GHME 2013 Conference
Session: Verbal autopsy
Date: June 18 2013
Presenter: Elisabeth Franca
Institution:
Federal University of Minas Gerais, Brazil
PAHO/WHO
Oswaldo Cruz Foundation
Strengthening vital statistics in Brazil: investigation of ill-defined causes of death and implicatoins on mortality statistics
1. Strengthening vital statistics in
Brazil: investigation of ill-defined
causes of death and implications
on mortality statistics
GPEAS
UniversidadeFederaldeMinasGerais
Conference “Global Health Metrics and Evaluation:
Data, Debates, Directions”
Seattle, June 17-19 2013
Elisabeth França1
, Fátima Marinho2
Lenice H Ishitani1
, Renato Teixeira1
,
Celia L Szwarcwald3
1.Federal University of Minas Gerais,
Brazil; 2. PAHO/WHO; 3. Oswaldo Cruz
Foundation
2. GPEAS/UFMG
Introduction (1)
Mortality Information System (MIS) in Brazil
. Created in 1975 by the Ministry of Heath
. Death certificate(DC): international form (ICD)
→ should be issued by the attending physician (or
a coroner under suspicious conditions)
. Automatic selection program (SCB) for the underlying
cause of death
. Cleaning and compilation: Municipal, States and
National levels electronic data file transfer
. Available on the Internet: www.datasus.gov.br
.
3. Quality of causes of death data in the MIS:
. North and the Northeast regions: 20%-
30% of IDCD until 2003;
. South and Southeast: <10%.
Political – administrative division:
•5 regions
•27 states
•5540 municipalities
Introduction (2)
4. GPEAS/UFMG
Introduction (3)
• In 2005, the Brazilian government implemented a project to
decrease the ill-defined causes of death (IDCD) from chapter
XVIII ICD-10, mainly in states and municipalities in the
poorest regions;
• In this project, deaths due to IDCD were investigated and
reclassified into a defined cause;
• A question: Is cause of death distribution among IDCD similar
to those of recorded data apart from injuries?
• This study aims to compare the distribution of causes of
death between the investigated IDCDs and the recorded
data.
5. GPEAS/UFMG
Investigation of IDCD in Brazil
• In the MoH project, for each death certificate with IDCD all
attempts were made by health-service professionals to
trace existing information about the final disease and
cause of death;
• Sources of information: hospital records, municipality
health departments, autopsies, family health program
records, linkage between health information systems (e.g.
hospital information system), and verbal autopsy
questionnaires (2009-2010);
• Underlying cause of death: assigned using the available
evidence.
6. Source: CGIAE/DASIS/SVS-MoH
Percentage of IDCD
2004
2008
2011
Reduction of deaths due to ill-defined
causes, by municipalities
2004, 2008 and 2011
Investigation of IDCD in Brazil –Results (1)
12. GPEAS/UFMG
Investigation of ill-defined causes of death in Brazil –
Results in 2010
Cause-of-death
(n) (%) n %
Infectious diseases 676 3.5 48,488 4.7 0.7
Neoplasms 1,792 9.3 176,402 17.0 0.5
Diseases of the blood and immune system 92 0.5 6,290 0.6 0.8
Endocrine diseases 1,974 10.2 67,863 6.5 1.6
Mental and behavioural disorders 1,232 6.4 11,374 1.1 5.8
Diseases of the nervous system 613 3.2 24,468 2.4 1.3
Chapter 7 8 0 28 0 15.4
Chapter 8 4 0 122 0 1.8
Diseases of the circulatory system 8,182 42.4 317,674 30.6 1.4
Diseases of the respiratory system 1,526 7.9 119,334 11.5 0.7
Diseases of the digestive system 828 4.3 57,239 5.5 0.8
Chapter 12 54 0.3 3,282 0.3 0.9
Chapter 13 80 0.4 4,396 0.4 1.0
Diseases of the genitourinary system 194 1.0 24,613 2.4 0.4
Maternal conditions 76 0.4 1,422 0.1 2.9
Perinatal conditions 105 0.5 23,723 2.3 0.2
Congenital malformations 72 0.4 9,869 1.0 0.4
Chapter 19 0 0 20 0 0
External causes 1,796 9.3 141,360 13.6 0.7
Total 19,304 100.0 1,037,967 100.0 1.0
*IDCD: ill-defined causes of death
**All causes apart from IDCD (n=97,314) and unclassified deaths (n=1,666)
***%IDCD / %defined causes
Ratio***
Distribution of causes of death among ill-defined causes and recorded data. Brazil, 2010.
IDCD investigated
and reclassified*
Defined causes (DC)
among recorded data**
13. GPEAS/UFMG
Investigation of ill-defined causes of death in Brazil –Results
in 2010
Age Ratio
n % n %
< 01 316 1.6 38,103 3.7 0.4
01-04 126 0.7 6,341 0.6 1.1
05-09 70 0.4 3,833 0.4 1.0
10-14 92 0.5 5,284 0.5 0.9
15-19 240 1.2 17,899 1.7 0.7
20-24 320 1.7 26,431 2.5 0.7
25-29 410 2.1 27,002 2.6 0.8
30-34 504 2.6 27,562 2.7 1.0
35-39 623 3.2 29,250 2.8 1.1
40-44 844 4.4 36,702 3.5 1.2
45-49 1,085 5.6 47,871 4.6 1.2
50-54 1,132 5.9 59,595 5.7 1.0
55-59 1,237 6.4 70,070 6.8 0.9
60-64 1,462 7.6 78,437 7.6 1.0
65-69 1,475 7.6 87,602 8.4 0.9
70-74 1,769 9.2 103,357 10 0.9
75-79 1,904 9.9 109,025 10.5 0.9
80-84 1,968 10.2 108,696 10.5 1.0
85 + 3,661 19.0 151,738 14.6 1.3
Total*** 19,304 100.0 1,037,967 100.0 1.0
* Ill-defined causes
**Defined causes (DC)
***Missing age included (n=65 for IDCD and n=3,169 for recorded data)
IDCD Recorded data (%)
Comparing age distribution among reclassified IDCD* and recorded
data (DC)**. Brazil, 2010.
14. GPEAS/UFMG
Investigation of ill-defined causes of death in Brazil –Results
in 2010
Garbage codes Recorded data (DC) Reclassified IDCD
Ill-defined circulatory codes/all circulatory diseases 6.0 7.1
Ill-defined cancercodes/all cancer 5.6 7.7
Ill-defined injury codes/all injuries 14.8 16.3
Ill-defined digestive codes/all digestive diseases 17.5 11.8
Ill-defined respiratory codes/all respiratory diseases 7.7 8.7
*Codes ICD-10 I47.2,I49.0,I46,I50,I51.4,I51.5,I51.6,I51.9,I70.9 (Cardiovascular disease)
*Codes ICD-10 C76, C80, C97 (Cancer)
*Codes ICD-10 Y10-Y34 and Y87.2 ( Injury)
*Codes ICD-10 K92, K76.9, K92.0-K92.2 (Digestive)
*Codes ICD-10 J96.9, J98 ( Respiratory)
Percentage of garbage codes* among recorded data (DC) and reclassified IDCD. Brazil, 2010.
15. GPEAS/UFMG
Investigation of ill-defined causes of death in Brazil
Conclusion
The distribution of the cause of death after investigation
of IDCD was different from those in the observed (DC)
data;
There are regional differences;
In order to minimize bias, these differences must be
taken into consideration when redistributing ill-defined
causes;
The investigation of IDCD may provide the basis for a
methodology to be used in the correction of the data and
also leads to an improvement in the quality of the
mortality data.