Social Gradients in Health in Indonesia

1
Social Gradients in Health in Indonesia

• Presented by Pam Stoddard
• June 5, 2006
• CHS 286 Doctoral Roundtable

2
Background

• Social gradient
• Higher SES better health at each level
  of the socioeconomic ladder
• Extends to health behaviors and health outcomes
  like blood pressure

3
Background cont.

• Health in developing countries may follow a
  reverse social gradient
• Some health indicators may worsen rather than
  improve as SES rises

4
Previous findings

• Limited comparative research
• Numerous single-country studies show reverse
  gradients (esp. nutrition-related)
• SES-body mass index ()(Sobal and Stunkard 1989
  The 1994 Colhoun, Hemingway et al. 1998 Adair
  2004 Islam, Akhtaruzzaman et al. 2004 Griffiths
  and Bentley 2005 Xu, Yin et al. 2005)
• SES-raised blood pressure () (Siegrist,
  Bernhardt et al. 1990 Singh, Niaz et al. 1998
  Sayeed, Banu et al. 2002).
• Smoking follows patterns in developed countries
• SES-Current smoker (-) (Singh, Niaz et al. 1998
  Pan 2004)
• SES-number of cigs per day (mixed)

5
Aims of Study

• Examine overall pattern of SES-health gradients
  in Indonesia
• Examine specific gradients for
• Smoking
• Body mass index (including lower cutpoints)
• Blood pressure
• Chest pain
• Explore interactions between SES and
  socio-demographic characteristics

6
Hypotheses

• Hypothesis 1 SES will have an inverse
  relationship with smoking
• Hypotheses 2-4 SES will be positively related to
  body mass index, blood pressure, and chest pain
  (reverse gradient)

7
Social mechanisms

• Smoking and SES
• Individual, interpersonal, organizational factors
  may mean low SES more likely to smoke
• Perceived control over health, future
  orientation, awareness may make high SES less
  likely to smoke

8
Social mechanisms

• Body mass index
• Healthy diets currently more affordable and
  available in Asia
• Processed foods more costly
• Poor less sedentary
• Westernized diet and sedentary are privileges of
  the wealthy

9
Conceptual Framework (Kim et al, 2005)
10
Data

• Indonesian Family Life Survey
• Longitudinal household survey
• Representative of 83 of Indonesian population in
  13 of 27 provinces
• Multistage, clustered sample design
• Attempted to interview all members of the
  original households
• In split-off households, any person who had been
  an original IFLS1 household member, their spouse
  and biological children

11
Data for this study

• IFLS Wave 3 (2000)
• Cross-sectional analysis
• Analytic sample adults aged 35 and older with
  complete information on all variables
• N 11,754 for smoking, body mass index, and
  blood pressure
• N 5,534 for chest pain (adults gt50)

12
Measures Dependent variables

• Smoker (binary)
• Currently smoke cigarettes or cigars
• Non-smokers includes former smokers, pipe
  smokers, tobacco chewers
• Body mass index (ordinal)
• Underweight (lt18.5)
• Normal (gt18.5 and lt25.0)
• Pre-obese/obese/very obese (gt25.0)

13
Measures Dependent variables

• Blood pressure (ordinal)
• Normal systolic of less than 120 and diastolic
  less than 80
• Prehypertension systolic of 120-139 or diastolic
  of 80-89
• Hypertension systolic of 140 or higher or
  diastolic of 90 or higher
• Chest pain (binary)
• ever felt chest pain

14
Measures Independent variables

• Main predictors
• Years of completed education
• Per capita logged household expenditures
• Sociodemographic variables (controls)
• age, sex, urban/rural, religious affiliation,
  province
• Interactions

15
Analyses

• Weighted
• univariate and bivariate frequencies
• used IFLS3 individual-level sample weights to
  provide population estimates and account for
  complex sample design
• Unweighted
• binomial and ordinal logistic regression
• included variables for sampling stratification
  urban/rural and province

16
(No Transcript)
17
(No Transcript)
18
(No Transcript)
19
(No Transcript)
20
(No Transcript)
21
(No Transcript)
22
(No Transcript)
23
(No Transcript)
24
Limitations

• Cross sectional data
• Some self-reports
• Endogeneity

25
Conclusions

• Only education-smoking gradient (not expenditure)
  follows same pattern in Indonesia as in developed
  countries
• Positive gradient (both SES indicators) for body
  mass index - opposite of pattern in developed
  countries
• No gradient for blood pressure (both SES
  indicators) does not follow developed pattern
  
• SES-gradients mixed for chest pain

26
Future directions

• Health patterns can set groundwork for research
  on underlying mechanisms
• What explains reverse gradients?
• Why do the advantaged forgo some unhealthy
  behaviors and not others?
• Short-term transition vs. long-term pattern

27
Policy implications

• Health implications chronic disease may rise as
  education and household resources rise
• Educational expansion may not automatically
  translate to health improvements (for chronic
  disease)
• Focus on subpopulations

28
Thank you for your feedback

• pstodda_at_ucla.edu