Challenges related to records and quality of information in the Amazon

One of the United Nations’ sustainable development goals is the establishment of high-quality, valid, and reliable civil registry and vital statistics systems necessary for the design, evaluation, and implementation of social, economic and health programs, especially in the context of changes in the pattern of mortality that many countries have been experiencing. Although several methods developed in demography 1,2 allow for following the evolution of the quality of the records and to indirectly estimate the levels of mortality in different localities, the quality of the system of vital statistics allows tracking the development of health conditions in a population, evaluating the public health policies, and assisting the health managers 3,4,5. Recently, the quality of mortality data in Brazil has improved substantially. Estimates of the degree of coverage of death registries in Brazil increased from about 80% in the 1980s to about 95% in 2010 6,7,8. However, regional variability in the quality of the degree of coverage of the death registry and the quality of information on the causes of death in the country is still high 6,7,9. Since 2010, the South and Southeast regions show a complete coverage of adult mortality records. The states of the Northeast and North, even with the trends of improvements in recent years, have locations with low coverage 6,10. The average coverage of mortality in the North Region rose from 65% to 76% from 1980 to 2010, whereas coverage in the South Region increased from 95% to 98% in the same period 10. In the case of smaller areas, such as mesoregions, the impacts of data quality can be even greater. The estimated degree of coverage for the São Paulo metropolitan area is 100%, high quality of data, with the probability of adult death of 0.2364, for individuals aged from 15 to 60 years. In the case of the South Amazonian mesoregion, the estimated degree of coverage is 68%. The observed data indicate a probability of adult death of 0.1102; however after applying the correction it is estimated at 0.1621 10. An exercise with data from the 2010 Demographic Census indicates, for infant mortality, coverage of the events registration of practically 100% in the São Paulo metropolitan area compared with about 50% in the mesoregion of the South Amazon. In summary, directly analyzing the data, without applying correction methods, can lead to erroneous results directly affecting health policies. An important conclusion, however, is that the different estimation methods applied to limited data may present quite different results, complexifying the definition of health strategies. A study comparing different estimates of mortality for Brazil and its regions shows the great discrepancy of these estimates and highlights the importance of continuous investment in the quality of records and in the replicability of the methods used 11. 1 Departamento de Demografia, Universidade Federal de Minas Gerais, Belo Horizonte, Brasil.