Skip to main content

Table 3 Redistribution algorithm derived under alternate matching algorithms.

From: Deaths from heart failure: using coarsened exact matching to correct cause-of-death statistics

  USA Mexico Brazil
  Base
(%)
Demographic (%) Autopsy (%) Base (%) Demographic (%) Autopsy (%) Base (%) Demographic (%) Autopsy (%)
Lower respiratory infections 1 1 2 3 2 2 3 3 1
Diabetes 6 5 2 13 15 12 9 9 3
Cancers 4 4 4 6 8 4 3 3 2
Ischemic heart disease 53 54 54 26 28 33 22 24 41
Cerebrovascular disease 2 2 0 3 3 2 4 4 1
Hypertensive heart disease 7 6 9 16 16 14 23 22 19
Cardiomyopathy 4 4 6 1 1 1 9 9 17
Other cardiovascular diseases 10 11 12 8 8 11 5 6 6
Chronic obstructive pulmonary disease (COPD) 5 5 3 11 8 8 9 8 3
Digestive diseases 1 1 3 4 3 4 2 2 2
Other diseases 7 7 6 9 8 10 11 9 6
  1. Match variables are shown in Table 2. In addition, potential matches were restricted as follows: Demographic: matches (controls) were selected from demographic groups that have the best access to health care (US: non-Hispanic white college graduates; Mexico: secondary school graduates covered by a formal health insurance system; Brazil: individuals with at least seven years of schooling). Autopsy: matches (controls) were selected only from deaths that were autopsied.