Redistribution of heart failure deaths using two methods: linkage of hospital records with death certificate data and multiple causes of death data

• Main Authors: Ana Luiza Bierrenbach, Gizelton Pereira Alencar, Cátia Martinez, Maria de Fátima Marinho de Souza, Gabriela Moreira Policena, Elisabeth Barboza França
• Format: Article
• Language: English
• Published: Escola Nacional de Saúde Pública, Fundação Oswaldo Cruz
• Series: Cadernos de Saúde Pública
• Subjects: Insuficiência Cardíaca; Atestado de Óbito; Mortalidade; Registro Médico Coordenado; Classificação Internacional de Doenças
• Online Access: http://www.scielosp.org/scielo.php?script=sci_arttext&pid=S0102-311X2019000605010&lng=en&tlng=en

Heart failure is considered a garbage code when assigned as the underlying cause of death. Reassigning garbage codes to plausible causes reduces bias and increases comparability of mortality data. Two redistribution methods were applied to Brazilian data, from 2008 to 2012, for decedents aged 55 years and older. In the multiple causes of death method, heart failure deaths were redistributed based on the proportion of underlying causes found in matched deaths that had heart failure listed as an intermediate cause. In the hospitalization data method, heart failure deaths were redistributed based on data from the decedents’ corresponding hospitalization record. There were 123,269 (3.7%) heart failure deaths. The method with multiple causes of death redistributed 25.3% to hypertensive heart and kidney diseases, 22.6% to coronary heart diseases and 9.6% to diabetes. The total of 41,324 heart failure deaths were linked to hospitalization records. Heart failure was listed as the principal diagnosis in 45.8% of the corresponding hospitalization records. For those, no redistribution occurred. For the remaining ones, the hospitalization data method redistributed 21.2% to a group with other (non-cardiac) diseases, 6.5% to lower respiratory infections and 9.3% to other garbage codes. Heart failure is a frequently used garbage code in Brazil. We used two redistribution methods, which were straightforwardly applied but led to different results. These methods need to be validated, which can be done in the wake of a recent national study that will investigate a big sample of hospital deaths with garbage codes listed as underlying causes.