Surprising SES gradients in mortality,health, and biomarkers in a Latin American population of adults

1
Surprising SES gradients in mortality,health, and
biomarkers in a Latin American population of
adults

• Luis Rosero-Bixby
• University of Costa Rica
• William H. Dow
• University of California at Berkeley
• Support from the Wellcome Trust

2
Rosero-Bixby, 1993
3
life expectancy vs. gross domestic product
4
Infant Mortality Trend, 1960-1990
5
(No Transcript)
6
(No Transcript)
7
Some Possible Explanations for Good Health

• Health care Good access? Equitable access?
  High quality? Primary health focus? Insurance?
• Public health Clean water? Sanitation, air
  quality?
• Health behaviors Good diet? Smoking, exercise,
  obesity? Modern health beliefs?
• Historical Accident Temperate climate? Genes?
• Social determinants High female education? Low
  poverty? Social equity and inclusion? Low stress
  society?

8
(No Transcript)
9
Costa Rica middle income country, high social
development, strong public sector, advanced
demographic transition
Indicator CR Mexico USA
Per capita income (US2007) 5,560 8,340 46,040
Life expectancy at birth (years, 2006) 79 74 78
Total fertility rate (births) 2.00 2.20 2.04
Population ages 65 () 5.6 5.1 12.3
Seniors with health insurance ( of population 65) 96 62 100

Public health expenditure ( of health expenditures) 78.8 46.4 44.6
Source World Bank
10
SES differentials shed light on good health in CR?

• Does public health lead to less exposure among
  low SES, despite few resources?
• Does health care access buffer effects of
  exposures from low SES?
• Is there a smaller gradient, consistent with
  stress stories?
• First step document differentials

11
Small SES gradients in CR health?

• Research elsewhere finds SES gradients persist
• Into old age (though shrink with age)
• Even with good health care access
• But previous work shows little CR adult mortality
  gradients.
• Is this true of other health indicators?

12
Previous workInsurance and other determinants
of elderly longevity in a Costa Rican panel

• Rosero-Bixby, Dow, and Lacle
• Journal of Biosocial Science
• 2005

13
Mortality data

• Panel of 876 individuals aged 60 in 1984
• Semi urban community near San Jose (100 sample
  from the 1984 census)
• Observed from June 1984 to December 2001
• Interview data from the 1984 census and visits in
  1985 and 1986
• Survival from 1988 and 2002 contacts, and
  computer follow up in the civil registration.

14
Result 3. No clear SES effect
15
New data from CRELES Costa Rican study of
Longevity and Healthy Aging

• National sample of 8,000 born before 1946, from
  the 2000 census.
• 6-year survival follow up
• Sub-sample of 3,000 interviewed in 2005-6
• First wave of a panel (resurvey 2007,2009)
• 90 minute interview and 10 minute diet
• Anthropometry, fasting blood and overnight urine
  samples

16
Study framework 3 levels of health indicators
Level 3
Level 1
Level 2
All indicators are poor-health dummies Control
demography with logistic regression
17

18
Health Outcomes by Age
19
Poor-health biomarkers by age
20
Poor-health lifestyles
21
The low old-age mortality in Costa Rica
challenges the notion of an SES gradient
22
The puzzling SES gradientmortality vs.
self-reported health(controlling for age, sex,
marital)
23
The puzzling SES gradient 2
24
Health-SES gradients from logistic
OR(Controlling age, sex, marital. plt.05)
25
Biomarker-SES gradients from logistic
OR(Controlling age, sex, marital. plt.05)
26
Lifestyle-SES gradients from logistic
OR(Controlling age, sex, marital. plt.05)
27
(No Transcript)
28
Summary

• Flat mortality gradient contrasts other measures.
• Quality of life shows strong gradient.
• CVD is major cause of death, so lack of mortality
  gradients reflects mixed CVD risk factors
• Smoking, low exercise worse for low SES
• Diabetes and hypertension not related to SES
• Cholesterol and obesity worse for high SES (worse
  diets)
• gt Lack of mortality gradient not imply Costa
  Rica has eliminated SES-health gradient

29
Reflects nutritional transition?

• Possible that Costa Rica is early in nutritional
  transition, and SES gradients in
  nutrition-related indicators will flip.
• But external surveys show female obesity (BMIgt25)
  rising for decades
• 1982 56 women age 45-59 overweight
• 1996 75 women age 45-59 overweight

30
What Next?

• New data
• 1984 census-mortality linkage to measure SES
  trends over time.
• Younger cohort 1945-55 birth cohorts.
• Further analyses
• Compare gradients to other countries.
• Rehkopf comparison with U.S. NHANES
• Test if stress has small relation to health.
• Gersten life stressor and neuroendocrine
  allostatic load
• Modrek inequality and health
• Investigate role of health care hypertension.

31
Differences in the association of cardiovascular
risk factors with education a comparison of
Costa Rica (CRELES) and the United States
(NHANES)

• David H. Rehkopf, University of California, San
  Francisco,
• Department of Epidemiology and Biostatistics
• William H. Dow, University of California,
  Berkeley,
• Department of Health Policy and Management
• Luis Rosero-Bixby, Universidad de Costa Rica,
• Centro Centroamericano de Poblacion

32
Objectives of this paper

• Compare risk factor levels across countries
• Compare education gradients in risk
• Inexact education comparison so focus on
  direction of gradients

33
data

• Costa Rica (Costa Rican Healthy Aging Study)
• 2000-2006, n 2827, age 60, 17 outcomes
• education 0-2, 3-6, 7
• United States (National Health and Nutrition
  Examination Survey)
• 1999-2004, n 5607, age 60, 17 outcomes
• education lt12, 12, 12

34
17 outcomes

• behavioral
• current smoking, lifetime smoking, sedentary,
  high saturated fat, high carbohydrates, high
  calorie diet
• Anthropometric
• obese, severely obese, large waist, body mass
  index
• biomarkers
• HDL cholesterol, LDL cholesterol, triglycerides,
  hemoglobin A1c, fasting glucose, C-reactive
  protein, systolic blood pressure

35
(No Transcript)
36
Table 1/Figure 1 Comparing means

• Age and marital distributions roughly similar.
  Education not easily comparable.
• Smoking
• Men Similar.
• Women Lower in CR.
• Diet Comparability concerns, but CR appears
  lower fat, maybe worse other dimensions.
• Obesity
• Men CR much lower
• Women CR only slightly lower.For men, CR much
  lower for women CR only slightly lower than US.
• Hypertension, cholesterol, diabetes
• Men CR lower than US (diabetes same)
• Women CR similar (diabetes higher than US)

37
comparison of means of biological risk factors
for cardiovascular risk factors
38
Costa Rica United
States men women
men women

• Current smoking
• Lifetime smoking
• Sedentary
• Saturated fat
• carbohydrates
• High calorie diet
• obese
• Severely obese
• Large waist

39
Costa Rica United
States men women
men women

• HDL cholesterol
• LDL cholesterol
• Triglycerides
• Hemoglobin A1c
• Fasting glucose
• C-reactive protein
• Systolic blood pressure
• Body mass index

40
Figures 2/3 Education gradients (from
regressions controlling for age)

• Smoking
• Males gradient both US and CR
• Females gradient only in CR
• Diet High calorie
• Reverse gradient especially in CR
• Obesity
• Males in CR have reverse gradient.
• Females have expected gradient (both US, CR).
• Cholesterol
• HDL only US women have expected gradient
• LDL CR men have gradient
• Blood pressure gradient only in CR men
• HbA1c Expected gradients, except none in CR men
• C-reactive protein Expected gradient in US, but
  none in CR

41
Summary

• Mixed gradients tell complex story, raise more
  questions.
• C-reactive protein why no CR gradient?
  (obesity, or buffers?)
• Obesity worrisome in CR
• Women already close to US levels.
• Male reverse gradient low SES may rise next.
• Next steps
• Study time trends in mortality by SES and cause
  of death (1984 census-mortality linkage)
• Examine treatment for hypertension, cholesterol,
  diabetes. Why are levels so high when medicines
  can help control? Why are there gradients in
  (male) blood pressure and LDL control in CRs
  vaunted system?

42
Education differentials in coronary heart disease
mortality among those 60 and older

• Costa Rica United States

43
CRELES all ages, all-cause
44
Life stressors and neuroendocrine allostatic load
in Costa Rica

• Omer Gersten, Ph.D.
• Academia Sinica
• Population Association of America
• Detroit, Michigan
• May 1, 2009

45
Research question

• Is greater AL predictive of worse health
  outcomes?
• Are various indicators of life stress linked to
  greater AL?
• Year 2004-6
• Earlier life history/ ------------------------gt
  Biomarker
• Current situation collection/survey

46
Research hypothesis

• Early life events
• low edu. of mother
• live w/out biological father
• econ. problems (index)
• health problems (index)
• Loss
• death of children
• widowhood/years widowed 2004-2006
• Social deprivation ---------------------gt High
  AL
• low/no church attendance
• lives alone
• Spousal characteristics
• low edu.
• poor health
• Demographic
• low edu.
• rural residence
• Economic
• low household wealth

47
Data dependent variable

• Neuroendocrine allostatic load (NAL)
• Biomarkers Physiologic sub-systems
  Physiologic system
• Epinephrine ----------gt Sympathetic nervous
  
• Norepinephrine system (SNS)
  
• ----gt Neuroendocrine
• Cortisol ----------------gt Hypothalamic-pituitary
  -
• DHEAS adrenal (HPA) axis
• Epi., norepi., cortisol initiate bodys most
  immediate stress response
• Survey measures resting, nonstressed levels

48
(No Transcript)
49
Conclusions

• Q Are early and other negative life events
  linked to riskier neuroendocrine allostatic load
  (NAL) levels?
• A No.

50
Gracias!

• Visit the CRELES web pages
• http//ccp.ucr.ac.cr