Measuring Socioeconomic Inequality in Obesity

Looking Beyond the Obesity Threshold
Marcel Bilger1, Eliza Kruger1, Eric Finkelstein1, 2
1 Signature
Program in Health Services & Systems Research
Duke-NUS Graduate Medical School, Singapore
2 Lien Centre for Palliative Care, Duke-NUS, Singapore
April 17th, 2013
Inaugural Conference of the Singapore
Health Economics Association
Why study socioeconomic inequality in obesity
Our proposed method to measure socioeconomic
Trends in socioeconomic inequality in obesity in the US
Decomposition of socioeconomic inequality according to
gender, race and other factors
Obesity Rates
One of the most prominent health concerns facing
the US is the growing prevalence of obesity
Rates for adults have more than doubled in the past
30 years (CDC 2010)
 66%
of adults are overweight
 34% are obese
Singapore is also affected as obesity rates have
increased from 6% in 1998 to 11% in 2010 (MOH
Distribution of Obesity
Whether the rich or the poor are disproportionally
affected by obesity matters too
If the poor are disproportionally affected:
 This
will contribute to increase health inequality
 The poor will not be able to financially cope with the
increased health risks they face
 Ultimately, this will affect the public health care sector,
through increased subsidies and/or increased social
insurance premiums
Concentration Index (CI)
Measures degree of
socioeconomic inequality in
health (e.g. obesity)
CI<0: the poor are more
affected by obesity
CI>0: the rich are more
affected by obesity
CI is normalized between -1
and +1
Source: Zhang Q, Wang Y (2007)
Foster-Greer-Thorbecke (FGT) measures
( − c)∝
  ≥ c
is the obesity threshold (i.e. BMI>30)
 α is the parameter that sets the sensitivity of the
measure above the threshold
α = 0: Y measures obesity status
α = 1: Y measures depth of obesity
α = 2: Y measures severity of obesity
National Health and Nutrition Examination Survey (NHANES)
Inclusion criteria – age 20 to 65, non-pregnant, data on income
and BMI available
We separately analyze data from 5 time periods (sample size
in brackets);
NHANES 1 1971-1974 (2,986)
NHANES 2 1976-1980 (9,101)
NHANES 3 1988-1994 (12,275)
Continuous NHANES 1999-2004 (8,533)
Continuous NHANES 2005-2010 (7,969)
Variable Definitions
Poverty income ratio
 calculated
by dividing family income by the poverty
lines established by the federal Register
Education (5 levels)
Race (Mexican American, Non-Hispanic White, NonHispanic Black)
Immigration status (born outside the US or not)
Marital status (5 levels)
Concentration Index of Status, Depth and
Severity of Obesity (BMI>30)
Decomposition of Overall Socioeconomic
Inequality in Obesity (2005-2010)
Decomposition of the CI by race
Decomposition of the CI by Gender
• We propose measures of socioeconomic
inequality in obesity that account for excess
weight above the obesity threshold
• In the US, there is no socioeconomic inequality
in the prevalence of obesity anymore
• But the poor are still disproportionally more
severely obese than the rich
• Socioeconomic inequality in obesity greatly
varies according to race, gender, and many
other factors
• Decomposition analysis permits finer analysis
• Poor obese individuals will not be able to
financially cope with the consequences of their
increased health risks
• Socioeconomic inequality thus increases the
burden on the public health sector
• Given the increase in obesity prevalence in
Singapore, a similar analysis would be useful
• The measures we propose are not limited to
obesity but could be applied to any quantitative
variables with a threshold or ceiling
– e.g. hypertension, diabetes