In this section we review economics terms and ideas that are central to understanding the issue of cost-effectiveness of ART. Table provides a road map of the key types of cost analysis in health. These include descriptive methods (e.g. cost estimation), as well as analytic techniques designed to produce a measure of the efficiency of health interventions and programmes (e.g. cost-effectiveness).
Key Types of Cost Analysis in Health
The most basic economic measure is the cost of health care interventions or services. The cost represents the mix and number of input resources required to deliver these services, with attached costs. Specifically, intervention cost is the sum of the product of resources required to implement the intervention and their unit costs. The resources typically include personnel, supplies (consumables, e.g. medications and condoms), equipment, services (such as advertising or electricity), training, and facility space (e.g. rent). Each resource has a unit cost, such as the hourly wage for a nurse or the cost of a single test kit. When all resources are tallied and unit costs assigned, the sum is the cost. Most cost analyses evaluate “economic” costs, which often differ from financial flows. Economic costs represent the true value to society of those resources, regardless of what the programme actually paid. Thus, donated resources (e.g. volunteer time, test kits provided gratis from the government) would be valued at fair market value. The goal is to quantify true costs - the value of resources consumed - not the monetary transactions which depend on the idiosyncrasies of how organisations obtain resources from collaborating or funding agencies. “Financial” costs are useful for understanding short-term budgetary implications and are sometimes also reported. They represent what the implementing agency paid, regardless of true societal value.
Net cost reflects the cost of delivering services (as above), adjusted for offsetting savings due to disease averted. For example, starting ART is likely to decrease opportunistic infections and other medical problems, thus averting some future health care costs. Similarly, HIV infections averted by prevention reduce the net costs of that prevention programme by obviating the health care requirements of those who would otherwise have become ill. (Diagnostic tests may induce health care costs, which would add to net costs.) Offsetting savings can be greater than programme costs, in which case there are net savings (and no need to calculate a cost-effectiveness ratio; see below).
Cost-effectiveness analysis (CEA) is a technique that has been widely used in health care for several decades. A CEA compares the net cost with the units of health benefit gained, expressed in ratio form, such as the net cost per death averted. This ratio is called the “Incremental Cost-effectiveness Ratio” (ICER), because both the costs and health benefits are incremental - i.e. the added cost and added health benefits versus a less expensive and less effective intervention approach (or no intervention).
There is a specific type of CEA that has in recent years been widely accepted as the standard approach for health. The net cost component of the ratio, in the numerator, is the same as above. However, the health benefits in the denominator are translated into a standardised single metric of change in health, combining into one number both mortality and morbidity effects. In global health, this metric is “Disability Adjust Life Years” (DALYs) (Table ). DALYs are a measure of disease burden: the LY represents premature mortality, and the DA represents disability due to morbidity. Thus, if an individual loses two years of life due to illness, and also has a 20% disability compromise while alive, for five years, the DALY burden of that disease would be 2 + 0.2 * 5 = 3.0 DALYs. (If occurring over multiple years, DALYs are discounted to reflect time preferences, but that is beyond the scope of this review.) Thus interventions avert DALYs, and the ICER is the net cost per DALY averted.
Explanation of Disability-Adjusted Life Years (DALYs) and Quality-Adjusted Life Years (QALYs)
“Quality-Adjusted Life Years” (QALYs) were developed before DALYs (Table ). They are still used for CEAs in the US and Europe. The QALY is a measure of health - essentially the negative of the DALY. Thus, an illness which shortens life by 2 years and lowers “health status utility” by 20% for 5 years would decrease QALYs by 3. Interventions are designed to increase QALYs, and the ICER is the net cost per QALY gained.
According to the World Health Organization, the attractiveness of the ICER can be determined by comparison with the country’s annual gross domestic product (GDP) per capita
. An intervention with an ICER below the annual GDP per capita
is considered “very cost-effective”. An ICER below three times the annual GDP per capita
is considered “cost-effective” [10
For any cost-effectiveness analysis, if there are net savings, the ICER is not used. This is because the ICER - intended to show the cost of health gains - has no intuitive meaning when costs drop as health rises. The ICER has no useful meaning when programmes result in net savings because there is no trade-off between costs and benefits. Instead, the convention is to report that the intervention is “dominant” - both cheaper and better, and to report the absolute magnitude of savings and health gains, with no ratio.
Some economists expand the cost component (the numerator) of CEA to reflect costs beyond medical care. They include “willingness to pay” for the health benefits, as an additional savings. This extends CEA into a realm traditionally left to cost-benefit analysis (see below), and is controversial because others consider it to duplicate the economic productivity element of the denominator (DALYs or QALYs).
Resource allocation is the assignment of available funds (e.g. a budget) to a mix of interventions to maximise a specified health goal (e.g., HIV infections or DALYs averted), subject to certain constraints, typically equity and political concerns. Thus, for example, HIV prevention funding might be allocated in part based on cost-effectiveness, and in part based on equity across groups and attention to different risk behaviours. The technically best solution (i.e. maximising health value for money) can be calculated based on a list (or “league table”) of ICERs, combined with information about how much each intervention can be scaled up (reflecting the availability of demand for services and of needed input resources, such as health care workers).
The league table is a listing of intervention options, from best to least performing (i.e. like a sports league list of team standings). It typically includes the name of the intervention, the programme cost for a specified number of individuals (e.g. 1000), the net cost, the health gain (in DALYs or QALYs), and the cost-effectiveness (CE) ratio compared with current standard of practice. The comparison may also be with a less intensive intervention (also in the table) addressing the same health problem, e.g. comparing an HIV prevention strategy for the entire adult population versus only those at high risk of HIV acquisition. Money would be allocated to each successively less cost-effective intervention until the budget was fully committed.
However, equity imperatives mean that in practice, budgets are allocated partly according to cost-effectiveness criteria, and partly to ensure that all important population groups (ethnicity, sex, geographic location, risk behaviours) are included. Political dynamics may also complicate the way funds are allocated and can produce funding outcomes that deviate substantially from health maximisation.
Equity and other non-economic factors can be incorporated into resource allocation in various ways. One is to attach an explicit cost to failure on non-economic standards, e.g. inequity, in the CE ratio. A strategy of reaching 1000 individuals in a particular group, to the exclusion of 1000 individuals in other population groups, may be deemed to “cost” society a certain amount - the value placed on equity. This cost is reflected in the numerator - the intervention net cost is increased accordingly. Alternatively, and more commonly, equity is addressed outside the league table. Interventions are characterised as less or more equitable. Choosing a more equitable strategy that is not as cost-effective is justified on non-economic criteria. Finally, this choice of equity (or any non-economic criterion) over efficiency can often be characterised in terms of its “shadow cost”. Shadow costs are the health or economic gains foregone by a decision based on other criteria. For example, if allocating $10 million in program funds, funding in part due to equity considerations may mean that 10 (or 50, or 100) HIV infections are not averted, as compared with the most efficient set of interventions. Or, $1 million (or $5 million, etc.) of economic savings are foregone.
Cost-benefit analysis (CBA) is a much less common type of analysis than CEA in health care. In CBA, the benefits that derive from improved health are monetised, rather than being expressed as DALYs or other health metrics. The CBA result is expressed, usually, as benefits minus costs. Thus, for example, the value of longer life and lower morbidity due to ART might be expressed as the sum of medical costs averted and economic productivity added, or in terms of the “willingness to pay” for ART of all affected individuals. This would be compared with the cost of delivering ART.
In CBA, all health outcomes including human life are assigned a dollar value. If benefits exceed costs, an activity is considered economically efficient. The power of CBA is that it permits comparisons of widely disparate funding alternatives. The value of spending on a new airport could be compared with spending on a maternal health initiative. However, the technical and philosophical underpinnings of the monetary valuation of health outcomes are controversial (even among some economists). The practical requirements of mustering the necessary data can also be onerous, especially for small agencies. For these reasons, CBA is rarely used in the economic assessment of health programmes.
CBA results are sometimes expressed as ratio of benefits to costs. This has the virtue of being unitless - the ratio is the same whether the intervention is 10% or 90% scaled up (assuming no economies or diseconomies of scale). In contrast, an arithmetic difference depends crucially on the scale selected. The disadvantage of a ratio is that it may be very sensitive to the placement of cost effects on the cost or benefit side. For example, with ART, are the reduced medical costs associated with decreased infections an adjustment to the cost, or a benefit? This somewhat arbitrary decision affects the final ratio.
CBA can answer the question, is it economically worthwhile to undertake the intervention? The advantage of CBA is that it can include a wider range of benefits not explicitly evident in an ICER, such as increased economic productivity (though the disability in DALYs and the utility in QALYs implicitly incorporate these). There are no standard rules regarding the kinds of benefits due to improved health to be counted (and monetised). CBA is much less common in health economics, due to its heterogeneous scope, lack of standardisation, de-emphasis of health outcomes, and, perhaps related, origin in welfare economics (whereas CEA grew from medical decision analysis).