You are on page 1of 23

Supplementary appendix

This appendix formed part of the original submission and has been peer reviewed.
We post it as supplied by the authors.

Supplement to: Baggaley RF, Irvine MA, Leber W, et al. Cost-effectiveness of screening
for HIV in primary care: a health economics modelling analysis. Lancet HIV 2017;
published online July 30. http://dx.doi.org/10.1016/S2352-3018(17)30123-6.
Supplementary Appendix

Supplement to: Baggaley RF, Irvine MA, Leber W, Cambiano V, Figueroa J, McMullen H,
Anderson J, Santos AC, Terris-Prestholt F, Miners A, Hollingsworth TD, Griffiths CJ. Cost-
effectiveness of screening for HIV in primary care: a health economics modelling analysis

Supplementary Methods
Costs for the RHIVA2 intervention
Start-up activities for the team included the developing of protocols, contacting practices and offering financial incentives
for trial participation, education and training of health professionals including nurse follow-up activities. Thus, start-up
costs referred to activities performed by the RHIVA2 team, the time spent by health professionals taking part in the initial
training sessions, training material, supplies, transport and equipment used to conduct the training, and enrolment
incentives paid to the surgeries.

Recurrent costs included costs incurred with rapid HIV testing, quality assurance, wastage of HIV test kits, and with the
increase in serology testing due to the RHIVA2 intervention. In total, seven additional patients were newly diagnosed as
HIV positive in intervention practices and we have assumed these new diagnoses were as a result of the RHIVA2
intervention. Thus, the resulting extra cost for these additional serology tests was assumed as part of the RHIVA2
recurrent costs. Costs for the quality assurance programme were payments made to the local laboratory (NHS Barts Health
NHS Trust) for provision of their services. We assumed that recurrent education and training to practices was part of the
quality assurance programme, which was therefore not included as a separated line of costing. Costs of CD4 count,
serology, and viral load tests included laboratory staff time.

The salaries of RHIVA2 staff involved in the planning of the trial, running the initial training sessions and any follow up
activity with the practices were assumed as pro rata of their stated salaries. Staff employed by the NHS and responsible
for HIV testing were assumed to be paid average salaries for their scales, according to their qualifications and positions,
using the NHS pay rates website.1 An additional 20% was added to all salary costs to allow for the costs of pension and
National Insurance contributions incurred by the NHS.

Staff time was estimated as cost per minute and was calculated as the staff annual salary divided by 99,000 minutes
(220 days multiplied by 7·5 hours of work per day multiplied by 60 minutes). We assumed that staff effectively worked
220 days per year instead of 365 days. This was to account for annual leave, bank holidays and sick leave. We also
assumed a week of 37·5 hours of work for five days, working 7·5 hours per day. Staff salaries varied by scale and
qualification, and costs were weighted by the number of professionals.

HIV screening cost-effectiveness model: further details


For simplicity, migration is not explicitly modelled, but the constant incidence rate used in the model, taken from Public
Health England (PHE) data, implicitly includes infections imported into Hackney from elsewhere.

HIV screening cost-effectiveness model: state variables


For individuals in HIV infection stage :
HIV-infected individuals, undiagnosed stage 1. .4
HIV-infected individuals, diagnosed: first (short-term) phase: diagnosed when infection was at stage
1. .5
HIV-infected individuals, diagnosed: second (long-term) phase: diagnosed when infection was at stage
1. .5
Secondary HIV infections, from undiagnosed HIV-infected individuals in stage 1. .5
Secondary HIV infections, from diagnosed HIV-infected individuals in the first (short-term) phase who were
diagnosed when infection was at stage 1. .5
Secondary HIV infections, from diagnosed HIV-infected individuals in the second (long-term) phase who were
diagnosed when infection was at stage 1. .5

HIV screening cost-effectiveness model: model equations


The model structure using mathematical notation is shown in Figure S1. Following infection, individuals progress
through the three phases of infection (undiagnosed, diagnosed short-term and diagnosed long-term) and at a certain
point in their infection, defined by the diagnosis rate, their infection is diagnosed and they move from undiagnosed to
the short-term diagnosis phase. The corresponding differential equations for the model are shown below:

 
⁄ 1
⁄ 2. .4
⁄ 1. .4

⁄ 1 1. .5
where is HIV incidence in Hackney, is proportion of HIV infections diagnosed at a general practice, is progression
to the next undiagnosed infection stage for 1. .4 and is set to zero for 5 because it is assumed all patients are
diagnosed before (HIV-related) death. is background death rate, is rate of diagnosis in the absence of the intervention
for those in HIV infection stages 1 to 4, which is set to zero for stage 1 (it is assumed that no diagnosis occurs during
acute infection, and that diagnosis accompanies progression to stage 5 [CD4<200 cells/μL]). Diagnoses rates to are
constrained to be equal. is the progression rate from the first (short-term) phase of diagnosed HIV infection to the
second (long-term) phase, which is set to one year and does not depend on the infection stage during which an individual
is diagnosed. is the proportion who are diagnosed when in CD4 band who die before reaching the long-term phase of
diagnosed infection ( 1. .5), which is set to zero for those with CD4 counts ≥350 cells/μL at diagnosis. For the
intervention scenario, the rate of diagnosis is changed from .. to , where represents the diagnosis rate due to rapid
HIV testing.

Secondary infections
Cumulative number of secondary infections over the time horizon is calculated as follows:
⁄ 1 1 1 1 1
1 1 1

⁄ 1 1 1 1 1 1 1
1 1 1

⁄ 1 1 1 1
1 1 1
1 1 1

where , and refer to HIV transmission probabilities per partnership for male-to-female, female-to-male
and MSM transmission respectively, from HIV-infected undiagnosed individuals in HIV stage . Baseline infectiousness
of long-term diagnosed (i.e. before discounting due to ART) is set to be the infectiousness of the CD4 >500 and 350-500
cells/μL groups. , and are proportions of those that are infected that are heterosexual males, heterosexual
females and MSM, respectively. and are mean number of sexual partnerships formed per year for heterosexuals
and MSM, respectively. and are proportion of sexual partnerships involving consistent condom use, for
heterosexuals and MSM, respectively. represents prevalence of circumcision amongst male partners of those infected
and represents the reduction in susceptibility to HIV acquisition for males that are circumcised.
represents the reduction in sexual behaviour among persons identified as HIV positive; represents the reduction in
sexual infectivity due to ART. The proportion of individuals on ART is higher in lower CD4 bands, defined by parameter
for short-term and for long-term diagnosed. HIV prevalence in the Hackney 15-59 year population is assumed to
remain constant over the time horizon (assumed values shown in Table S2) and is defined as , for
MSM, heterosexual women and heterosexual men, respectively.

The number of secondary infections generated in the model population over the time horizon was estimated by integrating
the secondary infection equations over the time horizon. Number of secondary infections averted was calculated as the
difference between this integral by running the model with diagnosis rates and proportion diagnosed in general practices
estimated using the control arm data, and then using diagnosis rates fitted to the intervention arm data.

Estimates of proportions within each risk group are taken from reported exposure category and gender of newly diagnosed
HIV patients in London, 2012.2 Data for new diagnoses were chosen over those for people living with HIV/AIDS because
they would better represent the future spectrum of diagnosed infections by risk group, rather than the past. As assumed
elsewhere, due to data limitations, we did not include preferential mixing by HIV status, ethnicity or immigration status,
nor did we consider differential condom use by HIV status.3 We have not included potential protection through pre-
exposure prophylaxis because despite its recently proven effectiveness in reducing transmission,4 it is not currently funded
by the NHS and its use in the UK remains low.

Impact of screening on HIV transmission


 
A successful screening intervention resulting in earlier diagnosis (as demonstrated in this trial by the larger number
diagnosed, at higher CD4 counts, for the intervention arm) implies a public health impact of the intervention on onwards
HIV transmission going beyond benefits to the diagnosed individual. Upon diagnosis, individuals are likely to modify
their risk behaviour to a certain extent (fewer partners, increased condom use etc.) and have lower infectiousness
(treatment reduces viral load and thus infectiousness,5 lower rates of other sexually transmitted infections [which may
increase infectiousness] as a result of reductions in risk behaviour). A comprehensive evaluation of this impact requires
a full understanding of the HIV epidemic amongst all relevant risk groups (men who have sex with men [MSM],
heterosexual migrants and non-migrants, injecting drug users [IDUs]), using all relevant and available data on sexual
behaviour and HIV epidemiology, as specific to Hackney as possible. Such an analysis has been attempted for The
Netherlands6, 7 but a similar exercise specific to the UK or even to London, is beyond the scope of this article.

As an alternative, our analysis captures transmission from those diagnosed to their sexual partners (secondary infections).
We therefore do not capture onward transmissions from these secondary infections and consequently, our results will
underestimate the full impact of improved HIV detection on onward HIV transmission to some extent.

Model fitting
Diagnosed HIV incidence and prevalence for Hackney have been relatively stable since 2008,8 therefore we assumed a
constant HIV incidence in the model. We first fitted the equilibrium state of the model to incidence in the control arm of
the trial, using the calculation that the proportion of HIV-infected individuals in the model diagnosed in control and
intervention general practices is . is the proportion of all HIV infections that are diagnosed in practices, is the
number of control practices (equal to the number of intervention surgeries, n=20), and is total number of practices in
Hackney. Using this number of HIV-positive individuals in the model, diagnosis rates for control and intervention
practices were fitted to the number of infections diagnosed and the proportions diagnosed at each CD4 band for
intervention and control practices. The model was fitted using maximum likelihood, with 95% credible intervals (95%CIs)
from the profile likelihood. Given the short duration of acute infection compared to the overall long duration of
asymptomatic HIV infection, it was assumed that no infections were diagnosed during the acute phase. It was also
assumed that diagnosis rates for CD4 bands >500 cells/μL, 350-500 cells/μL, and 200-349 cells/μL are equal, since those
who are diagnosed cannot know their CD4 status at this stage. For those with CD4 counts <200 cells/μL, we initially
proposed a model with a diagnosis rate for this group. When fitting the model, the fitted values were equivalent to almost
instantaneous diagnosis, with large 95%CIs, which we interpreted as reflecting rapid diagnosis in this stage due to high
levels of symptomatic presentations. We therefore altered the model to allow for rapid diagnosis once this stage was
reached. Once the model had been fitted to the control arm, a diagnosis rate for the intervention arm was estimated by
fitting to the data on the number and CD4 counts of new diagnoses in the intervention arm in a similar way to that in the
control arm.

Calculation of HIV healthcare costs


Cost data used are shown in Table S3. ART costs were calculated from data on the cost of healthcare for HIV patients
in London provided as part of a Freedom of Information (FOI) request made by Kevin Kelleher for the financial year
2013-14 (personal communication). FOI data were preferred for antiretroviral costs because the NHS pays less for
antiretrovirals than indicated in the British National Formulary (BNF). The FOI dataset also provided information on
healthcare costs for HIV patients other than antiretrovirals, but we used estimates from Beck et al9 for these because
more detail is provided on cost by CD4 count. It is assumed that each diagnosed HIV patient has four CD4 count and
four viral load tests per year, and one ART resistance test preceding ART initiation. This one-off resistance test is
averaged over an assumed 45 years lived with HIV infection. The cost of the HIV diagnostic test is omitted because this
cost is part of the intervention and is another one-off cost of HIV infection which, when averaged over the patient’s
lifespan to calculate an estimate of annual HIV healthcare cost, is a very small cost. Future unrelated healthcare costs
(i.e. non-HIV-related costs) are excluded from the analysis as they can lead to uninformative results (at the extreme,
their inclusion can lead to the abandoning of all medical care being the most cost-effective approach10). Costs were
discounted at a rate 3·5% per year11 using the standard discounting model where discounting factors are equal to
1 where is the discount rate, is the year in which the event occurs and is the baseline year (set to
2012).12 Costs were adjusted to 2012 prices.

⁄ 1 1

where is healthcare and treatment costs for undiagnosed in stage ; are costs for diagnosed in the short-term phase,
not receiving ART; are costs for diagnosed in the short-term phase who are receiving ART; are costs for
diagnosed in the long-term phase, not receiving ART; and are these costs for diagnosed in the long-term phase who
are receiving ART. and are the proportions in stage on ART for diagnosed short-term and long-term phases,
respectively.
There is relatively little information available on how long-term HIV healthcare costs are related to CD4 count at
diagnosis. Therefore we explored three plausible scenarios, where those diagnosed with late infection have higher

 
healthcare costs. These are summarised in Table S4 and are based on findings reported by Beck et al 20119 (UK, Scenario
1) and those of Krentz and Gill 201213 (Canada, Scenarios 2 and 3). Beck et al9 reported 21% higher costs for patients on
ART who were diagnosed at CD4 ≤200 cells/μL. Scenario 1 has 25% rather than 21% higher costs for long-term
diagnosed patients on ART and who were diagnosed at CD4 ≤200 cells/μL because our cost calculations include data
from other sources for price of ART and HIV-related tests (details shown in Table S3). Krentz and Gill reported that, for
patients accessing care in South Alberta between 1995 and 2010, direct medical costs remained 1.6 times higher for late
presenters diagnosed with CD4 count ≤350 cells/μL many years after diagnosis. We therefore explored two scenarios: 2)
treatment and care of long-term diagnosed at CD4 ≤350 cells/μL was weighted to be 1.6 times the cost of those diagnosed
at >350 cells/μL; and 3) treatment and care of long-term diagnosed at CD4 ≤350 cells/μL was weighted to be 1.6 times
the cost of those diagnosed at >350 cells/μL, with costs for those diagnosed at CD4 ≤200 cells/μL twice that for those
diagnosed at 200-349 cells/μL. In each case, weights were calculated so that total costs of all long-term diagnosed should
be the same as in Scenario 1, assuming proportions diagnosed in each CD4 band were as provided by PHE (unpublished
data). Parameter values are shown in Table S4.

Calculation of QALYs
QALYs gained under the intervention scenario, by reducing HIV-related mortality, improving the quality of life by
diagnosing infected individuals earlier and averting secondary infections through earlier diagnosis, are calculated by
estimating and comparing cumulative QALYs by running the model with intervention and control arm-specific
parameters over the time horizon. QALYs are calculated by multiplying each state variable in the model by its associated
QoL estimate (see parameter values, Table S2) and then integrating over the time horizon and comparing QALYs for the
model population for control and intervention arm-fitted HIV diagnosis rates:

where and are QoL measures for undiagnosed and diagnosed infected individuals respectively, in CD4 band
where 1. .4. QoL for the symptomatic (CD4<200 cells/μL) short-term diagnosed state variable is weighted by the
proportion receiving ART, for QoL of those with identified symptomatic HIV and QoL of those with symptomatic HIV
treated with ART: 1 , where is the proportion of short-term diagnosed with
CD4<200 cells/μL receiving ART. is QoL for infected individuals in the second (long-term) diagnosed phase which
is assumed to be independent of CD4 band at diagnosis. QALYs were discounted at a rate 3·5% per year 11 using the
standard discounting model where discounting factors are equal to 1 where is the discount rate and is the
year in which the event occurs.12 A scenario of 1·5% per year was undertaken in sensitivity analysis14 (Table S7).
To evaluate change in QALYs because of secondary infections averted due to the intervention, the QoL of secondary
infections in the control scenario is compared with QoL of secondary infections in the intervention scenario plus the
QoL of individuals who remain uninfected in the intervention scenario (i.e. secondary infections averted), who are
assumed to have QoL=1. The model does not incorporate ageing and so the comparison QoL for uninfected individuals
must be fixed as 1, and therefore does not account for reduction in QoL over time, with ageing.15 This leads to a slight
overestimation of QALYs gained to the intervention.

Choice of time horizon


A lifetime time horizon has not been chosen for this analysis because for this intervention, its benefits continue to accrue
indefinitely (because the reduction in onward HIV transmissions has long-lasting effects). The cost-effectiveness model
only tracks secondary HIV transmission events, so all benefits would be accrued over the lifetime of the sexual partners
of those screened as part of the intervention. NICE recommends that the time horizon for an analysis should be, “long
enough to reflect all important differences in costs or outcomes between the technologies being compared” 11. In view of
the atypical nature of this intervention, we have chosen 40 years as an appropriate time horizon, but the sensitivity of
cost-effectiveness results to this assumption is explored in sensitivity analysis.

Sensitivity analysis
Baseline values provided in Table S2 are used for model runs to determine central outcome estimates, while the ranges
provided in brackets are used for one-way sensitivity analysis for tornado plots and probabilistic sensitivity analysis which
uses Latin Hypercube Sampling (LHS).16 LHS parameters were varied independently of one another over 1000
simulations using uniform distributions for each varied parameter. Scenario analysis was used to explore the impact of
parameters which may substantially affect model outcomes. These parameters are: reduction of sexual behaviour among
persons upon diagnosis, HIV-related healthcare costs per year (to reflect the possibility of price reductions as antiretroviral
drug patents expire), time horizon and QALY discount rate. A summary of parameters varied in the one-way and
probabilistic sensitivity analyses and scenario analyses is given in Table S7.

Supplementary Results
Costs for the RHIVA2 intervention


 
Only 32 of 40 participating surgeries claimed the enrolment incentive. Ninety-two health professionals attended the
initial training sessions, including 29 general practitioners (GPs), 20 nurses, and 22 healthcare assistants (HCAs); and
87 staff completed competency-based training. On average, three rapid test kits were used per trainee per session, and
two additional kits were used for demonstration purposes.

Model fit
The values of fitted model parameters are shown in Table S5 and evaluation of model fit is shown in Table S6. Table S5
shows the fitted parameter value used as the base case, with 95%CI generated by drawing 1000 samples from the posterior
distribution and selecting the 2.5th and 97.5th percentile. These values were subsequently used in model sensitivity analysis
as the plausible parameter range. For the control arm, the fitted parameters are the underlying incidence rate in Hackney
and a single diagnosis rate for all those with CD4 counts greater than 200 cells/μL. We initially fitted the model with a
separate diagnosis rate for those with low CD4 counts, but this parameter was not identifiable, with fits converging on
ever-larger diagnosis rates at this stage of infection. This suggested that individuals with low CD4 counts are diagnosed
almost immediately in this population, presumably due to other symptoms. When we fitted these two parameters (Table
S5), the model had a good fit to the data (Table S6). For the intervention arm we then fitted an additional diagnosis rate,
or an increase in the baseline diagnosis rate, which led to more rapid diagnosis for individuals with CD4 counts greater
than 200 cells/μL. This meant that the proportion of people diagnosed with low CD4 counts inevitably fell, resulting in a
less good fit for this group (Table S6), making a less good fit than in the control arm. The model structure cannot capture
the full complexity of the infection and diagnosis process, and therefore cannot capture all the behaviour of the system,
but the overall fit to the data is good. HIV stage duration parameters were not varied as part of the univariate sensitivity
analysis because these parameters interact with HIV incidence and diagnosis rates, so change in the duration of any HIV
infection stage would have resulted in altered incidence and diagnosis rates. Hence the sensitivity analysis was performed
only on those parameters that would not alter the quality of the fit to the data.

Sensitivity to potential ART price reductions and cost of the intervention


Figure S4 shows that reductions in ART costs do not meaningfully impact RHIVA2 ICER predictions. RHIVA2 becomes
cost-effective and eventually cost-saving because diagnosing patients earlier averts the higher long-term healthcare costs
associated with late-diagnosed infection. The cost of ART is only part of the healthcare costs associated with managing
an HIV diagnosed patient, so ART prices do not heavily influence model predictions. Figure S4 also shows that setting
the costs of testing to zero increases cost-effectiveness.

Supplementary Discussion
Strengths and weaknesses
Our study has a number of strengths. We have undertaken a full uncertainty analysis involving Tornado plots for
univariable analysis and probabilistic sensitivity analysis (PSA) in order to provide 95%CIs, to provide a range for each
model output to reflect the uncertainty in our model inputs. However, the input parameters varied for the PSA do not
include every input for which there is uncertainty, because if all of these were varied simultaneously, there would be too
much “noise” in the model outputs to differentiate the different results for our intervention and control scenarios (for
example, durations of each HIV stage and rates of sexual partner change are model inputs that are not known precisely
and likely to be heterogeneous, but these properties would not differ between intervention and control scenarios, and so
these parameters are kept constant). For some of the more potentially influential model inputs, we conducted scenario
analyses to explore how different assumptions changed the model outputs (Figure 4).

Our study also has some limitations, reflecting uncertainties regarding the values of inputs for modelling the transmission
of a sexually transmitted disease. There is a paucity of data to inform HIV transmission parameters such as HIV
infectiousness. Sexual behaviour data, while easier to obtain, is often subject to bias. Therefore our model structure, which
only tracks secondary infections, as opposed to a full transmission model which accounts for all HIV transmissions, is
probably justified but will slightly underestimate benefits of the intervention as tertiary HIV infections onwards are not
included We have attempted to handle uncertainty in model inputs by exploring sensitivity in the univariate, multivariate
and scenario analyses described above. In addition, the precise timing of antiretroviral price reductions due to generics is
difficult to define. Finally, the QoL of uninfected individuals, required in the model to account for impact of reducing
onward transmission, assumes a constant value of 1, which does not account for reduction due to ageing. This leads to a
slight overestimation of the QALYs gained due to the intervention.

Our analysis is designed to evaluate the cost-effectiveness of the RHIVA2 trial. Evaluation of the impact of the
intervention with roll out, and in different contexts and for different populations, is beyond the scope of this paper but is
important future work. Costs of the intervention on roll out will decrease, as costs such as incentives may no longer apply.
In regions where HIV prevalence is lower than in Hackney, the positive predictive value of the test will decrease and
therefore the cost per additional patient newly diagnosed by the intervention will increase. Cost-effectiveness of the
intervention may increase if rolled out to higher HIV prevalence areas, for example areas with large MSM populations,
and conversely may be lower where HIV prevalence is lower.


 
References
1. Weinhardt LS, Carey MP, Johnson BT, Bickham NL. Effects of HIV counseling and testing on sexual risk
behavior: a meta-analytic review of published research, 1985-1997. American journal of public health 1999; 89(9):
1397-405.
2. Public Health England (PHE). Country and PHE Region HIV Data Tables. Available at:
http://www.hpa.org.uk/Topics/InfectiousDiseases/InfectionsAZ/HIV/HIVData/#4._Diagnosed_HIV_Prevalence_by_Up
per_Lower_Tier_Local_authority. Accessed 7th August 2014. 2013.
3. Long EF, Mandalia R, Mandalia S, Alistar SS, Beck EJ, Brandeau ML. Expanded HIV testing in low-
prevalence, high-income countries: a cost-effectiveness analysis for the United Kingdom. PLoS One 2014; 9(4):
e95735.
4. McCormack S, Dunn DT, Desai M, et al. Pre-exposure prophylaxis to prevent the acquisition of HIV-1
infection (PROUD): effectiveness results from the pilot phase of a pragmatic open-label randomised trial. Lancet 2016;
387(10013): 53-60.
5. Cohen MS, Chen YQ, McCauley M, et al. Prevention of HIV-1 infection with early antiretroviral therapy. The
New England journal of medicine 2011; 365(6): 493-505.
6. Bezemer D, de Wolf F, Boerlijst MC, van Sighem A, Hollingsworth TD, Fraser C. 27 years of the HIV
epidemic amongst men having sex with men in the Netherlands: an in depth mathematical model-based analysis.
Epidemics 2010; 2(2): 66-79.
7. Bezemer D, de Wolf F, Boerlijst MC, et al. A resurgent HIV-1 epidemic among men who have sex with men
in the era of potent antiretroviral therapy. AIDS 2008; 22(9): 1071-7.
8. Yin Z, Brown AE, Hughes G, Nardone A, Gill ON, Delpech VC. HIV in the United Kingdom 2014 Report:
data to end 2013. November 2014. Public Health England, London. Available at:
https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/401662/2014_PHE_HIV_annual_report_
draft_Final_07-01-2015.pdf Accessed 19th April 2015. 2014.
9. Beck EJ, Mandalia S, Sangha R, et al. The cost-effectiveness of early access to HIV services and starting
cART in the UK 1996-2008. PLoS One 2011; 6(12): e27830.
10. Miners A. Estimating 'costs' for cost-effectiveness analysis. PharmacoEconomics 2008; 26(9): 745-51.
11. NICE. NICE National Institute for Health and Care Excellence Process and Methods Guides: Guide to the
methods of technology appraisal. Available at: http://www.nice.org.uk/article/pmg9/resources/non-guidance-guide-to-
the-methods-of-technology-appraisal-2013-pdf Accessed 15th October 2014. 2013.
12. Economic Evaluation: Understanding Public Health. Fox-Rushby, J and Cairns, J eds, London School of
Hygiene & Tropical Medicine. Oxford University Press. 2005.
13. Krentz HB, Gill MJ. The Direct Medical Costs of Late Presentation (<350/mm) of HIV Infection over a 15-
Year Period. AIDS research and treatment 2012; 2012: 757135.
14. Methods for the development of NICE public health guidance (third edition). Section 6.3.4: Discounting.
Available at: http://www.nice.org.uk/article/pmg4/chapter/6-Incorporating-health-economics Accessed 6th May 2015.
2012.
15. Miners A, Phillips A, Kreif N, et al. Health-related quality-of-life of people with HIV in the era of combination
antiretroviral treatment: a cross-sectional comparison with the general population. Lancet HIV 2014; 1(1): e32-e40.
16. Blower SM, Dowlatabadi H. Sensitivity and uncertainty analysis of complex models of disease transmission:
an HIV model, as an example. Int Stat Rev 1994; 62(2): 229-43.
17. ONS. Office for National Statistics (ONS). Life expectancy at birth and at age 65 by local areas in the United
Kingdom, 2006-8 to 2010-12 data table. Available at: http://www.ons.gov.uk/ons/publications/re-reference-
tables.html?edition=tcm%3A77-354758 Accessed 26th August 2014. 2014.
18. Ellis S, Curtis H, Ong EL, British HIVA, Audit BC, Standards s-c. HIV diagnoses and missed opportunities.
Results of the British HIV Association (BHIVA) National Audit 2010. Clin Med 2012; 12(5): 430-4.
19. Whittle A. Increasing opportunities for HIV diagnosis in primary care: A borough wide evaluation of HIV
testing and pre-diagnosis care in general practice. Presentation at 19th Annual Conference of the British HIV
Association (BHIVA), Manchester, April 2013. Available at:
http://www.bhiva.org/documents/Conferences/2013Manchester/Presentations/130417/AliceWhittle.pdf Accessed 26th
August 2014,. 2013.
20. Hollingsworth TD, Anderson RM, Fraser C. HIV-1 transmission, by stage of infection. J Infect Dis 2008;
198(5): 687-93.
21. Lodi S, Phillips A, Touloumi G, et al. Time from human immunodeficiency virus seroconversion to reaching
CD4+ cell count thresholds <200, <350, and <500 Cells/mm(3): assessment of need following changes in treatment
guidelines. Clin Infect Dis 2011; 53(8): 817-25.
22. Pilcher CD, Joaki G, Hoffman IF, et al. Amplified transmission of HIV-1: comparison of HIV-1 concentrations
in semen and blood during acute and chronic infection. AIDS 2007; 21(13): 1723-30.
23. Delpech V, Brown AE, Croxford S, et al. Quality of HIV care in the United Kingdom: key indicators for the
first 12 months from HIV diagnosis. HIV medicine 2013; 14 Suppl 3: 19-24.
24. Davis DH, Smith R, Brown A, Rice B, Yin Z, Delpech V. Early diagnosis and treatment of HIV infection:
magnitude of benefit on short-term mortality is greatest in older adults. Age and ageing 2013; 42(4): 520-6.

 
25. Fiebig EW, Wright DJ, Rawal BD, et al. Dynamics of HIV viremia and antibody seroconversion in plasma
donors: implications for diagnosis and staging of primary HIV infection. . AIDS 2003; 17: 1871-9.
26. Patel P, Mackellar D, Simmons P, et al. Detecting acute human immunodeficiency virus infection using 3
different screening immunoassays and nucleic acid amplification testing for human immunodeficiency virus RNA,
2006–2008. Arch Intern Med 2010; 170: 66-74.
27. Pinkerton SD. How many sexually-acquired HIV infections in the USA are due to acute-phase HIV
transmission? AIDS 2007; 21: 1625-9.
28. Prabhu VS, Hutchinson AB, Farnham PG, Sansom SL. Sexually acquired HIV infections in the United States
due to acute-phase HIV transmission: an update. AIDS 2009; 23: 1792-4.
29. Abbas UL, Anderson RM, Mellors JW. Potential impact of antiretroviral therapy on HIV-1 transmission and
AIDS mortality in resource-limited settings. J Acquir Immune Defic Syndr 2006; 41: 632-41.
30. Downs AM, de Vincenzi I. Probability of heterosexual transmission of HIV: relationship to the number of
unprotected sexual contacts. European Study Group in Heterosexual Transmission of HIV. J Acquir Immune Defic
Syndr Hum Retrovirol 1996; 11: 388-95.
31. Kaplan EH, Heimer R. A model-based estimate of HIV infectivity via needle sharing. J Acquir Immune Defic
Syndr 1992; 5: 1116-8.
32. Mastro TD, de Vincenzi I. Probabilities of sexual HIV-1 transmission. AIDS 1996; 10 Suppl A: S75-82.
33. Nicolosi A, Correa Leite ML, Musicco M, Arici C, Gavazzeni G, Lazzarin A. The efficiency of male-to-
female and female-to-male sexual transmission of the human immunodeficiency virus: a study of 730 stable couples.
Italian Study Group on HIV Heterosexual Transmission. Epidemiology 1996; 5: 570-5.
34. Padian NS, Shiboski SC, Glass SO, Vittinghoff E. Heterosexual transmission of human immunodeficiency
virus (HIV) in northern California: results from a ten-year study. Am J Epidemiol 1997; 146: 350-7.
35. Quinn TC, Wawer MJ, Sewankambo N, et al. Viral load and heterosexual transmission of human
immunodeficiency virus type 1: Rakai Project Study Group. New Engl J Med 2000; 342: 921-9.
36. Wawer MJ, Gray RH, Sewankambo NK, et al. Rates of HIV-1 transmission per coital act, by stage of HIV-1
infection, in Rakai, Uganda. J Infect Dis 2005; 191: 1403-9.
37. Caceres CF, van Griensven GJ. Male homosexual transmission of HIV-1. AIDS 1994; 8: 1051-61.
38. Jacquez JA, Koopman JS, Simon CP, Longini IMJ. Role of the primary infection in epidemics of HIV
infection in gay cohorts. J Acquir Immune Defic Syndr 1994; 7: 1169-84.
39. McDaid LM, Li J, Knussen C, Flowers P. Sexually transmitted infection testing and self-reported diagnoses
among a community sample of men who have sex with men, in Scotland. Sex Transm Infect 2013; 89: 223-30.
40. Robinson C, Nardone A, Mercer CH, Johnson AM. Sexual health; 2011.
41. Carter M. Inconsistent condom use by UK’s gay men, with many putting themselves or others at risk of HIV.
In: http://www.aidsmap.com/Inconsistent-condom-use-by-UKs-gay-men-withmany-putting-themselves-or-others-at-
risk-of-HIV/page/1427035/. NAA, editor. London; 2007.
42. Phillips AN, Cambiano V, Nakagawa F, et al. Increased HIV incidence in men who have sex with men despite
high levels of ART-induced viral suppression: analysis of an extensively documented epidemic. PLoS One 2013;
8(e55312).
43. Dave SS, Fenton KA, Mercer CH, Erens B, Wellings K. Male circumcision in Britain: findings from a national
probability sample survey. Sex Transm Infect 2003; 79: 499-500.
44. Doerner R, McKeown E, Nelson S, Anderson J, Low N, Elford J. Circumcision and HIV infection among men
who have sex with men in Britain: the insertive sexual role. Archives of sexual behavior 2013; 42(7): 1319-26.
45. Auvert B, Taljaard D, Lagarde E, Sobngwi-Tambekou J, Sitta R, Puren A. Randomized, controlled
intervention trial of male circumcision for reduction of HIV infection risk: the ANRS 1265 Trial. PLoS Med 2005; 2:
e298.
46. Desai K, Boily MC, Garnett GP, Masse BR, Moses S, Bailey RC. The role of sexually transmitted infections in
male circumcision effectiveness against HIV-insights from clinical trial simulation. Emerg Themes Epidemiol 2006;
3(19).
47. Siegfried N, Muller M, Deeks JJ, Volmink J. Male circumcision for prevention of heterosexual acquisition of
HIV in men. Cochrane Database Syst Rev: CD003362; 2009.
48. HIV epidemiology in London, 2011 data. Public Health England Field Epidemiology Services (Victoria). May
2013. Available at: http://www.hpa.org.uk/webc/HPAwebFile/HPAweb_C/1317138999825. Accessed 15th August
2014. 2013.
49. Holtgrave DR, Pinkerton SD. Updates of cost of illness and quality of life estimates for use in economic
evaluations of HIV prevention programs. J Acquir Immune Defic Syndr Hum Retrovirol 1997; 16(1): 54-62.
50. Honiden S, Sundaram V, Nease RF, et al. The effect of diagnosis with HIV infection on health-related quality
of Life. Quality of life research : an international journal of quality of life aspects of treatment, care and rehabilitation
2006; 15(1): 69-82.
51. Schackman BR, Goldie SJ, Freedberg KA, Losina E, Brazier J, Weinstein MC. Comparison of health state
utilities using community and patient preference weights derived from a survey of patients with HIV/AIDS. Medical
decision making : an international journal of the Society for Medical Decision Making 2002; 22(1): 27-38.


 
52. Tengs TO, Lin TH. A meta-analysis of utility estimates for HIV/AIDS. Medical decision making : an
international journal of the Society for Medical Decision Making 2002; 22(6): 475-81.
53. Sanders GD, Bayoumi AM, Sundaram V, et al. Cost-effectiveness of screening for HIV in the era of highly
active antiretroviral therapy. The New England journal of medicine 2005; 352(6): 570-85.
54. Department of Health. HIV Adult Outpatient Services, Financial Data Analysis (based on HIV Clinical Care
PathwayV8 and 2008/09 costs). 2010.


 
Supplementary Figure Legends

Figure S1 Model structure with mathematical notation. Numbers refer to stages of HIV infection: 1 = acute infection,
2= CD4 >500 cells/μL, 3= CD4 350-500 cells/μL, 4=200-349 cells/μL, 5= <200 cells/μL. Diagnosis rate for
compartment 1 is set to zero i.e. it is assumed that no diagnoses occur during acute infection.

Figure S1
λ

d1(=0) ϕD
U1 D1 L1

ϕU1 μ μ μ
d2 ϕD
U2 D2 L2

ϕU2 μ μ μ
d3 ϕD
U3 D3 L3

ϕU3 μ μ μ
d4 ϕD
U4 D4 L4
ν4
ϕU4 μ μ μ
ϕD
D5 L5
ν5
μ μ
D – diagnosed HIV infection (short‐term phase); L – diagnosed HIV infection (long‐term phase); U – undiagnosed HIV infection. 
Subscripts for undiagnosed and diagnosed short‐term phase individuals refer to stage of infection (1 = acute infection, 2= CD4 >500 
cells/μL, 3= CD4  350‐500 cells/μL, 4=200‐349 cells/μL, 5= <200 cells/μL). For descriptions of parameter symbols, see web extra 
material Methods section.


 
Figure S2 Cumulative costs accrued by the RHIVA2 trial, over time, plot using baseline parameters as detailed in Table
S2. The intervention is continued for 28 months only (the duration of the RHIVA2 trial). “Diagnosed” individuals
(green line) accrue costs for the trial due to increased HIV healthcare costs because of earlier diagnosis, while some
costs are saved by preventing secondary infections (blue line).  

 
 

10 
 
Figure S3 Ribbon plots of incremental cost-effectiveness ratio (ICER), cost in GBP(£) per QALY gained due to the
RHIVA2 trial, over time. The central, black line is the plot using base case values of all model parameters. The blue
shaded ribbon is the area between ICER estimates plotted using the lowest and highest plausible parameter range for A)
incidence (new infections in Hackney per year, range: 69-143); and B) QoL multiplier for acute stage HIV infection
(range: 0.6-0.95).

 
Figure S3A 

11 
 
Figure S3B 

12 
 
Figure S4 Scenario analysis exploring the sensitivity of model results to different assumptions regarding reduction in
cost of ART upon patent expiry and a scenario of zero HIV screening costs. The NICE threshold for cost-effectiveness
of £20,000-£30,000 is shown in yellow.

13 
 
Supplementary Tables
Table S1 Total start-up and recurrent costs of RHIVA2 intervention (2012 prices)

Resource type Unit cost Number of units Total cost


£ (US$*)
Start-up costs
Cost for RHIVA2 staff time (clinicians, economists, social £0·38016 69,750 minutes £26,516 (US$42,805)
scientists, technicians) including follow-up activities; per minute (US$0·6137)
Cost of GP/nurse time for attending the initial training sessions; £0·58 (US$0·94) 17,484 minutes £10,141 (US$16,370)
per minute
Cost of rapid HIV test kits for training, including gloves £4·05 (US$6·54) 301 units of tests and £1,219 (US$1,968)
gloves
Cost with practice HIV lead nurse follow-up training £0·77 (US$1·24) 600 minutes £462 (US$746)

Cost with training new nurses/HCAs £0·58 (US$0·94) 630 minutes £365 (US$590)

Cost of training materials (e.g. folders), translation of educational - - £7,623 (US$12,306)


materials, supplies, transport, and equipment
Cost of incentives to practices: £300 per educational training
session per practice £300 (US$484) 32 practices £9,600 (US$15,497)

Total start-up costs - - £55,926 (US$90,282)

Start-up costs assumed to be utilised by RHIVA2 trial‡ - - £26,099 (US$42,132)

Recurrent costs
Costs for performing rapid HIV testing

Cost of GP/nurse time for performing a rapid HIV test, per minute £0·58 (US$ 0·94) 4,978 tests £8,662 (US$13,983)
(3 minutes per patient)
Cost of rapid HIV test kit† £4·00 (US$6·46) 4,978 tests £19,912 (US$32,144)

Incentive cost per rapid HIV test performed £10·00 (US$16·14) 4,978 tests performed £49,780 (US$80,360)

Total recurrent costs for performing rapid HIV testing - - £78,354 (US$126,486)

Additional cost associated with reactive rapid HIV test results

Cost of GP/nurse time for post-test discussion, per reactive test £0·77 (US$ 1·24) 14 patients £108 (US$174)
result, per minute (10 minutes per patient)
Cost of confirmatory HIV test for patients with a reactive result £7·72 (US$12·46) 14 patients £108 (US$174)

Cost of GP/senior nurse time for confirmatory follow up £0·77 (US$1·24) 14 patients £108 (US$174)
appointment; per minute (10 minutes per patient)
Cost of CD4 count test for patients confirmed HIV positive £36·90 (US$59·57) 11 patients £406 (US$655)

Cost of viral load test for patients confirmed HIV positive £50·74 (US$81·91) 11 patients £558 (US$901)

Total recurrent costs associated with reactive rapid HIV test - - £1,288 (US$2,078)
results

Resource type Unit cost Number of units Total cost


£ (US$*) £ (US$*)
Recurrent costs
Additional cost associated with indeterminate/twice invalid rapid
HIV test results
Cost of confirmatory HIV test to 3 twice invalid and 6 £7·72 (US$12·46) 9 patients £69 (US$112)
indeterminate test results
Cost of GP/nurse time for post-test discussion, per twice £0·77 (US$ 1·24) 9 patients £69(US$112)
invalid/indeterminate test result, per minute (10 minutes per
patient)

14 
 
Resource type Unit cost Number of units Total cost
£ (US$*)
Cost of GP/senior nurse time for confirmatory follow up £0·77 (US$1·24) 9 patients £69 (US$112)
appointment; per minute (10 minutes per patient)
Total recurrent costs associated with indeterminate/twice - - £207 (US$336)
invalid rapid HIV test results
Costs of quality assurance and additional rapid HIV test kits used

Cost of quality assurance - - £13,749(US$ 22,195)

Cost of nurse/HCA for processing external quality assurance £0·58 (US$0·94) 1,470 minutes £853 (US$1,376)
samples
Cost of additional nurse/HCA training required for quality £0·58 (US$0·94) 360 minutes £209(US$337)
assurance
Costs of rapid HIV test kits expired or not used2 £4·00 (US$6·46) 1,561 tests £6,244 (US$10,080)

Total recurrent cost of quality assurance and additional rapid - - £21,054 (US$33,988)
HIV test kits used
Costs of marginal increase in serology testing due to the RHIVA2
intervention
Cost of additional serology tests performed due to the RHIVA2 £7·72 (US$12·46) 7 tests £54 (US$87)
intervention
Cost of GP/nurse time for post-test discussion; per minute (10 £0·77 (US$1·24) 7 patients £54 (US$87)
minutes per patient)
Cost of CD4 count test to confirmed HIV positive patients £36·90 (US$59·57) 7 patients £258 (US$417)

Cost of viral load test to confirmed HIV positive patients £50·74 (US$81·91) 7 patients £355 (US$573)

Total recurrent cost of marginal increase in serology testing - - £721 (US$1,165)


due to the RHIVA2 intervention

Total recurrent costs - - £101,625 (US$164,054)

TOTAL START-UP AND RECURRENT COST FOR £127,724 (US$206,185)


RHIVA2 INTERVENTION

* Approximated figures. †Cost does not include cost of gloves. ‡Trial duration: 28 months.
Exchange rate: US$1·6143 (average annual exchange rate for 2012, www.oanda.com); slight inconsistencies in costs
are due to rounding errors.

15 
 
Table S2 Model parameter values, with sources.

Symbol Description Estimate and/or range Source


Demography
8
-* Hackney resident population aged 181,000
15-59 years
Duration in model population 45 years Assumption that average age at infection
acquisition is 35 years (80 years is life
expectancy in Hackney17)
Epidemiology
8
-* Diagnosed HIV prevalence in 8·11 per 1000
Hackney among 15-59 year olds
Proportion of HIV infections 19% Consideration of several sources: 10%
diagnosed in general practice in from a 2010 national audit18 contrasting
Hackney with a 28% level from a Hackney
general practice audit of patients
registered with a general practice.19 The
latter may overestimate the proportion
for the total population of Hackney, as
some may not be registered with a
general practice, and so a central
estimate was used.
Trial data
-** Duration of trial 28 months RHIVA2 trial data
Number of participating practices 20 in control arm RHIVA2 trial data (45 GPs in borough
20 in intervention arm of Hackney)
HIV disease stage duration
20-22
1/ Acute stage (untreated) 12 weeks
21, 22
1/ Chronic stage >500 cells/μL 0·935 years
(untreated)
21, 22
1/ Chronic stage 350-500 cells/μL 3·0 years
(untreated)
21, 22
1/ Chronic stage 200-350 cells/μL 3·75 years
(untreated)
1/ Duration of HIV first (short-term) 1 year Assumption to complement changes in
phase post HIV diagnosis QoL estimates over duration with
diagnosed infection (see later in table)
Proportion dying before entering CD4 (cells/μL) %
long-term diagnosed phase Acute 0·0% Assumption
>500 0·0% Assumption
23
350-500 0·52%
24
200-349 0·9%
(0·8-1·0%)
24
<200 6·1%
(5·6-6·6%)
Secondary infection parameters
Proportion HIV infected in 58·7% PHE 20122: exposure category and sex
Hackney that are MSM of new HIV diagnoses in London, 2012
data
Proportion HIV infected in 15·7% PHE 20122: exposure category and sex
Hackney that are heterosexual men of new HIV diagnoses in London, 2012
data
Proportion HIV infected in 25·6% PHE 20122: exposure category and sex
Hackney that are heterosexual of new HIV diagnoses in London, 2012
women data
HIV transmission probabilities per partnership:
Heterosexual (female to male)
20, 25-28
Acute HIV 0·20 (0·10-0·30)
20, 29-36
Asymptomatic HIV† 0·02 (0·01-0·04)
20, 29-36
Symptomatic HIV§ 0·04 (0·01-0·06)
Heterosexual (male to female)
20, 25-28
Acute HIV 0·30 (0·10-0·40)
20, 29-36
Asymptomatic HIV† 0·03 (0·02-0·05)
20, 29-36
Symptomatic HIV§ 0·06 (0·02-0·10)
Homosexual (male to male)
20, 25-28
Acute HIV 0·40 (0·20-0·50)
37, 38
Asymptomatic HIV† 0·04 (0·03-0·06)
37, 38
Symptomatic HIV§ 0·08 (0·03-0·15)
Annual number of partners
39
MSM 4·2 (2-10)
40
Heterosexuals 1·5 (1-2)
Condom use, %
41, 42
MSM 56% (25-75%)
40
Heterosexuals 20% (7-41%)

16 
 
Symbol Description Estimate and/or range Source
43, 44
Fraction of men who are 16% (5-25%)
circumcised
45-47
Reduction in heterosexual HIV 50% (48-60%)
transmissibility due to male
circumcision, %
Reduction in sexual behaviour 25% (0-50%) Assumptions regarding sexual behaviour
among persons identified as HIV change upon diagnosis (no change, 25%
positive, % and 50% reductions) explored through
scenario analysis
5
Reduction in sexual infectivity due 96% (50-99%)
to ART, %
Proportion of short-term HIV- CD4 (cells/μL) %
diagnosed on ART Acute 0 PHE, unpublished data
>500 54 (52-56) PHE, unpublished data
350-500 60 (57-62) PHE, unpublished data
200-349 79 (77-82) PHE, unpublished data
<200 85 (83-88) PHE, unpublished data
Proportion of long-term HIV- CD4 (cells/μL) %
diagnosed on ART Acute 0 PHE, unpublished data
>500 93 (92-93) PHE, unpublished data
350-500 90 (89-90) PHE, unpublished data
200-349 94 (94-95) PHE, unpublished data
<200 94 (93-94) PHE, unpublished data
HIV prevalence among partners: 13·2% PHE 2014, HIV prevalence among MSM
MSM living in London8
HIV prevalence among partners: 0·35% Estimated HIV prevalence in pregnant
female heterosexuals women in London48
HIV prevalence among partners: 0·35% Assumed to be same as for pregnant
male heterosexuals women in London48
Quality of life multipliers
49-52
Acute HIV 0·89 (0·60-0·95)
49-52
.. Unidentified asymptomatic HIV 0·91 (0·85-0·95)
49-53
.. Identified asymptomatic HIV, first 0·84 (0·80-0·90)
year
49-53
Identified asymptomatic HIV, 0·89 (0·85-0·95)
subsequent years
49-52
Identified symptomatic HIV, not 0·72 (0·70-0·80)
on ART
49-52
Symptomatic HIV treated with 0·83 (0·82-0·87)
ART
Uninfected individuals 1 Assumption
Discount rate for QALYs 3·5% NICE11
HIV-related healthcare costs
9
.. Undiagnosed, all HIV stages (1-4) £2,457‡
.. Diagnosed short-term, not on £2,892‡ Calculations based on costs shown in
ART: CD4>200 cells/μL (stages 1- Table S3
4)
Diagnosed short-term, not on £6,466‡ Calculations based on costs shown in
ART: CD4≤200 cells/μL (stage 5) Table S3
.. Diagnosed short-term, on ART: £9,688‡ Calculations based on costs shown in
CD4>200 cells/μL (stages 1-4) Table S3
Diagnosed short-term, on ART: £12,133‡ Calculations based on costs shown in
CD4≤200 cells/μL (stage 5) Table S3
.. Diagnosed long-term, not on ART: Varied, 3 scenarios See Table S4 for details
CD4>200 cells/μL (stages 1-4)
Diagnosed long-term, not on ART: Varied, 3 scenarios See Table S4 for details
CD4≤200 cells/μL (stage 5)
.. Diagnosed long-term, diagnosed at Varied, 3 scenarios See Table S4 for details
CD4>200 cells/μL on ART
Diagnosed long-term, diagnosed at Varied, 3 scenarios See Table S4 for details
CD4≤200 cells/μL on ART
Discount rate for healthcare costs 3·5% NICE11
ART – antiretroviral therapy; NICE – National Institute for Health and Care Excellence; PHE – Public Health England.
Values in brackets denote ranges used in the probabilistic sensitivity analysis.
* Parameter used in the fitting procedure to estimate HIV incidence in the borough of Hackney.
** This trial duration parameter changes the diagnosis rates within the model when modelling the intervention scenario.
For the first 28 months that the model runs, diagnosis rates are set to , the HIV diagnosis rate for the intervention
scenario, and then revert to , the control HIV diagnosis rate (see Table S5 for values).
† “Asymptomatic” is here used to define the status of individuals in CD4 bands >500 cells/μL, 350-500 cells/μL and
200-349 cells/μL.
‡HIV-related healthcare costs quoted are for year 2012-13.
§ Symptomatic refers to individuals with CD4 count <200 cells/μL.
17 
 
ǁ 95% credible interval for proportion of long-term diagnosed on ART 92·1%-92·5%. Therefore we did not include this
narrow range in the univariate and probabilistic sensitivity analyses.

18 
 
Table S3 HIV-related healthcare cost parameters

Healthcare cost item Cost assumed* Source


CD4 measurement £41 Department of Health54
Viral load measurement £67 Department of Health54
Resistance test £234 Department of Health54
Use of Undiagnosed £2,457 No data available for healthcare costs of undiagnosed. Model assumes same
healthcare- cost as diagnosed, not on ART CD4>200 cells/μL, using the rationale that
centre services HIV infection raises the risk of common clinical conditions.
per year Diagnosed, not on ART CD4>200 cells/μL £2,457 Includes the “average use of services cost” and the “other drug cost” from
Diagnosed, not on ART CD4 ≤200 cells/μL £6,032 Beck et al.9
On ART, CD4>200 cells/μL £3,994
On ART, CD4 ≤200 cells/μL £6,439
Average annual ART cost £5,212 Cost of ARVs for 2013-14 financial year: £186,718,672 for 35,427 patients
accessing HIV treatment in London (FOI request, Kevin Kelleher, personal
communication).
* Adjusted to 2012/13 prices.
 

19 
 
Table S4 HIV-related healthcare cost scenarios for long-term diagnosed patients.

.. ..
Scenario 1: based on observations by Beck et al9* £2892 £2892 £6466 £9688 £9688 £12,133
Scenario 2: based on observations by Krentz and £2966 £4746 £4746 £8195 £13,113 £13,113
Gill13 (costs are 1.6 times as high for patients
diagnosed at CD4 ≤350 cells/μL**)
Scenario 3: based on observations by Krentz and £2966 £3471 £6943 £8195 £8914 £17,828
Gill13 (costs are 1.6 times as high for patients
diagnosed at CD4 ≤350 cells/μL compared to
those with higher CD4 cell count, and costsfor
patients diagnosed at CD4 ≤200 cells/μL are twice
as high as those diagnosed at CD4 200-349
cells/μL **)
Mathematical symbols: C – healthcare costs for long-term diagnosed, not on ART; C – healthcare costs for
long-term diagnosed, on ART; 1..5 – HIV stages 1 to 5 (acute infection and CD4 bands ≥500, 350-499, 200-349,
<200 cells/μL respectively).
* Beck et al9 reported 21% higher costs for patients on ART who were diagnosed at CD4 ≤200 cells/μL. Scenario
1 has 25% higher costs for long-term diagnosed patients on ART rather than 21% because our cost calculations
include data from other sources for price of ART and HIV-related tests (details shown in Table S3).
** Calculations are based on the following proportions in each CD4 band at diagnosis: 18.7% (5.6%), 21.0%
(9.7%), 21.8% (26.1%), and 38.6% (58.6%) in CD4 bands <200, 200-349, 350-499 and ≥500 cells/μL for patients
on ART (not on ART) with CD4 count recorded within one year of diagnosis (PHE, unpublished data 2013).
 

 
 

20 
 
Table S5 Values of model parameters fitted to empirical data.

Model parameter Symbol Fitted parameter value (95%CI)


Incidence (new infections per year in Hackney) 100 (69-143) infections per year
HIV diagnosis rate (excluding acute infection*) .. 0·10 (0·046-0·19) per year
HIV diagnosis rate for the intervention scenario 0·23 (0·12-0·45) per year
95%CI – 95% credible interval. *Acute infection diagnosis rate is set to zero ( 0).

Table S6 Model validation results.

Model estimate Empirical data


Fit to control data

Number diagnosed in control practices 17·7 14 (RHIVA2 trial data)

Proportion diagnosed CD4 >500 cells/μL 9·4% 7·1% (RHIVA2 trial data)

Proportion diagnosed CD4 350-500 cells/μL 22·0% 28·6% (RHIVA2 trial data)

Proportion diagnosed CD4 200-349 cells/μL 18·8% 21·4% (RHIVA2 trial data)

Proportion diagnosed CD4 <200 cells/μL 49·9% 42·9% (RHIVA2 trial data)
Fit to intervention data
Number diagnosed in intervention practices 26·3 32 (RHIVA2 trial data)
during trial duration
Proportion diagnosed CD4 >500 cells/μL 13·7% 20·0% (RHIVA2 trial data)
Proportion diagnosed CD4 350-500 cells/μL 30·5% 23·3% (RHIVA2 trial data)

Proportion diagnosed CD4 200-349 cells/μL 25·7% 20·0% (RHIVA2 trial data)

21 
 
Table S7 Description and values of parameters varied in the sensitivity and scenario analyses.

Parameter description Symbol Parameter values Source


One-way and probabilistic sensitivity analyses
Incidence (new infections per year in Hackney) 69-143 infections/year Fitted (Table S5)
HIV diagnosis rate (excluding acute infection) .. 0·046-0·19/year Fitted (Table S5)
Additional HIV diagnosis rate for the intervention scenario 0·12-0·45/year Fitted (Table S5)
Proportion dying before entering long-term diagnosed
phase:
23
Chronic stage 350-500 cells/μL 0·25-0·79%
24
Chronic stage 200-349 cells/μL 0·8-1·0%
24
Symptomatic stage <200 cells/μL (untreated) 5·6-6·6%
HIV transmission probabilities per partnership:
20, 25-28
Acute HIV (female to male) 0·10-0·30
20, 29-36
Asymptomatic HIV (female to male) 0·01-0·04
20, 29-36
Symptomatic HIV(female to male) 0·01-0·06
20, 25-28
Acute HIV (male to female) 0·10-0·40
20, 29-36
Asymptomatic HIV (male to female) 0·02-0·05
20, 29-36
Symptomatic HIV(male to female) 0·02-0·10
20, 25-28
Acute HIV (male to male) 0·20-0·50
37, 38
Asymptomatic HIV (male to male) 0·03-0·06
37, 38
Symptomatic HIV(male to male) 0·03-0·15
39
Annual number of partners: MSM 2-10
40
Annual number of partners: heterosexuals 1-2
41, 42
Condom use: MSM 25-75%
40
Condom use: heterosexuals 7-41%
43, 44
Fraction of men who are circumcised 5-25%
45-47
Reduction in heterosexual HIV transmissibility due to male 48-60%
circumcision
5
Reduction in sexual infectivity due to ART 50-99%
Proportion of short-term HIV-diagnosed on ART:
>500 cells/μL 52-56% PHE unpublished data
350-500 cells/μL 57-62% PHE unpublished data
200-349 cells/μL 77-82% PHE unpublished data
<200 cells/μL 83-88% PHE unpublished data
Quality of life multipliers:
49-52
Acute HIV 0·60-0·95
49-52
Unidentified asymptomatic HIV .. 0·85-0·95
49-53
Identified asymptomatic HIV, first year .. 0·80-0·90
49-53
Identified asymptomatic HIV, subsequent years 0·85-0·95
49-52
Identified symptomatic HIV, not on ART 0·70-0·80
49-52
Symptomatic HIV treated with ART 0·82-0·87
Scenario analysis (shown in Table 2 and Figures 4 and S3)
1. Time horizon N/A 30,40,50 years Assumption
14
2. Discount rate for costs and QALYs and 1·5%,3·5%,5·0%,
3. Reduction in sexual behaviour among persons upon 0,25%,50% Assumption
diagnosis
4. ART costs reduce by 50% (80%): upon patent .. , .. £5518 (£7963) Assumption
expiry, 2018 or from model start , £7082 (£9527)

HIV screening costs set to zero N/A 0 Assumption


5. QoL is unaffected by diagnosis
Unidentified asymptomatic HIV and diagnosed .. and 0·88 Assumption
asymptomatic HIV, first-year ..
ART – antiretroviral therapy; MSM – men who have sex with men; N/A – Not applicable; QoL – Quality of Life.
* Scenarios represent a 50% and an 80% reduction in ART costs upon patent expiry, respectively. Reduction in ART
costs is estimated to occur the year after patent expiry (assumed to occur in 2018).

22