Cost Effectiveness of Screening Strategies for Early Identification of HIV and HCV Infection in Injection Drug Users

Abstract

Objective

To estimate the cost, effectiveness, and cost effectiveness of HIV and HCV screening of injection drug users (IDUs) in opioid replacement therapy (ORT).

Design

Dynamic compartmental model of HIV and HCV in a population of IDUs and non-IDUs for a representative U.S. urban center with 2.5 million adults (age 15–59).

Methods

We considered strategies of screening individuals in ORT for HIV, HCV, or both infections by antibody or antibody and viral RNA testing. We evaluated one-time and repeat screening at intervals from annually to once every 3 months. We calculated the number of HIV and HCV infections, quality-adjusted life years (QALYs), costs, and incremental cost-effectiveness ratios (ICERs).

Results

Adding HIV and HCV viral RNA testing to antibody testing averts 14.8–30.3 HIV and 3.7–7.7 HCV infections in a screened population of 26,100 IDUs entering ORT over 20 years, depending on screening frequency. Screening for HIV antibodies every 6 months costs $30,700/QALY gained. Screening for HIV antibodies and viral RNA every 6 months has an ICER of $65,900/QALY gained. Strategies including HCV testing have ICERs exceeding $100,000/QALY gained unless awareness of HCV-infection status results in a substantial reduction in needle-sharing behavior.

Discussion

Although annual screening for antibodies to HIV and HCV is modestly cost effective compared to no screening, more frequent screening for HIV provides additional benefit at less cost. Screening individuals in ORT every 3–6 months for HIV infection using both antibody and viral RNA technologies and initiating ART for acute HIV infection appears cost effective.

Introduction

Approximately 16% of new HIV diagnoses and two-thirds of new hepatitis C virus (HCV) diagnoses in the U.S. are in injection drug users (IDUs) [1], [2]. Co-infection among IDUs is common, affecting progression rates and treatment effectiveness for both diseases [3], [4], [5], [6], [7], [8]. During the acute infection phase, standard antibody testing either cannot or has low sensitivity to detect these diseases; however, they can be detected with viral RNA tests [9], [10]. Identification of individuals during this phase of infection may be important in averting infections and improving patient outcomes.

The acute phase of HIV infection, lasting approximately 3 months, is characterized by high viral load and high infectivity [11]. The proportion of new infections attributable to individuals with acute HIV infection is unknown, with estimates ranging from 11–50% of new sexually transmitted HIV infections [12], [13]. Identification of individuals during the period of acute infection may reduce HIV transmission through behavior change and initiation of combination antiretroviral therapy (ART) which can reduce infectivity [14]. Additionally, initiating ART during acute infection may slow disease progression [14], [15], [16], [17].

Treatment of chronic HCV with pegylated-interferon and ribavirin (PEG-IFN+RBV) is potentially curative but has high rates of undesirable side effects and is ineffective in 40–60% of patients [8], [18], [19], [20]. Recent clinical trials demonstrated that combination therapy with a HCV protease inhibitor (PEG-IFN+RBV+PI) has higher efficacy in mono-infected genotype 1 patients who are not active IDUs [21], [22], [23]. In a non-IDU population, treatment with PEG-IFN+RBV+PI is cost effective in patients with moderate fibrosis [24]. During the acute phase of HCV infection, estimated to last up to 6 months, PEG-IFN+RBV treatment has substantially higher rates of sustained viral response than when treatment is initiated later in the course of the disease [25], [26], [27], [28], [29], [30], [31], [32], [33] and therefore it is possible that treatment during this phase of the disease may result in important benefits to patients and society.

Previous studies have found that HIV prevention and treatment programs targeted to IDUs, including opioid replacement therapy (ORT) and expanded access to ART, are cost effective and reduce transmission [34], [35], [36], [37], [38], [39], [40]. Although individuals in ORT reduce their risky behaviors, they continue to be at high risk for HIV and HCV [41]. Individuals in ORT are a readily accessible population for frequent screening and treatment initiation because of frequent interactions with health services. Screening for the short acute phase of HIV and HCV infection may identify enough individuals, resulting in improved health outcomes and reduced transmission, to be good value for the additional costs of viral RNA testing. We used a mathematical model to evaluate the potential population-level impacts–costs, effectiveness, and cost effectiveness–of various protocols and frequencies of screening IDUs in ORT for acute and chronic HIV and HCV infection. We considered two HIV and HCV screening technologies, conventional antibody testing and combined antibody and viral RNA testing, and several screening frequencies: once upon entry to ORT only; or upon entry to ORT and routinely thereafter, every 3, 6, or 12 months.

Methods

Model Overview

We developed a deterministic dynamic compartmental model to simulate the population of a representative large U.S. city with 2.5 million persons aged 15 to 59. We estimated values for all model parameters based on published literature, expert opinion, and model calibration (Table 1, Table S1). We validated the model’s estimates of HIV and HCV incidence rates and the proportion of sexually transmitted HIV infections attributable to transmission from an individual in the acute phase of HIV infection to literature estimates (details in Appendix S1). We considered a 20-year time horizon, with calculations in monthly increments. We calculated expected survival, quality-adjusted survival, and expected lifetime health care costs by tracking the time spent in each health state and compared multiple scenarios. We took a societal perspective, considered costs and benefits over a lifetime horizon, and discounted outcomes at 3% annually [42]. We calculated incremental cost-effectiveness ratios (cost per life year (LY) and quality-adjusted life year (QALY) gained) by comparing each strategy to the next best non-dominated strategy. We conducted extensive sensitivity analysis to assess the robustness of model results.

Table 1. Key input parameters.

Variable	Base value	Range		Source
Total population size, age 15–59	2,500,000
Fraction of population that is IDU	1.2%	0.7%	1.8%	* [43]
Fraction of IDUs in ORT	7%	5%	15%	[55], [136]
HIV Prevalence
Overall (age 15–59)	0.47%			[45]
IDU	6.5%	2%	15%	* [137]
Non-IDU	0.40%	0.30%	0.45%	Calculated
Hepatitis C (HCV) Prevalence
Overall (age 15–59)	1.7%	1.4%	2.0%	[46]
IDU	35%	14%	51%	[44]
Non-IDU	1.3%	1.2%	1.4%	Calculated
HCV Treatment Response
Genotype 1 or 4:
Acute HCV	62%	50%	70%	[25], [26], [27], [28], [29]
Acute HCV, HIV+	70%	50%	80%	[30], [31], [32], [33]
Chronic HCV	PEG-IFN+RBV: 40%	30%	60%	[8], [18], [19], [20]
	PEG-IFN+RBV+PI: 65%	40%	80%	[21], [22], [23]
Chronic HCV, HIV+	PEG-IFN+RBV: 30%	20%	50%	[8]
	PEG-IFN+RBV+PI: 65%	40%	80%	Assumed
Genotype 2 or 3:
Acute HCV	62%	50%	70%	[25], [26], [27], [28], [29]
Acute HCV, HIV+	70%	50%	80%	[30], [31], [32], [33]
Chronic HCV	82%	60%	88%	[19], [20]
Chronic HCV, HIV+	66%	50%	80%	[8]
SEXUAL BEHAVIOR PARAMETERS
Average number of sexual partners per year
NON-IDU	2	1.1	3	[58]
IDU	4.3	2	8	[58], [59]
HIV transmission (rate per partner-year)
Acute HIV	0.20	0.10	0.70	Calculated
Asymptomatic HIV (CD4>500 cells/mm3)	0.025	0.02	0.03	[79]
Symptomatic HIV (CD4<500 cells/mm3)	0.05	0.04	0.075	[79]
Effect of ART on infection risk	0.1	0.01	0.5	[79], [80], [81], [82], [83], [84], [85], [86]
HCV transmission (rate per partner-year)
Acute and chronic HCV	0.0003	0	0.002	[138], [139], [140], [141], [142]
Effect of PEG-IFN+RBV or PEG-INF+RBV+PI on infection risk	0.1	0.01	0.5	Estimated, [143], [144]
INJECTING BEHAVIOR PARAMETERS
Average number of injections per year	700	500	1500	[65], [145], [146], [147], [148], [149], [150]
Fraction of injections that are shared	13%	10%	60%	[52], [62], [149], [150], [151], [152], [153], [154], [155]
Relative risk of shared-injecting behavior, in ORT	30%	50%	100%	[61], [62]
HIV transmission (per injection with an HIV+ IDU)
Acute HIV	1.0%	0.8%	1.2%	Assumed the same relative risk of transmission as for sexual contact
Asymptomatic HIV (CD4>500 cells/mm3)	0.12%	0.09%	0.15%	[156], [157]
Symptomatic HIV (CD4<500 cells/mm3)	0.3%	0.25%	0.04%	[156], [157]
Effect of ART on infection risk	0.50	0.1	1.0	[79]
HCV transmission (per injection with an HCV+ IDU)
Acute and chronic HCV	0.4%	0.1%	4.0%	[158], [159]
Effect of PEG-IFN+RBV or PEG-IFN+RBV+PI on infection risk	0.5	0.1	1.0	Estimated, [143], [144]
COSTS
Screening costs
Counseling
Pre-test counseling	12.76			[73]
Post-test, negative result	7.14			[73]
Post-test, positive result	13.84			[73]
HIV diagnostics (testing protocol details are described in Table S2)
Antibody (negative)	12.96			CMS [94], CPT4 86701
Antibody (positive)	67.14			CMS [94], CPT4 86701 (3 times) +86689
RNA amplification (negative)	124.24			CMS [94], CPT4 87535
RNA amplification (positive)	276.74			CMS [94], CPT4 87535 (2 times) +86689
HCV diagnostics
Antibody (negative)	20.84			CMS [94], CPT4 86803
Antibody (positive)	85.13			CMS [94], CPT4 86803 (3 times) +86804
RNA amplification (negative)	62.54			CMS [94], CPT4 87521
RNA amplification (positive)	147.69			CMS [94], CPT4 87521 (2 times) +86804

ART – antiretroviral therapy; HIV – human immunodeficiency virus; HCV – hepatitis C virus; ORT – opioid replacement therapy; CMS – Center for Medicare and Medicaid Services; CPT4 - Current Procedural Terminology, 4th Edition.

^*The proportion of the population that is IDU and the HIV prevalence among IDUs was estimated as the unweighted average of the 21 Metropolitan Statistical Areas (MSAs) with populations between 1.5 and 5 million. Across these cities there is very wide variation in both parameters, so we performed extensive sensitivity analysis on these inputs. The cities included were (Population; % of population that are IDU; Prevalence of HIV in IDU): Boston–Brockton–Nashua, MA–NH (4.2 million, 1.6%, 4.5%), Washington, DC–MD–VA–WV (3.6 million, 0.8%, 9.0%), Philadelphia, PA–NJ (3.4 million, 1.7%, 8.8%), Atlanta, GA (3.0 million, 0.5%, 14.9%), Houston, TX (3.0 million, 1.1%, 6.4%), Detroit, MI (3.0 million, 0.9%, 6.4%), Dallas, TX (2.6 million, 1.3%, 3.4%), Phoenix–Mesa, AZ (2.3 million, 1.2%, 3.6%), Riverside–San Bernardino, CA (2.3 million, 0.9%, 3.5%), Minneapolis, MN (2.1 million, 0.5%, 3.3%), Orange County, CA (2.0 million, 1.0%, 2.4%), San Diego, CA (2.0 million, 1.3%, 3.4%), Nassau–Suffolk, NY (1.8 million, 0.7%, 12.3%), St. Louis, MO–IL (1.8 million, 0.6%, 3.1%), Baltimore, MD (1.7 million, 3.4%, 11.7%), Seattle–Bellevue–Everett, WA (1.7 million, 1.6%, 2.9%), Oakland, CA (1.7 million, 1.3%, 4.2%), Tampa–St. Petersburg–Clearwater, FL (1.6 million, 1.1%, 6.1%), Miami, FL (1.5 million, 0.6%, 22.8%), Denver, CO (1.5 million, 1.4%, 3.1%), Pittsburgh, PA (1.5 million, 0.9%, 3.9%), Cleveland–Lorain–Elyria, OH (1.5 million, 0.8%, 4.2%). We excluded the three MSAs with populations over 5 million: Los Angeles–Long Beach, CA (6.5 million, 1.5%, 3.8%), New York, NY (6.4 million, 1.4%, 21.2%), Chicago, IL (5.7 million, 0.6%, 8.4%).

Population Groups

We subdivided the population into three risk groups based on IDU status: current IDU, IDU in ORT, and non-IDU (Figure 1). Based on current estimates from large U.S. cities, we assumed that approximately 1.2% of the modeled population are IDUs, with 6.5% HIV prevalence [43] and 35% HCV prevalence [44] among IDUs. We estimated HIV and HCV prevalence among non-IDUs using the U.S. adult population prevalence of 0.47% [45] and 1.7% [46], respectively. We calibrated the model to match estimates of HIV and HCV prevalence and incidence in IDUs and the general population (details in Appendix S1, Figure S1, Figure S2, and Figure S3).

Figure 1. Model schematic.

Each compartment is described by three characteristics: (A) risk group (IDU category), (B) HIV status, and (C) HCV status. In each cycle, individuals within any compartment may stay in the same compartment or may change in any or all of these dimensions. Rates of movement between levels of disease severity are conditional on the current state of the individual (including IDU status and presence of co-infection). Rates of movement between status of uninfected and infected are conditional on risk group, the number of infected individuals, and the sufficient contact rate.

We divided HIV infection status into uninfected, acute HIV infection, asymptomatic HIV, and symptomatic HIV/AIDS. We divided HCV infection status into uninfected, acute infection, asymptomatic chronic, symptomatic chronic, and end-stage liver disease. We grouped the four most common HCV genotypes into two groups based on similarity of treatment protocol and treatment response: genotypes 1 and 4 and genotypes 2 and 3. Further, we considered whether an individual is aware of his/her HIV or HCV infection status or is on HIV and/or HCV treatment. The model includes a compartment for every combination of IDU, HIV, and HCV status, and treatment and awareness, for a total of 756 compartments. Individuals transitioned between compartments according to rates defined by the dynamics of disease transmission and progression.

Data Sources and Assumptions

Population Dynamics

All individuals enter the model at age 15 as non-injection drug users (non-IDUs) without HIV or HCV infection. Individuals exit the population due to maturation (at age 60) or death. Annual baseline death rates vary by risk group to account for variation in drug-use-related mortality [47]. We estimated the mortality rate among non-IDUs using the average mortality rate for the 15–59 year old United States (U.S.) population [48], [49]. We estimated the mortality rate among IDUs not in ORT to be 31.1 per 1000 person-years and estimated that IDUs in ORT have a 60% lower mortality rate than IDUs not in ORT [47], [50], [51].

Disease Progression and Mortality

We estimated HIV and HCV progression and mortality rates, and the impacts of co-infection on progression and treatment effectiveness from previous models of their natural history and progression as well as clinical and observational trials (Table 1, Table S1). We assumed that individuals with a CD4 count <500 cells/mm³ were eligible to receive combination ART and that treatment with ART slowed the progression of HIV and reduced HIV infectivity. The duration of HCV therapy and treatment effectiveness differed by HCV genotype category and treatment type [2], [22], [23]. The effectiveness of a PEG-IFN+RBV+PI regimen to cure chronic genotype 1 HCV infection in mono-infected individuals was estimated from recent trials [22], [23]. Treatment effectiveness of PEG-IFN+RBV for treatment of chronic HCV infection for genotypes other than type 1 and during the acute phase of HCV in mono- and HIV co-infected individuals was estimated based on recent trials [25], [26], [27], [28], [29], [30], [31], [32].

Risk Behaviors

We estimated IDU risk behaviors using published reports from the Collaborative Injection Drug Users Study (CIDUS) [52], [53], [54]. We assumed that the injection-drug-using population would remain a stable proportion of the total population over the 20-year intervention horizon and that the proportion of the IDU population in ORT would be constant at 7% [55]. Without incremental interventions, we assumed that HIV-negative IDUs have a 4.0% annual probability and HIV-positive IDUs have a 6.7% annual probability of stopping injection behaviors [56]. We estimated that the annual rate of leaving ORT and stopping injection drug use was 1.8% and that each year 44.1% of individuals in ORT would quit ORT and return to drug injection [57]. Using these assumptions and estimates, we calculated the rate at which non-IDUs become IDUs and the rate at which IDUs enter ORT.

Disease Transmission

We incorporated HIV and HCV transmission from sexual partnerships and injection equipment sharing through risk-structured mass action. In each month, the number of sexual partnerships, using and not using condoms, and the number of injection equipment sharing partnerships, using and not using bleach, were calculated based on risk-group-specific average number of sexual and injection equipment sharing partners, condom rates, and bleach use rates [58], [59], [60], [61], [62]. We assumed preferential sexual mixing of IDUs with other IDUs (40% of IDU sexual partners were other IDUs) [54], [63], [64], [65]. We assumed that the viral load reductions that occur during treatment for HIV and HCV resulted in reductions in infectivity. In the base case, regardless of how diagnosis occurred, we assumed that awareness of HIV-positive disease status resulted in an increase in condom use [63], [66], [67] and, among IDUs, a 20% reduction in needle sharing [68]. We assumed that awareness of HCV-positive disease status did not result in a reduction in needle sharing behavior [53], [69], [70], [71]. We varied these assumptions in sensitivity analysis.

Screening Strategies

We assumed that individuals may learn of their HIV and/or HCV status through symptomatic case finding, an existing screening program, or a new screening intervention. We estimated baseline rates of diagnosis via existing screening programs through calibration to current rates of under-diagnosis of HIV and HCV among IDUs and non-IDUs (Appendix S1).

We considered two HIV and HCV screening technologies, conventional antibody testing and combined antibody and RNA testing. The HIV and HCV test sequence and confirmatory follow-up are based on those implemented in screening programs [72], [73] and the CDC recommendations for suspected cases, respectively (Table S2) [2]. In the base case, we considered a 3^rd generation HIV antibody test which we assumed identifies one-third of individuals infected in the past 3 months (acutely infected individuals); we considered HIV antibody tests with greater sensitivity in the acute infection period (such as a 4^th generation HIV antibody and p24 antigen test) in sensitivity analysis. In scenarios with HIV RNA testing, individuals who did not test HIV antibody positive were subsequently tested for HIV RNA. The individuals screened are clients of an ORT program, so we assumed that 100% of individuals receive their test results. We considered several screening frequencies: once upon entry to ORT only; or upon entry to ORT and routinely thereafter, every 3, 6, or 12 months.

In the base case, we assumed 50% of individuals identified with acute HIV [74], individuals with a negative antibody test and a positive RNA test, and 40% of individuals identified with acute HCV would initiate treatment. The optimal duration of therapy for patients with acute HIV infection is unknown. We assumed that individuals who initiated ART during acute HIV infection continued ART after the acute phase even with a CD4 count >500 cells/mm³ [75], [76], [77], [78]. We assumed that ART reduces sexual infectivity by 90% and infectivity from injection transmission by 50% [79], [80], [81], [82], [83], [84], [85], [86]. In the base case, we did not consider any change in the rate of HIV disease progression caused by ART initiation during acute or early HIV infection. We estimated the probability of sustained virologic response in patients who initiate PEG-IFN+RBV during acute HCV infection based on recent clinical trials [25], [26], [27], [28], [29], [30], [31], [32]. Consistent with current evidence [28], [87], [88], we assumed that acute HCV treatment would be equally effective for IDUs in ORT and for non-IDUs.

Costs

Individuals accrued health care costs based on their health state each month and for transitions between states or events within a cycle such as screening and diagnosis. We expressed all costs in 2009 U.S. dollars using the U.S. GDP deflator [89].

Baseline costs

We estimated annual baseline health care expenditures for non-IDUs using age-specific averages for the U.S. population [90], [91] and we increased this by $2,021 for HIV- and HCV-negative IDUs [92]. We estimated the annual cost of ORT to be $5,171 [93]. We estimated the cost of death for an IDU for causes other than HIV or HCV to be $8,350 based on Medicare reimbursement rates for an emergency room visit and hospitalization from drug overdose with major complications [94].

Disease-attributable HIV and HCV costs

We assumed that following diagnosis with HIV or HCV, all patients would have their disease staged and characterized to assist with treatment decisions; we assumed that this included assessment of viral load and genotyping and cost $500 and $438 per HIV and HCV diagnosis, respectively, based on the Medicare reimbursement schedule [94].

We used a recent modeling study to estimate the costs of HIV health states [95]. We assumed that asymptomatic HIV-infected individuals who are unaware of their disease incur no additional health care costs, while individuals with symptomatic disease incur additional costs regardless of whether their disease has been diagnosed. We assumed that the annual cost of ART is approximately $22,000 and the remainder of the HIV-associated health care cost is for disease monitoring, opportunistic infection prophylaxis, and other outpatient care [95]. We estimated the cost of health care in the last month of life with HIV to be $33,480 which is the cost of death from an opportunistic infection [95].

We used a prior cost-effectiveness analysis evaluating screening for HCV in the general population to inform our estimates of the HCV attributable costs [96]. We assumed that the weekly cost of PEG-IFN+RBV was $471 ($11,304 for 24-week course of treatment and $22,608 for a 48-week course of treatment) [97], [98]. We estimated that combination therapy with a protease inhibitor cost an additional $1,100 per week which would add an average cost of $40,000 per patient. We assumed the incremental end-of-life costs associated with HCV to be the same as those accruing from non-HCV death.

Screening program costs

For screening costs, we used CDC estimates for pre- and post-test counseling and 2009 Medicare reimbursement rates for laboratory tests [73], [94]. We assumed testing protocols as described by guidelines and in descriptions of practice [2], [72], [73], [99] and assumed HIV and HCV antibody and RNA test costs based on the Medicare reimbursement schedule [94]. We assumed that 100% of screened individuals would obtain their results and receive the appropriate post-test counseling [73].

Quality of Life

We assumed a baseline quality-of-life weight of 0.9 for healthy non-IDUs using age-specific values for the U.S. population and averaging based on the distribution of individual ages [100], [101]. We estimated a baseline quality-of-life weight of 0.747 for IDUs after adjusting for the average age of the population in the model [102].

Additionally, we incorporated multiplicative quality-of-life weights for individuals with HIV [103], [104], [105], [106] and HCV [107], [108] based on their disease stage. Awareness of HIV and HCV status affects quality of life, so we included this in the model [109], [110]. In addition, we included a decrement in quality of life associated with PEG-IFN+RBV(+/−PI) treatment [107].

Results

HIV and HCV Infections Averted

With no screening targeted to individuals in ORT (referred to as ‘no screening’), we estimate that 7371 HIV infections and 25,704 HCV infections will occur over the next 20 years (discounted at 3% annually) in a population of 2.5 million with 26,100 IDUs entering ORT (2100 IDUs in ORT at any one time). Screening only for chronic HIV infection averted 13.8 to 27.6 HIV infections (depending on screening frequency) and, primarily through risk-reducing behavior changes associated with awareness of HIV-positive status, a very small number of HCV infections (Figure 2). Screening only for chronic HCV infection averted 18.0 to 20.0 HCV infections and 2.3 to 2.5 HIV infections. HIV infections were averted by HCV screening because all individuals newly diagnosed with one infection were screened for the other during follow-up; due to its relatively high prevalence (35%) and low rate of awareness (25%), HCV screening results in a large absolute number of diagnoses and, therefore, HIV tests.

Figure 2. Estimated number of HIV and HCV infections averted for each screening strategy over a 20-year time horizon compared to a strategy of no screening of IDUs in ORT (discounted at 3% annually).

Screening for HIV antibodies with increased frequency averted few incremental infections. For example, increasing screening frequency from annually to twice-annually averted only 3.3 additional HIV infections over 20 years. Incorporating HIV RNA testing to identify acute infections averted many more infections than increasing the frequency of HIV screening: for screening frequency of upon entry to ORT to every 3 months, including RNA detection averted 14.8 to 30.3 more HIV infections, respectively, than antibody screening alone. Across all screening strategies considered, approximately 52% of infections averted were averted in the non-IDU population. Identifying 1 IDU in ORT with chronic HIV with a CD4 count <500 cells/mm³ and initiating ART averted 0.1 HIV infections over 20 years. Diagnosis during the acute phase averted more HIV infections than later diagnosis even if ART is not initiated: over 20 years, diagnosing 1 IDU in ORT with acute HIV infection averted 0.4 HIV infections if ART was not immediately initiated and 1.3 HIV infections if ART was immediately initiated.

Compared to screening for HCV antibodies annually, screening twice annually averted no additional HCV infections over 20 years. Including HCV viral RNA detection averted an additional 3.7 to 7.7 infections over 20 years compared to antibody screening alone for screening frequency of upon entry to ORT to every 3 months, respectively. Early identification and treatment of HCV averts few infections primarily because not all acutely infected individuals will progress to chronic infection and HCV re-infection is common, absent behavior change.

HIV and HCV Prevalence

Screening of IDUs in ORT for HIV and HCV prevents infections but has little effect on overall HIV and HCV prevalence because the number of people targeted through screening in ORT is small. Compared to no screening, the relative change in HIV prevalence in the total population in year 20 is 0.20% and 0.23% lower with annual and twice-annual HIV antibody testing, respectively; whereas the relative change in HIV prevalence in year 20 is 0.43% and 0.51% lower with annual and twice-annual HIV antibody and RNA testing, respectively. In the IDU population, twice-annual screening for HIV antibody and RNA decreases HIV prevalence in year 20 by 1.1% (relative) compared to no screening. Across all strategies considered, the relative change in HCV prevalence in the total population in year 20 was reduced no more than 0.32% compared to a strategy of no screening.

Cost Effectiveness

Following current guidelines of annual HIV and HCV antibody screening for all IDUs in ORT costs $35,100/LY gained and $80,800/QALY gained when compared to no screening of IDUs in ORT. However, this strategy costs more and provides fewer benefits than strategies that screen more frequently for HIV only (Figure 3).

Figure 3. Cost-effectiveness plane presenting all non-dominated and selected dominated screening protocols and frequencies targeting injection drug users in ORT.

Table 2 reports the incremental cost-effectiveness ratio (ICER) of each strategy compared to the next-best alternative for strategies on the efficient frontier; Table S3 shows results for all strategies. Results differed depending on the measure of benefit (LY gained or QALY gained), largely because of the decrease in quality of life associated with awareness of asymptomatic HIV or HCV infection. Screening every 6 months for HIV antibodies and RNA costs $65,900/QALY gained compared to screening annually. Screening every 3 months for HIV antibodies and RNA costs $115,400/QALY gained. Further, including HCV antibody testing upon entry to ORT increases the ICER to $168,600/QALY. Screening every 6 months for HIV antibodies and RNA and for HCV antibodies upon entry to ORT costs $57,200/LY gained; further increasing the frequency of HCV antibody screening increases the cost to $71,400/LY gained. Screening every 3 months for HIV antibodies and RNA and annually for HCV antibodies costs $100,750/LY gained.

Table 2. Base case outcomes and incremental cost-effectiveness ratios for non-dominated strategies in a representative city of 2.5 million individuals age 15–59 years, with 1.2% of the population IDU, and 6.5% and 35% prevalence of HIV and HCV among IDU, respectively.^* .

Screening Protocol	Screening Frequency**	HIV Infections Averted	HCV Infections Averted	Incremental Cost	Incremental LYs	IncrementalQALYs	ICER ($/LY gained)	ICER ($/QALY gained)***
No screening****		Reference	Reference	Reference	Reference	Reference	Reference	Reference
Anti-HIV	Upon entry to ORT	13.78	0.01	1,580,365	169	141	9,365	11,191
Anti-HIV	Annual	20.22	0.00	2,874,166	245	206	16,938	20,075
Anti-HIV	6 months	23.55	0.02	3,832,733	281	237	26,436	30,713
Anti-HIV+RNA	Upon entry to ORT	28.54	(0.37)	5,509,497	337	287	30,323	33,503
Anti-HIV+RNA	Annual	41.51	(0.60)	11,200,954	487	416	37,900	44,141
Anti-HIV+RNA	6 months	49.34	(0.75)	16,207,602	574	492	Dominated	65,883
Anti-HIV; Anti-HCV	Annual	19.10	19.85	25,652,696	731	318	Dominated	Dominated
Anti-HIV+RNA	3 months	57.82	(0.96)	25,664,563	668	574	Dominated	115,429
Anti-HIV+RNA; Anti-HCV	Annual Upon entry to ORT	40.57	17.33	30,938,150	930	533	44,532	Dominated
Anti-HIV+RNA; Anti-HCV	6 months Upon entry to ORT	48.42	17.17	35,936,712	1,017	609	57,192	Dominated
Anti-HIV+RNA; Anti-HCV	6 months Annual	48.26	19.06	38,956,858	1,060	604	71,399	Dominated
Anti-HIV+RNA; Anti-HCV	3 months Upon entry to ORT	56.90	16.96	45,390,578	1,111	691	Dominated	168,600
Anti-HIV+RNA; Anti-HCV	3 months Annual	56.75	18.86	48,410,723	1,154	686	100,749	Dominated
Anti-HIV+RNA; Anti-HCV	3 months 6 months	56.75	18.82	49,421,140	1,156	683	489,639	Dominated
Anti-HIV+RNA; Anti-HCV+RNA	3 months Annual	56.72	23.45	55,246,297	1,162	681	905,133	Dominated
Anti-HIV+RNA; Anti-HCV+RNA	3 months	56.71	26.47	64,329,321	1,170	689	1,220,703	Dominated

HIV – human immunodeficiency virus; HCV – hepatitis C virus; LYs – life years; QALYs – quality-adjusted life-years; ICER – incremental cost-effectiveness ratio; IDU – injection drug user.

^*Outcomes for all strategies considered are shown in Table S3.

^**Frequencies considered were: Upon entry to ORT; “Annual” = Upon entry to ORT and annually while in ORT; “6 months” = Upon entry to ORT and every 6 months while in ORT; “3 months” = Upon entry to ORT and every 3 months while in ORT.

^***“Dominated” indicates that the strategy costs more and provides fewer benefits than another strategy or a combination of two strategies.

^****This strategy consists of baseline case detection rates in the IDU and non-IDU populations and no screening targeted to individuals in ORT.

Sensitivity Analysis

We considered alternate-city scenarios by varying the number of IDUs, the fraction of IDUs in ORT and the HIV and HCV prevalence among IDUs. Varying the number of IDUs, the fraction of IDUs in ORT, and the prevalence of HCV among IDUs had little impact on the cost effectiveness of the screening strategies (Table S4). When we increased the proportion of IDUs in ORT to 40%, the ICER of screening for HIV antibodies and RNA every 6 months increased from $65,900/QALY gained to $100,600/QALY gained because high rates of ORT use lower the average HIV risk of the population (in the economic sense, ORT and HIV screening are partial substitutes). Our results were sensitive to HIV prevalence among IDUs. In low (3.5% of IDUs) and high (17% of IDUs) HIV-prevalence scenarios, screening for HIV antibodies and RNA every 6 months costs $107,000/QALY gained and $23,000/QALY gained, respectively. Results were not sensitive to the effectiveness of ORT or to the average time spent in ORT within realistic ranges (Table S5).

Results were robust to clinically relevant changes in the HIV natural history and ART effectiveness parameters, but sensitive to rates of HIV treatment initiation (Table S6). However, even with low uptake of ART (25%) among individuals identified with acute HIV infection, screening every 6 months for HIV antibodies and RNA cost $77,200/QALY gained. In general, our results were not sensitive to changing access to or effectiveness of HCV treatment (Table S7). We considered scenarios in which initiation of ART in individuals with CD4 counts >500 cell/mm³ slowed HIV progression. These additional benefits increase the cost effectiveness of acute HIV screening strategies: screening every 6 months for HIV antibodies and RNA cost between $61,500 and $65,200/QALY gained depending of the reduction in progression rate (Table S6).

Results were sensitive to the length of time after infection until HIV is detectable (Table S8). As newer 4^th generation HIV tests which combine sensitive HIV antibody technologies with p24 antigen tests become more widely available, fewer acute infections are identified by the addition of RNA testing to the screening protocol. If the window period of detection for the 4^th generation HIV test is 1 month, screening every 6 months with a 4^th generation test and RNA costs $116,000/QALY gained (compared to $65,900/QALY gained if the window is 2 months).

We also explored scenarios in which awareness of HCV status changed needle-sharing behavior. Assuming that awareness of HCV-positive status decreases needle-sharing by 5% substantially improved the cost-effectiveness of HCV screening. For example, screening every 6 months for HIV antibodies and RNA and for HCV antibodies upon entry to ORT costs $67,400/QALY gained. However, even with high rates of behavior change, screening for acute HCV infection always has very high ICERs (>$200,000 per QALY gained).

Assumptions relating to quality of life were important drivers in the difference between the results in terms of per LY gained and per QALY gained. However, varying the quality of life weights within clinically reasonable ranges that maintain the rank ordering of health states did not substantially change the conclusions, with one notable exception: the reduction in quality of life associated with HCV diagnosis. When we considered no reduction in quality of life associated with awareness of HCV-positive status in an asymptomatic individual, screening for HCV antibodies became increasingly attractive: screening for HIV antibodies and RNA annually and for HCV antibodies upon entry to ORT costs $44,200/QALY gained, screening for HIV antibodies and RNA every 6 months and for HCV antibodies upon entry to ORT costs $65,740/QALY gained, and screening for HIV antibodies and RNA every 6 months and for HCV antibodies annually costs $69,400/QALY gained (similar strategies in the base case analysis cost more than $100,000/QALY gained).

Discussion

Using a model which was calibrated to empirical data and expert estimates of trends if the status quo were continued, our analysis indicates that screening IDUs in ORT as frequently as every 6 months for HIV antibodies and RNA is likely to be a cost-effective means of reducing the spread of HIV among IDUs and non-IDUs. Although screening annually with antibodies to HIV and HCV is moderately cost effective relative to no screening, this strategy is less effective and more costly than strategies that include more frequent HIV screening. The cost effectiveness of HCV screening strategies improves when awareness of HCV-positive status is associated with a reduction in needle-sharing behavior and is not associated with a decrement in quality of life.

Initiation of treatment during the highly infectious acute period of HIV may be influential in reducing HIV transmission [9], [14]. Our results demonstrate the importance of being able to distinguish between acute and chronic infections because it facilitates targeted treatment during the highly infectious acute phase. Thus, when 4^th generation HIV tests are used, the preferred strategy is HIV antibody screening every 3 months (ICER of $38,000/QALY gained) and strategies that include HIV RNA testing have ICERs above $100,000/QALY gained. This tradeoff between more sensitive 4^th generation HIV antibody and p24 antigen tests and the ability to distinguish between acute and chronic HIV infections has also been observed in other analyses comparing HIV RNA testing combined with 3^rd or 4^th generation HIV antibody tests [10]. As of 2012, ART is recommended for all HIV-infected individuals [78]. If, as a result, all patients initiate ART at diagnosis, distinguishing between acute and chronic infections will be less important.

Cost has been identified as a key factor preventing expanded access to acute HIV testing [111]. Pooling samples to reduce cost has been proposed and implemented in pilot projects of acute HIV testing [72], [111], [112], [113]. Importantly, we find that twice-annual acute HIV screening costs less than $50,00/QALY gained even when each sample is tested individually at a cost of $51.25 per sample (the Medicare reimbursement level [94]), much higher than the average pooled cost per specimen of $3.53 reported elsewhere [72].

Initiation of PEG-IFN+RBV during acute and early HCV infection appears more likely to result in a sustained viral response than when treatment is initiated later in the course of disease [25], [26], [27], [28], [29]. However, our analysis indicates that relatively few HCV infections are averted per acute HCV infection treated because the lifetime risk of HCV infection remains very high among IDUs. Also, the prolonged asymptomatic phase of HCV infection results in a small present value of benefits to each treated patient from early intervention.

Recommendations for chronic HCV screening in high-risk individuals are a subject of debate [114]. The U.S. Preventive Services Task Force found the evidence supporting screening insufficient to make a recommendation [99] but the CDC and NIH recommend routine HCV screening of high-risk individuals [2], [115]. How the recommendations will change with the availability of a more effective treatment for chronically infected genotype 1 patients is uncertain. While our analysis does not find acute HCV testing to be cost effective in any scenario, we do find that HCV antibody testing upon entry to ORT with subsequent treatment with PEG-IFN+RBV+PIs or PEG-IFN+RBV to have an ICER of just over $100,000/QALY gained when access to treatment is high. Further, the quality-of-life reduction associated with awareness of HCV-positive status was an important but highly uncertain parameter: with little to no quality-of-life reduction, HCV screening upon entry to ORT or annually is moderately cost effective. Additionally our results highlight the importance of behavior change, especially after HCV diagnosis, for achieving reduced HIV and HCV transmission, underscoring the need for effective counseling and access to clean needles and injection equipment.

Our findings are broadly consistent with prior studies of the cost effectiveness of HIV screening and treatment expansion [35], [116], [117] and screening for chronic HCV infection in IDUs [118], [119], [120], [121]. We find, as have others [34], [35], [36], [37], that HIV prevention strategies targeted to IDUs can substantially reduce the number of new HIV infections among non-IDUs. To our knowledge, no previous study has considered the cost effectiveness of routine screening for acute HIV infection in IDUs. Our results differ from the one study that considered the cost effectiveness of screening IDUs for acute/early HCV infection; that study found antibody screening every 6 months and initiation of treatment to be highly cost effective and potentially cost-saving [122]. However, that study assumed that 100% of identified cases among IDUs would be eligible for PEG-IFN+RBV treatment and did not include the possibility of re-infection, which is known to occur [123].

Our analysis has several limitations. Our ‘representative city’ does not perfectly represent the HIV-HCV co-epidemic in IDUs in any specific U.S. city. However, via sensitivity analysis of key ‘city-specific’ parameters we attempted to demonstrate the fairly wide generalizability of our model findings and to show how results change for cities with very high rates of ORT use or relatively low rates of HIV in IDUs. We only capture new infections among adults aged 15 to 59. Including older individuals would minimally impact the results as few new infections occur in persons over age 60. We did not include benefits from maternal transmissions averted or from contact tracing. Inclusion of these benefits may increase the cost effectiveness of screening. We did not consider screening for other diseases that also occur frequently in this population such as hepatitis B virus infection. We did not consider HIV screening technologies including rapid or oral tests, or the recently approved at-home HIV test. We did not include the risks of poor ART adherence resulting in drug-resistant HIV and the increase in costs associated with treating drug-resistant infections. We did not include many of the potential effects on behavior–either positive or negative–that might accrue from very frequent screening and counseling such as increased condom use or increases in serosorting [124], [125], [126]. Finally, we estimated the lifetime costs, LY, and QALYs for all individuals in the model at the end of the intervention horizon (20 years) based on their terminal health state using a model in which we did not continue the screening intervention and did not allow for any additional disease transmission. Although these two assumptions may have resulted in overestimations of the LYs and QALYs gained in this period, these estimates had little influence on the cost effectiveness of strategies.

Currently, testing for acute HIV is not widely available outside of pilot programs [9], [72], [111], [127], [128], [129], [130], [131], and access to HIV and HCV counseling, testing, and treatment varies widely across drug treatment programs [132], [133], [134]. Fewer than 50% of IDUs receive the recommended annual testing for HIV and HCV [132], [133], [134]. For acute HIV screening to be effective, testing of samples, reporting of results, and initiation of treatment must occur quickly. Infrastructure changes and education of substance abuse workers and associated health professionals may be required [13], [134], [135]. Our analysis indicates that not testing IDUs in ORT frequently for acute and chronic HIV infection is a missed public health opportunity. Such screening could reduce the number of new HIV infections and would be cost effective.

Supporting Information

Figure S1

Results of calibration to total population and IDU rates of undiagnosed HIV (Figure S1a) and HCV (Figure S1b).

(TIF)

Figure S2

Results of calibration to prevalence of HIV in IDUs (Figure S2a) and the total population (Figure S2b) and calibration to prevalence of HCV in IDUs (Figure S2c) and the total population (Figure S2d).

(TIF)

Figure S3

Results of validation to total population HIV incidence (Figure S3a) and HCV incidence (Figure S3b).

(TIF)

Table S1

Base case parameter values and range for sensitivity analysis.

(DOCX)

Table S2

Description of screening protocols.

(DOCX)

Table S3

Base case results for all strategies considered.

(DOCX)

Table S4

Sensitivity analysis on city-specific epidemic characteristics. Incremental cost-effectiveness ratio ($/QALY gained) for selected strategies on the efficient frontier compared to the next-best strategy.

(DOCX)

Table S5

Sensitivity analysis on ORT effectiveness parameters. Incremental cost-effectiveness ratio ($/QALY gained) for selected strategies on the efficient frontier compared to the next-best strategy.

(DOCX)

Table S6

Sensitivity analysis on HIV parameters. Incremental cost-effectiveness ratio ($/QALY gained) for selected strategies on the efficient frontier compared to the next-best strategy.

(DOCX)

Table S7

Sensitivity analysis on HCV parameters. Incremental cost-effectiveness ratio ($/QALY gained) for selected strategies on the efficient frontier compared to the next-best strategy.

(DOCX)

Table S8

Sensitivity analysis on the length of the HIV antibody test detection window. Incremental cost-effectiveness ratio ($/QALY gained) for selected strategies on the efficient frontier compared to the next-best strategy.

(DOCX)

Appendix S1

Supplemental results and sensitivity analysis and supplemental model details.

(DOCX)

Funding Statement

This work was supported by grant R01-DA15612 from the National Institute on Drug Abuse. LEC is supported by a doctoral fellowship from the Social Science and Humanities Research Council of Canada (http://www.sshrc-crsh.gc.ca) and the Seth Bonder Scholarship for Applied Operations Research in Health Services (http://www.informs.org). DKO and MH are supported by the Department of Veterans Affairs. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.