Overview of the measure
In developing its framework for health systems performance assessment in the 1990s, the WHO established a common set of measures for evaluating healthcare services in systems (World Health Organization, 2000; Murray and Evans, 2003). The primary motivation of the work was to provide analysts with standardized tools to assess the performance of healthcare systems across diseases and countries. In the WHO framework, maintaining and improving population health was defined by the WHO as the ultimate goal of healthcare services. DALYs is proposed as the operational indicator of the extent to which the health goal is being achieved. It defines population health in negative terms taking into account both premature death caused by disease and the loss of healthy life due to disability (Murray, 1996; Murray et al., 2002). As a broad indicator of the adequacy of healthcare services, DALYs reflect a number of factors including disease incidence, disability and mortality effects, and the availability, accessibility, and effectiveness of treatments.
DALYs are the sum of years of life lost to premature death (YLL) and years lived with a disabling condition (YLD) among those who survive. YLL is calculated as the number of deaths attributable to the condition, multiplied by the years that would have been lived in the absence of a premature death. Similarly, YLD is calculated as the years of life lived with the condition, adjusted for the perceived severity of the condition. For a given calendar year and condition, YLL and YLD are calculated for age- and gender-specific cohorts. YLL and YLD are then summed across cohorts to yield global DALYs per condition.
YLL and YLD that occur in the relatively young and relatively old are discounted using an age-weight function that reflects the broader societal impact of losses of life and health in the more productive middle years. YLL and YLD that occur in the future are discounted to reflect the greater social value of preventing near-term death and disability. As discounting reflects nonuniversal value judgments, results may be presented with and without age-weighting and discounting.
Evaluation methods and results
The Commission's examination of the DALY was guided by Dr. Chisholm who attended the Commission's meetings and conferences and led us in obtaining and translating literature on this WHO measure. With Dr. Chisholm's assistance, we performed an exercise to test the applicability of DALYs as an indicator of epilepsy healthcare in five developed countries (Canada, France, Italy, United Kingdom, and the United States) and one developing country (South Africa) using basic epidemiological and health care data. The goals of the exercise were to: (1) test the feasibility of calculating country-specific DALYs, using published epidemiologic data; (2) identify gaps in the data required for estimating DALYs for specific regions, and (3) identify aspects of the DALY method that require qualification or further refinement to ensure that it validly captures the burden of epilepsy. We did not set out to provide definitive burden estimates for each country, which would have required more systematic and precise data estimates than resources permitted.
Calculations of epilepsy-related DALYs were made using a modified version of the WHO spreadsheet (available from http://www.who.int/entity/healthinfo/bodreference dalycalculationtemplate.xls). We identified region-, condition-, age-, and gender-specific data on population, incidence, prevalence, duration, mortality, and disability via literature review. We chose the most representative data, based on consultation with epidemiological experts and used the standard WHO age-weights and a 3% discount rate to facilitate comparisons. Of necessity, we included data only from studies whose age strata overlapped the strata in the WHO worksheet.
Country-specific parameter estimates are shown in Table 1. Country-specific data were not available in all cases, so estimates based on regional studies or studies of other countries were applied where populations, economic conditions, or healthcare systems were considered sufficiently similar. Epidemiologic estimates for epilepsy are not available for South Africa, for example, so parameters were selected from studies of other sub-Saharan countries. In the absence of country-specific data, standardized mortality ratios (1.1–4.0, depending on age) obtained through literature review were applied to all countries.
Table 1. Epidemiologic parameters for calculating country-specific DALYs
|Country||Population death rate (per 1,000)||Incidence (per 1,000)||Prevalence (per 1,000)||Treated (%)||Controlled (%)|
In the WHO calculation of DALYs, different disability weights due to epilepsy depend only on treatment status. We modified the WHO worksheet to reflect both the proportion of patients in each age and gender stratum who were treated versus the untreated and the proportion with controlled versus uncontrolled seizures. Remission rates were estimated to be 0.05–0.09 per year, depending on age, based on the aforementioned review. We used disability weights for treated and untreated cases from the original WHO study as well as additional weights derived from our own literature review to illustrate the importance of accurately measuring disability associated with epilepsy.
Epilepsy-related DALYs for the five developed countries are shown in Fig. 1, expressed as DALYs per 1,000 population. For comparison, dashed lines show DALYs per 1,000 for different conditions in the Established Market Economy region, as estimated in the 1996 WHO Global Burden of Disease study (Murray et al., 2002). The analysis indicates the sensitivity of the burden estimates to the epidemiologic parameters chosen. For example, the original WHO estimates that in the Established Market Economy countries the burden of epilepsy (0.51/1,000) was between that of multiple sclerosis (0.26/1,000) and Parkinson's disease (0.62/1,000)(not shown in Fig. 1) whereas the parameters for France used in our exercise equate epilepsy's burden with that of diabetes. The variability between the five countries is mainly due to differences in the estimated incidence (higher in France), treatment gap (higher in the United States), and population death rates (higher in the United States, United Kingdom, and France)(Table 1).
Key methodological differences account for our higher estimates of burden, compared to the 1996 WHO study. Disability weights are particularly influential, because much of the epilepsy burden is accounted for by YLDs. Disability weights from the literature were elicited directly from persons with epilepsy, using validated preference-based methods. The range for persons with uncontrolled seizures (0.29–0.44) (Messori et al., 1998; Wiebes et al., 2002; Langfitt et al., 2006), and with relatively well-controlled seizures (0.12–0.19) (Stavem et al., 2001), considerably exceeds original WHO disability weights (0.15 for untreated cases, 0.065 for treated cases,) that were derived from a person-trade-off protocol administered to panels of disease experts (Murray, 1996).
Epidemiologic estimates used in the exercise included idiopathic, cryptogenic, and symptomatic epilepsies. The WHO excluded symptomatic cases, resulting in considerably lower estimates of incidence, prevalence, mortality, and percent uncontrolled. The WHO assigns one primary condition per case in order to avoid overestimating global burden of disease by double-counting cases with multiple conditions. Under this approach, the burden of symptomatic epilepsies should theoretically be incorporated into the burden estimates for the primary conditions (e.g., stroke, traumatic brain injury, etc.). While the WHO explicitly accounts for epilepsy as a consequence of meningitis, it does not do so for those conditions that account for a large proportion of symptomatic epilepsies. The WHO's exclusion of symptomatic cases likely leads to an underestimate of total epilepsy burden.
Due to a different scale of burden, results for South Africa are shown separately in Fig. 2, along with WHO-estimated DALYs/1,000 for different conditions in the sub-Saharan African region. Uncertainty about the incidence, prevalence, and the appropriate disability weights produces even wider variation in burden estimates for South Africa. Epilepsy burden estimates using the lower and higher end incidence and prevalence estimates for sub-Saharan Africa (Preux and Druet-Cabanac, 2005) are compared to other conditions. Depending on the estimates used, the burden of epilepsy may be comparable to that of Alzheimer's disease (0.66/1,000) or greater than that of war.