Comments on: Adult Mortality in East Asia: Trends and Patterns Y' Choi and D' You Noreen Goldman Adu

1
Comments onAdult Mortality in East Asia
Trends and PatternsY. Choi and D. YouNoreen
GoldmanAdult Mortality in Developing
CountriesMarconi Center, CAJuly 9-11, 2004
2
Objectives Strengths

• Identify trends and patterns of adult mortality,
  based on data from China, S. Korea and Taiwan (
  Japan)
• Large time spans of data (beginning in early or
  mid 20th century) with deaths by age and sex
  (generally vital registration) and population
  (census)
• Variant that combines two indirect estimation
  procedures (growth balance synthetic extinct
  generation) and is more robust to error patterns
• Produced a wealth of estimates of
• Completeness of death recording
• Levels trends of adult mortality
• Changes in adult mortality by age sex over time
• Conformity to Coale Demeny regional life tables
  over time
• Sex ratios of age-specific mortality over time
• Adult mortality and sex ratios by major causes
  over time

3
What Is New?

• Rich set of estimates of completeness of death
  recording
• Some interesting findings include the higher
  completeness for Korean males in earlier years
  and what appear to be very incomplete data for
  Taiwanese males during/post WWII
• Far Eastern pattern (see figure)
• Disappearance in Korea in 1980s (although not
  shown in graphs)
• Different pattern in China (no sex difference)
• Excess mortality of young adult males in Korea
  and Taiwan (but not China) since the 1980s that
  resembles earlier pattern in Japan

4
Far Eastern Pattern
Source Age-specific death rates are based on
death registration data for 1971-1975, corrected
for omissions, A.J. Coale, L.J.
Cho and Noreen Goldman, Estimation of Recent
Trends in Fertility and Mortality in
the Republic of Korea. Report prepared for the
Committee on Population and Demography, National
Academy of Sciences, Washington,
1979, adjusted for omission of 21 per cent of
male and 31 per cent of female
deaths.
Goldman, 1980, Population Studies
5
What is Missing?

• Indirect estimation is a complex puzzle that
  requires verification fine tuning
• There is a need to identify integrate the
  relevant literature
• Evaluate goodness of fit of indirect estimation
  techniques do the assumptions appear to be met?
• Comparisons with other sources of data and
  existing literature
• Estimates of completeness do they agree with
  whatever bits are available from other sources?
  This is important information for demographers!
• Coale, Cho Goldman indicate that deaths in
  Korea for 1971-75 were 79 complete for men and
  69 for women. This paper shows 94 and 81 for
  same intercensal period.
• On the other hand, Banister Hill (2004) have
  similar estimates for China (except for females
  1990-2000)

6
What is Missing, cont

• Do estimates of adult mortality agree with
  external estimates? If not, is this due to
  failure to correct data, alternative indirect
  estimation procedures, or what?
• This paper reports adult female mortality
  disadvantage in China in earlier periods, whereas
  Hill and Bannister report higher death rates for
  males throughout 1964-2000. Are estimates in
  general agreement?
• Different pattern for China excess older age
  mortality for both men and women
• This has been noted in various places, including
  UN model life tables Far Eastern pattern, Zhao
  (2003), Campbell, and is not confined to East
  Asian populations

7
What is Missing, cont

• Examination of male-female ratios in asmrs
  demonstrates increase in excess male mortality
  for young adults in Korea and Taiwan see figure
• What does excess mean here since virtually all
  industrialized countries have higher male
  mortality
• Is this pattern unusual? Does it occur outside of
  East Asia? Did it occur in earlier time periods?
• There is considerable data on sex ratios of
  mortality e.g., Zhao (2003), Kruger Nesse
  (2004). These reveal
• substantial variation over time countries
• increases peaks for young adult ages are common
• what seems atypical is high rates at later ages
  in Korea
• Coale Demeny model life tables are probably not
  a very useful standard for recent mortality

8
Male-to-female ratios of the age-specific
mortality rate
Source Choi, Yoonjoung and Danzhen You. Adult
Mortality in East Asia Trends and Patterns.
Presented at the Adult Mortality in
Developing Countries Conference, July 9, 2004,
Marconi Center, CA.
9
Sex ratio of death rates for Korea, 1925-95
Source Zhao, Zhongwei. 2003. On the Far
Eastern pattern of mortality. Population
Studies,57(2) 131-147.
10
Sex ratio of death rates for Taiwan, 1910-90
Source Zhao, Zhongwei. 2003. On the Far
Eastern pattern of mortality. Population
Studies,57(2) 131-147.
11
Sex ratio of death rates for Portugal, 1920-91
Source Zhao, Zhongwei. 2003. On the Far
Eastern pattern of mortality. Population
Studies,57(2) 131-147.
12
Sex ratio of death rates for Singapore, 1957-97
Source Zhao, Zhongwei. 2003. On the Far
Eastern pattern of mortality. Population
Studies,57(2) 131-147.
13
Sex ratio of death rates in 2000 for 20 countries
Source Kruger, DJ and Nesse, RM. 2004. Sexual
selection and the MaleFemale Mortality Ratio
Evolutionary Psychology, 266-85
14
USA 1931 -1995
Source Kruger, DJ and Nesse, RM. 2004. Sexual
Selection and the MaleFemale Mortality Ratio
Evolutionary Psychology, 266-85
15
FRANCE 1946 -1995
Source Kruger, DJ and Nesse, RM. 2004. Sexual
Selection and the MaleFemale Mortality Ratio
Evolutionary Psychology, 266-85
16
JAPAN 1951 -1995
Source Kruger, DJ and Nesse, RM. 2004. Sexual
Selection and the MaleFemale Mortality Ratio
Evolutionary Psychology, 266-85
17
Cause-of Death Data

• The authors are legitimately concerned about the
  potential poor quality of the information,
  especially from Korea in earlier years
• It is not surprising that these 3 causes account
  for much of the excess male mortality in Korea
  since together they account for the majority of
  deaths for these ages. Alas, this is a common
  finding.
• And, given the rapid decline in TB morbidity and
  mortality, it would be difficult to test earlier
  hypotheses.
• Can we learn more from cause-of-death data???