Title: Religious Involvement and Mortality Risk among PreRetirement Aged U.S. Adults Robert A. Hummer Popul
1Religious Involvement and MortalityRisk among
Pre-Retirement Aged U.S. AdultsRobert A.
HummerPopulation Research Center andDepartment
of Sociology,University of Texas at Austin
Presentation prepared for Heritage Foundation
Conference on Religion and Health, December 3,
2008
2Acknowledgements and Citation
- Co-Authors Maureen R. Benjamins, Christopher G.
Ellison, Richard G. Rogers - This paper will appear as Chapter 14 in the
upcoming volume entitled Religion, Families, and
Health Population-Based Research in the United
States, edited by C.E. Ellison and R.A. Hummer.
Rutgers University Press, 2009.
3(No Transcript)
4(No Transcript)
5PREVIOUS LITERATURE, I
- Handbook of Religion and Health (Koenig et al.,
2001, Oxford University Press) provides
comprehensive summary and assessment of the
literature. - 1200 studies over the 20th century 400 of which
are theoretical/review - Many of the studies are cross-sectional, have
poor religion measures, and lack appropriate
controls others, methodologically very weak - Many are at the ecological level (e.g., county)
level - Many are of limited geographic areas, specific
denominational groups - To date, only 3 studies focusing on religious
involvement and mortality risk among individuals
at the national level in the United States - Few national-level data sets that will allow such
investigation!
6Previous Literature, II
- Hummer et al. (1999) in Demography
- Graded association between religious
attendance and adult mortality risk in U.S. - No statistically significant differences in
the association across demographic groups - NHIS data from 1987, mortality links through
1995 - Ages 18-99 at baseline
7Previous Literature, III
- Ellison et al. (2000) in Research on Aging
- Among African Americans, strong and pervasive
risk of non-attendance on mortality among
different demographic subgroups - Associations stronger and graded among younger
(lt55 at baseline) adults association somewhat
weaker and only only non-attenders among older
(55 at baseline) adults - Also used NHIS data from 1987 linked to mortality
risk through 1995 - Ages 18-99 at baseline
8Previous Literature, IV
- Musick et al. (2004), Journal of Health and
Social Behavior - Strong association between non-attendance and
adult mortality riskbut not graded (no
differences between infrequent and frequent
attenders). Other religion measures not related
to mortality risk. - Much stronger relationship between attendance
and mortality risk among younger (lt60 at
baseline) adults than among older (60 at
baseline) adults. - Americans Changing Lives dataset from 1986
mortality links through 1994 - Ages 25 and above
9GUIDING RESEARCH QUESTIONS
- What is the relationship between religious
involvement and adult mortality risk among a
specific birth cohort (1931-1941, ages 51-61 at
baseline) of individuals in the United States? - Is this relationship influenced by controls for
demographic, health, socioeconomic, and
behavioral factors? - Does the religion-mortality relationship among
this cohort vary by sex, race/ethnicity,
education level, and marital status?
10Why Focus on This Age Range?
- Previous evidence suggests that the
religion-mortality relationship may be quite
strong among younger adults (at least in a
relative sense) - Much literature in this area focuses on the
elderly - Deaths in this age range are clearly premature in
the context of current U.S. life expectancy
11CONCEPTUAL FRAMEWORK
Stress
Health Factors
Demographic Factors
Survival Or Die (Probability)
Social Support/Integration
Religious Involvement
Health Behavior
Socioeconomic Factors
Psychological Factors
Health Care
12DATA
- Large, nationally-representative health surveys
of U.S. adults - Health and Retirement Study original cohort
(1992), individuals born 1931-1941 (ages 51-61 at
baseline) - N 9,423, with 834 identified as dying during
follow-up - Data includes information on religious
involvement and important correlates - Statistically linked to death information from
the National Death Index (NDI) and through
follow-up interviews with spouses or other family
member contacts - HRS individuals statistically followed for
mortality risk for 8 years
13Key Variables, I
- Religious Attendance
- Frequent (usually once per week or more 35.3
of sample) 6.0 died during follow-up - Infrequent (usually less than once per week
36.7 of sample) 7.7 died during follow-up - Never (very infrequent or never 28.1 of
sample) 11.1 died during follow-up
14Key Variables, II
- Demographic Controls Age, Gender,
Race/Ethnicity, Region, Religious Denomination - Socioeconomic Controls Education, Marital Status
- Health Controls Self-Rated Health, Activity
Limitations - Behavioral Factors Smoking, Drinking, Exercise
15HAZARD RATIOS ESTIMATING THE RELATIONSHIP BETWEEN
RELIGIOUS VARIABLES AND SUBSEQUENT MORTALITY
RISK, U.S. ADULTS, AGE 51-61, 1992
16Supplementary Analysis, NHIS-LMF (Rogers,
Krueger, Hummer Chapter 15 in forthcoming RUP
Volume)
- NHIS Cancer Risk Factor Supplement from 1987,
with mortality follow-up through end of 2002 - Ages 45-64 at baseline
- N 4,906 individuals, 1,041 of whom were
identified as dying during follow-up - Follow-up exclusively through linkages to the
National Death Index
17Hazard Ratios of Mortality, NHIS-LMF (Ages 45-64
at Baseline)
- Model 1 Model 2 Model 3
- Attendance (gt1 week)
- Never 1.73 1.43 1.35
- Less than once/week 1.27
1.29 1.21 - Once per week 0.97 1.02 1.00
- Controls Demog Demog, Demog,
- SES, Soc, SES, Soc,
- Behavior Behavior,
- Health
-
18Conclusions
- Strong association between religious attendance
and mortality risk among middle-aged U.S. adults - Non-attenders particularly stand out (31-34
higher mortality risk than frequent attenders, in
most completely specified models) upwards of 60
higher risk in less completely specified models - Moderate-to-modestly higher mortality (depending
on model specification) among less frequent
attenders, compared with frequent attenders - No national-level evidence that the
attendance-mortality relationship varies by
gender, race/ethnicity, educational level, or
marital status among this age range of
individuals in the 1931-1941 birth cohort
19CONTINUED CRITIQUES AND NEED FOR RESEARCH
- Literature has relied very heavily on
self-reported public attendance as a measure of
religious involvement. Substantial data needs - Religious life histories
- Cross-national evidence
- Second, some have criticized the quality of
empirical work in the religion-health and
religion-mortality areas, focusing on - The lack of appropriate controls for confounders
(selection into religious attendance/involvement
still a very important concern) - Perception of inconsistent findings
- Third, even greater attention to population
subgroups needed - By race/ethnicity, socioeconomic status, birth
cohort, etc - Fourth, better measurement of mediators needed
- Fifth, more attention to cause of death warranted