The authors hereby confirm that Dr. Eduard J. Beck is a PLOS ONE Editorial Board member but this does not alter the authors' adherence to PLOS ONE Editorial policies and criteria. The authors have declared that no competing interests exist.
The authors have declared that no competing interests exist.
Conceived and designed the experiments: EFL EJB MLB. Performed the experiments: RM SM EFL SSA. Analyzed the data: EFL EJB RM SM EFL SSA. Contributed reagents/materials/analysis tools: EFL EJB MLB. Wrote the paper: EFL EJB.
In many high-income countries with low HIV prevalence, significant numbers of persons living with HIV (PLHIV) remain undiagnosed. Identification of PLHIV via HIV testing offers timely access to lifesaving antiretroviral therapy (ART) and decreases HIV transmission. We estimated the effectiveness and cost-effectiveness of HIV testing in the United Kingdom (UK), where 25% of PLHIV are estimated to be undiagnosed.
We developed a dynamic compartmental model to analyze strategies to expand HIV testing and treatment in the UK, with particular focus on men who have sex with men (MSM), people who inject drugs (PWID), and individuals from HIV-endemic countries.
We estimated HIV prevalence, incidence, quality-adjusted life years (QALYs), and health care costs over 10 years, and cost-effectiveness.
Annual HIV testing of all adults could avert 5% of new infections, even with no behavior change following HIV diagnosis because of earlier ART initiation, or up to 18% if risky behavior is halved. This strategy costs £67,000–£106,000/QALY gained. Providing annual testing only to MSM, PWID, and people from HIV-endemic countries, and one-time testing for all other adults, prevents 4–15% of infections, requires one-fourth as many tests to diagnose each PLHIV, and costs £17,500/QALY gained. Augmenting this program with increased ART access could add 145,000 QALYs to the population over 10 years, at £26,800/QALY gained.
Annual HIV testing of key populations in the UK is very cost-effective. Additional one-time testing of all other adults could identify the majority of undiagnosed PLHIV. These findings are potentially relevant to other low-prevalence, high-income countries.
Following the recognition that antiretroviral therapy (ART) has both preventive and therapeutic benefits
In 2011, an estimated 96,000 people (95% credible interval 90,800–102,500) were living with HIV in the United Kingdom (UK), with 24% (19%–28%) unaware of their infection status
Identifying undiagnosed PLHIV through a voluntary counseling and testing (VCT) program could offer timely access to ART, thereby reducing morbidity and mortality, and could decrease unsafe behaviors
Recent recommendations from the British HIV Association, the British Association for Sexual Health and HIV, and the British Infection Society propose expanding VCT provision to all health settings, and offering routine testing to all adults aged 15–64 years where HIV prevalence exceeds 0.2%
However, with most HIV testing still concentrated in genitourinary medicine and antenatal clinics
To date, few costing studies of population-based testing programs in the UK have been performed. A 2011 costing exercise estimated that the cost of detecting a PLHIV in the UK ranged between £2,222 and £3,793 (2010 prices)
We populated a previously published dynamic HIV epidemic model with epidemiological, behavioral, and cost data from the UK. All model parameters are provided in
Variable | Base Value | Range | Source |
People from HIV-endemic countries | |||
Men | 420,000 | 300,000–500,000 | |
Women | 406,000 | 300,000–500,000 | |
PWID (men and women) | 200,000 | 100,000–400,000 | |
MSM | 800,000 | 600,000–1,200,000 | |
All others | |||
Men | 30,643,254 | - | |
Women | 31,618,713 | - | |
People from HIV-endemic countries | |||
Men | 2.5% | 2–8% | |
Women | 5.0% | 3–8% | |
PWID (men and women) | 1.2% | 0.6–4% | |
MSM | 5.0% | 3–6% | |
All others | |||
Men | 0.033% | 0.02–0.05% | |
Women | 0.033% | 0.02–0.05% | |
Men | 0.00312 | 0.0030–0.0035 | |
Women | 0.00192 | 0.0015–0.0025 | |
PWID (excess) | 0.01 | 0.001–0.02 | |
Men | 0.0188 | 0.01–0.025 | |
Women | 0.0185 | 0.01–0.025 | |
Men | 0.0261 | 0.02–0.05 | |
Women | 0.0251 | 0.02–0.05 | |
Men | 0.025 | 0.01–0.05 | |
Women | 0.025 | 0.01–0.05 | |
Men | 2.5% | 2–8% | Assumed |
Women | 5.0% | 3–8% | Assumed |
Heterosexual (female to male) | |||
Acute HIV | 0.20 | 0.10–0.30 | |
Asymptomatic HIV | 0.02 | 0.01–0.04 | |
Symptomatic HIV | 0.03 | 0.01–0.04 | |
AIDS | 0.05 | 0.03–0.06 | |
Heterosexual (male to female) | |||
Acute HIV | 0.30 | 0.10–0.40 | |
Asymptomatic HIV | 0.03 | 0.02–0.05 | |
Symptomatic HIV | 0.04 | 0.02–0.05 | |
AIDS | 0.08 | 0.05–0.10 | |
Homosexual (male to male) | |||
Acute HIV | 0.40 | 0.20–0.50 | |
Asymptomatic HIV | 0.04 | 0.03–0.06 | |
Symptomatic HIV | 0.05 | 0.03–0.06 | |
AIDS | 0.10 | 0.08–0.15 | |
MSM | 4.2 | 2–10 | |
MSM | 56% | 25–75% | |
PWID | 3.0 | 2–5 | |
All other heterosexuals | 1.5 | 1–2 | |
PWID | 17% | 10–25% | |
All other heterosexuals | 20% | 7–41% | |
Fraction of men who are circumcised | 16% | 5–25% | |
Reduction in heterosexual HIV transmission due to male circumcision, % | 50% | 48–60% | |
Acute HIV | 0.016 | 0.008–0.040 | |
Asymptomatic HIV | 0.002 | 0.001–0.005 | |
Symptomatic HIV | 0.003 | 0.001–0.005 | |
AIDS | 0.003 | 0.001–0.005 | |
Average number of injections per year | 432 | 300–564 | |
Fraction of injections that are shared, % | 17% | 16–22% | |
6 | 4–12 | ||
Untreated | 0.13 | 0.10–0.20 | |
Treated | 0.08 | 0.05–0.10 | |
Untreated | 0.33 | 0.20–0.50 | |
Treated | 0.06 | 0.05–0.10 | |
Untreated | 0.40 | 0.25–0.50 | |
Treated | 0.25 | 0.10–0.50 | |
Acute HIV | 1% | 0–5% | Calculated |
Asymptomatic HIV | 40% | 30–50% | |
Symptomatic HIV | 16% | 10–20% | |
AIDS | 43% | 30–50% | |
High-risk persons | |||
PWID | 77% | 50–80% | |
MSM | 25% | 10–50% | |
People from HIV-endemic countries | 25% | 10–60% | |
All other persons | 10% | 5–50% | |
HIV | 10% | 0–20% | |
AIDS | 20% | 8–50% | |
Reduction in sexual behavior among persons identified as HIV-positive, % | 25% | 0–50% | Assumed |
High-risk persons | |||
PWID | 6% | 0–20% | |
MSM | 46% | 25–60% | |
People from HIV-endemic countries | 22% | 10–40% | |
All other persons | |||
Men | 75% | 50–90% | |
Women | 23% | 20–60% | |
Annual ART entry rate if CD4 <350 cells/mm3, % | 37% | 0–50% | |
Retention in care 12 months after diagnosis, % | 86% | 50–95% | |
Reduction in injection infectivity due to ART, % | 50% | 25–75% | |
Reduction in sexual infectivity due to ART, % | 96% | 50–99% | |
Uninfected | 1.00 | – | |
Acute HIV | 0.89 | 0.60–0.95 | |
Unidentified asymptomatic HIV | 0.91 | 0.85–0.95 | |
Identified asymptomatic HIV at 1 year | 0.84 | 0.80–0.90 | |
Identified asymptomatic HIV at 2+ years | 0.89 | 0.85–0.95 | |
Unidentified symptomatic HIV | 0.79 | 0.70–0.80 | |
Identified symptomatic HIV | 0.72 | 0.70–0.80 | |
Symptomatic HIV treated with ART | 0.83 | 0.82–0.87 | |
Unidentified AIDS | 0.72 | 0.60–0.75 | |
Identified AIDS | 0.72 | 0.60–0.75 | |
AIDS treated with ART | 0.82 | 0.82–0.87 | |
PWID (multiplier) |
0.90 | 0.80–1.00 | |
Untreated | £1,862 | £1,220–£2,505 | |
Treated | £7,793 | £5,119–£10,467 | |
Untreated | £5,447 | £3,889–£7,005 | |
Treated | £9,305 | £7,093–£11,517 | |
Untreated | £13,457 | £10,163–£16,750 | |
Treated | £16,892 | £13,962–£19,823 | |
Annual non–HIV-related health care costs | £2,571 | £1,950–£3,213 | |
Annual cost of ART | £14,294 | £11,770–£16,140 | |
Cost of HIV ELISA antibody test | £8 | £3–£12 | |
Cost of behavior counseling/hour |
£36 | £26–£46 | |
Annual cost of ancillary PWID services | £7,700 | £3,000–£10,000 | |
Annual discount rate | 3% | 0%–5% |
ART = antiretroviral therapy; MSM = men who have sex with men; PWID = people who inject drugs.
* Quality of life for all PWID is multiplied by this factor, across all health states.
** Under current UK guidelines
The length of the post-test counseling session depends on the outcome of the HIV test.
The model (schematically illustrated in
Individuals entered the population through maturation or immigration to the UK from HIV-endemic countries. We assumed that 2.5% of male immigrants and 5.0% of female immigrants entering from HIV-endemic countries were HIV-infected, similar to the proportion living in the UK who were HIV-infected. Individuals exited the population due to age maturation or death.
HIV transmission occurred via sexual contact among MSM, sexual contact between men and women, and needle sharing among PWID. For each type of sexual contact we estimated the annual number of partnerships, rate of condom use, and transmission probability per partnership. For needle sharing contacts, we estimated the annual number of injections, average needle sharing rates, and probability of transmission per shared needle. Each sexual transmission probability depended on the infected person's gender, disease status, and treatment status and the uninfected person's gender and circumcision status, if male. Upon acquiring HIV infection, individuals passed through the disease states at a rate inversely proportional to the time spent in each state
Under the status quo scenario, we assumed that 25% of MSM, 25% of people from HIV-endemic countries, 77% of PWID, and 10% of other adults in the population received an HIV test in the last 12 months
Current ART initiation rates vary widely, ranging from only 6% of PWID to 46% of MSM and 75% of diagnosed men living with HIV beginning ART at CD4 count <350 cells/mm3. Each year thereafter, an additional 37% of diagnosed individuals commence ART at lower CD4 counts. In our model, we considered a scenario where 75% of PLHIV start ART at CD4 count <350 cells/mm3, with additional initiation thereafter. Attrition and loss to follow-up also reduce the proportion of HIV-infected people with viral suppression, with 86% adherent after 12 months. Finally, we assumed that effective ART reduced HIV transmission probabilities via needle-sharing (by 50%) and sexual contact (by 96%)
For each strategy, we projected HIV prevalence and incidence over a 10-year time horizon, and lifetime quality-adjusted life years (QALYs) accrued in the population. We calculated lifetime healthcare costs for all individuals in the population, which depended on gender, HIV infection and treatment status, and injection drug use status, and we included the cost of VCT and ART per person. We estimated HIV infections averted and the incremental cost-effectiveness ratio (ICER) for each strategy. Taking a societal perspective, we considered all health care costs and savings regardless of source or beneficiary
We compared our model-projected outcomes to available data on HIV prevalence, incidence, and diagnosis trends, and found close approximations (
Epidemic Outcome (2011) | Model Estimate | Data |
Total annual new HIV infections | 3,471 | 3,640 |
Total annual new HIV diagnoses | 6,125 | 6,280 |
Men | 4,386 | 4,470 |
Women | 1,739 | 1,810 |
People living with HIV/AIDS | ||
Men who have sex with men (MSM) | 40,000 | 40,100 |
People who inject drugs (PWID) | 2,400 | 2,300 |
Men from HIV-endemic countries | 10,500 | 10,500 |
Women from HIV-endemic countries | 20,300 | 20,300 |
All other adults | 20,546 | 20,200 |
Total | 93,746 | 93,400 |
HIV-infected people aware of status | 71,013 | 73,400 |
HIV-infected people receiving ART | 59,000 | 61,510 |
* Source: Health Protection Agency. HIV in the United Kingdom: 2012 Report
If current HIV testing and treatment levels persist, we projected that overall HIV incidence will remain stable at approximately 3,500 new infections per year, and increase slightly by around 3% by year 2022 (
The six graphs correspond to six different risk groups in the UK, with projected annual HIV incidence per 100,000 people shown under current testing and treatment levels (black solid line), universal annual testing of all adults (blue dashed), or universal annual testing coupled with antiretroviral therapy (ART) initiation of 75% at CD4 <350 cells/mm3 (red dotted). The cumulative number of new HIV infections over 10 years is given in
A comprehensive VCT campaign offered to all adults identified a substantial number of previously undiagnosed PLHIV and prevented a substantial number of new infections, but the aggregate impact depended significantly on the effectiveness of testing and counseling at reducing sexual partnerships (
Each bar corresponds to modeled estimates of new HIV infections over the next 10 years, assuming a 25% reduction in sexual partnerships following HIV diagnosis. For each strategy, the higher estimate (top of black line) corresponds to the scenario with no partnership reduction, and the lower estimate (bottom of black line) corresponds to a 50% partnership reduction following diagnosis. The time in parentheses corresponds to the testing interval, and “once” refers to one-time testing of individuals not in the key populations we considered. “ART (75%)” refers to 75% antiretroviral therapy initiation of at CD4 <350 cells/mm3.
HIV Infections (Prevented) over 10 years |
|||||||
Scenario | HIV+People Identified over 10 Years | No Reduction in Sexual Behavior | 25% Reduction in Sexual Behavior | 50% Reduction in Sexual Behavior | Incremental Costs (billions) |
Incremental QALYs |
Incremental Cost/QALY Gained |
Status quo | 48,932 | 34,524 | 28,324 | 22,410 | – | – | – |
Universal testing every 3 years | 54,046 | (325) | (521) | (639) | £1.25 | 13,000 | £96,200 |
Universal testing every 2 years | 59,424 | (1,032) | (1,705) | (2,181) | £2.18 | 32,900 | £66,300 |
Universal testing every 1 year | 65,598 | (1,887) | (3,180) | (4,081) | £4.61 | 57,400 | £80,300 |
Universal testing every 1 year+ART | 62,640 | (5,411) | (5,709) | (5,749) | £7.41 | 161,700 | £240,000 |
High-risk testing every 1 year, Low-risk testing every 2 years | 64,017 | (1,768) | (2,986) | (3,847) | £2.37 | 53,100 | £44,700 |
High-risk testing every 1 year, Low-risk testing one time | 58,391 | (1,461) | (2,555) | (3,368) | £0.75 | 42,900 | £17,500 |
High-risk testing every 1 year, Low-risk testing one time+ART | 56,293 | (4,920) | (5,066) | (5,044) | £3.49 | 145,300 | £26,800 |
* Model-projected new HIV infections in the UK between 2013 and 2022 under status quo testing and treatment levels, and projected number of averted infections with increased testing and treatment in parentheses. Three scenarios for the reduction in risky sexual behavior following HIV diagnosis are given.
** Incremental costs and quality-adjusted life years (QALYs) discounted to the present, are relative to the status quo, assuming a 25% reduction in risky sexual behavior following HIV diagnosis.
Strategies that are dominated, i.e., have higher costs but generate fewer health benefits than a combination of other strategies.
Although universal HIV testing has the potential to diagnose the greatest number of people and link them to lifesaving treatment, its relative cost-effectiveness may impede its implementation. The population-wide gain in health benefits with annual testing ranged from 46,000–65,000 additional discounted QALYs over 10 years. Our model projected that annual testing of all adults cost £67,400/QALY gained under the optimistic assumption of a 50% reduction in sexual partnerships following diagnosis, and £106,000/QALY gained with no partnership reduction (
The incremental costs (x-axis) and QALYs (y-axis) of different HIV testing and treatment scenarios are shown, relative to status quo levels. The blue points correspond to universal HIV testing strategies for all adults, with testing every one, two, or three years. The green points correspond to targeted strategies, with annual testing for high-risk persons and testing every two years or one-time for all other persons. The red points correspond to an expanded HIV testing program coupled with 75% antiretroviral therapy initiation of at CD4 <350 cells/mm3. The solid black line marks the cost-effectiveness frontier, or the set of strategies that is most economically efficient, and the corresponding incremental cost-effectiveness ratios are given. Costs and QALYs are discounted at 3% annually, and include the direct costs of the programs over 10 years, as well as the lifetime costs and QALYs of all individuals in the population. HR = high-risk, and includes men who have sex with men (MSM), people who inject drugs (PWID), and men and women from HIV-endemic countries. LR = low-risk, and includes men and women who do not belong to the identified key populations. ART = antiretroviral therapy. QALY = quality-adjusted life year.
A program that provided annual HIV testing to MSM, PWID, and people from HIV-endemic countries, coupled with one-time screening of all other adults, averted nearly as many infections as universal screening, for a fraction of the cost (
A slightly more comprehensive program offering repeat HIV testing every two years to lower-risk persons prevented 5–17% of new HIV infections over 10 years. However, the small marginal benefit of repeat testing suggests that a one-time testing program identified most undiagnosed PLHIV not in the key populations we considered; more frequent testing of these individuals only identified people who later acquired HIV. This strategy cost £36,000–£62,000/QALY gained and was more cost-effective than universal testing.
The above analyses assumed that ART initiation rates remained at current levels. If ART access increased concomitantly with expanded testing, the health benefits increased dramatically. Universal annual HIV testing with 75% ART initiation for infected individuals with CD4 <350 cells/mm3 prevented up to 26% of new HIV infections, adding up to 165,000 QALYs over 10 years at a cost of approximately £240,000/QALY gained, relative to annual testing alone.
If expanded ART was instead linked with targeted HIV testing of high-risk persons every year and one-time testing of all other adults, up to 23% of infections were prevented, adding 148,000 QALYs – or nearly 90% of the benefits accrued with universal testing and ART – at a cost of only £26,800/QALY gained (
As with any extensive HIV screening program, the risk of false diagnoses should be considered. Given an HIV testing sequence with 99.5% sensitivity and 99.9994% specificity
We varied all model parameters in sensitivity analyses (
The epidemic's baseline trajectory also affected cost-effectiveness estimates, although the relative cost-effectiveness ranking remained unchanged. If HIV prevalence in each risk group was 50% greater than initially assumed, then annual testing of high-risk persons along with one-time testing of all other adults in the population cost £15,000/QALY gained, compared to £17,500 in the base case. With 50% lower prevalence, this strategy exceeded £25,000/QALY gained, because more people must be screened to identify each PLHIV. Similarly, if HIV transmission probabilities were 50% higher than initially assumed, the cost-effectiveness of this strategy improved to £7,000/QALY gained; with 50% lower transmission probabilities, the cost-effectiveness worsened to £36,000/QALY gained. Increased testing and counseling costs, increased annual ART costs, reduced adherence levels, and reduced ART effectiveness all slightly mitigated the benefits of expanded HIV testing, but again the relative cost-effectiveness of the strategies under consideration remained unchanged.
The HIV epidemic in the UK currently remains relatively concentrated among MSM and people born in HIV-endemic countries and related communities, and efforts to identify undiagnosed PLHIV and link them to care have been insufficient. With one-quarter of PLHIV unaware of their HIV status, and approximately 3,500 people continuing to acquire HIV every year, the need for improved HIV testing and counseling is evident. Who to test – and how frequently – had not yet been systematically evaluated in the UK. Our study addresses these important policy questions and provides strong support for increased HIV testing of adults.
Our analysis suggests that periodic testing of all adults in the UK, regardless of risk status, could prevent 18% of new infections with existing treatment levels, or nearly 30% of new infections if treatment levels increase simultaneously. Nevertheless, universal testing is less efficient than targeted testing, requiring four times as many tests to identify each PLHIV. A targeted testing approach that offers annual HIV testing to MSM, PWID, and people from HIV-endemic countries, along with one-time screening of all other adults, could offer 80% of the benefits of universal testing for only 14% of the cost over 10 years. At £17,500/QALY gained, this strategy is well below established UK cost-effectiveness thresholds of £20,000–£30,000/QALY gained
As with any risk-based disease screening program, a key challenge of the proposed targeted HIV testing strategy is identifying high-risk individuals and incentivizing VCT uptake. An advantage of one-time universal testing of all adults is that it may reduce stigma and reticence among key populations to get tested. Of course, any wide scale VCT program should focus efforts on linking PLHIV to care and ensuring effective counseling aimed at reducing risky sexual behavior following diagnosis.
Our analysis has several limitations. As with many epidemic models, we simplified the complex dynamics of HIV disease progression, development of resistance, and changes in viral suppression. Although our model captured the reduction in primary transmission to the partners of persons diagnosed with HIV, as well as secondary transmission to those partners' partners, we assumed a standard proportional mixing model of partnership selection. Due to data limitations we did not include preferential mixing by HIV status, race or immigration status, nor did we consider differential condom use by HIV status. We assumed similar HIV prevalence levels for newly arriving immigrants and those already living in the UK, due to a lack of data on HIV infection rates of those just arriving. Improved data on baseline demographics, sexual behavior and other risk behaviors would allow for more refined estimates of testing impact. Finally, we estimated costs on a per person basis using current estimates of the costs of HIV testing and counseling and treatment for HIV infection. If expansion of HIV testing coverage were linked with a broad national campaign or with significant changes in delivery of health care services then costs could be higher than we have estimated.
Despite its limitations, in particular the lack of current accurate UK epidemic data, our study showed that targeted HIV testing of specific key populations is cost-effective in the UK. This finding is potentially relevant to other low-prevalence, high-income countries
The success of any country's response to its HIV epidemic rests on ‘know your epidemic’ – the key populations that drive the epidemic – and ‘know your response’ – especially interventions that directly involve those populations
One way to reduce potential stigma or discrimination for such populations is to link HIV testing to other health checks, such as those for hypertension, diabetes or other conditions, thereby removing a direct link with HIV. This has been done in some middle-income countries
Without increased HIV testing, HIV will continue to spread in key populations in the UK and other countries, increasing the number of PLHIV, and necessitating ongoing expansion of HIV services. With a growing number of people projected to receive ART and benefit from prolonged life expectancy
(DOCX)