The European Journal of Public Health Advance Access originally published online on November 23, 2007
The European Journal of Public Health 2008 18(2):195-200; doi:10.1093/eurpub/ckm109
Health Monitoring |
Sensitivity analysis in summary measure of population health in France
Annabelle Lapostolle1, Agnès Lefranc2,3, Isabelle Gremy2 and Alfred Spira1
1 Inserm U822/Service de Santé Publique, France
2 Observatoire Régional de Santé Ile-de-France, 21-23 Rue Miollis, 75015 Paris, France
3 Institut de Veille Sanitaire – DES, 12, Rue du Val d’Osne, 94415 Saint-Maurice cedex, France
Correspondence: Annabelle Lapostolle, Hôpital de Kremlin Bicêtre, 82 rue du Général Leclerc, 94276 Le Kremlin Bicêtre cedex, France, tel: +33 1 45 21 23 39, fax: +33 1 45 21 20 75, e-mail: lapostol{at}vjf.inserm.fr
Received December 4, 2006, accepted October 25, 2007
 |
Abstract |
|---|
 Top Abstract MethodsResultsDiscussionAcknowledgementsReferences |
|---|
Background: The aim of this study is to provide estimates of the French burden of disease, using the WHO Global Burden of Disease methodology and to perform sensitivity analysis on different set of mortality data. Methods: The burden of disease is measured by disability-adjusted life years (DALYs) that take into account both mortality and morbidity data. Results were obtained using French mortality data for the years 2000 and 2001 and morbidity data estimated by WHO for France. Sensitivity analyses were conducted using different mortality data sets and various life tables as mortality norms. Calculations were also performed with and without discounting and age-weighting. Results: In France, the annual burden of disease was about 12.4 million DALYs. Depending on the mortality data set and the choice of social values used for calculation, results could be quite different. The use of WHO estimates for mortality resulted in an underestimation of 2.6% of total DALYs with respect to French data. Changes of the mortality norm imply changes in the number of years of life lost (YLLs), whereas the use of discounting and age-weighting mainly modifies the ranking of diseases. Conclusion: DALYs constitute a summary measure of population health, which is a powerful tool for the grading of health problems, allowing to compare fatal and non-fatal diseases. Nevertheless, the validity of results obtained depends primarily on the validity of the input data. Collecting morbidity data (mainly incidence) at the national level is hence an important step in order to assess more accurately the specific burden of diseases in France.
Keywords: burden of disease, disabled person, France, mortality
Summary measures of population health combine information about mortality and morbidity, as a means of describing the health of the population in a simple, easily comprehensible way. This type of indicator was developed following the major decrease in mortality in developed countries, as the traditional indicators of mortality previously used to describe the population (life expectancy and infant mortality) were no longer appropriate to describe the health of a population. In the same time, with the development of curative and preventive medicine, public health authorities formulated an increasing need for tools allowing to assess the overall performance of health actions, both in terms of mortality and morbidity.1–6
Disability-adjusted life years (DALYs) were developed in the 1990s by the World Health Organization (WHO) and the World Bank as part of the Global Burden of Disease (GBD) Project.7–9 DALYs were developed to respond to diverse health policy goals: the consideration of non-fatal diseases, the creation of an objective, independent measure for health systems evaluation and the quantification of disease burden with an indicator that can also be used for economic evaluation as age-weighting and/or discounting is included in the calculation.
Such indicators have been the subject of many critical and contradictory analyses.10–18 The main question concerns choosing the most appropriate indicator for the desired usage.1 In particular, these indicators can be used to assist public decision-makers in their choices as the allocation of resources for prevention, health care and research in the domain of health.1,19,20 For this type of use, it must be possible, with the tool chosen, to rank diseases in order of contribution to the burden of disease and injury.
The use of DALY permits such a ranking.7 For this reason, this indicator was used to select health problems that were addressed by specific objectives attached to the public health policy law adopted in 2004 in France. With this goal in mind, DALY must be estimated at regional and national levels, and the validity of these estimates must be evaluated. We present here an assessment of the burden of disease in France, based on the GBD method of WHO. These first results provide a ranking of diseases and injuries, which can be used to identify a list of eligible diseases (about 10) for a more detailed burden of disease analysis in future studies.
|
Methods |
|---|
|
TopAbstractMethodsResultsDiscussionAcknowledgementsReferences |
|---|
We calculated DALY for France using the GBD method described by WHO.8,9,21 This indicator combines information about mortality and morbidity to describe the health of the population. DALYs are therefore calculated by summing up two components for all diseases: years of life lost (YLLs) that takes into account the burden of mortality, and years lived with disability (YLDs) that takes into account the burden of morbidity.
|
|
Deaths from unknown reasons were distributed between categories according to the GBD method recommendations: deaths by disease of unknown aetiology were redistributed proportionally among other diseases according to sex and age; deaths by injuries of unknown cause were redistributed in a similar manner among other injuries.8,23
We also redistributed a subset of ill-defined cardiovascular deaths (ICD-10 codes I47.2, I49.0, I46, I50, I51.4–51.6, I51.9 and I70.9) to ischaemic heart diseases (I20–I25) using the method recommended by WHO in order to solve miscoding problems for these diseases.24
YLLs were calculated by multiplying, for each cause, age group and sex, the number of deaths by standard life expectancy for the corresponding age and sex. The mortality norm initially chosen for comparison was the one used in the GBD analysis which is based on the West Level 26 reference table of Coale and Demeny.7 This norm is characterized by a life expectancy of 82.5 years for women and 80 years for men. For each cause, sex and age:
|
|
YLLs were also calculated from WHO estimates of mortality data for France, obtained by the GBD method: WHO estimated deaths by causes using french vital registration data from 1980 to 1999. Death rates by age and sex for specific causes were calculated and used in order to project rates for the year 2002.25 These rates were then applied to the UN Population Division estimates of population for France for 2002 to provide number of deaths by cause.
The calculation of YLD required data for disease incidence, duration and disability distributed by sex and age, for each sequelae of a disease or injury. The number of YLDs for a given disease is obtained by summing the YLDs attributable to the sequelae of this disease. For each cause, sex and age:
|
|
As such data are difficult to obtain, we chose to use in the present study the incidence, duration and disability weights used for France by WHO in the GBD study.25 WHO first estimated incidence, duration and severity of conditions based on available epidemiological data at a regional level (EurA for France). WHO then applied an incidence/mortality ratio to French mortality to estimate specific incidences at national level. Consistency was checked using Dismod II.26
Age-weighting and discounting were not applied to the calculation of YLL and YLD in the first instance. However, we also calculate DALYs using a discounting rate of 3% and the same age-weighting as in the GBD in order to allow comparisons with other burden of disease studies.
Finally, sensitivity was assessed by changing life table used as norm: the life table for France in 2001 and the projected life table for France in 2020.27 The DALY indicator was calculated as described above, varying only the mortality norm to measure the impact of this norm on the final indicator.28
|
Results |
|---|
|
TopAbstractMethodsResultsDiscussionAcknowledgementsReferences |
|---|
The 10 leading causes of DALYs by sex are presented in table 1. In men, 6 of the 10 leading causes of DALYs correspond mostly to YLLs: trachea, bronchus and lung cancer, ischaemic heart diseases, road traffic accidents, self-inflicted injuries, cerebrovascular disease and cirrhosis of the liver. In women, disability makes a greater contribution to the burden of disease than mortality: only 3 of the 10 leading causes of DALYs make a greater contribution to YLLs than to YLDs—cerebrovascular diseases, ischaemic heart disease and breast cancer.
|
In women, Alzheimer's disease and other forms of dementia head the ranking, accounting for 7.5% of all DALYs. Unipolar depressive disorders were second, accounting for 7.1% of DALYs, whereas these diseases accounted for only 3.4% of deaths (eighth position).
In France, in the 2000–2001 period, there were a mean of 528 503 deaths per year, 20% of which (109 830 deaths) occurred before the age of 65.
Considering all causes of death together, we counted 6.8 million years of life lost, 60% of which were lost by men (4.1 million). The 10 most important causes of YLLs are presented in table 2.
|
The leading causes of YLLs differed according to sex. For men, the leading causes of YLLs were ischaemic heart disease (9.2% of YLLs), followed by trachea, bronchus and lung cancers (9.0%), self-inflicted injuries (6.0%) and road traffic accidents (5.6%). For women, the leading causes of YLLs were breast cancer (7.8%), followed by ischaemic heart disease (6.9%) and cerebrovascular diseases (6.0%). Alzheimer's disease and other types of dementia, which were not included among the first 10 leading causes of YLLs for men, were ranked sixth for women, accounting for 3.2% of all YLLs.
Major differences exist between WHO estimated of mortality for France and actual French mortality data. WHO estimates of mortality in France were 499 040 deaths in 2002 (Global Burden of Disease 2002), corresponding to an underestimation of 5.6% with respect to French data.29 Table 3 shows a classification of diseases according to their contribution to the total number of deaths using both data sources. For example, for cerebrovascular diseases, the difference is 3676 deaths, implying a 9% underestimation for WHO evaluations of the number of deaths actually observed for this disease. Conversely, deaths linked to chronic obstructive pulmonary disease (COPD) were the seventh leading cause of death and accounted for 3.2% of all deaths according to WHO estimates. According to French mortality data, this disease accounted for 1.5% of deaths and was ranked 16th, corresponding to a more than 100% overestimation of deaths due to this disease by WHO. For some cause-of-death categories, the difference between observational and estimated data exceeded 25%: skin diseases, skeletal muscle diseases, endocrine disorders and respiratory infections.
|
Table 4 shows the results obtained if the life table used to define the mortality gap is changed. The use of the standard life table led to an estimation of 6.8 million YLLs and 12.4 million DALYs. Using the French life table for 2001 provided 2.6% more YLLs and 1.4% more DALYs than estimated from the standard table recommended by WHO. This difference depends on the sex: use of the French life table decrease the number of YLLs (–2.8%) and of DALYs (–1.7%) for males, whereas for females YLLs and DALYs are increased (+10.7% of YLLs). The use of the projected life table for 2020 further increased this difference: values were 14.7% higher for YLLs and 8.1% higher for DALYs than estimated with WHO mortality norms. Number of YLLs is increased by using the projected life table for both males and females (+9.2 and +23.1%).
|
Figure 1 presents the impact of the use of discounting and age-weighting on the 10 leading causes of DALYs. We can see that for both males and females, diseases occurring at active ages of life reach higher rank by the use of such weights: unipolar depressive disorders and alcohol use disorders. For males alcohol use disorders and unipolar depressive disorders are at the first ranks with the use of discounting and age-weighting whereas such diseases were ranking 3rd and 8th without such weights. For females, unipolar depressive disorders represents 6.4% of total DALYs without any weights and 13.7% with discounting and age-weighting.
|
On the contrary, diseases occurring at older ages like Alzheimer's disease, ischaemic heart disease, cerebrovascular disease and trachea, bronchus and lung cancers tend to decrease in rank, whatever the sex.
|
Discussion |
|---|
|
TopAbstractMethodsResultsDiscussionAcknowledgementsReferences |
|---|
The use of DALY makes it possible to express mortality and morbidity in the same units (years of life), highlighting diseases, usually not considered in studies of mortality alone. This study of DALY in France in 2000 and 2001 highlighted the importance of non-fatal diseases. For example, in mortality analysis, no death was attributed to diseases of the sense organs, whereas such diseases accounted for about 5% of DALYs. The choice of an indicator for the description of the health of the population is very important and depends heavily on its intended use. Several lines of evidence suggest that the DALY indicator provides a more appropriate measure of the health of the population than the simple use of mortality or premature death statistics.
The quality of DALY estimates also depends on the quality of data used for calculation. The use of WHO estimates to measure morbidity does not seem to be the most appropriate solution for a study of the burden of disease. Concerning mortality, we have shown that there were many differences between WHO estimates data and observed data for France. These differences could be explained by the fact that we used french mortality for 2000 and 2001 and that WHO estimates for 2002 were based on 1999 data that were processed differently. Indeed, French vital registration moved from ICD-9 to ICD-10 between 1999 and 2000 and from manual coding to automatic coding. According to a double coding study carried out by the CepiDC, these changes induced a division by approximately two of the number of deaths for which COPD is the primary cause.30 This has been discussed in a recently published paper.31 However, for most groups of causes used in the GBD and in the present study, changes in the revision of the classification used for death causes coding did not change the deaths statistics drastically.
Indeed, given the observed differences for mortality, it seems likely that, for a certain number of diseases, estimates of morbidity are not very reliable. A certain number of local studies of disease burden have made use of YLD/YLL ratios to estimate the morbidity component, in cases in which local mortality data were available.5,32,33 This method presents the advantage of not requiring local incidence data. However, the use of these ratios can be questioned, because they also depend heavily on WHO estimates.
The ideal solution for obtaining a reliable indicator would therefore be to collect data for disease incidence, duration and disability for each of the health states defined in the global burden of disease. This task has proved to be a major undertaking, and very few teams have carried out complete analyses of the burden of disease at a national level.34,35. In France, the data required to calculate YLD are difficult to obtain for the entire range of diseases. For this reason, we intend to determine incidences, durations and disability weights for only a selection of about 10 diseases and injuries in a subsequent work. This work will provide information about the validity of WHO estimates for the calculation of DALY in a repetitive manner in France, with the aim of following the health of the population over time and of contributing to evaluate the public health law in addition to other indicators.
The DALY indicator is easy to understand and provides an overall picture of health problems. Studies of the burden of disease based on this indicator lead to a ranking of diseases and injuries in their totality, providing a complete description of the health status of a population at a given time. This indicator, which is not limited to mortality, incidence or prevalence data, provides a more complete picture of health as defined by WHO.
While the estimates are based on objective data (mortality, incidence of disability and duration of the disease), the results obtained also depend on the arbitrary choice of parameters used in the calculation of DALY: the mortality norm chosen, the use or not of age-weighting and discounting, and the choice of disability weight. The use of this indicator requires total clarity in the choice of these parameters, to make it clear what is being calculated. As part of the current work, we first choose not to use discounting nor age-weighting. We then calculated DALYs using the same values for these parameters as the ones used by WHO in the GBD study. Comparison of the two sets of results illustrates the fact that diseases occurring at active ages see that their relative importance increase when such parameters are introduced in the calculation. This reflects the consequences of social preferences, that one may consider as non-justified.12,13
The choice of the mortality norm has already been the subject of much debate in the literature.12,28 In this article, we chose to present results obtained using the standard life table of the GBD, which reduces considerably the difference between males’ and females’ life expectancy at birth. In France, this difference was about 7 years in 2001 instead of 2.5 years with the standard life table used in GBD. Use of the GBD standard life table leads to an overestimation of DALYs for men compared with the French mortality norm, increasing by the same way the gender gap of the burden of disease. Each choice has its own advantages: the use of standard life table allows comparisons between countries whereas the life table of the population provides an objective norm.
Comparison of DALY between countries requires the use of the same parameters in the calculation of this indicator. In studies of national burden of disease in developed countries with same methodological choices, a similar hierarchy of main causes of DALYs is obtained. However, if we compare French results with those for Australia, we find that France has a higher frequency of DALYs for Alzheimer's disease, depression, hearing loss, problems associated with alcohol and injuries (road traffic accidents and suicides).34 Conversely, ischaemic heart diseases, cerebrovascular and respiratory diseases (asthma and COPD) were ranked lower in France than in Australia on the basis of DALY. A Dutch study ranked anxiety second and visual problems fourth.35 These two diseases were not found to be particularly important in France. A similar classification was obtained for Los Angeles, with the exception of diseases and injuries inherent to the urban environment: murders, drug use and problems associated with drug use and HIV/AIDS.32
|
Acknowledgements |
|---|
|
TopAbstractMethodsResultsDiscussionAcknowledgementsReferences |
|---|
We would like to acknowledge the French Ministry of Health that funded this project and Alain Fontaine for helpful comments. We would also like to thank the city of Paris for its material help.
Conflicts of interest: None declared.
Key points
|
|
References |
|---|
|
TopAbstractMethodsResultsDiscussionAcknowledgementsReferences |
|---|
1 Van der Maas PJ. How summary measures of population health are affecting health agendas. Bull World Health Organ (2003) 81:314.[Web of Science][Medline]
2 Mathers CD, Murray CJ, Ezzati M, et al. Population health metrics: crucial inputs to the development of evidence for health policy. Popul Health Met (2003) 1.
3 Robine J-M, Jagger C. Allongement de la vie et état de santé de la population. In: Démographie: Analyse et synthèse—Casselli G, Vallin J, Wunsch G, eds. (2004) Paris: Ined. 51–84.
4 Sermet C, Cambois E. Mesurer l'état de santé. In: Démographie: analyse et synthèse III - Les déterminants de la mortalité—Casselli G, Vallin J, Wunsch G, eds. (2002) Paris: Ined. 25–52.
5 Granados D, Lefranc A, Reiters R, et al. Les Années de Vie Ajustées sur l'Incapacité, un outil d'aide à la définition des priorités de santé publique? Rev d'Epidemiol Santé Pub (2005) 53:111–25.
6 Veerman JL, Barendregt JJ, Mackenbach JP. Quantitative health impact assessment: current practice and future directions. J Epidemiol Commun Health (2005) 59:361–70.
7 Murray CJ. Quantifying the burden of disease: the technical basis for disability-adjusted life years. Bull World Health Organ (1994) 72:429–45.[Web of Science][Medline]
8 Murray CJ, Lopez AD. The Global Burden of Disease: a comprehensive assessment of mortality and disability from diseases, injuries, and risk factors in 1990 and projected to 2000. In: Harvard University Press edn (1996) Boston: World Health Organisation, Harvard School of Public Health, World Bank.
9 Murray CJ, Lopez AD, Jamison DT. The global burden of disease in 1990: summary results, sensitivity analysis and future directions. Bull World Health Organ (1994) 72:495–509.[Web of Science][Medline]
10 AbouZahr C, Vaughan JP. Assessing the burden of sexual and reproductive ill-health: questions regarding the use of disability-adjusted life years. Bull World Health Organ (2000) 78:655–66.[Web of Science][Medline]
11 Allotey P, Reidpath D, Kouame A, Cummins R. The DALY, context and the determinants of the severity of disease: an exploratory comparison of paraplegia in Australia and Cameroon. Soc Sci Med (2003) 57:949–58.[CrossRef][Web of Science][Medline]
12 Anand S, Hanson K. Disability-adjusted life years: a critical review. J Health Econ (1997) 16:685–702.[CrossRef][Web of Science][Medline]
13 Arnesen T, Nord E. The value of DALY life: problems with ethics and validity of disability adjusted life years. Br Med J (1999) 319:1423–5.
14 Bastian H. A consumer trip into the world of the DALY calculations: an Alice-in-Wonderland experience. Reprod Health Mat (2000) 8:113–6.[CrossRef]
15 Gold MR, Stevenson D, Fryback DG. HALYs and QALYs and DALYs, OH MY: Similarities and Differences in Summary Measures of Population Health. Annu Rev Public Health (2002) 23:115–34.[CrossRef][Web of Science][Medline]
16 Nygaard E. Is it feasible desirable to measure burdens of disease as a single number ? Reprod Health Mat (2000) 8:117–25.[CrossRef]
17 Reidpath DD, Allotey PA, Kouame A, Cummins RA. Measuring health in a vacuum: examining the disability weight of the DALY. Health Policy Plan (2003) 18:351–6.
18 Williams A. Calculating the global burden of disease: time for a strategic reappraisal? J Health Econ (1999) 8:1–8.[Medline]
19 Mahapatra P. Priority-setting in the health sector and summary measures of population health. In: Summary measures of population health—Murray CJL, Salomon JA, Mathers CD, Lopez AD, eds. (2002) Geneva: World Health Organisation. 53–60.
20 Van der Maas PJ. Applications of summary measures of population health. In: Summary measures of population health—Murray CJL, Salomon JA, Mathers CD, Lopez AD, eds. (2002) Geneva: World Health Organisation. 53–60.
21 Murray CJ, Acharya AK. Understanding DALYs (disability-adjusted life years). J Health Econ (1997) 16:703–30.[CrossRef][Web of Science][Medline]
22 Inserm. accessed on 5 September 2006. Centre d'épidémiologie sur les causes médicales de décès. Available from: http://www.cepidc.vesinet.inserm.fr/.
23 Mathers CD, Vos T, Lopez AD, Salomon JA, M E. National burden of disease studies: a practical guide (2001) 2.0 edn. Geneva: World Health Organisation.
24 Lozano R, Murray CJL, Lopez AD, Satoh T. Miscoding and misclassification of ischaemic heart disease mortality (2001) Geneva: World Health Organization.
25 Mathers C, Loncar D. Updated projections of global mortality and burden of disease, 2002–2030: data sources, methods and results (2005) Geneva: World Health Organization.
26 Barendregt JJ, Van Oortmarssen GJ, Vos T, Murray CJ. A generic model for the assessment of disease epidemiology: the computational basis of DisMod II. Popul Health Met (2003) 1:1.[CrossRef]
27 Vallin J, Meslé F. Tables de mortalité françaises pour les XIXe et XXe siècles et projections pour le XXIe siècle. (2001) Paris: Ined.
28 Bonneux L. How to measure the burden of mortality? J Epidemiol Commun Health (2002) 56:128–31.
29 Lopez AD, Mathers CD, Ezzati M, Jamison DT, Murray CJL, eds. Global burden of disease and risk factors. (2006) Washington (DC): World Bank and Oxford University Press.
30 Pavillon G, Boileau J, Renaud G, et al. Conséquences des changements de codage des causes médicales de décès sur les données nationales de mortalité en France, à partir de l'année 2000. Bull Epidémiol Hebdomad (2005) 4:13–6.
31 Fuhrman C, Jougla E, Nicolau J, et al. Deaths from chronic obstructive pulmonary disease in France, 1979–2002: a multiple cause analysis. Thorax (2006) 61:930–4.
32 Kominski GF, Simon PA, Ho A, et al. Assessing the burden of disease and injury in Los Angeles County using disability-adjusted life years. Public Health Rep (2002) 117:185–91.
33 Schopper D, Pereira J, Torres A, et al. Estimating the burden of disease in one Swiss canton: what do disability adjusted life years (DALY) tell us? Int J Epidemiol (2000) 29:871–7.
34 Mathers CD, Vos ET, Stevenson CE, Begg SJ. The burden of disease and injury in Australia. Bull World Health Organ (2001) 79:1076–84.[Web of Science][Medline]
35 Melse JM, Essink-Bot ML, Kramers PG, Hoeymans N. A national burden of disease calculation: Dutch disability-adjusted life-years. Dutch Burden of Disease Group. Am J Public Health (2000) 90:1241–7.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  C. Terschuren, O. C. L. Mekel, R. Samson, T. K. D. Classen, C. Hornberg, and R. Fehr Health status of 'Ruhr-City' in 2025 - predicted disease burden for the metropolitan Ruhr area in North Rhine-Westphalia Eur J Public Health, October 1, 2009; 19(5): 534 - 540. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||