Nativity and Racial/Ethnic Inequalities in Smoke-free Policy, in Support for Smoke-free Policy, and in State Patterns of Tobacco Use
Theresa L. Osypuk, SD SM Assistant Professor, Northeastern University June 9, 2009 TUS-CPS Data Users Workshop
1
Acknowledgements
o My collaborators at Harvard School of Public Health
o Dolores Acevedo-Garcia o Ichiro Kawachi o SV Subramanian
o Funding from American Legacy Foundation, Association of Schools of Public Health, and Robert Wood Johnson Foundation Health & Society Scholars program
Theresa L. Osypuk 2
Agenda
o Why I use TUS-CPS for inequality analyses
o Illustrate the use of the TUS-CPS to examine tobacco control inequality research questions with 4 examples from my own work
• Nativity (immigrant) and racial/ethnic differences in smoking patterns, and related to tobacco-control related policy
Theresa L. Osypuk 3
Why use the TUS-CPS for nativity inequality analyses?
4
Why use the TUS-CPS for nativity inequality analyses?
o TUS-CPS has better information on immigration related variables and tobacco use than any other survey
o TUS-CPS is a large representative survey
o TUS-CPS also has good information on demographic and socioeconomic control variables
Theresa L. Osypuk 5
Immigration variables in the CPS (since 1994)
o Country of origin of the sample person o Country of origin of his/her mother and father o Citizenship status o Year of entry into the United States o Interview language
Theresa L. Osypuk 6
Example 1: What are the patterns of daily smoking, by immigrant generation & assimilation?
Acevedo Garcia, Pan, Jun, Osypuk, Emmons (2005). “The Effect of Immigrant Generation on Smoking”. Social Science and Medicine, 61(6): 1223-1242. 7
Significance: Smoking patterns by nativity
o Demographic Imperative
o Epidemiologic Paradox
Theresa L. Osypuk
Acevedo Garcia, Pan, Jun, Osypuk, Emmons (2005). “The Effect of Immigrant Generation on Smoking”. Social Science and Medicine, 61(6): 1223-1242. 8
Immigrants are less likely to smoke than US-born, with protective effect in 2nd generation.
Daily Smoking Prevalence by Immigrant Generation, 1995/96 TUS-CPS.
20.7% Gen 3: U.S. Born 20%
16.9% 16.2% Gen 2: US Born, FB Mom 15% Gen 2: US Born, FB Dad
11.0% Gen 2: US Born, FB 9.5% 10% Parents Gen 1: Foreign Born
5%
0%
9 Theresa L. Osypuk Acevedo Garcia, Pan, Jun, Osypuk, Emmons (2005). “The Effect of Immigrant Generation on Smoking”. Social Science and Medicine, 61(6): 1223-1242..
puk
Patterns persist after adjustment; generational gradient in smoking, foreign born is protective for smoking
Relative Odds of Daily Smoking by Immigrant Generation. Multiple Logistic Regression, 1995/96 TUS
1.05
Odd
s R
atio
of D
aily
Sm
okin
g (G
en 3
R
EF)
0.95
0.85
0.75
0.65
0.55
0.45
0.96 0.98
0.70 *
Gen 3: US Born (REF)
Gen 2: US Born, FB Mom 0.78 * Gen 2: US Born, FB Dad
0.74 * Gen 2: US Born, FB Parents 0.58 * Gen 1: Foreign Born
0.47 * 0.40 *0.35 * p<.05 Adjusted Crude
(demogr. & SES)
10
Acevedo Garcia, Pan, Jun, Osypuk, Emmons (2005). “The Effect of Immigrant Generation on Smoking”. Social Science and Medicine There, 61(6): 1223-1242.sa L. Osy
Example 2: Is Workplace Smoking Policy Equally Prevalent and Equally Effective for Immigrants?
Osypuk, Subramanian, Kawachi, and Acevedo-Garcia. (In Press, 2009). “Is Workplace Smoking Policy Equally Prevalent and Equally Effective for Immigrants?” Journal of Epidemiology & Community Health.
11
The proportion of indoor workers in smokefree workplaces has increased substantially
Smoke-free Workplace Prevalence: U.S. 1993-2003 Among Indoor Workers, TUS
80%
% in
door
wor
kpla
ces
smok
efre
e
77% U.S. 70% 71%69%
63%60%
50% 46%
40%
30%
1993 1996 1999 2002 2003
Theresa L. Osypuk Source: NCI 2006, TUS-CPS
12
Significance: What do we know about smoking bans & smoking ban coverage? o Smoking bans prevent ETS exposure, and reduce
smoking among smokers
o Tobacco use regulations do not protect everyone equally.
o Uneven coverage due to voluntary passage, tobacco control patchwork at multiple levels
Theresa L. Osypuk 13
Significance: Why examine inequalities among immigrants?
o Demographic imperative o Immigrant status (nativity) is often conflated with
race/ethnicity o Immigrants disproportionately in low-wage and
low-skilled jobs o Immigrants less likely to enjoy workplace benefits
Theresa L. Osypuk 14
Are immigrants equally covered by workplace smokefree policies?
15
Immigrants are less likely to be covered by a workplace smokefree policy, Hispanic men particularly so. (2001/02)
76%
65% 69%
64% 66%
72%71% 66%
58%
72% 75% 73%
81%
72% 76%
67%
0%
10%
20%
30%
40%
50%
60%
70%
80%
White Black Asian Hispanic % o
f ind
oor e
mpl
oyee
s in
sm
okef
ree
wor
kpla
ces
US-Born men
Foreign-Born men
US-Born women
Foreign-Born women
US Ave 70.5%
16
Osypuk, Subramanian, Kawachi, and Acevedo-Garcia. (In Press, 2009). “Is Workplace Smoking Policy Equally Prevalent and Equally Effective for Immigrants?” Journal of Epidemiology & Community Health. Theresa L. Osypuk
Although immigrants less likely to be in smokefree workplaces, disparity accounted for by occupation & industry
O
dds
Rat
io o
f For
eign
Bor
nvs
. US-
born
Multiple Logistic Regression, Odds of Working in Smokefree Workplace, Foreign Born vs. US-Born (01/02 TUS)
1.20
1.15 1.02 1.02 1.031.10
0.90 * 1.05
1.00
0.95
0.90 Crude Adjusted for Adjusted for Fully Adjusted
0.85 Occupation Industry
17
Osypuk, Subramanian, Kawachi, and Acevedo-Garcia. (In Press, 2009). * p<.05“Is Workplace Smoking Policy Equally Prevalent and Equally Effective for Immigrants?” Journal of Epidemiology & Community Health. Theresa L. Osypuk
Conclusions & Implications
o Industry and occupation are important mediators for why immigrants (or other groups) have lower coverage by a smokefree policy
o Policies often exempt jobs with higher concentrations of minorities
o A voluntary policy regime will be less effective for covering certain minority groups because occupational segregation will induce systematic disparities.
Theresa L. Osypuk 18
Do workplace smoking bans have equal associations with smoking among immigrants and the US-born?
19
The effect of non-smokefree workplaces with current smoking is weaker for immigrants
Multiple Logistic Regression, Odds of Smoking, Comparing Non-Smokefree to Smokefree Workplaces
2.00 1.83 *
1.80 * p<.05
Crude Fully Adjusted
1.60 1.39 *1.36 * 1.40
1.15 * 1.20
1.00 US Born Foreign Born
Odd
s R
atio
of S
mok
ing
Osypuk, Subramanian, Kawachi, and Acevedo-Garcia. (In Press, 2009). “Is Workplace Smoking Policy Equally Prevalent and Equally Effective for Immigrants?” Journal of Epidemiology & Community Health. Theresa L. Osypuk 20
Conclusion
o Immigrants exhibited weaker associations between workplace smokefree policy & smoking
o Why?• Differential industries accounted for 16% of the
weaker policy effects among immigrants• Workplace smoking policies may be more effective for
those with higher smoking prevalence/consumption • Other causes may be more important for smoking
among immigrants
Theresa L. Osypuk 21
Example 3: Who doesn’t support tobacco control policy?
Osypuk and Acevedo-Garcia. (In Press) American Journal of Public Health. “Who Doesn’t Support Smokefree Policies? A Nationwide Analysis of Immigrants, Native Born, and Other Demographic Groups 1995-2002”.
22
Significance: Support for Smoking Bans
o The population’s opinions about where smokingshould be allowed are a general indicator ofsupport for tobacco control policy
o Public opinion• one catalyst for enacting tobacco control policy• e.g. state ballot initiatives/referenda
o Knowledge of support by demographic group mayaid advocates to identify voter constituencies forcoalition building in election strategies
Osypuk and Acevedo-Garcia. (In Press) American Journal of Public Health. “Who Doesn’t Support Smokefree Policies? A Nationwide Analysis of Immigrants, Native Born, and Other Demographic Groups 1995-2002”. Theresa L. Osypuk 23
Variables: Support for Smoking Bans
o Outcome: summary measure of support for total smoking bans in 6 venues • Individual questions asked whether smoking should be allowed
in “all areas, some areas, or not allowed at all” in (1) restaurants, (4) bars and cocktail lounges, (2) hospitals, (5) indoor sports venues, (3) indoor work areas, (6) indoor shopping malls.
• Support for smoking ban based on answering that smoking should “not be allowed at all” for each venue
• Summary measure of support created by summing # of venues out of 6 where a person supports a ban, & dichotomizing.
o Strong support = support for banning smoking in 4 of 6 venues (based on Gilpin et al. 2004).
Theresa L. Osypuk 24
Variables
o Main independent variable: Immigrant generation• 1st generation (foreign born);• 2nd generation (US-born of foreign-born parents);• 3rd generation (US-born of US-born parents)
Theresa L. Osypuk 25
Who is less likely to support smoking bans? (95-02 TUS)
% supporting a smoking ban in 4 of 6 venues
U.S. Total
1995/9
6
1998/9
9
2001/0
2
1stGen Im
m(FB)
2nd Gen
3rdGen
(US-born)
FB, Non Citizen
FB, Naturaliz
ed Citizen
US Born, Citiz
en
60%
75%76%
59% 66%
76%
55% 62% 61%
69%
30%
40%
50%
60%
70%
80%
% s
uppo
rtin
g ba
n
Osypuk and Acevedo-Garcia. (In Press) American Journal of Public Health. “Who Doesn’t Support Smokefree Policies? A Nationwide Analysis of Immigrants, Native Born, and Other Demographic Groups 1995-2002”. Theresa L. Osypuk 26
Who is less likely to support smoking bans? (95-02 TUS)
% supporting a smoking ban in 4 of 6 venues, 1995-2002 TUSCPS
U.S. T
otal
NHwhit
eNH
indian
NH
black
NH
APIHisp
anic
Male
Female
Nev
erSmok
er
Form
erSm
oker
Curre
ntSmok
er
ome no
t smkfr
ee
Smokefr
eeho
me W
ork no
t smkfr
ee
Smokefr
eewor
k
74%
66%
72%
62%
34% 39%
76%
59%
68%
75%
57%59%62%
57% 61%
25%
35%
45%
55%
65%
75%
% s
uppo
rtin
g ba
n
HOsypuk and Acevedo-Garcia. (In Press) American Journal of Public Health. “Who Doesn’t Support Smokefree Policies? A Nationwide Analysis of Immigrants, Native Born, and Other Demographic 27Groups 1995-2002”. Theresa L. Osypuk
Support for smoking bans increased for all groups w/time; 3rd generation/US born least likely to support & foreign born most likely
Support for smoking bans by generation across time
gen
3
gen 2,
1parF
B
gen 2,
2parF
Bgen
1
gen 3
gen2,
1parF
B
gen2,
2parF
Bgen
1
gen 3
gen2,
1parF
B
gen 2,
2parF
Bgen
1
gen 3
gen 2,
1parF
B
gen2,
2parF
Bgen
1
90% 80% 70% 1996 60% 1999 50% 2002 40% 30% 20% 10% 0%
% s
uppo
rtin
g ba
n
Summary 4 of 6 Bars Work Hospitals
Theresa L. Osypuk 28
What accounts for stronger smoking ban support among 1st & 2nd generation immigrants vs. 3rd generation?
Odds of Support for Smoking Ban in 4 of 6 Venues, MultipleLogistic Regression Results, TUS 95-02
Odd
s R
atio
of S
mok
ing
2.162.20 2.00 1.80 1.67
1.501.60 1.39 1.31 1.40
1st generation
2nd generation
3rd generation (REF)
1.20 1.09 1.03 1.00
1.061.11
1.32
Model 1 Model 2 Model 3 Model 4 Model 5 Crude +Demogr & +Smoking +Work & +Cali
SES home ban
Theresa L. Osypuk 29
Osypuk and Acevedo-Garcia. (In Press) American Journal of Public Health. “Who Doesn’t Support Smokefree Policies? A Nationwide Analysis of Immigrants, Native Born, and Other Demographic Groups 1995-2002”.
Example 4: Are state patterns of smoking different for different racial/ethnic groups?
Osypuk, Kawachi, Subramanian, Acevedo-Garcia. (2006) Public Health Reports, “Are State Patterns of Smoking Different for Different Racial/Ethnic Groups? An Application of Multilevel Analysis.” 121(5):563-577. 30
Significance: Why look at state-specific patterns of smoking?
o Descriptive analysis• Surveillance
o Prioritizing resources for programs and policy
• Implicates state level causal explanations
• Influential states may drive national prevalence
• Etiologic hypothesis generation Theresa L. Osypuk 31
Analysis: Multilevel Modeling
o 2-level multiple logistic regression predictingcurrent smoking• Predicted smoking prevalence from state-level
residuals, from multilevel multiple logistic regressionmodels controlling for covariates
• Markov Chain Monte Carlo estimation o To adjust for complex survey design
o Mapped state predicted probability of smokingfor each race/gender group• Relative adjusted (regression-based) estimates
significantly higher or lower than the nationalprevalence
Theresa L. Osypuk 32
33
NH White Men Adjusted Smoking PrevalenceRelative Map: 1995-96 CPSTUS Residuals 4-Aug 05
Insufficient Data (0)
>5 pts lower than national: from 0.3412 to 0.3911 (0)
1-5 pts lower than national: from 0.3912 to 0.4400 (0)
No different from national: 0.4412 (44)
1-5 pts higher than nationa : from 0.4501 to 0.4911 (7)
>5 pts h gher than nat onal: from 0.4912 to 0.5 (0)
NH Black Men Adjusted Smoking PrevalenceRelative Map: 1995-96 CPSTUS Residuals 4-Aug 05
Insufficient Data (0)
>5 pts lower than national: from 0.3383 to 0.3882 (0)
1-5 pts lower than national: from 0.3883 to 0.4300 (0)
No different from national: 0.4383 (50)
1-5 pts higher than nationa : from 0.4401 to 0.4882 (0)
>5 pts h gher than nat onal: from 0.4883 to 0.5 (1)
Hispanic Men Adjusted Smoking PrevalenceRelative Map: 1995-96 CPSTUS Residuals 4-Aug 05
Insufficient Data (0)
>5 pts lower than national: from 0.2329 to 0.2828 (0)
1-5 pts lower than national: from 0.2829 to 0.3300 (1)
No different than national: 0.3329 (50)
1-5 pts higher than nationa : from 0.3401 to 0.3828 (0)
>5 pts h gher than nat onal: from 0.3829 to 0.50 (0)
NH White Women Adjusted Smoking PrevalenceRelative Map: 1995-96 CPSTUS Residuals 4-Aug 05
Insufficient Data (0)
>5 pts lower than national: 0.23 to 0.3270 (2)
1-5 pts ower than national: 0.3271 to 0.3700 (2)
Not s gnificantly different from nat onal .3767 (41)
1-5 points higher than nat onal: 0.3801 to 0.427 (6)
>5 points h gher than nat onal: 0.4270 to 0.5000 (0)
NH Black Women Adjusted Smoking PrevalencRelative Map: 1995-96 CPSTUS Residuals 4-Aug-05
Insufficient Data (0)
>5 pts lower than national: from 0.1900 to 0.2298 (4)
1-5 pts lower than national: from 0.2299 to 0.2700 (2)
Not different from national: 0.2799 (44)
1-5 pts higher than nat onal: from 0.2801 to 0.3299 (0)
>5 pts higher than nat onal: from 0.3300 to 0.5 (1)
Hispanic Women Adjusted Smoking PrevalenceRelative Map: 1995-96 CPSTUS Residuals 4-Aug 05
Insufficient Data (0)
>5 pts lower than nat onal: from 0.18 to 0.1996 (1)
1-5 pts lower than national: from 0.1997 to 0.24 (0)
No d fferent from national: 0.2497 (49)
1-5 pts higher than national: from 0.2501 to 0.2996 (0)
>5 pts h gher than national: from 0.2997 to 0.3100 (1)
Theresa L. Osypuk
White
Black
Hispanic
-
l
i i
-
l
i i
-
l
i i
-
l
i i
i
i i
e
i
i
-
i
i
i
Women Men55++ ppooiinnttss hhiigghheerr 0-5 points higher No different from U.S. rate 0-5 points lower 55++ ppooiinnttss lloowweerr
Adjusted State Smoking Prevalence Relative to U.S. Group Average (95/96 TUS)
Osypuk, Kawachi, Subramanian, Acevedo-Garcia. (2006) Public Health Reports, “Are State Patterns of Smoking Different for Different Racial/Ethnic Groups? An Application of Multilevel Analysis.” 121(5):563-577
Conclusions
o A race-specific pattern of smoking among the 50 states
o State tobacco variables (taxation and agriculture) did not account for remaining state smoking variance.
o Descriptive regression-based mapping may be valuable for place-patterned tobacco use surveillance
Osypuk, Kawachi, Subramanian, Acevedo-Garcia. (2006) Public Health Reports, “Are State Patterns of Smoking Different for Different Racial/Ethnic Groups? An Application of Multilevel Analysis.” 121(5):563-577
Theresa L. Osypuk 34
Citations
Osypuk and Acevedo-Garcia. (In Press) “Who Doesn’t Support Smokefree Policies? A Nationwide Analysis of Immigrants, Native Born, and Other Demographic Groups 1995-2002”. American Journal of Public Health.
Osypuk, Subramanian, Kawachi, and Acevedo-Garcia. (In Press, 2009). “Is Workplace Smoking Policy Equally Prevalent and Equally Effective for Immigrants?” Journal of Epidemiology & Community Health. Published online first 8 April 2009.doi: 10.1136/jech.2008.079475.
Osypuk, Kawachi, Subramanian, Acevedo-Garcia. (2006) “Are State Patterns of Smoking Different for Different Racial/Ethnic Groups? An Application of Multilevel Analysis.” Public Health Reports, 121(5): 563-577.
Acevedo Garcia, Pan, Jun, Osypuk, Emmons (2005). “The Effect of Immigrant Generation on Smoking”. Social Science and Medicine, 61(6): 1223-1242.
Theresa L. Osypuk 35
Thank you!
36