviewed studies was variable; of the 84 articles, 70% were of high quality, 19 were of medium quality (23%) and six were of low quality (7%) (Table S8). Risk of bias was low in 42 studies (50%), medium in 36 studies (43%) and high in six studies (7%) (Table S8). Risk of bias was low in 42 studies (50%), medium in 36 studies (43%) and high in six studies (7%) (Table S8).

Testing provision strategies

Novel HIV testing technologies were employed by 40 studies in a wide range of clinical settings to increase testing coverage; the majority utilized rapid testing (n = 36) 18, 19, 22, 24, 25, 34, 36, 45, 46, 48, 49, 50, 56, 57, 59, 65, 75, 77, 81, 83, 85, 88, 89, 99, 100, 101, 107, 113, 117, 129, 132, 134, 137, 139, 141 while four utilized self‐sampling (n = 3) 44, 52, 102 and self‐testing strategies (n = 1) 108. Two of the four self‐sampling studies used oral fluid sampling while the self‐test required a blood sample. Novel testing approaches were particularly applied to improve testing coverage in HIV risk groups including MSM (n = 9) 44, 45, 46, 52, 57, 81, 85, 129, 134, migrants (n = 7) 22, 45, 46, 57, 85, 129, 132, and people who use/inject drugs (PWUD/PWID) (n = 5) 50, 57, 85, 89, 129.

Other testing strategies included routine testing (n = 32) 20, 21, 24, 25, 32, 35, 36, 43, 61, 64, 65, 69, 74, 77, 78, 92, 93, 94, 95, 96, 98, 104, 111, 115, 116, 117, 120, 125, 130, 135, 137, 147, provision of HIV testing as a component of an integrated testing programme (n = 29) 22, 23, 32, 35, 37, 38, 47, 50, 52, 55, 62, 63, 70, 72, 74, 79, 92, 93, 94, 96, 97, 102, 103, 119, 122, 123, 134, 137, 143, IC testing (n = 14) 26, 38, 47, 57, 88, 110, 114, 116, 120, 125, 126, 127, 128, 135 and partner notification (n = 4) 54, 60, 104, 133. Routine testing was most commonly implemented in hospital departments including emergency departments (n = 15) 20, 21, 25, 35, 36, 64, 69, 78, 92, 93, 115, 116, 117, 137, 147, inpatient units (n = 10) 24, 43, 61, 95, 98, 104, 111, 125, 130, 135 and outpatient departments (n = 4) 21, 74, 113, 116. Similarly, IC testing programmes were predominantly instigated in hospitals (n = 8) 109, 114, 116, 125, 126, 127, 128, 135 and primary care (n = 10) 26, 38, 47, 57, 88, 110, 114, 116, 127, 128. In contrast, integrated testing for HIV with other infections such as hepatitis B and C and STIs was adopted in diverse settings including prisons (n = 3) 70, 103, 122, STI clinics (n = 4) 23, 52, 63, 134, drug services (n = 2) 50, 79 and pharmacies (n = 1) 102.

The majority of the 94 testing interventions were directed to the general population (74%). Testing strategies directed to risk groups included studies among migrants and black and minority ethnic groups (n = 12) 22, 23, 32, 45, 46, 57, 62, 80, 85, 129, 132, 143, MSM (n = 11) 44, 45, 46, 52, 57, 63, 81, 85, 129, 134, 136, young people (n = 1) 102, PWUD/PWID (n = 7) 50, 57, 79, 85, 89, 103, 129, and mental health patients (n = 1) 123. Often, these studies targeted multiple risk groups without presenting group‐specific results.

HIV test coverage and positivity differed considerably between health care settings (Table 1). HIV test coverage varied from 2.9 to 94% in primary care and from 3.9 to 66% in emergency departments. HIV positivity ranged from 0 to 25% in STI clinics, with the higher rates achieved when partner notification was used to identify cases. In general, positivity rates were lower in studies set in emergency departments (0–1.3%) and antenatal services (0–0.05%) than in those set in other hospital departments (e.g. up to 5.3% in inpatient units).

Table 1

HIV testing and positivity rates by health care testing venue

Testing venue	Number of people tested	Test offered (%)	Test accepted (%)	Test coverage (%)	Positivity rate (%)	References
Primary care	3–7706*	12–97	45–99.7	2.9–94	0–4.7	25, 32, 37, 44, 45, 46, 54, 55, 56, 61, 71, 74, 76, 79, 81, 86, 87, 93, 98, 100, 106, 115
STI clinic	4–3738, 15–62 kits returned		63	7–78	0.6–4.5 21–25 (PN)	22, 33, 43, 51, 59, 62, 99, 132, 135
Inpatient services	10–4122*	48–80	70–100	17–73	0–5.3	23, 27, 42, 60, 94, 97, 103, 110, 115, 122, 124, 125, 129, 134
Emergency department	275–27 632	6.2–74	30–95	3.9–66	0–1.3	19, 24, 34, 35, 63, 68, 77, 91, 92, 96, 112, 115, 116, 146
Outpatient services	55–166	53	32–68	35–98	0–1.9	21, 73, 82, 142
Prison	357–1932			51–67	0.3–3.9	18, 69, 102, 121
Pharmacies	2168*–24 151, 96 kits returned			45	0.9	47, 48, 58, 101
Drug services	146–211	33–69	40–99	13–52	0–2.5	49, 88
Antenatal services	430–561 158*	100	35–99	18	0–0.05	36, 95
Other health care sites (e.g. TB services)	71–3881	31–100	76–99	24–99	0–2.0	31, 64, 118, 119, 131, 140
Combined health care settings†	141–9471	14	63	56–89	0.3–5.4 12–21 (PN)	20, 53, 109, 112, 113, 126, 127, 133, 138
Combined health care and non‐health care settings†	119–11 549		54		0.7–2.5	80, 84, 99, 107, 128

PN, partner notification; TB, tuberculosis; STI, sexually transmitted infection.

Includes number of tests performed, where there is more than 1 test per person.

Where combined health care and non‐health care settings could not be separated.

Testing provision strategies that increase testing

Thirty studies evaluated the impact of a testing intervention by comparing the intervention data with baseline data (n = 24) 20, 26, 34, 44, 47, 65, 72, 75, 78, 80, 83, 87, 95, 96, 98, 102, 111, 113, 125, 126, 130, 133, 135, 141 or with a control group (n = 6) 22, 77, 81, 82, 120, 136 (Table 2). Twelve studies employed novel testing (10 rapid testing and two self‐sampling) in diverse settings, of which one reported an increase in HIV test coverage from 2% before the intervention to 45% after 65, while others reported increases in HIV diagnoses 77, 141, testing 44, 75, 83, 102 and test acceptance 81 and higher positivity rates 34 after the intervention. The use of rapid tests also resulted in 98% of people obtaining their results compared to 64% in the standard serology group 22. One study reported a decline in numbers of tests performed 113. A further six studies conducted in inpatient services, TB services and primary care reported the impact of IC testing: HIV test coverage changed from 3.9–72% before to 12–85% after its implementation, with most studies reporting a 10–20% increase 26, 120, 125, 126, 135 and the median number of tests also increased 47. Twelve studies, of which half were set in inpatient services, examined the impact of universal routine testing on test coverage. These studies reported an increase in coverage from 2–28% before to 17–80% after the intervention 65, 78, 98, 111, 125, 130, 135 and higher coverage in the intervention group compared to the control group (85% versus 72%, respectively) 120. Other indicators included increases in test acceptance 20, numbers tested (although the increase was small) 95 and numbers diagnosed 77 and a reduction in vertical transmission 96. Only six studies measured the impact of the intervention in at least one risk group; five in MSM 44, 81, 133, 136, 141, one in young people 102 and one in migrants 141.

Table 2

Impact of HIV testing interventions on HIV tests, coverage and positivity

Author (year)	Testing venue	Testing strategy	Number of people tested	Tests performed	Offered and/or accepted a test (%)	Testing coverage (%)	Positivity (%) and/or HIV diagnoses	QA score	Risk of bias
Bath et al. (2015) 19	Emergency department	NTS; UT	Before, 72; after, 2828		Accepted before, 2.4%; after, 30%			High	Low
Bottero et al. (2015) 21	Outpatient services	BR	Control, 115; intervention, 159			Control, 71%; intervention, 98%		High	Low
Cayuelas‐Redondo et al. (2015) 25	Primary care	ICT; NTS	Before, 22; after, 78			Before, 3.9%; after, 12%	Before, 12 diagnoses; after, 13	High	Medium
Cuesta et al. (2012) 33	STI clinic	RT	After, 1011				Before, 0.33%; after, 1.1%	Medium	medium
Elmahdi et al. (2014) 43	STI clinic	OFSS; risk group (MSM)	Men returning kits, 15	Median tests per person per year prior, 1 (1–2); during study, 2 (1–3)				NA	NA
Fagard et al. (2014) 46	Primary care	I; ICT; NTS	1338	Median GP prescribed tests week before, 2; week of, 16	Offered after, 68%; accepted, 76%			NA	NA
Herbert et al. (2012) 64	Emergency clinic, hospital for tropical diseases	BR; NTS; UT	Before, 38; post‐UT, 183; post‐BR, 1261			Before, 2%; post‐UT, 23%; post‐BR, 45%	Before, 0%; post‐UT, 1.1%; post‐BR, 0.4%	Medium	Medium
Kelly et al. (2014) 71	Primary care	NTS; I	Before, 5; after, 61			Before, 4.8%; after, 43%	Before & after, 0%	Low	High
Kuttner‐May (2015) 74	Primary care	R; NTS				Increase in HIV testing rates: 13%	Before, 1.0%; after, 1.2%	NA	NA
Lascar et al. (2015) 82	Outpatient services	OFR; NTS	148	Before, 420; after, 676				Medium	Low
Leber et al. (2015) 76	Primary care	BR; UT; NTS	Control GP practices, 2465; intervention, 7706 (BR, 4978)				Rate of HIV diagnosis/10 000 patients/years: control, 0.07; intervention, 0.3	Medium	Medium
Leber et al. (2017) 81	Primary care	NTS; BR				Increase in HIV testing rates: 85%		NA	NA
Lim et al. (2014) 77	Emergency department	NTS; UT				Before, 16%; after: 33%		NA	NA
Loos et al. (2014) 79	Primary care	NTS		GPs testing no sub‐Saharan Africa migrant patients: before, 28%; after, 8%				High	Medium
Lorente et al. (2013) 80	VCT community testing services	BR; NTS; risk group (MSM)	Control (VCT): 119; intervention: 211		Accepted: control, 54%; intervention, 89%		Control, 2.5%; intervention, 1.4%	Medium	Medium
Martín‐Cabo et al. (2012) 86	Primary care	NTS; UT		Before, 22; after, 212		Before, 3.7%; after, 27%		High	Medium
Onen et al. (2015) 94	Inpatient services	NTS; UT	Before, 1; post‐education, 0; post‐UT, 4					NA	NA
Op de Coul et al. (2011) 95	Antenatal services	UT, I		561 158			Children born with HIV annually: before, 5–10; after, 1	High	Low
Palfreeman et al. (2013) 97	Inpatient services	NTS; UT	Before, 205; during, 938; after, 1399			Before, 3.7%; during, 17%; after, 23%	Before, 2.0%; during, 1.1%; after, 1.1%	Medium	Medium
Peacham et al. (2015) 101	Pharmacy	SS; risk group (YP); I		Kits returned, 96; increase in STI tests (incl. HIV) from before to after, 700%		Kit return rate, 45%		NA	NA
Raman et al. (2015) 110	Inpatient services	NTS; UT	Before, 20; during, 49; after, 34			Before, 9%; during, 28%; after, 17%		NA	NA
Rayment et al. (2013) 112	Emergency department	OFR; NTS		Pre‐automation of OF testing, 3721; after, 2960			Pre‐automation, 0.1%; after, 0.1%	High	High
Roy et al. (2013) 119	TB services	UT; ICT	Group A*, * control, 183; intervention, 462 group B*, * control, 91; intervention, 172		Offered: group A control, 76%; intervention, 87%; group B control, 89%; intervention, 96%; accepted: group A control, 84%; intervention, 87%; group B control, 81%; intervention, 87%	Group A control, 72%; intervention, 82%; group B control, 76%; intervention, 85%		High	Low
Sharvill et al. (2015) 124	Inpatient services	NTS; UT; ICT	Before, 19; after, 43			Before, 28% within 24 h; after, 73%		Medium	Low
Sokhi et al. (2015) 125	Inpatient services	NTS; ICT	Pre‐protocol, 9; post, 11; post‐proforma, 20; after 1 year, 17			Pre‐protocol, 22%; post, 37%; post‐proforma, 83%; after 1 year, 65%		High	Medium
Thornhill et al. (2014) 129	Inpatient services	NTS, UT	After, 465			Before, 6%; after, 52%		High	Low
van Aar et al. (2015) 132	STI clinic	PN; risk group (MSM)	MSM partners, 136				21%; of all HIV infections, those detected through PN before, 19%: after, 34%	High	Medium
Wallis et al. (2015) 134	Inpatient services	NTS; UT; ICT		Before, 4; after, 22		Before, 5%; after, 26%		Medium	Low
Whitlock et al. (2015) 135	STI clinic	RT; risk group (MSM)	Control, 19; intervention (SMS reminder), 44			Control, 19%; intervention, 44%	Control, 5.3%; intervention, 4.5%	NA	NA
Wouters et al. (2014) 140	Low threshold centre	BR; NTS; risk groups (MSM, African migrants)	3881 (MSM, 1173; migrants, 454)	Pilot, 219; implementation, 4806			Before, 0%; after, 1.5% (MSM†, 4.0%; migrants†, 2.2%)	High	Low

IQR, interquartile range; GP, general practitioner; RT, rapid test; NTS, nontraditional settings; BR, blood rapid test; UT, universal testing; ICT, indicator condition testing; MSM, men who have sex with men; PN, partner notification; SMS, short message service; SS, self‐sampling; OFR, oral fluid rapid test; OFSS, oral fluid self‐sample; I, integrated test; QA, quality assessment; VCT, voluntary counselling and testing; YP, young people; R, rapid test.

NA, not applicable: conference proceedings were not assessed for quality and bias.

Group A, centres using selective testing; group B, centres using UT.

Denominator includes pilot.

Other HIV testing strategies targeted to providers

There were 30 studies using other strategies (campaigns, education and use of clinical decision‐making tools) directed to providers. Three campaigns targeted providers to increase awareness using posters, social media (e.g. Twitter) and promotional materials 21, 111, 146. A significant number of educational intervention studies targeted providers (n = 19), with the majority providing HIV testing training sessions to health care professionals including hospital doctors, general practitioners (GPs), medical students and nurses 30, 47, 67, 68, 77, 82, 83, 95, 105, 112, 126, 133, 138, 139, 146 and pharmacists 48, 49. One employed the plan, do, study, act (PDSA) methodology, which increased physicians’ willingness to test but did not increase testing 84. Another used serious incident reporting to improve testing awareness within clinics 140. Where assessed, education provision resulted in an increase in the offer of HIV testing from 2 to 11% 146 and in the number of individuals tested from 11–13 before to 16–20 after the intervention in two smaller studies, and from 420–1056 before to 676–2333 after the intervention in two larger studies 83, 105, 112, 126. Two studies reported a decline in test offer from 8–15% to 0–10%, which may have been attributable to the small numbers of patients included in the studies (n = 4–26) 68, 95. There were eight studies using clinical decision‐making tools to aid providers in identifying populations that should be tested for HIV. A variety of tools were developed: addition of HIV tests to the blood test ‘set’ requests or checklist 31, 39, 76, 77, computer prompts for higher risk populations 27, 38 and risk assessments 118, 144. Where recorded, the above interventions were successful at increasing testing 27, 31, 39.

Other HIV testing strategies targeted to patients

Other than interventions where testing was provided, there were 23 studies using other interventions directed to patients. The majority of these interventions were campaigns that promoted local testing using social media, posters, digital media and websites 19, 78, 83, 89, 146, campaigns to promote National HIV Testing Week 21, 71, 111, 124 and other regional campaigns to promote testing 48, 49, 85, 97, 142. National HIV Testing Week increased testing from 4–9% before to 8–28% during the week 111, 124, and it also resulted in half of those having blood samples collected at a hospital being tested for HIV 21. During the regional Go Viral campaign, 27% of patients were tested for blood‐borne viruses (BBVs) 97.

There were two educational interventions targeting patients; one for pregnant women 90, which resulted in an increase in testing coverage (from 87 to 92%) after provision of a patient information leaflet and one for patients admitted to a hospital inpatient unit, where there was a decline in test offer (from 8 to 0%) 95. All four communication technology studies were directed to patients, with two providing videos on HIV testing 24, 106, one utilizing text messages to recall MSM for testing 136 and one using online platforms for partner notification 60. The only study measuring intervention impact reported an increase in the re‐testing rate among MSM who were actively recalled (from 19 to 44%) 136. Two studies adopted decision‐making tools that helped individuals determine if an HIV test would be recommended 42, 118. One of these examined whether computer‐assisted self‐interviewing resulted in increased HIV testing when compared to interviews with clinicians 118 and found significantly less testing in the self‐interviewing population (63% versus 69%, respectively). Finally, in one study, the STI clinic moved location to be in a higher MSM density area, which resulted in a large increase in the number of HIV diagnoses from 175 in 2008 to 381 in 2013 after relocation 73.

Feasibility and acceptability

HIV testing interventions were generally acceptable to patients and providers in health care settings (Table 3). Results also suggest that rapid testing is acceptable to both groups (patient studies, n = 5; provider studies, n = 6), with providers willing to use rapid tests 24, 53, 56, 57, 107 and finding their interpretation easy 50. Some feasibility studies highlighted that nontraditional health care settings can target populations not previously tested for HIV (n = 4) 18, 85, 132, 145, with the reported percentage of first‐time testers in such settings ranging from 51 to 75%. One study suggested that provider‐initiated testing is unlikely to be acceptable when specific populations are targeted (in this case, sub‐Saharan African patients) 86.

Table 3

Selection of feasibility and acceptability indicators for HIV testing in health care settings by testing venue

Testing venue	Sample size	Feasibility/acceptability (patient)	Feasibility/acceptability (provider)	References
Primary care	62–3314	GP testing acceptable, 77–93% First time testers, 75% MSM who strongly agreed that the clinic environment was friendly to MSM, 66%	GPs willing to use/continue to use rapid HIV testing in their daily practice, 59–77%	17, 46, 52, 55, 56, 57, 61, 65, 79, 87, 98, 106, 130
STI clinic	50–337	Self‐sampling kits acceptable, 30–62.5% (MSM) Reported self‐sampling really easy, 66% First time testers, 94% MSM who would recommend service to a friend, 100%	‐	28, 43, 50, 51, 62, 144
Inpatient services	10–478	Inpatient testing acceptable, 84–100% Rapid HIV testing in inpatients acceptable, 97%	Clinical staff who thought HIV rapid testing disrupted their job, 0%	23, 42, 94, 103
Emergency department	19–5657	Emergency department testing acceptable, 50–96%	Routine HIV testing should be rolled out permanently in the emergency department, 95% Patients not offered testing because the physician forgot to ask, 6.5%	19, 35, 63, 92, 116
Outpatient services	166–246	Outpatient service testing acceptable to migrants, 72% Preference for rapid tests over standard serological tests, 76%	‐	21, 142
Pharmacies	806–2168	Pharmacy testing acceptable, 100% Pharmacy quick and convenient, 31–71% Pharmacy accessible, 4.7–20% Young people who were very or quite satisfied with the service and were very or quite likely to use the service again, 100%	‐	47, 58, 101
Drug services	12	‐	Rapid test and test interpretation easy or very easy, 100%	49
Antenatal services	1243–2123	Antenatal screening acceptable, 81% HIV testing of partners of pregnant women acceptable, 35%	‐	36, 108
Other single sites combined	21–825	Rapid HIV testing acceptable to migrants, 99% First‐time testers among migrants, 71%	Health care providers often felt untrained and unconfident giving the result Not acceptable to include behavioural survey as part of HIV test for migrants	90, 131
Combined health care settings*, *	20–5329	Testing as part of routine care acceptable, 71–92% Acceptability of providing: mouth swab, 95%; blood test, 89%; finger prick blood test, 90% IC testing coverage in: primary care, 12%; hospitals, 92% PN case‐finding effectiveness, 18%	Provider‐initiated testing of sub‐Saharan African migrants acceptable, 35% Identified need for training for physicians, 72%	39, 40, 53, 85, 112, 115, 120, 127, 133
Combined health care and non‐health care settings*, *	128–264	First time testers: 51% Preference for rapid tests over standard serological tests, 84% Self‐test acceptable, 92% Successful performance of a finger‐stick whole‐blood HIV self‐test, 99%	‐	84, 107

GP, general practitioner; IC: indicator condition; MSM, men who have sex with men; PN, partner notification.

Where combined health care and non‐health care settings could not be separated.

Discussion

This systematic review has identified a significant body of evidence on HIV testing in health care settings in Europe. Testing has been provided in a range of clinical settings with results suggesting that it is feasible to achieve high testing coverage, and that it is acceptable for providers and service users. HIV testing positivity ranged widely, from 0 to 25%, with higher positivity rates observed with certain strategies such as partner notification, IC testing and testing among risk groups compared to strategies that offered testing to the general population, including lower risk groups. However, the data also highlight that there is considerable room to increase the offer of testing in health care settings, particularly in primary care and emergency departments.

Evidence from systematic reviews shows that, in primary care, barriers to testing are related to the clinician’s knowledge 17, 148, as well as the clinician’s anxiety associated with raising the topic of HIV testing with patients 149. Improving testing in primary care is important because in many countries more testing takes place in primary care than in specialized services, as those countries have a testing strategy that primarily uses GPs (e.g. the Netherlands and Germany) 12. Furthermore, an evaluation of the impact of the guidance highlighted that, while over half the respondent countries reported that their national testing guidelines were closely aligned with the 2010 European guidance, only 35% included the relevant recommendations on routine offering in primary care 150. Testing in primary care is highly acceptable to all patients, with one study among MSM also reporting primary care as an acceptable setting, and therefore interventions that increase knowledge and provide training for primary care staff could be successful at increasing test coverage. The knowledge and capability of health care staff could be enhanced through education interventions. One study in this review gave GPs training in sexual health clinical skills and achieved large increases in testing rates from 1056 tests before to 2333 after the intervention 105, which emphasizes the impact on testing once GPs are enabled to offer testing.

In addition to testing in primary care, there are other strategies identified in this review that could improve testing coverage in health care settings in Europe, including: scaling up IC testing across all settings; introducing testing in high‐prevalence areas, although the majority of studies implementing this strategy are from the UK; and implementing integrated testing for BBVs in settings such as drug services and prisons and HIV and STI testing in STI clinics. Although IC testing is an effective strategy to diagnose HIV infection and results in high positivity rates, it is not always included in national or speciality testing guidelines for specific ICs and, where it is included, the scale of implementation is quite variable 151, 152. The concept of testing based on high diagnosed sero‐prevalence areas assumes that areas of high diagnosed prevalence are likely to also have high rates of undiagnosed infections. This strategy removes the need to target specific populations, which was found to be unacceptable to primary care providers in one study in this review. The adopted strategy will, however, depend on the health care setting. Finally, integrated testing for BBVs was recently recommended in prisons by the ECDC/ European Monitoring Centre for Drugs and Drug Addiction guidance 153.

There were only three studies set in community‐based drug services and pharmacies. These venues potentially have an important role in HIV testing by reaching populations that do not necessarily attend traditional health care venues and by acting as a bridge between health care and the community. Community‐based pharmacies are acceptable venues for HIV testing where testing can be implemented with basic teaching and skills provision to the pharmacist 48, 49. Pharmacies are already an established model of delivery for chlamydia testing 154, which only highlights the growing role of pharmacies in public health provision. Rapid testing or self‐sampling kits were the adopted testing strategies in this setting. The added benefit of self‐sampling kits is the possibility of providing integrated testing for BBVs and STIs 102.

The distribution of self‐sampling and self‐testing kits for HIV from other health care settings (e.g. STI clinics) has been found to be acceptable among MSM 44, 52 and people attending free and anonymous testing services 108. However, further evidence is needed to understand whether this strategy can increase testing frequency in high‐risk groups including MSM.

There were some limitations to this systematic review. There was only one study from Eastern Europe, which could have implications for the applicability and reproducibility of the review findings to this region. There were methodological concerns relating to measuring increases in test coverage. Although this review aimed to document the impact of intervention on increasing test coverage, a very small proportion of studies included a baseline measure or a comparator group to allow assessment of improvements in testing. Where comparisons could be made, the timeframe over which the impact of the intervention was assessed varied between studies, with most being assessed immediately after the intervention. Future studies should consider implementing interventions over longer timeframes. We restricted the review to the EU/EEA, so we may have missed studies that were applicable to and reproducible in the European setting. Finally, studies with positive findings or using novel approaches are more likely to be published. The inclusion of conference abstracts and reports (35%; 46 of 130) is therefore important to reduce publication bias; however, the quality of these studies has not been assessed. Without this assessment, we cannot know the reliability and reproducibility of the presented results. It is important that all findings including those from conference abstracts are published in peer‐reviewed journals. There are three important strengths to this review. The review employed the robust PRISMA methodology which is standardized and reproducible. Secondly, the scope of the questions facilitated a broad and all‐encompassing review of the literature on HIV testing in health care settings. Thirdly, the review included papers not in English, which were translated.

Conclusions

HIV testing is an integral component of combination HIV prevention, treatment and care. HIV testing is the first step in the global 90‐90‐90 target set by UNAIDS, where 90% of all people with HIV infection should be diagnosed, 90% of those diagnosed should receive HIV treatment and 90% of those on treatment should have a suppressed viral load by 2020 5. Therefore, strategies that diagnose individuals as early as possible are essential to achieve the ultimate goal of ending the HIV epidemic. This review has identified that, although there are considerable missed opportunities for earlier HIV diagnosis, there are also several acceptable and feasible strategies to achieve high HIV testing coverage across a variety of health care settings and populations in Europe.

Author contributions

All authors were involved in the evidence synthesis and contributed important intellectual content to this paper. SD drafted the manuscript and was responsible for submission; all authors commented on the manuscript and approved the final draft. SD and SC coordinated the systematic review process: developed the protocol and search terms (with VD, DR and LT), ran the searches and compiled the results. SC, SD, LT, AKS, LC, DR, SFJ and VD contributed to systematic review study screening, data extraction and quality assessment along with the wider review working group (as listed in the Acknowledgements). AJAG and LT provided ECDC quality control and directed the ECDC‐PHE‐CHIP collaboration.

Table S1 Search terms for OVID Medline – (17/03/2017).

Table S2 Search terms for OVID Embase (20/03/2017).

Table S3 Search terms for OVID PsycINFO (20/03/2017).

Table S4 Search terms for the Cochrane library (17/03/2017).

Table S5 Search terms for Scopus (20/03//2017).

Table S6 List of the 30 EU/EEA countries included in the systematic review.

Table S7 Systematic review quality assessment.

Table S8 Overview of included studies and their approach to HIV testing in healthcare settings.

Click here for additional data file.