|Home | About | Journals | Submit | Contact Us | Français|
As the global community evaluates the unprecedented investment in the scale-up of HIV therapy and considers future investments in HIV care, it is crucial to identify those HIV interventions that maximize the benefit realized from each dollar spent. The use of laboratory monitoring assays – CD4 cell count and HIV RNA – in decisions about when to initiate and switch antiretroviral therapy may offer substantial clinical benefit, but their economic value remains controversial. Cost-effectiveness analysis can be used to evaluate the value for money of strategies for HIV care, including alternative approaches to laboratory monitoring. Five published cost-effectiveness analyses address the question of CD4 and HIV RNA monitoring for HIV-infected patients in Africa, with differing conclusions. We describe the use of cost-effectiveness analysis in resource-limited settings and review the cost-effectiveness literature with regard to CD4 and HIV RNA monitoring in Africa, highlighting some of the most critical issues in this debate.
Since the introduction of antiretroviral therapy (ART) in resource-limited settings, debate has mounted regarding the value of laboratory monitoring with CD4 cell count and HIV RNA for HIV disease in these settings. While these technologies are the standard of care in well-resourced countries, the benefits of these tools must be weighed against their costs in resource-limited countries [1-4]. Some argue that complicated monitoring technologies impede the overall goal – to deliver antiretroviral therapy to as many patients as possible. This view was exemplified in a 2006 quote by Anthony Harries, MD, in the Lancet: “the best is the enemy of the good…If we complicate the [ART] plan with technical accessories, it will be in great danger of failing” .
Cost-effectiveness analysis is a methodology used to examine the clinical benefit of interventions and their value for money. Several cost-effectiveness analyses, most based on mathematical models, have examined the value of CD4 count and HIV RNA monitoring for patients on ART in sub-Saharan Africa [6-10]. We further inform this debate by critically reviewing this diverging literature with focused attention to differences in methods, input parameters, and assumptions.
The 2006 World Health Organization (WHO) treatment guidelines and recently published 2009 brief recommendations emphasize two key roles for laboratory testing in HIV-infected patients: 1) to inform decisions regarding eligibility for ART initiation, and 2) after patients initiate ART, to identify treatment failure and inform the timing of switching patients to another available ART regimen [2, 4]. Without widely available laboratory infrastructure, the WHO guidelines generally recommend clinical assessment and CD4 testing to determine eligibility for ART initiation and to monitor patients on ART. CD4 count monitoring is recommended biannually, and HIV RNA monitoring is suggested biannually as a “conditional recommendation” in settings where HIV RNA tests are “routinely available.”
In many countries, national treatment guidelines reflect locally available resources and differ from the WHO guidelines. In Malawi, for example, where CD4 counts are not widely accessible, the 2008 revised recommendations suggest clinical monitoring alone, with CD4 prioritization (for use in ART initiation) for pregnant women, children, and those with WHO stage 2 disease . In Tanzania, national recommendations suggest CD4 monitoring every 6 months and HIV RNA, when available, noting that the capacity for HIV RNA testing is limited largely to tertiary referral centers . In contrast, the South Africa guidelines are more consistent with those of the WHO, suggesting CD4 monitoring every 6 months and CD4 and HIV RNA monitoring every 6 months during the first ART regimen .
A critical component in determining the value of laboratory tests is their cost, including the cost of the test kits; test administration; specimen transport; purchase or rental of laboratory equipment; laboratory reagents; personnel time, training, and retention; specimen processing; laboratory information systems; and ongoing quality assurance. In most resource-limited settings, a CD4 count test costs about $5-$31 (2007 USD) and an HIV RNA assay by PCR about $26-$92 (2007 USD) [6-10]. However, test costs alone do not convey a complete picture of the costs and/or savings associated with the use of these assays. Although the use of clinical monitoring alone to guide ART initiation or switching is often considered to be “free” of cost, this assumption ignores the costs associated with the increased likelihood of developing an opportunistic disease, which confers substantial morbidity and mortality, prompting the use of costly health care services. A more comprehensive assessment of the value of laboratory tests takes into account both economic and health outcomes and incorporates test costs and costs of care required or avoided by their use.
To assert that an intervention is cost-effective does not mean that it is cheap or that it saves money . Most interventions that improve health and extend survival add costs to care. By standard definition, a strategy of care may be considered “cost-effective” if its additional clinical benefit, relative to another strategy, is felt to be “worth” its additional cost . Cost-effectiveness analysis is a formal methodology that includes both costs (current and future) and effectiveness (short- and long-term), either per person or as a total amount for a defined population. Costs are measured in a specific currency (often US or international dollars), and effectiveness is most often quantified in either years of life saved (YLS) or quality-adjusted life-years saved (QALY). The latter outcome assigns quality-of-life “weights” to health conditions and values each year lived in imperfect health as worth less than one year in perfect health .
From cost and effectiveness outcomes for two alternative strategies, an incremental cost-effectiveness ratio is calculated. The difference in costs between the competing strategies is the numerator, and the difference in effectiveness comprises the denominator. Thus, the cost-effectiveness ratio values interventions by examining the benefits they confer, compared to the benefits that might be obtained if resources were allocated to another intervention, acknowledging both that resources are limited and that there may be other claims to them.
In comparing interventions, any strategy that costs more but produces fewer YLS or QALYs than a competing strategy is said to be “strongly dominated,” or an economically irrational choice, and is removed from consideration . Incremental cost-effectiveness ratios are then calculated for all remaining strategies, comparing each to the next less expensive strategy. Strategies that represent economically inefficient uses of resources (i.e., have higher cost-effectiveness ratios than less expensive but less effective interventions) are described as “weakly dominated” and are also eliminated from consideration . The remaining strategies have cost-effectiveness ratios that increase with increasing costs.
The Commission on Macroeconomics and Health (sponsored by the WHO) has suggested that an individual country’s per capita Gross Domestic Product (GDP) be used to guide whether the intervention is affordable in that country [18, 19]. According to these recommendations, interventions may be considered “very cost-effective” if they have cost-effectiveness ratios less than the country’s per capita GDP and “cost-effective” if they have cost-effectiveness ratios less than 3 times the per capita GDP.
Inherent in this threshold, however, is the differential capacity for countries to pay for healthcare, according to its per capita GDP (Table 1). These GDP-based thresholds were developed for and may be appropriate when a given country must prioritize among multiple health demands within its budget. However, HIV/AIDS interventions are often partially supported by international organizations in collaboration with individual governments, raising important issues concerning the use of GDP-based cost-effectiveness criteria among countries [21, 22]. Despite these limitations, the WHO-recommended GDP thresholds remain the most commonly used criteria by which published cost-effectiveness ratios are contextualized for resource-limited settings [18, 19].
Five published model-based cost-effectiveness analyses have examined the value of laboratory monitoring in resource-limited settings [6-10]. The data used as model input parameters for each of these studies vary by country, each with differences in the natural history of HIV disease progression as well as incidence of tuberculosis and other opportunistic diseases. In Tables Tables2a2a and and2b,2b, we provide an overview of some of the differences in methods, assumptions, costs, and time horizons utilized across these studies. What follows is a brief examination of each study. We included only strategies relevant to the question of laboratory monitoring for the purpose of ART initiation and switching. A composite summary of results from all studies is provided in Table 3, where we have repeated the calculations in each study using conventional cost-effectiveness methodology, including elimination of dominated strategies that may have been reported in the individual studies. A more detailed version of these calculations is provided in Table TA1 of the Technical Appendix. To allow for appropriate comparison across studies, we have expressed all cost-effectiveness results in per-person costs and benefits and have updated all ratios to 2007 US dollars.
In this Côte d’Ivoire analysis, Goldie et al. examine multiple strategies: no ART, ART with clinical monitoring alone where alternative numbers of opportunistic diseases are used for ART initiation (and discontinuation), as well as ART initiation and discontinuation guided by CD4 monitoring. This was one of the earliest analyses of laboratory monitoring; HIV RNA was scarcely used in resource-limited settings at the time and is not examined in this analysis. Only a single antiretroviral therapy regimen is assumed to be available (51% virologic suppression at 52 weeks).
Compared to no ART, ART initiation with development of two opportunistic diseases and discontinuation upon one results in a cost-effectiveness ratio of $660/YLS, very cost-effective by WHO standards for Côte d’Ivoire (Tables (Tables11 and and3).3). The description of the wide range of clinical strategies examined in this analysis is excluded here for simplicity. When CD4 monitoring is available and ART is initiated at CD4<200/μl (or <350/μl with a severe opportunistic disease), there is a resultant increase in life expectancy of more than 2 years, from 41.37 months to 69.63 months. Compared to the clinical monitoring described above, the incremental cost-effectiveness ratio for the CD4-based strategy is $1,140/YLS. Thus, CD4 monitoring would be considered cost-effective for Côte d’Ivoire.
Using data from Kenya, Zimbabwe, and Botswana, the authors examine four strategies over a 10-year time horizon: no ART, ART with no laboratory monitoring, ART with CD4, and ART with CD4 and HIV RNA monitoring. All strategies are evaluated under scenarios of either one or two lines of ART. For comparison with other studies, we have eliminated discussion of the total lymphocyte strategy, which is no longer recommended for routine monitoring by the WHO .
After elimination of dominated and total lymphocyte count strategies, the results reveal that CD4 monitoring has a cost-effectiveness ratio ranging from $640/QALY when only first-line ART is available, to $5,960/QALY when a second-line is also available (Table 3). The latter ratio is considered very cost-effective for Botswana but exceeds the one-times-the-GDP threshold for both Kenya and Zimbabwe (Table 1). An important assumption in this paper is that when CD4 counts are utilized, clinical criteria alone are insufficient to start or switch ART (Table 2a). The authors note, “The costs of the CD4 cell count tests are offset significantly by eliminating costly drug treatment for patients who meet criteria on clinical grounds, but whose CD4 cell counts remain adequate.” This modeling assumption conflicts with current WHO recommendations, which are to initiate ART with the development of any stage 3 or 4 opportunistic disease regardless of CD4 count . Compared to CD4 monitoring, viral load monitoring was less economically attractive, with incremental cost-effectiveness ratios ranging from $16,860/QALY (first-line only) to $15,250/QALY (second-line available) compared to CD4 monitoring alone.
In this study from South Africa, the authors examine the cost-effectiveness of a developed-world approach to ART, including laboratory monitoring, in a resource-limited setting. Two strategies, with two lines of ART available, are examined. The first is ART initiation and monitoring according to “WHO criteria,” where CD4 monitoring occurs every 6 months, ART is initiated at a threshold of CD4<200/μl, and ART is switched based on CD4 or clinical criteria. The second is a “developed world strategy,” where CD4 and HIV RNA monitoring occur every 3 months; ART is initiated at a threshold of CD4<350/μl or HIV RNA>100,000 copies/ml; and ART is switched for CD4 decrease (to <200/μl or to 50% of its peak on-treatment value), HIV RNA increase to >400 copies/ml, or clinical criteria.
When applied to South Africa, the developed world strategy increases both costs and life expectancy compared to the WHO strategy. The incremental cost-effectiveness ratio of the developed world strategy, with quarterly CD4 and HIV RNA monitoring, is $5,780/LYS, very cost-effective by South African criteria (Tables (Tables11 and and33).
In this analysis, which uses data from several lower income countries, the authors consider strategies for switching and stopping ART over a 20-year horizon. The benefits of laboratory monitoring for ART initiation are not addressed; all patients in this analysis are assumed to be ART-eligible (Table 2a). Switching strategies examined include clinical observation alone, CD4 monitoring, or CD4 and HIV RNA monitoring. The results – amended in Table 3 to show only the undominated strategies – suggest that CD4 monitoring without HIV RNA testing is incrementally less cost-effective than the more expensive strategy of switching with a new stage 3 or 4 event. These results define CD4 monitoring in ART switching decisions as a “weakly dominated” strategy. HIV RNA monitoring, compared to switching based on a new stage 3 or 4 event (with the CD4-dominated strategy eliminated), has a cost-effectiveness ratio of $3,610/QALY. This incremental cost-effectiveness ratio is higher than that originally reported by Phillips et al. (Technical appendix, Table TA1) but is below the one-times-the-GDP very cost-effective threshold for some countries in sub-Saharan Africa (Table 1).
This analysis, set in southern Africa, examines alternative strategies for starting, switching, and stopping ART according to symptom-based monitoring, CD4 monitoring alone, and CD4 with HIV RNA monitoring. Two lines of ART are available, and the authors consider alternative thresholds for ART initiation, based on first opportunistic disease and/or a CD4 count threshold <200/μl or <350/μl. ART switching criteria are based on second or third opportunistic disease, 50% CD4 count decline, and/or an increase in HIV RNA, when such monitoring is available.
Among all of the strategies considered, the authors find the following four undominated strategies and their associated incremental cost-effectiveness ratios: 1) biannual CD4 monitoring with ART initiation at <200/μL (comparator, no ratio calculated), 2) biannual CD4 monitoring with ART initiation at <350/μl ($90/YLS), 3) biannual CD4 and HIV RNA monitoring with ART initiation at <350/μl ($4,140/YLS), and 4) quarterly CD4 and HIV RNA monitoring with ART initiation at <350/μl ($124,000/YLS) (Table 3). The authors highlight that, for any given ART initiation strategy, symptom-based monitoring approaches are more expensive but less effective than CD4-based strategies. This finding that CD4 monitoring alone is cost-saving compared to symptom-based approaches may be due to the previously described averted high cost of opportunistic diseases when ART is initiated based on CD4 cell counts.
While the discussion above describes the results from each of the individual studies, it is important to recognize that the incremental cost-effectiveness ratios in each of these studies depend not only on the costs of the laboratory monitoring interventions but also on the costs of ART and clinical care. Additionally, such costs vary depending not only on the year and the country in which the analysis takes place but also on how the cost is estimated. Table 2b summarizes the CD4 and HIV RNA costs per test in each of these studies; they range from $5-$31 per CD4 test and $26-$92 per HIV RNA test. Annual costs for ART regimens vary even more widely, ranging from $130-$429 for first-line and from $640-$1,432 for second-line regimens. Laboratory monitoring in each of these studies generally results in an expedited and timelier switch to a more expensive second-line ART regimen, on which patients often remain for many years. Thus, ART regimen costs, rather than the laboratory test costs themselves, are the primary determinant of the total costs in these analyses. As price negotiations render 2nd-line ART regimens less expensive worldwide , laboratory monitoring strategies may become more cost-effective. Sensitivity analyses may be helpful to allow one country’s policy makers to apply cost-effectiveness results of monitoring strategies conducted in another country. Generally, most studies report that results were sensitive to both ART and laboratory test costs.
This review has several limitations. First, we were bound by the strategies selected and information provided in the reviewed analyses. The decision regarding which monitoring strategies to examine is not uniform across studies. Furthermore, not all studies reported how their costs were derived, and, as a result, costs for this review could not be normalized across studies occurring in different settings and times. Additionally, the high cost of initial investment in the implementation of laboratory monitoring, including the costs of modifying existing infrastructure and of purchasing and maintaining laboratory information management systems, warrants careful consideration and is generally not commented on in these studies.
Finally, the recently reported DART study is among the most widely cited studies on laboratory monitoring in resource-limited settings . While results from DART suggest that quarterly CD4 monitoring bundled with other laboratory tests (hematology and biochemistry panels) provided only modest survival benefits compared to clinical monitoring alone, it did not address CD4 monitoring alone, CD4 monitoring at longer intervals, or CD4 monitoring for the purpose of ART initiation . Because cost-effectiveness results from DART remain unpublished, they were not available for this review. However, preliminary cost-effectiveness analyses from the DART study suggest that CD4 monitoring alone (in the absence of other laboratory monitoring) may be cost-effective in some settings (cost-effectiveness ratio of $2,146/QALY (2008 USD)) [26, 27].
Increasingly the WHO and other health-governing agencies are relying on cost-effectiveness analyses among their guiding principles . Current published studies on the cost-effectiveness of CD4 count and HIV RNA laboratory monitoring differ in design, setting, test cost, and specific strategies compared. While it may be desirable for cost-effectiveness analyses to be individualized to specific settings, relying on this approach may not be practical; some results may be generalizable across countries. Future cost-effectiveness analyses would be more comparable and generalizable if they clearly state the time horizon and year of currency, include critical components of test costs (personnel training, lab infrastructure, specimen transport, and quality assurance programs), and analyze strategies that are most reflective of current in-country clinical practice. We find that many, though not all studies, suggest that CD4 monitoring is cost-effective – and maybe cost-saving – in at least some resource-limited settings. The cost-effectiveness of HIV RNA monitoring, however, ranges widely. The lowest published values ($3,610-$4,140/YLS) suggest that biannual HIV RNA monitoring may be considered cost-effective, but generally in resource-limited countries with the highest per capita GDPs. Further studies are needed to evaluate newer algorithms of targeted HIV RNA monitoring upon meeting other clinical and/or immunologic criteria .
Model-based cost-effectiveness analysis is a critical tool in understanding the value of CD4- and HIV RNA-based laboratory monitoring strategies in resource-limited settings. The results of such cost-effectiveness analyses should be interpreted within the context of the strategies considered, the input parameters included, and the country examined. Published studies to date suggest that CD4 monitoring is likely cost-effective in many settings and that viral load monitoring, at current ART and laboratory costs, may be cost-effective, depending on available resources.
Financial support. National Institute of Allergy and Infectious Diseases (R01 AI058736, K24 AI062476, P30 AI060354, K01 AI078754), the Doris Duke Charitable Foundation (Clinical Scientist Development Award), and the Elizabeth Glaser Pediatric AIDS Foundation.
|Analysis and Strategy||Per person|
or QALE b
|Δ LE compared|
to next less
|Δ cost compared|
to next less
|ICER reported for|
each strategy in
($/YLS or $/QALY)
|Recalculated ICER j|
($/YLS or $/QALY)
|Goldie et al., N Eng J Med 2006 |
|Strategy (2002 USD)|
|ART (no CD4)||41.40||0.83||1,230||450||590||540|
|Start ART: 2 ODs|
|Stop ART: 1OD|
|Start ART: CD4<200/μl (or CD4<350/μl|
& 1 severe OD)
|Stop ART: 90% drop in CD4|
|Bishai et al., AIDS 2007 |
|One line of ART available|
|First line ART only d||54.93||2.31||1,580||1,450||630||(dominated)|
|Two lines of ART available|
|Two lines of ART only||54.81||2.30||1,700||1,570||680||680|
|Vijayaraghavan et al., JAIDS 2007 |
|Strategy (2005 USD) f|
|CD4 every 6 mo, ART initiation at CD4 <200/μl g||116.40||--||10,620||--||--||--|
|CD4/HIV RNA every 3 mo, ART at CD4<350/μl h||138.36||1.83||20,350||9,730||5,310||5,320|
|Phillips et al., Lancet 2008 [9, 30]|
|Strategy c i|
|New WHO stage 4 event||106.80||--||2,067||--||--||--|
|Multiple WHO stage 3 events/new stage 4 event||109.80||0.25||2,310||240||930||960 e|
|CD4 decline from peak||110.40||0.05||2,790||480||9680||(dominated)|
|New stage 3/4 event||112.44||0.17||2,870||80||470||2,550 e|
|CD4/HIV RNA (switch for HIV RNA>10,000/ml)||115.44||0.25||3,950||1,080||4,011||(dominated)|
|CD4/HIV RNA (switch for HIV RNA>500/ml)||116.40||0.08||4,060||1,110||1,500||3,610 e|
|Bendavid et al., Arch Intern Med 2008 |
|Strategy (2007 USD)|
|CD4 every 6 mo, ART at CD4<200/μl||68.41||--||3,610||--||--||--|
|CD4 every 6 mo, ART at CD4<350/μl||73.73||0.44||3,650||40||110||90 e|
|CD4 every 3 mo, ART at CD4<350/μl||74.04||0.03||3,970||320||(dominated)||(dominated)|
|CD4 and HIV RNA every 6 mo, ART initiation at|
|CD4/HIV RNA every 3 mo, ART at CD4<350/μl||75.87||0.01||5,790||1,240||101,250||124,000 e|
ART: antiretroviral therapy, ICER: incremental cost-effectiveness ratio; LE: life expectancy; mo: months; QALE: quality-adjusted life expectancy; QALY: quality-adjusted life-year;yrs: years; YLS: years of life saved
Potential conflicts of interest. All authors: no conflicts.