While QALYs have become the dominant outcome measure for cost-effectiveness analyses in health and medicine, there has been steady debate in the literature about the advantages and disadvantages of QALYs, touching on an array of issues – ethical, conceptual, and practical. A good list of references for these debates through the mid-1990s may be found in the volume by Gold and colleagues (Gold et al., 1996). For a flavor of more recent contributions see Williams (1996), Nord (1999), Tsuchiya (2000), and Bleichrodt and Pinto (2005).
One of the most widely discussed alternatives to QALYs is the healthy-years equivalent, or HYE, first proposed by Mehrez and Gafni (1989). Unlike the QALY, the HYE assigns utilities to an entire sequence of health states, rather than to each individual component of this sequence. This alternative approach relaxes some of the structure imposed by the assumptions built into QALYs – discussed in the section titled ‘Key concepts’ above – for example, the assumption that health states are additively independent across time, or the requirement of constant proportional tradeoffs. Consider two simple examples. First, imagine a chronic condition that may have varying durations. For QALYs, the valuation of a health state is considered to be independent of duration, so that knowing the value attached to 1 year lived in the state enables computation of QALYs for periods of 2 years, or 5 years, or 25 years, through simple multiplication. The HYE, on the other hand, allows that the valuation of a health state may depend on its duration, and therefore calls for direct valuation of 1-year, 2-year, 5-year, and 25-year periods in the health state, which need not relate linearly to each other.
As a second example, imagine an individual’s health experience over time as transitions through a sequence of health states. The assumption in QALYs is that the total utility attached to the sequence is the sum of the utilities of each individual component (characterized by a constant health level over some duration). Thus, to compute QALYs for the entire stream requires only the estimation of the duration of each component and the value attached to the health state that is experienced during each component. For the HYE, on the other hand, the value attached to the entire sequence need not be the simple sum of the value of the individual components. This more flexible formulation allows for the possibility that the value attached to time spent in a health state may be different depending on the particular experiences that preceded the state. The disadvantage of this more flexible approach is that valuation must be undertaken on all relevant sequences of health states, which may be numerous.
Another type of summary measure that has been regarded as an alternative to QALYs is the disability adjusted life year, or DALY. The DALY was first developed for the primary purpose of quantifying the global burden of disease. In this context, it was constructed as a summary measure of population health, specifically, to be used as an indicator of the relative magnitude of losses of healthy life associated with different disease and injury causes. The construction of summary measures of population health has much in common with the construction of measures of the benefits from health interventions. Indeed, the developers of the DALY explicitly intended that the measure could be used as both a unit of account for the burden of disease, and as a metric for health benefits in the denominator of cost-effectiveness ratios (Murray et al., 1994). The major debut of the DALY in the World Bank’s World Development Report 1993 introduced applications of the measure toward both ends (World Bank, 1993). More recently, guidelines from the World Health Organization on conducting ‘generalized cost-effectiveness analyses’ – with a particular focus on health policies in developing countries – have included an explicit recommendation to use DALYs as the measure of benefit in these analyses (Tan Torres et al., 2003).
The relationship between DALYs and QALYs has been characterized as follows by developers of the DALY: ‘‘DALYs can be considered as a variant of QALYs which have been standardized for comparative use’’ (Murray and Acharya, 1997: 704–705). There are certain key distinctions between DALYs and QALYs:
- Because DALYs are negative measures that reflect health losses, the scale used to quantify nonfatal health outcomes in DALYs is inverted compared to the scale used in QALYs; that is, numbers near 0 represent relatively good health levels (or small losses) in DALYs, while numbers near 1 represent relatively poor health levels (or large losses). The inverted scale means that interventions that improve health result in DALYs averted, whereas QALYs are gained.
- Disability weights in DALYs, which are the health state valuations analogous to the ‘quality’ adjustments in QALYs, are intended to reflect the degree to which health is reduced by the presence of different conditions, whereas at least one interpretation of the weights in QALYs is based on the individual utility derived from different states.
- The standard formulation of DALYs weights healthy life lived at different ages according to a variable function that peaks at young adult ages, while QALYs do not typically incorporate unequal age weights. (It should be noted that DALYs may also be computed with equal age weights.)
- For measuring the burden of disease, years of life lost due to premature mortality at different ages are computed in reference to a standard life table, defined by the period life expectancy at birth for Japanese females at the time of the original Global Burden of Disease study. For purposes of cost-effectiveness, this distinction is largely inconsequential, since the standard life expectancy essentially cancels out when benefits of interventions are computed as the change in DALYs. As a simplified example, imagine an intervention that defers one death from age 50 to age 70, and suppose that the normative target lifespan used as the yardstick for DALYs is 80 years (irrespective of one’s current age). Then the number of DALYs averted through intervention is a change from 80-50 = 30, to 80-70 = 10, for a net of 20 DALYs averted, which is the same as the number of QALYs gained through the intervention. (Note that in the actual standard life table that is used, as in most life tables, the target life span, equal to the number of years of remaining life expectancy at age x plus x, rises slightly with advancing adult ages rather than remaining constant as per the simple example here. This will produce a slight discrepancy between DALYs averted and QALYs gained, but this difference is usually negligible.)