Counting the dead and what they died from: an assessment of the global status of cause of death data.

171Bulletin of the World Health Organization | March 2005, 83 (3) Objective We sought to assess the current status of global data on death registration and to examine several indicators of data completeness and quality. Methods We summarized the availability of death registration data by year and country. Indicators of data quality were assessed for each country and included the timeliness, completeness and coverage of registration and the proportion of deaths assigned to ill-defined causes. Findings At the end of 2003 data on death registration were available from 115 countries, although they were essentially complete for only 64 countries. Coverage of death registration varies from close to 100% in the WHO European Region to less than 10% in the African Region. Only 23 countries have data that are more than 90% complete, where ill-defined causes account for less than 10% of total of causes of death, and where ICD–9 or ICD–10 codes are used. There are 28 countries where less than 70% of the data are complete or where ill-defined codes are assigned to more than 20% of deaths. Twelve high-income countries in western Europe are included among the 55 countries with intermediate-quality data. Conclusion Few countries have good-quality data on mortality that can be used to adequately support policy development and implementation. There is an urgent need for countries to implement death registration systems, even if only through sample registration, or enhance their existing systems in order to rapidly improve knowledge about the most basic of health statistics: who dies from what? Keywords Mortality; Cause of death; Data collection/standards; Registries/standards; Vital statistics (source: MeSH, NLM). Mots clés Mortalité; Cause décès; Collecte données/normes; Registre/normes; Statistique démographique (source: MeSH, INSERM). Palabras clave Mortalidad; Causa de muerte; Recolección de datos/normas; Registros/normas; Estadísticas vitales (fuente: DeCS, BIREME). Bulletin of the World Health Organization 2005;83:171-177. Voir page 176 le résumé en français. En la página 176 figura un resumen en español. Counting the dead and what they died from: an assessment of the global status of cause of death data Colin D. Mathers,1 Doris Ma Fat,1 Mie Inoue,1 Chalapati Rao,2 & Alan D. Lopez2 Research 1 Evidence and Information for Policy, World Health Organization, 1211 Geneva 27, Switzerland. Correspondence should be sent to Dr Mathers at this address (email: mathersc@who.int). 2 School of Population Health, University of Queensland, Brisbane, Australia. Ref. No. 04-015784 (Submitted: 29 June 2004 – Final revised version received: 8 October 2004 – Accepted: 11 October 2004) Introduction Policies and programmes to combat diseases and injuries should properly be based on current, timely information about the nature and extent of health problems, their determinants, and how the impact of such diseases and injuries is changing, both with respect to magnitude and distribution in populations. Where programmes to improve health are in place, they need to be routinely monitored and evaluated to ensure that their objectives and targets are being met. Equally, priorities for health research should, in part at least, be based on a thorough assessment of the relative importance of various diseases and injuries affecting the population’s health. Undoubtedly the most widely available and commonly used data for meeting these needs, and related needs for health policy, are statistics on the number of people who die, by age and sex, and on the causes of those deaths, classified according to a standard set of medical criteria. Almost all countries have legislation that establishes vital registration systems to collect and collate statistics on who dies from what cause. Yet the utility of the data generated from these systems for meeting basic health policy purposes varies substantially from country to country. Indeed, such systems are still inoperative for a large proportion of the world’s population, especially in countries with high burdens of disease. Editorials have drawn attention to the lack of informa- tion on causes of death in many developing countries and the urgent need for WHO and other international health agencies to take a lead in redressing this situation (1). Indeed, the Director- General of WHO has stated that one of his priorities is the strengthening of vital statistics registration systems (2). Over the past four years, WHO has intensified efforts to support the collection of vital registration information and other mortality data in developing countries. The number of .176 172 Bulletin of the World Health Organization | March 2005, 83 (3) Research Assessment of global cause of death data Colin D. Mathers et al. countries reporting data on recent death registrations to WHO annually has increased from around 65 in 1970, and 90 in 1999, to 115 in 2003. In addition, information from sample registra- tion systems, population studies and epidemiological analyses of specific conditions have been used to improve estimates of patterns of causes of death. This paper reviews the status of global death registration data as supplied to WHO by Member States and examines several indicators of data quality. Such a review is timely. These data form the cornerstone of analyses of the health situation and disease burden in countries and are a critical input into building the evidence base for health policies and programmes. Methods Death registration data supplied to WHO were analysed to determine the extent to which death registration systems cover deaths occurring in the six WHO regions, to summarize the timeliness of available data, and to examine indicators of data quality including completeness of registration, the coding system used for causes of death, and the proportion of deaths coded to various ill-defined categories. Completeness Considerable differences exist in the degree of completeness of the vital registration data submitted by countries. In some cases, the vital registration system functions only for part of the country (for example, only in urban areas or some provinces). In other countries, although the registration system attempts to cover the whole country, not all deaths are registered. The proportion of all adult deaths that is registered for the population covered by the vital registration system (referred to as “complete- ness”) has been estimated separately for each sex using standard demographic methods (3) for those Member States submitting data. The completeness of information on deaths occurring among infants and children may be lower in countries with incomplete death registration and can only be qualitatively assessed by comparing childhood mortality as calculated from the vital registration system with estimates derived from cen- suses and surveys (3). Each Member State also routinely reports population data for the population covered by the death registration sys- tem, which in some cases is a subset of the national population. Death registration data may be complete for the population specified but may not cover the entire population resident in the country. Additionally, the completeness of death registration may be less than 100% for the specified population. Registra- tion of vital events is often less complete in rural areas than in cities and, in general, may be worse in areas with poor living conditions. Coverage Coverage of the resident population by death registration data may be less than 100% not only because some geographical areas may be excluded but also because in some cases registration may be restricted to a subset of the resident population, such as citizens or permanent or legal residents, and may thus exclude deaths among groups such as guest workers or refugees. For our study, coverage was calculated by dividing the total number of deaths reported for a country–year by the total number of deaths occurring among the resident population as estimated for that year by WHO. The total number of estimated deaths for the resident population was calculated by applying best estimates of death rates by age and sex (3), which had been calculated from the reported deaths and reported population and adjusted for incompleteness, to the estimates of the resident population taken from the UN Population Division’s 2002 revision (4). WHO-estimated coverage for a Member State may be less than 100% due to incompleteness of registration, the fact that only some parts of the resident population are covered, or to differ- ences between the vital registration population and the UN’s estimate of the population. Coverage levels above 100% may occur due to small discrepancies between UN and national population estimates; these were rounded to 100%. Quality of information on causes of death The comparability of worldwide cause of death data has been facilitated through the development and successive revisions of the International statistical classification of diseases and related health problems (known as ICD). The 10th revision of the ICD came into effect in 1993 (5). Although each revision has pro- duced some discontinuities in cause of death data, the revision from ICD–9 to ICD–10 has resulted in more substantial changes than previous revisions. ICD–10 has considerably greater detail than ICD–9 (with almost twice the number of codes) and includes both conceptual and classification revisions as well as changes in the coding rules used to select the under- lying cause of death. Additional problems in comparing data on causes of death across countries arise from variations in the accuracy of diagnosing causes of death. In most developed countries, the underlying cause of death is certified by a medical practitioner although the practi- tioner may not always have had prior contact with the deceased or access to relevant medical records. In developing countries, a significant proportion of deaths may occur without medical attention, and they may be registered without a medical opinion about the cause of death. In both developing and developed countries, legal, societal and other reasons may lead to the un- derreporting of causes of a sensitive nature, such as suicide or HIV/AIDS (6, 7). Furthermore, the selection of a single underlying cause of death is frequently problematic in elderly people, who often have had several chronic diseases that con- currently led to death. Even in countries where causes are assigned by medically qualified staff, there is often substantial use of coding categories for unknown and ill-defined causes (6, 8). The proportion of deaths assigned to the ICD codes for “symptoms, signs, and ill-defined conditions” (ICD–9 codes 780–799 and ICD–10 codes R00–R99) can be used as one indicator of the quality of coding in the registration system (9). There are a number of other ICD codes that do not represent useful underlying causes from a policy perspective and whose inappropriate overuse compromises the usefulness of information on cause of death. These “garbage codes” include deaths from injuries where the intent is not determined (ICD–9 codes E980–989 and ICD–10 codes Y10–Y34 and Y872); cardiovascular disease categories lacking diagnostic meaning, such as cardiac arrest and heart failure (ICD–9 codes 427.1, 427.4, 427.5, 428, 429.0, 429.1, 429.2, 429.9, 440.9 and ICD–10 codes I47.2, I49.0, I46, I50, I51.4, I51.5, I51.6, I51.9, I70.9); and cancer deaths coded to categories for secondary or unspecified sites (ICD–9 codes 195, 199 and ICD–10 codes C76, C80, C97). 173Bulletin of the World Health Organization | March 2005, 83 (3) Research Colin D. Mathers et al. Assessment of global cause of death data Using the most recent year for which cause of death data were available for each Member State, we computed the proportion of deaths coded to each of these four groups of ill- defined codes. This provides one set of indicators of the quality of coding of causes of death. Taking into account other factors, such as the type of cause coding used, the completeness of death registration, and how recent the latest available data are, it is possible to broadly categorize the overall quality of death registration data. Among countries with at least 50% completeness or coverage of deaths, we have defined three broad categories. • High-quality data: In this category ICD–9 or ICD–10 cod- ing are used and there is > 90% completeness and ill-defined codes appear on < 10% of registrations. • Medium-quality data: In this category the completeness of data is 70–90% or ill-defined codes appear on 10–20% of registrations or completeness > 90% and ill-defined codes appear on < 10% of registrations but non-ICD codes are used. • Low-quality data: In this category completeness is < 70% or ill-defined codes appear on > 20% of registrations. The cut-off points used to define these categories are to some extent arbitrary, and timeliness of data has not been included beyond requiring that the most recent year be later than 1990. The completeness thresholds of 50% and 90% were chosen because it is likely that death registration data that are more than 90% complete will give a reasonably unbiased picture of the distribution of causes of death and, conversely, that data Table 1. Number of countries reporting data on cause of death from their vital registration system to WHO and countries with no recent data, December 2003 No. of countries WHO region Total Complete Useable No No data Member States with No. data data recent available no recent data available available dataa Africa 46 1 4 42 25 Algeria,b Angola, Benin, Botswana, Burkina Faso, Burundi, Cameroon, Cape Verde,c Central African Republic, Chad, Comoros, Congo, Côte d’Ivoire, Democratic Republic of the Congo, Equatorial Guinea, Eritrea, Ethiopia, Gabon, Gambia, Ghana, Guinea, Guinea-Bissau, Kenya, Lesotho, Liberia, Madagascar, Malawi, Mali, Mauritania, Mozambique, Namibia, Niger, Nigeria, Rwanda, Senegal, Sierra Leone, Swaziland, Togo, Uganda, United Republic of Tanzania, Sao Tome and Principe,c Zambia The Americas 35 14 33 2 0 Bolivia, Hondurasc Eastern 21 3 7 12 6 Afghanistan, Djibouti, Jordan,b,c Iraq, Libyan Arab Jamahiriya, Mediterranean Pakistan,b Somalia, Sudan, Saudi Arabia, Tunisia,b United Arab Emirates,b Yemen Europe 52 39 50 2 0 Andorra,b Monacoc South-East 11 0 4 7 4 Bangladesh,b Bhutan, Indonesia, Democratic People’s Asia Republic of Korea, Timor-Leste, Maldives,b Nepalb Western 27 7 17 10 4 Cambodia, Lao People’s Democratic Republic, Marshall Pacific Islands,b Micronesia,b Papua New Guinea,c Palau,b Samoa,b Solomon Islands,b Vanuatu, Viet Nam World 192 64 115 75 39 a Information available from 1990 or later. b Information on deaths is available but cause of death is not included. c Latest available year for information on causes of death is prior to 1990. that are less than 50% complete are unlikely to yield an un- biased picture (10). An exception to these assumptions is the case where a sample registration system has low coverage but has been explicitly designed to provide a representative sample of deaths. Results Global coverage of registration As of December 2003, 115 Member States had supplied death registration data to WHO. Table 1 summarizes the number of countries reporting these data by WHO region. Regional coverage of death registration varies from close to 100% in the European Region to less than 10% in the African region. How- ever, death registration is considered to be essentially 100% complete in only 64 of the 115 countries reporting data, and predominantly these are developed countries in Europe, the Americas and the Pacific regions. China and India, two countries with large populations, do not have complete vital registration but are reasonably well served by sample registration systems (see Discussion below). In terms of actual deaths recorded by registration systems, data are provided to WHO annually for about 18.6 million deaths, representing one-third of all deaths estimated to be occurring in the world (11). If the sample registration systems in India and China are considered to be sufficiently representative to provide information on their whole populations, then informa- tion on mortality is available for around 72% of the world’s population. 174 Bulletin of the World Health Organization | March 2005, 83 (3) Research Assessment of global cause of death data Colin D. Mathers et al. Table 1 also lists the Member States for which no recent data on death registration are available (that is, from 1990 or later). In 75 countries, no information is available on causes of death for any year after 1990. Completeness and coverage by country The estimated completeness of all cause-of-death registration for both sexes combined is given in Table 2 (web version only, available at: http://www.who.int/bulletin) for Member States reporting data. The overall level of coverage for the latest avail- able year for each country is also shown in Table 2. Although the coverage of the sample registration systems in India and China is low, in China the system provides representative information that can be used to assess causes of death among the national population. Timeliness For each Member State reporting data Table 2 shows the years for which death registration data have been provided together with several indicators of data quality. In 18 Member States, death registration systems provide recent data only on deaths due to all causes, without information on the underlying cause of death. A more detailed listing of the specific years for which data are available has been made available on the WHO web site (10). Interested readers can access these data via the online WHO mortality database (available at http://www.who.int/whosis), which contains data supplied by Member States but received after the publication of the 1996 edition of the World health statistics annual, the last to appear in printed form (12). Coding The number of countries using ICD–10 coding to report data to WHO has increased from 4 in 1995 to 75 in 2003. There are still around 40 countries reporting data using ICD–9, and one is still using ICD–8. The penultimate column of Table 2 shows the total per- centage of deaths coded to the four groups of ill-defined codes. It should be noted that deaths from other coded causes are not necessarily correctly coded; they are subject to diagnostic, certification and coding errors that cannot be detected except through specific studies. Table 3. Distribution of percentage of total deaths assigned to ill-defined codes for most recent available year for 105 WHO Member Statesa Percentage of deaths assigned to ill-defined codes Ill-defined Median 25th 75th Maxi- code group percentile percentile mum Symptoms 4.0 2.1 8.7 44.0 Injury 0.5 0.2 1.3 5.1b Cancer 1.0 0.5 1.5 2.7 Cardiovascular disease 5.3 2.7 7.7 23.4 Total ill-defined 12.0 7.0 17.2 48.8 a Includes those Member States supplying data on death registration for most recent year since 1990 and with at least 50% completeness or coverage. b These data exclude South Africa where 93% of deaths from external causes are coded to ill-defined injuries. The percentage of deaths coded to the aggregate of the groups of ill-defined causes varies from 4% in New Zealand to more than 40% in Sri Lanka and Thailand. The Ministry of Health in Thailand has recently conducted an investigation into cause-of-death certification and coding (using a sample of 35 000 deaths) to verify the underlying causes of death using verbal autopsy techniques. This sample included about 12 000 deaths from ill-defined causes of which, on investigation, about two-thirds were assigned to specific causes of death (13). Table 3 shows the distribution of deaths assigned to ill- defined codes for the 105 Member States reporting data on death registrations since 1990 with at least 50% completeness or coverage. The median percentage of deaths coded to ill- defined causes is 12%; the median percentage of symptoms, signs and ill-defined conditions is 4.0%; and the median of ill-defined cardiovascular causes is 5.3%. Such codes are of no use in guiding disease prevention and injury prevention and control programmes. Quality of death data Table 4 classifies death registration data into high quality, me- dium quality or low quality using the criteria described above. There are only 23 countries classified as having high-quality data. There are 28 countries classified as having low-quality data; these include some high-income countries, such as Greece and Portugal. Somewhat surprisingly, among the 55 countries included in the medium-quality category, there are more than 10 high-income countries from western Europe including Austria, Belgium, Denmark, France and Germany. In these countries the percentage of deaths assigned to the ill-defined codes ranges from 10% to 20%, not so much because of overuse of codes for symptoms, signs and ill-defined conditions, but because of excessive use of garbage codes for vascular disease, cancers and injuries. Discussion Ideally, vital registration systems should capture all deaths that occur in a given population, should reliably record the age and sex of each deceased person, and the cause of death recorded should be based on the opinion of a medically qualified person. Trained personnel are needed to ensure that information about the medical condition leading to death is coded appropriately so that the underlying cause of each death can be identified. To calculate age-specific death rates, which is the most useful output of a vital registration system for policy purposes, reliable estimates of the population exposed to the risk of dying are also required. Taken together the resources required to establish and maintain a functioning vital registration system, including the incorporation of periodic revisions of the ICD, are consider- able. As a result, good vital registration systems are generally more likely to be found in developed countries. Only 23 Member States were considered to have high- quality death registration data (using quality criteria based on timeliness, completeness, coverage and the sparing use of codes for ill-defined causes). About 10% or more of deaths are assigned to ill-defined codes in a surprisingly high number of developed countries. The use of these codes is a particular problem in some developing countries, exceeding 30% in coun- tries such as Thailand and Sri Lanka. These issues can be addressed by making greater efforts to increase coverage, by increasing the proportion of deaths 175Bulletin of the World Health Organization | March 2005, 83 (3) Research Colin D. Mathers et al. Assessment of global cause of death data Table 4. Quality of cause of death information for WHO Member States supplying death registration data for most recent year (1990 or later) and with at least 50% completeness or coverage. (China and India are not included in this table, although their sample registration systems provide population-representative data) Quality of data Criterion used Countries (No. of countries) High (23) ICD–9 or ICD–10 coding is Australia, Bahamas, Canada, Cuba, Estonia, Finland, Hungary, Iceland,a used and completeness is > 90% Ireland, Japan, Latvia, Lithuania, Malta, Mexico, New Zealand, Republic of and ill-defined codes appear on Moldova, Romania, Singapore, Slovakia, Slovenia, United Kingdom, < 10% of registrations United States of America, Venezuela Medium (55) Completeness is 70–90% OR Antigua and Barbuda,a Austria, Azerbaijan, Barbados, Belarus, Belgium,a ill-defined codes appear on 10–20% Belize, Brazil, Brunei Darussalam, Bulgaria, Chile, Colombia,a Cook Islands, of registrations OR non-ICD codes Costa Rica, Croatia, Czech Republic, Denmark,a Dominica, El Salvador,a used although completeness is France,a Georgia, Germany, Grenada,a Guatemala, Guyana,a Israel,a Italy, > 90% and ill-defined codes appear Kazakhstan, Kuwait, Kyrgyzstan, Luxembourg, Malaysia,a Mauritius, Mongolia, on < 10% of registrations Netherlands, Niue, Norway, Panama, the Philippines,a Republic of Korea, Russian Federation, Saint Kitts and Nevis,a Saint Lucia, Saint Vincent and the Grenadines, Serbia and Montenegro, Seychelles, Spain, Sweden, Switzerland,a former Yugoslav Republic of Macedonia, Trinidad and Tobago,a Turkmenistan,a Ukraine, Uruguay, Uzbekistan Low (28) Completeness < 70% OR Albania, Argentina, Armenia, Bahrain, Bosnia and Herzegovina,a Cyprus,a ill-defined codes appear on > 20% Ecuador, Egypt, Fiji, Greece,a Jamaica,a Kiribati, Nicaragua, Oman, Paraguay, of registrations Peru, Poland, Portugal, Qatar, San Marino, South Africa,a Sri Lanka,a Suriname,a Syrian Arab Republic, Tajikistan,a Thailand, Tonga,a Tuvalu a Latest year for which data are available is 1999 or earlier. certified by attending and non-attending physicians, by edu- cating physicians and other informants about the importance of accurate and complete reporting on death certificates, and by avoiding, unless absolutely necessary, the use of ill-defined codes. To support public health policy-making and evaluation, research is also needed to assist in improving analyses of causes of death. Priorities for research include assessing the accuracy of cause- of-death certification and improving knowledge about the true distribution of causes of death attributed to ill-defined codes. In 75 Member States, including more than 90% of African countries, no information on cause of death is available for any year after 1990. For these countries, health planning and priority setting is essentially proceeding on the basis of perceptions or survey-based information on overall levels of child mortality that are used together with model life tables, cause-of-death models and partial information from surveillance systems for some specific causes of disease. Experience in countries such as China, India and the United Republic of Tanzania suggests that sample registration, which is based on a representative set of surveillance sites, with appropriate controls and reporting procedures, can yield extremely useful information about levels, patterns and causes of mortality for large populations (14–17). Setel et al. (unpublished results, 2005) argue that for many low- income developing countries such sample registration systems provide a more cost-effective solution to collecting data than either full registration systems or survey-based methods. In all cases, the key to improving information on causes of death is to ensure that such data become integral to health policy-making at national, regional and district levels. WHO is working with Member States to evaluate the completeness of their registration systems, to investigate the quality of cause-of-death coding, and to develop and validate a standardized verbal autopsy reporting form. Yet much remains to be done if countries are truly to benefit from methodological advances in measuring mortality. WHO also has a critical part to play in transferring knowledge through the development of training materials and registration standards. Research is needed to identify the most cost-effective strategies for improving in- formation on causes of death in developing countries. These may range from establishing full death registrations systems to using low-cost sample registration systems or implementing regular surveys or collections for census data. O Acknowledgements The authors thank the staff of the former Epidemiology and Burden of Disease team in the Evidence and Information for Policy cluster at WHO headquarters, particularly Hongyi Xu who contributed to the collection, cleaning and analysis of death registration data from Member States. The authors also thank WHO Regional Offices for their valuable collaboration and assistance in the collection of death registration data. The named authors alone are responsible for the views expressed in this publication, which do not necessarily reflect the opinion of the World Health Organization or of its Member States, or of the sponsoring agency. Funding: The authors gratefully acknowledge financial support for this project from the National Institute on Aging (Research Grant No. PO1 AG17625). Conflicts of interest: none declared. 176 Bulletin of the World Health Organization | March 2005, 83 (3) Research Assessment of global cause of death data Colin D. Mathers et al. Résumé Recensement des décès et des causes de décès : une évaluation de l’état des données relatives aux causes de décès dans le monde Objectif Les auteurs se sont donnés pour objectif d’évaluer au niveau mondial l’état actuel des données figurant sur les registres des décès et d’examiner plusieurs indicateurs représentatifs de l’exhaustivité et de la qualité de ces données. Méthodes Ils ont récapitulé les données issues des registres des décès disponibles par année et par pays. Ils ont évalué des indicateurs de la qualité de ces données pour chaque pays, dont notamment la réalisation en temps voulu, la complétude et la couverture de l’enregistrement, et la proportion de décès attribués à des causes mal définies. Résultats A la fin de l’année 2003, on disposait de données concernant l’enregistrement des décès provenant de 115 pays, ces données n’étant toutefois pratiquement exhaustives que pour 64 pays. Le taux de couverture des registres des décès va d’une valeur proche de 100 % dans la région Europe de l’OMS à moins de 10 % dans la région Afrique. Seuls 23 pays disposent de données complètes à plus de 90 %, qui mentionnent une cause de décès mal définie dans moins de 10 % des cas et utilisent les codes ICD-9 et ICD-10. Dans 28 pays, moins de 70 % des données sont exhaustives ou plus de 20 % des décès ont donné lieu à l’affectation de codes indiquant une cause de décès mal définie. Douze pays à hauts revenus d’Europe occidentale figurent parmi les 55 pays collectant des données de qualité intermédiaire. Conclusion Peu de pays disposent de données de bonne qualité sur la mortalité, utilisables pour étayer solidement l’élaboration et la mise en œuvre des politiques. Certains pays ont besoin d’urgence de mettre en place des systèmes d’enregistrement des décès, ne serait-ce qu’un enregistrement par sondage, ou de perfectionner les systèmes existants, de manière à approfondir rapidement les connaissances au sujet des statistiques sanitaires les plus élémentaires : qui meurt de quoi ? Resumen Determinar el número de defunciones y las causas de defunción: evaluación de la situación mundial de los datos sobre las causas de mortalidad Objetivo Decidimos evaluar la situación actual de los datos mundiales sobre los registros de defunción y examinar varios indicadores de la integridad y calidad de esos datos. Métodos Resumimos la disponibilidad de los datos de los registros de defunción por año y país. Se evaluaron diversos indicadores de la calidad de los datos para cada país, entre ellos la prontitud de los registros, la integridad y cobertura de los mismos y la proporción de defunciones atribuidas a causas mal definidas. Resultados A finales de 2003 se disponía de datos de registros de defunción de 115 países, aunque sólo podían considerarse completos para 64 países. La cobertura de los registros varía desde cerca del 100% en la Región de Europa de la OMS hasta menos del 10% en la Región de África. Sólo 23 países disponen de registros con datos completos en más del 90% de los casos, donde las causas mal definidas de defunción representan menos del 10% de todas las causas de muerte, y donde se emplean los códigos de la CIE-9 o la CIE-10. Hay 28 países donde menos del 70% de los datos están completos o donde se asignan códigos mal definidos a más del 20% de las defunciones. Entre los 55 países con datos de calidad intermedia figuran doce países de ingresos altos de Europa occidental. Conclusión Son pocos los países que disponen de datos de mortalidad de buena calidad que puedan emplearse para respaldar adecuadamente la formulación y ejecución de políticas. Es urgente que los países implanten sistemas de registro de las defunciones, aunque sea mediante el registro de muestras, o mejoren los sistemas ya existentes, pues sólo así podremos conocer rápidamente con más detalle la información fundamental de la estadísticas sanitarias; a saber: cuántas personas mueren, y de qué mueren. 177Bulletin of the World Health Organization | March 2005, 83 (3) Research Colin D. Mathers et al. Assessment of global cause of death data References 1. Sibai AM. Mortality certification and cause-of-death reporting in developing countries. Bulletin of the World Health Organization 2004;82:83. 2. Lee JW. Global health improvement and WHO: shaping the future. Lancet 2003;362:2083-8. 3. Lopez AD, Ahmad O, Guillot M, Ferguson B, Salomon J, Murray CJL, et al. World mortality in 2000: life tables for 191 countries. Geneva: World Health Organization; 2002. 4. United Nations. World population prospects: the 2002 revision. New York: United Nations; 2003. 5. World Health Organization. International statistical classification of diseases and related health problems, tenth revision. Geneva: WHO; 1992. 6. Ruzicka LT, Lopez AD. The use of cause of death statistics for health situation assessment: national and international experiences. World Health Statistics Quarterly 1990;43:249-58. 7. King MB. AIDS on the death certificate: the final stigma. BMJ 1989;298:734-6. 8. Lopez AD. Mortality data. In: Elliot P, Cuzick J, English D, Stem R, editors. Geographical and environmental epidemiology methods for small-area studies. Oxford: Oxford University Press; 1992. p. 37-50. 9. Silvi J. On the estimation of mortality rates for countries of the Americas. Epidemiological Bulletin 2003;24:1-5. 10. Mathers CD, Bernard C, Iburg KM, Inoue M, Ma Fat D, Shibuya K, et al. Global burden of disease in 2002: data sources, methods and results. Geneva: World Health Organization; 2003 (GPE Discussion Paper No. 54). 11. World Health Organization. The world health report 2004 – Changing history. Geneva: WHO; 2004. 12. World Health Organization. World health statistics annual: 1996. Geneva: WHO; 1998. 13. Ministry of Public Health, Thailand. Burden of disease and injuries in Thailand: priority setting for policy. Bangkok: Government of Thailand; 2002. 14. Ministry of Health, United Republic of Tanzania. The policy implications of Tanzania’s mortality burden: a ten-year community-based perspective. Volume 1. Dar es Salaam: Ministry of Health, United Republic of Tanzania, University of Newcastle upon Tyne, UK Department for International Development; 2004 (Adult Morbidity and Mortality Project-2 Final Report). 15. Mahapatra P, Rao C. Cause of death reporting systems in India: a performance analysis. National Medical Journal of India 2001;14:154-62. 16. Registrar General of India. Sample registration system. New Delhi: Vital Statistics Division, Office of the Registrar General of India; 2002. 17. Yang GH, Huang ZJ, Tan J. [Priorities of disease control in China – analysis of mortality data of national disease surveillance points system]. Chinese Journal of Epidemiology 1996;17:199-202. In Chinese. ABulletin of the World Health Organization | March 2005, 83 (3) Research Colin D. Mathers et al. Assessment of global cause of death data Table 2. Summary of data on death registration available to WHO (December 2003) together with estimated completeness, coverage and percentage of deaths assigned to codes for ill-defined causes Member State Years of Complete- Years with ICD Cover- Deaths coded Qualityi mortality data nessb cause-of- revision aged to ill-defined (all causes) death data usedc codes (%)e Albania 1950–2001a 60 1987–2001a 9 69 19 Low Algeria 1950–2000a 69 Andorra 1950–2000a NAj Antigua and Barbuda 1950–1999a 90 1961–1995a 9 84 13g Medium Argentina 1950–2001a 100 1966–2001a 10 100 22 Low Armenia 1981–2001a 68 1981–2001a 9 91 4g Low Australia 1950–2001 100 1950–2001 10 100 5 High Austria 1950–2001 100 1955–2001 9 100 14 Medium Azerbaijan 1981–2001 74 1981–2001a 10 71 2h Medium Bahamas 1965–2000a 100 1969–2000a 10 93 5 High Bahrain 1980–2001a 100 1985–2001a 10 90 25 Low Barbados 1950–2001 100 1955–2000 10 100 12 Medium Belarus 1981–2001 100 1981–2001a 9 98 10g Medium Belgium 1950–2001 100 1954–1997 9 100 15 Medium Belize 1950–1998 100 1964–1998a 10 99 12 Medium Bosnia and Herzegovina 1985–1999a 100 1985–1999a 10 88 12h Low Brazil 1974–2000 84 1977–2000 10 79 20 Medium Brunei Darussalam 1950–2000a 84 1996–2000 10 100 7h Medium Bulgaria 1950–2002 100 1964–2002 9 100 18 Medium Canada 1950–2000 100 1950–2000 10 100 7 High Cape Verde 1955–1980a NA 1980 8 Chile 1950–2000 100 1954–1999 10 100 11 Medium Chinaf 1987–2000 54 1987–2000 9 <50 3h Colombia 1950–2001a 81 1953–1999a 10 79 7 Medium Cook Islands 1951–2001a 100 1995–2001 10 100 3h Medium Costa Rica 1950–2002a 90 1956–2002 10 79 6 Medium Croatia 1982–2001 100 1985–2001 10 98 16 Medium Cuba 1959–2001a 100 1959–2001a 10 100 9 High Cyprus 1950–2001a 100 1996–1999 10 70 44h Low Czech Republic 1982–2001 100 1985–2001 10 100 13 Medium Denmark 1950–2002 100 1951–1999 10 100 12 Medium Dominica 1950–1999a 100 1961–1999a 9 100 7g Medium Dominican Republic 1950–1999a <50 1950–1998a 10 <50 Ecuador 1954–2000a 70 1961–2000a 10 76 22 Low Egypt 1950–2000a 85 1954–2000a 10 80 40 Low El Salvador 1950–1999a 76 1950–1999a 10 73 19 Medium Estonia 1981–2001 100 1981–2001a 10 100 5 High Fiji 1950–2000a 100 1978–2000a 10 100 29 Low Finland 1950–2001 100 1952–2001 10 100 3 High France 1950–2001 100 1950–1999 9 100 14 Medium Georgia 1981–2000a 75 1981–2000a 10 66 8 Medium Germany 1969–2000 100 1969–2000 10 100 14 Medium Greece 1951–2001 100 1956–1999 9 88 26 Low Grenada 1950–1996a NA 1974–1996a 9 86 6g Medium Guatemala 1950–1999a 87 1958–1999a 9 86 14g Medium Guyana 1950–1996a 67 1975–1996a 9 74 8g Medium Haiti 1972–1999a <50 1980–1999a 10 <50 Honduras 1950–1983 NA 1966–1983a 9 Hungary 1950–2002 100 1955–2002 10 100 7 High Iceland 1950–2001 100 1951–1999 10 100 5 High Indiaf 1990–1999 88f 1996–1998 9 <50 18 Iran, Islamic Republic of 1983–2001a 56k 1999–2001 10 <50 Ireland 1950–2001 100 1950–2000 9 100 7 High Israel 1953–2000 100 1975–1999 10 100 11 Medium Italy 1950–2000 100 1951–2000 9 99 12 Medium B Bulletin of the World Health Organization | March 2005, 83 (3) Research Assessment of global cause of death data Colin D. Mathers et al. (Table 2, cont.) Member State Years of Complete- Years with ICD Cover- Deaths coded Qualityi mortality data nessb cause-of- revision aged to ill-defined (all causes) death data usedc codes (%)e Jamaica 1950–1999a 75 1960–1991a 9 60 13g Low Japan 1950–2002 100 1950–2000 10 97 9 High Jordan 1953–2001a 63 1959–1979a 8 Kazakhstan 1981–2001 89 1981–2001a 9 80 5g Medium Kiribati 1991–2002 86 1999–2002 10 76 35 Low Kuwait 1962–2002a 100 1972–2001a 10 96 13 Medium Kyrgyzstan 1981–2001 84 1981–2001a 10 73 6 Medium Latvia 1980–2002 100 1980–2001 10 100 5 High Lebanon 1997–1999 <50 1997–1999 10 <50 Lithuania 1981–2001 100 1981–2001a 10 97 5 High Luxembourg 1950–2002 100 1955–2002a 10 100 14 Medium Malaysia 1990–1998 90 1990–1998 9 <50 17 Medium Maldives 1978–1998 61 Malta 1950–2002 100 1955–2002 10 95 10 High Marshall Islands 1986–1997 NA Mauritius 1950–2001 100 1957–2000 9 100 14 Medium Mexico 1950–2002 100 1955–2001 10 96 5 High Micronesia, Federated 1986–1994 74 States of Monaco 1950–1987a NA 1986–1987 9 Mongolia 1987–2001 100 1990–2000 9 76 1h Medium Morocco 1990–1998 < 50 1990–1997 10 <50 Mozambique 1961–1973a NA Myanmar NA 1977–2000a 10 <50 Nauru 1965–1996a NA 1994–1996 Other <50 Netherlands 1950–2001 100 1950–2000 10 100 15 Medium New Zealand 1950–2001 100 1950–2000 10 100 4 High Nicaragua 1950–2000a 58 1959–2000a 10 55 9 Low Niue 1950–2000a 70 1995–2000 10 100 0 Medium Norway 1950–2001 100 1951–2001 10 98 12 Medium Oman 1997, 2000 NA 1997, 2000 2001 10 71 9h Low Palau 1985–1999a 100 Panama 1950–2000a 88 1954–2000a 10 86 12 Medium Papua New Guinea NA 1977, 1980 9 Paraguay 1950–2000a 74 1961–2000a 10 74 28 Low Peru 1950–2000a 50 1966–2000a 10 50 20 Low Philippines 1950–1998a 77 1963–1998a 9 85 12 Medium Poland 1950–2001 100 1959–2001 10 100 25 Low Portugal 1950–2001 100 1955–2000 9 100 21 Low Qatar 1981–2001a 100 1995–2001a 10 83 24 Low Republic of Korea 1957–2001a 89 1985–2001 10 88 14 Medium Republic of Moldova 1981–2001 100 1981–2001a 10 83 3 High Romania 1956–2002 100 1959–2002a 10 100 8 High Russian Federation 1980–2002 100 1980–2002 10 100 5h Medium Saint Kitts and Nevis 1950–2000 100 1961–1995a 9 100 2g Medium Saint Lucia 1950–2001a 100 1968–2001a 10 100 12 Medium Saint Vincent and the 1950–1999a 100 1970–1999a 9 99 2g Medium Grenadines Samoa 1955–2002a 51 San Marino 1962–2001a 100 1995–2000 9 76 35 Low Sao Tome and Principe 1955–1987a NA 1984–1987a 9 Serbia and Montenegro 1982–2000a 100 2000 10 97 8h Medium Seychelles 1952–2000a 94 1981–2000a 10 100 5h Medium Singapore 1950–2001 100 1955–2001 9 81 3 High Slovakia 1982–2001 100 1992–2001 10 100 4 High Slovenia 1982–2001 100 1985–2001 10 100 10 High South Africa 1980–1999a 88 1993–1996 10 <50 37 Low CBulletin of the World Health Organization | March 2005, 83 (3) Research Colin D. Mathers et al. Assessment of global cause of death data Member State Years of Complete- Years with ICD Cover- Deaths coded Qualityi mortality data nessb cause-of- revision aged to ill-defined (all causes) death data usedc codes (%)e Spain 1950–2000 100 1951–2000 10 100 11 Medium Sri Lanka 1950–1996a 91 1950–1996a 9 74 46 Low Suriname 1950–1997a 100 1963–1992a 9 66 17g Low Sweden 1950–2001 100 1951–2001 10 100 11 Medium Switzerland 1950–2001 100 1951–2000 10 100 13 Medium Syrian Arab Republic 1973–2001a 71 1973–2001a 10 100 21h Low Tajikistan 1981–1999a 60 1981–1999a 9 <50 9g Low Thailand 1950–2000 86 1955–2000a 10 89 49 Low The former Yugoslav 1982–2000 100 1991–2000 9 92 13 Medium Republic of Macedonia Tonga 1957–2002a 86 1998 Other 70 19h Low Trinidad and Tobago 1950–1998 100 1951–1998 9 93 2g Medium Tunisia 1960–2000a 64 Turkey 1967–2000a 89l 1987–1998 8 <50 Turkmenistan 1981–1998a 78 1981–1998a 9 76 3g Medium Tuvalu 1991–2000 77 1992–2000 Other 100 27h Low Ukraine 1981–2000 100 1981–2000a 9 95 5g Medium United Arab Emirates 2001–2002 NA United Kingdom 1950–2001 100 1950–2000 9 100 7 High United States of America 1950–2001 100 1950–2000 10 100 7 High Uruguay 1950–2000 100 1955–2000a 10 100 16 Medium Uzbekistan 1981–2000 91 1981–2000a 9 82 2g Medium Venezuela 1950–2000 96 1955–2000a 10 97 8 High Zimbabwe 1990–1995a <50 1990–1995a 9 <50 a Incomplete series with data unavailable for some years. b Completeness refers to the proportion of all deaths that are registered in the population covered by the vital registration system for a country. For completeness < 100% the figures refer to completeness for adult deaths. c ICD = International statistical classification of diseases and related health problems. “Other” indicates that other coding systems are used. d Coverage is calculated by dividing the total number of deaths reported from the vital registration system for a country–year by the total number of deaths estimated by WHO for that year for the national population. e Percentage of deaths coded to symptoms, signs and ill-defined conditions (ICD–9 codes 780–799 and ICD–10 codes R00–R99), injury deaths with undetermined intent (ICD–9 codes E980–989 and ICD–10 codes Y10–Y34 and Y872), secondary neoplasms and neoplasms of unspecified sites (ICD–9 codes 195, 199 and ICD–10 codes C76, C80, C97) and ill-defined cardiovascular causes (ICD–9 codes 427.1, 427.4, 427.5, 428, 429.0, 429.1, 429.2, 429.9, 440.9 and ICD–10 codes I47.2, I49.0, I46, I50, I51.4, I51.5, I51.6, I51.9, I70.9). Percentages reported are for the most recent year for countries with > 50% coverage or completeness and the latest available year from 1990 or later. f These countries have sample vital registration systems. Although coverage is low, the sample provides population-representative information. g These countries include only data coded to the ICD chapter on symptoms, signs and abnormal clinical and laboratory findings, not elsewhere classified and to ill-defined injuries. h These countries include only data coded to the ICD chapter on symptoms, signs and abnormal clinical and laboratory findings not elsewhere classified. i Quality was classified using criteria described in the Methods section for WHO Member States supplying data on cause-of-death registrations for most recent year (from 1990 or later) and with at least 50% completeness or coverage as estimated by WHO. j NA = not available or not applicable. k Based on 2001 data from vital registration in 18 provinces. l For urban areas only. (Table 2, cont.)