Conceived and designed the experiments: MR AT GE PR RJ AN MTF JA AKS. Performed the experiments: MR AT GE MTF AKS. Analyzed the data: MR AT SM MTF JA AKS. Contributed reagents/materials/analysis tools: MA. Wrote the paper: MR. Provided technical advice, and evaluated and authorized the HIV testing technology: MA. Reviewed the manuscript: MR AT GE SM MA PR RJ AN MTF JA AKS.
The authors have declared that no competing interests exist.
UK guidelines recommend routine HIV testing in healthcare settings if the local diagnosed HIV prevalence >2/1000 persons. This prospective study assessed the feasibility and acceptability, to patients and staff, of routinely offering HIV tests in four settings: Emergency Department, Acute Care Unit, Dermatology Outpatients and Primary Care. Modelling suggested the estimated prevalence of undiagnosed HIV infection in attendees would exceed 1/1000 persons. The prevalence identified prospectively was not a primary outcome.
Permanent staff completed questionnaires assessing attitudes towards routine HIV testing in their workplace before testing began. Subsequently, over a three-month period, patients aged 16–65 were offered an HIV test by study staff. Demographics, uptake, results, and departmental activity were collected. Subsets of patients completed questionnaires. Analyses were conducted to identify factors associated with test uptake.
Questionnaires were received from 144 staff. 96% supported the expansion of HIV testing, but only 54% stated that they would feel comfortable delivering testing themselves, with 72% identifying a need for training. Of 6194 patients offered a test, 4105 (66·8%) accepted (61·8–75·4% across sites). Eight individuals were diagnosed with HIV (0–10/1000 across sites) and all transferred to care. Younger people, and males, were more likely to accept an HIV test. No significant associations were found between uptake and ethnicity, or clinical site. Questionnaires were returned from 1003 patients. The offer of an HIV test was acceptable to 92%. Of respondents, individuals who had never tested for HIV before were more likely to accept a test, but no association was found between test uptake and sexual orientation.
HIV testing in these settings is acceptable, and operationally feasible. The strategy successfully identified, and transferred to care, HIV-positive individuals. However, if HIV testing is to be included as a routine part of patients’ care, additional staff training and infrastructural resources will be required.
In 2009, an estimated 26% of all adults living with HIV infection in the UK were thought to be undiagnosed.
Guidance from the National Institute for Health and Clinical Excellence (NICE) joins existing recommendations from the Chief Medical and Nursing Officers, the British Association for Sexual Health and HIV, the British HIV Association and the British Infection Society in calling for the expansion of HIV testing to settings beyond sexual health clinics and antenatal care.
Routine HIV testing in antenatal care and sexual health clinics in the UK has greatly increased HIV testing rates.
The HINTS study was an observational study conducted at four locations in London, UK: an Emergency Department (ED), an Acute Care (medical and surgical admissions) Unit (ACU), a Dermatology Outpatient Clinic (OPD), and a Primary Care Centre (PC). All are located within areas of diagnosed HIV prevalence >2/1000 persons as per the Health Protection Agency’s
Study Site | Primary Care Trust (PCT) | Diagnosed HIV prevalence per 1000 individuals 15–59 years living in PCT (ranking in UK) |
Model of routine HIV testingservice delivery: |
Testing phase: weeks (dates conducted) | HIV testing method: |
Chelsea and WestminsterHospital EmergencyDepartment (ED) |
Kensington and Chelsea | 8·33 (4th) | 15 (August-November 2009) | ||
Homerton UniversityHospital - Acute Care Unit( |
City and Hackney | 8·25 (5th) | 12 (January-April 2010) | ||
Kings College Hospital - Dermatology Outpatient Department ( |
Lambeth | 13·28 (1st) | 12 (July–September 2010) | ||
North End Medical Centre - Primary Care ( |
Hammersmith and Fulham | 8·15 (6th) | 14 (February-May 2010) | ||
The study was divided into two phases. During the pre-study phase, a questionnaire was administered to all clinical and support staff within each department, assessing attitudes and knowledge towards HIV infection and the provision of HIV testing in non-specialist settings. All staff were encouraged to respond, and the return rate was calculated. The majority of staff completing the questionnaire were not directly involved in delivering the subsequent HIV testing programmes (see below).
During the study phase, HIV testing was offered to all sequential patients accessing the healthcare setting, who fulfilled the inclusion criteria, as follows: aged between 16 and 65 years and (i) not known to be HIV positive, (ii) accessing the healthcare setting for the first time over the testing period, and (iii) able to consent to a test. A leaflet was provided and verbal consent was obtained prior to HIV testing. Delivery of HIV testing in these settings was in line with published National guidelines,
Oral fluid-based HIV testing was used in the PC, ED and OPD arms, and was performed using a fourth-generation assay on a modified platform to detect HIV-1 antibodies. Internal validation of this technique in the pre-study phase showed it to be robust.
Questionnaires were administered to sequential patients offered HIV tests at prospectively planned intervals throughout the testing period, in order to minimise selection bias. The questionnaire distribution strategy was designed in a pragmatic fashion, in order to maximise the amount of time during which HIV testing was the only additional intervention. As provision of the questionnaire required additional time, it impacted upon the number of tests that could be offered; therefore questionnaires were offered only on pre-determined days. On these days, questionnaires were sequentially offered to all patients who were eligible for the study (both those who accepted testing and those who declined). The questionnaire collected basic demographic data, HIV testing behaviour, and attitudes towards a series of statements on HIV testing, quantified by modified Likert scales. In the PC arm, active recruitment to the questionnaire study was also supplemented retrospectively by a mobile phone text recruitment campaign.
Full, verbal consent was sought prior to HIV testing, in line with good clinical practice (as per General Medical Council guidance) and in line with the published, national, clinical guidelines.
The primary outcome measures of the study comprised:
feasibility of delivering routine HIV testing in the non-specialist setting, as evaluated by measuring the proportion of eligible individuals offered tests
acceptability, to patients and staff, of screening method and site, as evaluated by test uptake and the results of the patient and staff questionnaires
proportion of individuals newly diagnosed with HIV infection in non-specialist settings successfully transferred to specialist HIV clinical care
A modelling exercise (described in the Methods section) was undertaken to estimate the prevalence of undiagnosed HIV in the populations using each setting. At a projected 2/1000 persons, measuring the true prevalence of undiagnosed HIV in the study settings to within +/−0·05% would require sampling of >30 000 subjects. Determining true prevalence was therefore not a primary outcome of the study. In order to answer our primary outcomes of feasibility of delivering testing, and of staff and patient acceptability, it was decided to time-limit the trial, with a minimum of twelve weeks testing at each site. We projected reaching 8000 patients for testing, and administering 1000 patient questionnaires.
Electronic patient records were used to capture age, sex, ethnicity, reasons for non-offer of HIV test (including ineligibility) and the outcome of the test offer. All data were anonymised prior to analysis. Data were held in line with the Data Protection Act 1998 and local Trust policies.
Operational (feasibility) data (populations attending, test offer rate, test uptake) were analysed by site, and differences were assessed for associations using the χ2–test, with Yates’ correction applied where appropriate.
Univariate logistic regression analysis was used to examine factors that were associated with the likelihood of test uptake, and this was stratified by clinical site, as differences in accepting an HIV test were found by site. The strata specific adjusted estimates found to be significant (p<0·2) in the univariate model were used to derive a multivariable logistic regression model. In order to ensure no degrees of freedom were lost when building a multivariable model, all variables with missing data were coded with a separate category. The final multivariable model presented has been stratified by clinical site. Stratified logistic regression analyses were repeated, based on data from the subset of patients who completed the questionnaire (the questionnaire collected additional variables: previous HIV testing behaviour and self-reported sexuality).
This study was registered on ClinicalTrails.gov with the Study ID: NCT01138878, in line with local requirements.
Pre-study staff questionnaires were completed by 146 members of staff in each of the departments (57 doctors, 56 nurses, 14 allied health professionals and 19 clerical staff). The return rate was: ED 91%; ACU 76%; OPD 82%; PC 92%. Staff members in all departments were supportive of the provision of HIV tests in non-specialist settings and of the provision of HIV testing within their department (
Proportion agreeing with statement (maximum number of respondents to question) [95% CI] | |||||
ACU (41) | ED (71) | PC (14) | OPD (18) | OVERALL (144) | |
|
|||||
“HIV testing should be available in services other than sexual health and antenatal clinics” | 98% | 94% | 100% | 100% | 96% [93–99] |
“HIV testing should be routinely offered to everyone” | 81% | 84% | 87% | 83% | 83% [77–89] |
“Offering HIV testing to all patients in this department is a good idea” | 79% | 80% | 79% | 83% | 80% [73–86] |
“I would feel comfortable offering HIV testing to all patients in this department” | 42% | 57% | 75% | 63% | 54% [45–63] |
|
|||||
“I don’t think this department provides enough privacy to routinely offer an HIV test to all patients” | 44% | 58% | 0% | 17% | 43% [35–51] |
“I am concerned that patients would have questions I could not answer” | 52% | 72% | 57% | 53% | 63% [55–70] |
“I don’t have time to include routine HIV testing as part of patients’ care in this department” | 40% | 53% | 38% | 27% | 44% [45–53] |
“I would require additional training before routinely offering HIV tests to patients” | 65% | 82% | 56% | 63% | 72% [64–80] |
Across all four sites, there were 13 855 age eligible attendees, of whom 7033 (50·7%) were approached by study personnel (
ED | ACU | PC | OPD | |
|
5541 | 1388 | 6337 | 1776 |
|
5505 (99·3%) [99·1%–99·6%] | 1298 (93·5%) [92·2%–94·8%] | 5352 (84·5%) [83·6%–85·4%] | 1700 (95·9%) [94·8%–96·7%] |
|
4070 (73·9%) [72·7%–75·1%] | 623 (48·0%) [45·3%–50·8%] | 1442 (26·9%) [25·8%–28·1%] | 898 (52·8%) [50·5%–55·2%] |
|
637 (15·7%) [14·5%–16·8%] | 75 (12·0%) [9·5%–14·6%] | 113 (7·8%) [6·5%–9·2%] | 14 (1·5%) [0·8%–2·4%] |
|
3433 | 548 | 1329 | 884 |
|
2121 (61·8%) [60·2%–63·4%] | 384 (70·1%) [66·2%–74·0%] | 1002 (75·4%) [73·1%–77·7%] | 598 (67·6%) [64·6%–70·7%] |
|
6 | 4 | 5 | 0 |
|
2 | 0 | 5 | 0 |
|
4 | 4 | 0 | 0 |
|
1·9 [0·05–3·8] | 10·1 [0·3–20·5] | ||
|
100% | 100% | n/a | n/a |
Of the 6194 patients offered an HIV test, 4105 accepted this offer: test uptake = 66·8% (
Uptake of HIV testing differed significantly by site; range: 61·8%–75·4% (ED vs. PC) (p<0·001). The multivariable model demonstrated that sex, age and the profession of the test provider were significantly associated with test uptake. Male patients and younger patients were more likely to accept tests. Patients were more likely to accept tests when offered the test by a medical student or doctor, rather than by a nurse or a trained tested from a non-clinical background (
A total of 1003 questionnaires were returned during the testing phase. The number of questionnaires received from each site (n) was not proportional (%) to the total numbers of patients offered testing: ED = 528 (15·3%), ACU = 107 (19·5%), OPD = 297 (33·6%), PC = 71 (5·3%). We are unable to calculate a return rate on questionnaires due to operational differences in the questionnaire study across the sites. In relation to the total population (n = 6194) questionnaire respondents were younger (mean age yrs (SD): 36·0 (12·2) vs. 37·6 (12·8)), more likely to be White (69·9% vs. 56·8%), and more likely to have accepted an HIV test (73·5% vs. 66·8% overall) (p<0·001 for all comparisons).
Using data derived from questionnaire respondents only, previous HIV testing behaviour was shown to be associated with test uptake: 48·5% of respondents had never tested for HIV infection, and this group was more likely to accept an HIV test than those who had previously tested (unadjusted RR 1·00 versus 0·65 (95% CI 0·49–0·87) p = 0·007). A trend was observed amongst previous testers with those having tested recently (less than one year ago) being the least likely to accept an HIV test, and those having tested between two and five years previously the most likely (p<0·001; χ2 for trend). The multivariable analysis is shown in
Questionnaire respondents who had declined an HIV test (n = 265) were asked to cite reasons for doing so. Respondents could cite more than one reason. The most commonly cited reasons, across all four sites, were: “I have tested recently” (54%), “I do not think I am risk of HIV infection” (47%), and “I have other health concerns today” (24%).
In response to the statement:
The delivery of routine HIV testing in non-specialist medical settings in areas of high HIV prevalence (more than 2 diagnosed cases/1000 persons) is advocated by National Guidelines in the UK. Whilst data exists in the US, there is, to date, no published evidence on the acceptability or feasibility of delivering wide-scale, routine HIV testing in general medical settings within the context of the UK health service. The HINTS Study was conceived to generate an evidence base to inform the development of testing services in the UK.
Our study shows that it is acceptable to the majority of patients to be offered an HIV test in a non-specialist setting. Uptake of testing is high across all socio-demographic categories, assuaging concerns that higher risk individuals may be less likely to accept a test in such settings. Patients declining to test still found it acceptable to be offered the test. Many patients accepting a test had never tested for HIV infection before. Cases of undiagnosed HIV were identified, and all patients were successfully transferred to clinical care. Staff members from the departments involved in the study were supportive of the initiative, but many identified educational needs and operational barriers that would need to be addressed if they were to deliver routine HIV testing in the long term.
The HINTS study is the first published data regarding routine, non-targeted testing within secondary care in the UK, and provides evidence on how such a strategy may operate within the UK health service. More than 6000 patients were offered an HIV test, and more than 4100 consented. The uptake of HIV testing across the sites of 66·8% compares favourably with (and in many cases is substantially superior to) uptake rates observed in US settings, and from preliminary reports from other pilot studies in operation in the UK and France.
The coverage rates are substantially higher than those reported in most US settings. In ED-based programmes in the US, programmes employing near-patient testing augmented by counsellors have managed to reach much smaller fractions (typically <5%) of all eligible patients passing through.
There are several differences between the sites that merit special mention. The seropositivity of previously undiagnosed HIV is significantly different across the sites, with 1.9/1000 and 10.1/1000 HIV-infected persons identified in the ED and ACU settings, and no new diagnoses made in the OPD and PC settings. It is essential to interpret these figures with due caution, as seropositivity was not a primary outcome of the study (as it was not possible to sufficiently power the study to demonstrate this with statistical confidence) but the significant difference between the ED and ACU populations is of interest; it may be related to the acute nature of the settings and the possibly greater severity of presenting complaints, and certainly warrants further investigation. Differences in prevalence across healthcare settings may have clinical and cost benefit implications. The higher case finding rate in inpatient settings may be tempered by differences in the clinical status of patients diagnosed here, as evidenced by UK surveillance data which demonstrate that patients diagnosed as inpatients have lower median CD4 counts and worse outcomes.
The proportions of patients offered testing and the proportions deemed ineligible to participate are also different. This may reflect differences in the patient populations attending each setting, but may also reflect operational differences such as different staffing arrangements. Large patient throughput in the ED, for example, meant two testers needed to be present to offer testing at all times, whereas in the ACU a lone tester making a daily round was deemed sufficient (see
Finally, there were seven false reactive HIV screening tests, all related to the use of oral fluid. With a total of 11 reactive oral fluid tests of 3717 oral fluid tests done, the specificity of the oral fluid test is 99.8% with a positive predictive value of 36%. The positive predictive value will remain a function of the prevalence of true HIV infection in the population tested, and both this figure and the test specificity are acceptable in this context. All patients with reactive screening tests were successfully informed of their results, and all but one returned had their confirmatory testing undertaken in local sexual health service. All of this group found this process acceptable.
Concerning HIV test uptake, the two adjusted models (overall data and questionnaire respondents only) show age, sex, previous testing behaviour and test provider to be independent predictors of likelihood of test uptake, when stratified by clinical site. However, uptake across all socio-demographic categories was good, specifically amongst MSM and minority ethnic groups. This assuages our concerns that higher risk individuals may be less likely to accept an HIV test in a non-specialist setting. Never having previously tested was also positively associated with test uptake. This suggests offering HIV tests in non-specialist settings brings the opportunity of an HIV test to a population who might not otherwise present for one, and that this group welcomes the opportunity. It is important not to assume that patients who decline an HIV test do so because they find it unacceptable. Many patients declining an HIV test did so for understandable reasons. There remained a significant proportion of test decliners who felt that they “were not at risk of HIV infection” and this merits investigation in future work. However, the majority of questionnaire respondents (92%) found the offer of an HIV test in the non-specialist setting acceptable, including 85% of test decliners. Few differences in attitude were observed between test accepters and decliners. The overall level of acceptability is greater than the 72% reported by Drayton
Whilst staff are accepting of expanding HIV test provision, barriers to implementation at the staff level are evident. Some of the barriers identified are operational in nature, and echo those identified in US ED-based research.
There are limitations to the study. Targeting of test offer at the PC and ED sites may have introduced selection bias, which the multivariable analyses and per-site stratification cannot eliminate. We are unable to report a return rate on the questionnaire study, and the differences in the respondent population strongly suggest that questionnaire distribution (and completion) was not systematic. The proportion of questionnaire respondents across the sites, as a fraction of the total number offered testing, was widely different, for example. The questionnaire was available in English only, which may explain some of the ethnic disparity, but other differences are likely to represent offer bias by study staff. As such, generalisability of the questionnaire data is limited. The unadjusted data and adjusted data must be interpreted within the populations studied. The fact that clinical site does not appear to predict test uptake or acceptability of testing in the adjusted analyses suggests that these results could be generalisable in a broader population setting. However, as previously discussed, many of the differences observed between the sites may have been driven by local operational factors, specific and unique to each situation. Broader conclusions and interpretations ought to be limited to the general, cross-site observations that, we believe, confidently demonstrate that testing is acceptable and efficacious. Whilst we believe we have shown that routine HIV testing in non-specialist settings is effective at diagnosing infected individuals, we have yet to prove such an approach is cost effective. This work is ongoing. However, the 2/1000 persons diagnosed prevalence threshold for routine testing advocated in UK guidelines is drawn from US modelling estimates that suggest a lower threshold of 1/1000 persons is likely to be highly cost effective.
In conclusion, the HINTS study demonstrates that routine HIV testing in non-specialist settings in the UK is highly acceptable to patients and staff. Furthermore, it shows such a strategy is feasible to deliver in a diverse range of settings, and is effective at diagnosing HIV-infected individuals and transferring them to care. As such, the study offers a substantial evidence base to underpin strong recommendations from specialist societies, clinical effectiveness bodies, and public health agencies in the UK that routine HIV testing in secondary and primary care settings in areas of high HIV prevalence, ought to be developed and implemented as a matter of urgency. The specific logistics of local implementation will need to be developed within the local context, after appropriate stakeholder engagement.
We would like to acknowledge the contribution of all members of the HINTS Study Group in the work leading to this publication. The group comprised all authors of this manuscript, plus: Rebecca Cridford, Eric Doku, Robert Ghosh, Shweta Gidwani, Rachel Morris-Jones, Sifiso Mguni, Danna Millett, David Mummery, Matthew Pearn, Karen Phekoo, Caroline Rae, Malur Sudhanva and Tanya Welz (Patient and Public Engagement: Gus Cairns and John Holland). All members were involved in the conduct of the study: provision of HIV testing, distribution of patient and staff literature and questionnaires, or laboratory procedures. GC and JH reviewed the protocol and all study literature, and participated in regular study review meetings.
All original data is available from the Corresponding Author on request.