Improved Health and Insurance Status Among Cigarette Smokers After Medicaid Expansion,2011-2016

Abstract

Objectives:

The high concentration of smokers among subgroups targeted by the Affordable Care Act and the historically worse health and lower access to health care among smokers warrants an evaluation of how Medicaid expansion affects smokers. We evaluated the impact of Medicaid expansion on smoking behavior, access to health care, and health of low-income adults, and we compared outcomes of all low-income people with outcomes of low-income current smokers by states’ Medicaid expansion status.

Methods:

We obtained data from the Behavioral Risk Factor Surveillance System (2011-2016) for low-income adults aged 18-64. We estimated multivariable linear ordinary least squares probability models using a quasi-experimental difference-in-difference approach to compare smoking behavior, access to health care, and health between people in expansion states and nonexpansion states and, specifically, on low-income adults and the subgroup of low-income current smokers.

Results:

Compared with low-income smokers in nonexpansion states, low-income smokers in expansion states were 7.6 percentage points (95% confidence interval [CI], 5.7-9.6; P < .001) more likely to have health insurance, 3.2 percentage points (95% CI, 1.3-5.2; P = .001) more likely to report good or better health, and 2.0 percentage points (95% CI, –3.9 to –0.1; P = .044) less likely to have cost-related barriers to care. Health and insurance gains among current smokers in expansion states were larger relative to health gains (1.6 percentage points; 95% CI, 0.5-2.7; P = .003) and insurance gains (4.6 percentage points; 95% CI, 3.5-5.8; P < .001) of all low-income adults in these states.

Conclusions:

Greater improvements among low-income smokers in Medicaid expansion states compared with nonexpansion states could influence future smoking behaviors and warrant longer-term monitoring. Additionally, health and insurance gains among low-income smokers in expansion states suggest the potential for Medicaid expansion to improve health among smokers compared with nonsmokers.

Keywords

smoking smoking cessation insurance coverage Patient Protection and Affordable Care Act Medicaid expansion

In 2010, President Obama signed the Patient Protection and Affordable Care Act into law, with a primary goal of reducing the large number of people lacking health insurance, estimated then at approximately 50 million people.^1,2 The Affordable Care Act contains multiple programs and initiatives aimed at improving the affordability, accessibility, and quality of health care services in the United States. One of the most important components of the Affordable Care Act is the option for states to expand their Medicaid coverage to families with income levels ≤138% of the federal poverty level, with the cost of health insurance expansion to state governments covered 100% by the federal government during the first 3 years (2014-2016), dropping to 95% during the following 3 years (2017-2019), and dropping again to 90% from 2020 and thereafter.^2,3 After the 2012 landmark Supreme Court decision to allow states to opt out of Medicaid expansion, 19 states opted out.^4,5

Medicaid expansion is largely responsible for the unprecedented reduction in the number of uninsured Americans. This reduction, an estimated 20 million people, includes an estimated 15 million newly insured people in expansion states.⁶ Rates of having no insurance in expansion states dropped 26 percentage points compared with a 10 percentage-point drop in rates of having no insurance in nonexpansion states in the 2 years after state Medicaid expansion (2014 and 2015) relative to the 2 years before expansion (2012 and 2013).⁷ Initial evaluations of state Medicaid expansion found substantial improvements in health insurance coverage among vulnerable populations, including those with lower socioeconomic status and those in racial/ethnic minority groups.^7
-9 Overall, evaluations of state Medicaid expansion indicate improvements in access to and use of health care services as a result of increased insurance coverage.^10

-13 Evaluating the impact of changes in insurance coverage, especially increased access to preventive services, among vulnerable populations may identify mechanisms for improving the health outcomes of these populations.^14,15

One subgroup of the low-income population targeted by the Affordable Care Act is cigarette smokers. Cigarette smokers were disproportionately poor and less likely to be insured than their nonsmoking counterparts before passage of the Affordable Care Act.^16,17 Access to health insurance coverage for current smokers increases the use of not only smoking cessation resources but also other preventive health care services.^16,18

Despite smoking rates in the United States dropping from 43% in 1965 to 18% in 2014, exposure to tobacco smoke is the leading preventable cause of death in the United States, and it results in nearly $333 billion in medical expenses and lost productivity annually. Tobacco smoking leads to an increased risk of disease in every organ system, and it is associated with a 10-year shorter life expectancy among smokers compared with nonsmokers.¹⁹ Evaluating the gains in insurance and health among current smokers after state Medicaid expansion can provide information about the ability for Medicaid expansion to improve the high rates of morbidity and mortality among smokers. A 2017 study examined the potential for expanded access to insurance among low-income populations to influence the use of preventive services, rates of cigarette smoking and other risky behaviors, and health; however, the study did not control for state-level tobacco control policies.²⁰ To date, evaluations of how insurance expansion affected the insurance status and health of low-income smokers are lacking. This study compared states that expanded Medicaid with states that did not expand Medicaid, with a primary goal of providing evidence about the impact of health insurance expansion on smoking behavior, access to health care, and health for low-income people in general and current cigarette smokers in particular.

Methods

Data

Data for this study came from the Behavioral Risk Factor Surveillance System (BRFSS),²¹ a state-based, telephone-administered, cross-sectional survey supported by the Centers for Disease Control and Prevention that contains questions about individuals’ health-related behaviors, health care service access and use, and health status.^22,23 The use of random-digit–dialing methods paired with its complex survey weighting design make the BRFSS representative of adults aged 18 or older in the United States. We used responses to the core questionnaire administered through landline telephone and cellular telephone calls to all 50 states and Washington, DC, from 2011 through 2016. Because of a change in the weighting methodology introduced in 2011, it is advised to not compare data before 2011 with data after 2011.²⁴ Analyses in this study used data weighted by year according to the specified BRFSS methodology.

Analytic Sample

We limited the analytic sample to adults aged 18-64. Older adults aged ≥65 may not be affected by the Medicaid expansion because of Medicare eligibility.^13,25,26 Although Medicaid expansion in the Affordable Care Act targets people living at ≤138% of the federal poverty level, the income-related data provided by the BRFSS do not allow such specificity. Those who reported an annual household income of <$25 000 were considered to be low income and were included in the sample population, as in previous studies using BRFSS data.^27

-31 Using the entire low-income sample, we created an analytic subsample of low-income smokers by using the computed smoking status variable in the BRFSS data set. In this subsample, we included adults who responded as “current smoker—now smokes every day” or “current smoker—now smokes some days.” The analytic sample included 390 938 low-income adults and 130 537 low-income smokers.

Outcomes

We examined 4 main outcome measures: health insurance coverage, having good or better health, smoking behavior, and cost-related barriers to receiving health care (ie, needing to see a physician in the past 12 months but being unable to because of cost). Our analysis on smoking behavior among all low-income adults used a smoking behavior outcome that measured whether an adult reported being a current smoker (ie, “current smoker—now smokes every day” or “current smoker—now smokes some days”). For the subgroup analysis of smoking behavior among current smokers, our smoking behavior outcome measured whether the current smoker reported smoking every day. We used everyday smoking as an outcome measure among the current smoker subpopulation to assess changes in smoking intensity among current smokers.

Primary Exposure Variable

The primary exposure variable in our study was the expansion of Medicaid under the Affordable Care Act. The expansion states were states that expanded Medicaid during 2014 or before January 1, 2014.^5,32 We categorized states (and Washington, DC) that expanded Medicaid before January 1, 2014 (Delaware, Massachusetts, New York, and Vermont), as expansion states, as in previous studies.^7,32 Expansion states were Washington, DC, and 26 states: Arizona, Arkansas, California, Colorado, Connecticut, Delaware, Hawaii, Illinois, Iowa, Kentucky, Maryland, Massachusetts, Michigan, Minnesota, Nevada, New Hampshire, New Jersey, New Mexico, New York, North Dakota, Ohio, Oregon, Rhode Island, Vermont, Washington, and West Virginia. Whereas 2 states, Michigan and New Hampshire, did not expand Medicaid until later in 2014, they were included as expansion states, as in previous classifications.³¹ The nonexpansion states included 24 states: Alabama, Alaska, Florida, Georgia, Idaho, Indiana, Kansas, Louisiana, Maine, Mississippi, Missouri, Montana, Nebraska, North Carolina, Oklahoma, Pennsylvania, South Carolina, South Dakota, Tennessee, Texas, Utah, Virginia, Wisconsin, and Wyoming. We conducted sensitivity analyses to assess 2 alternate definitions³³ of expansion and nonexpansion states. In the first sensitivity analysis, we removed states (and Washington, DC) from our analyses that expanded before 2014 (Delaware, Massachusetts, New York, and Vermont), later in 2014 than the January 1, 2014, date (Michigan and New Hampshire), and states that expanded after 2014 (Alaska, Indiana, Louisiana, Montana, and Pennsylvania). In the second sensitivity analysis, we included states (and Washington, DC) that expanded Medicaid before January 1, 2014 (Delaware, Massachusetts, New York, and Vermont), as nonexpansion states.³³ However, results did not differ significantly by using the alternate definitions.

Statistical Analysis

We tabulated the numbers of records from expansion and nonexpansion states before and after Affordable Care Act Medicaid expansion (January 1, 2014). We tabulated the unweighted count of individual records after pooling records for the 3 years before (2011, 2012, and 2013) and the 3 years after (2014, 2015, and 2016) the Medicaid expansion date (Table 1). We also tabulated data on demographic characteristics for the population and subpopulation across the study period.

Table 1.

Number of low-income adults and current smokers aged 18-64 before and after implementation of the Affordable Care Act on January 1, 2014, by state Medicaid expansion status, Behavioral Risk Factor Surveillance System, United States, 2011-2016^a

Expansion Status	Low-Income Adults^b			Low-Income Current Smokers^b,c
Expansion Status	Before^d Implementation	After^d Implementation	Total	Before^d Implementation	After^d Implementation	Total
Expansion states^e	105 386	86 178	191 564	34 739	27 424	62 163
Nonexpansion states^e	114 731	84 643	199 374	40 231	28 143	68 374
Total	220 117	170 821	390 938	74 970	55 567	130 537

^aData source: Behavioral Risk Factor Surveillance System—annual survey data.²¹

^bLow income was defined as an annual household income of <$25 000 to best align with the Affordable Care Act state Medicaid eligibility of an income ≤138% of the federal poverty level.

^cCurrent smokers were adults who responded “current smoker—now smokes every day” or “current smoker—now smokes some days” to the smoking behavior question in the Behavioral Risk Factor Surveillance System survey.

^d“Before” indicates an observation from the time period before the Affordable Care Act state Medicaid expansion date on January 1, 2014 (ie, 2011, 2012, or 2013). “After” indicates an observation from the time on or after the Affordable Care Act state Medicaid expansion date on January 1, 2014 (ie, 2014, 2015, or 2016).

^eExpansion states included Washington, DC, and 26 states: Arizona, Arkansas, California, Colorado, Connecticut, Delaware, Hawaii, Illinois, Iowa, Kentucky, Maryland, Massachusetts, Michigan, Minnesota, Nevada, New Hampshire, New Jersey, New Mexico, New York, North Dakota, Ohio, Oregon, Rhode Island, Vermont, Washington, and West Virginia. Nonexpansion states included 24 states: Alabama, Alaska, Florida, Georgia, Idaho, Indiana, Kansas, Louisiana, Maine, Mississippi, Missouri, Montana, Nebraska, North Carolina, Oklahoma, Pennsylvania, South Carolina, South Dakota, Tennessee, Texas, Utah, Virginia, Wisconsin, and Wyoming.

We conducted a quasi-experimental analysis by using the difference-in-difference approach.³⁴ This approach corrects many of the endogeneity problems encountered in nonrandomized experimental designs, and it is an excellent approach for measuring the impact of policy changes such as the Affordable Care Act on patient health outcomes. It compares the difference in an outcome variable between nonexpansion states and expansion states before and after January 1, 2014, to determine the impact of Medicaid expansion absent external secular trends by using nonexpansion states as a counterfactual.

We used multivariable linear probability ordinary least squares regression to analyze each defined outcome measure, controlling for individual- and area-level factors that have been found to influence smoking behavior and have been used in other Medicaid expansion studies.^7,32,35

-38 Linear probability models are frequently used in analyses using the difference-in-difference approach because of the ability to directly interpret the interaction coefficients.³⁹ We multiplied our difference-in-difference interaction coefficients by 100 to provide easy-to-interpret values; therefore, all difference-in-difference values are presented as percentage points. Covariates were age (18-24, 25-34, 35-44, 45-54, or 55-64), marital status (married, divorced, or other), sex (male or female), race/ethnicity (white, black, Hispanic, or other), metropolitan status (a binary variable equal to 1 if the individual resided in the county containing the center city of a metropolitan statistical area or 0 otherwise), education (<high school, high school, or ≥college), and annual household income (<$15 000 or $15 000-<$25 000), as well as state-level tobacco control policies, including state smoking control appropriations and indoor clean air regulations.

We used smoking, insurance, and health status as covariates when they were not the outcome variable of interest. For state tobacco control appropriations, we used a state’s appropriations in 2014 as a percentage of recommendations by the Centers for Disease Control and Prevention’s Best Practice for Comprehensive Tobacco Control published in 2007, calculated elsewhere.⁴⁰ Specifically, we created a variable to indicate the quartile rank for each state’s appropriation percentage with quartiles equal to 0.0%-7.2%, 7.3%-17.9%, 18.0%-36.0%, and >36.0%. Indoor clean air regulations included state bans on smoking in the private worksite, categorized by type and based on each state’s 2014 regulations.⁴¹ “No provision” included states with no ban on smoking; “banned” included states with a ban on smoking; and “partial ban on smoking” included states with designated areas, separate ventilated areas, and designated smoking areas according to the Centers for Disease Control and Prevention’s tobacco legislation database.⁴¹ We used adjusted Wald χ² tests to test for differences between the difference-in-difference coefficients for the low-income population and for the low-income smoker subpopulation for rates of health insurance coverage, reported health, and cost-related barriers to care. We considered P < .05 to be significant.

An assumption of difference-in-difference analysis is that the trends in the treatment (ie, expansion) and control (ie, nonexpansion) groups are parallel before the intervention and that they will remain parallel if the intervention did not occur.⁴² We created plots for the unadjusted trends in each outcome variable for the expansion and nonexpansion states during the study period, and we used interaction terms between expansion status and a time variable to evaluate the parallel trends assumption in the 3 years before Medicaid expansion (ie, 2011-2013). We determined the rates of each outcome using BRFSS weighting procedures but unadjusted for other variables. We excluded people with missing responses on any of the included variables from the final data set so that analyses included all people from the sample population or subpopulation. We conducted all analyses by using subpopulation survey analyses in STATA version 12.⁴³ We used the centered specification, which uses the population mean rather than the strata mean, to calculate strata variances in strata with only 1 sampling unit. Because this study used publicly available deidentified data, it was considered exempt from institutional review board review according to the University of Arkansas for Medical Sciences.

Results

Adults in expansion and nonexpansion states differed by demographic and health characteristics as well as smoking policies (Table 2). Compared with adults in expansion states, low-income adults in nonexpansion states had a higher likelihood of being black (24.1% vs 14.5%), not receiving health care in the past 12 months because of costs (36.9% vs 29.3%), and lacking health care insurance (58.7% vs 69.1%). Additionally, compared with expansion states, nonexpansion states were less likely to have bans on indoor smoking and had a lower percentage of suggested spending devoted to tobacco control. In expansion states, low-income smokers were less likely than the entire low-income population to complete college (4.9% vs 9.1%), have an annual household income of $15 000-<$25 000 (46.6% vs 44.7%), or have good or better health (63.2% vs 69.6%), and they were more likely to not receive health care in the past 12 months because of costs (34.5% vs 29.3%).

Table 2.

Demographic characteristics of low-income adults and low-income current smokers aged 18-64 and Medicaid expansion status of the state in which they reside, Behavioral Risk Factor Surveillance System, United States, 2011-2016^a

Characteristic	Low-Income Adults^b		Low-Income Current Smokers^b,c
	Expansion States^d (n = 191 564; Weighted n = 142 708 243)	Nonexpansion States^d (n = 199 374; Weighted n = 130 311 197)	Expansion States^d (n = 62 163; Weighted n = 41 594 076)	Nonexpansion States^d (n = 68 374; Weighted n = 43 372 864)
	Weighted % (95% CI)^e	Weighted % (95% CI)^e	Weighted % (95% CI)^e	Weighted % (95% CI)^e
Sex
Male^f	46.6 (46.2-47.0)	44.9 (44.4-45.3)	52.4 (51.6-53.1)	50.2 (49.5-50.9)
Female	53.4 (53.0-53.8)	55.1 (54.7-55.6)	47.7 (46.9-48.4)	49.8 (49.1-50.5)
Metropolitan status
Resided in the county containing the center city of MSA^f	29.4 (29.0-29.8)	20.2 (19.8-20.5)	26.1 (25.4-26.8)	18.8 (18.2-19.4)
Did not reside in the county containing the center city of MSA	70.6 (70.2-71.0)	79.9 (79.5-80.2)	73.9 (73.2-74.6)	81.2 (80.6-81.7)
Age, y
18-24^f	20.2 (19.8-20.6)	19.6 (19.2-19.9)	15.0 (14.4-15.6)	14.1 (13.6-14.7)
25-34	24.3 (24.0-24.7)	23.2 (22.8-23.6)	25.7 (25.0-26.5)	24.8 (24.2-25.5)
35-44	18.5 (18.1-18.8)	18.3 (17.9-18.6)	19.1 (18.5-19.8)	19.4 (18.9-20.0)
45-54	19.0 (18.6-19.3)	19.8 (19.5-20.2)	22.5 (21.9-23.1)	23.1 (22.5-23.7)
55-64	18.1 (17.8-18.3)	19.2 (18.9-19.5)	17.7 (17.2-18.2)	18.5 (18.0-19.0)
Marital status
Married^f	25.9 (25.5-26.2)	26.8 (26.5-27.2)	20.3 (19.7-20.9)	21.7 (21.2-22.3)
Divorced	15.0 (14.7-15.2)	17.2 (16.9-17.5)	19.9 (19.4-20.5)	22.1 (21.6-22.7)
Other	59.2 (58.7-59.6)	56.0 (55.6-56.4)	59.7 (59.0-60.5)	56.1 (55.4-56.8)
Race/ethnicity
White^f	44.0 (43.6-44.4)	49.5 (49.1-50.0)	58.7 (57.9-59.5)	61.5 (60.7-62.2)
Black	14.5 (14.2-14.8)	24.1 (23.8-24.5)	16.6 (15.9-17.2)	21.8 (21.2-22.5)
Hispanic	32.7 (32.2-33.1)	21.3 (20.9-21.7)	17.7 (17.0-18.4)	11.6 (11.0-12.2)
Other	8.8 (8.5-9.1)	5.0 (4.9-5.2)	7.0 (6.6-7.5)	5.1 (4.8-5.4)
Education
<High school diploma^f	29.7 (29.2-30.1)	27.3 (26.8-27.7)	32.0 (31.2-32.8)	33.2 (32.5-34.0)
High school diploma	61.3 (60.9-61.7)	64.6 (64.2-65.1)	63.1 (62.3-63.9)	62.6 (61.9-63.3)
≥College	9.1 (8.9-9.2)	8.1 (7.9-8.3)	4.9 (4.6-5.1)	4.2 (4.0-4.4)
Annual household income, $
<15 000^f	44.7 (44.3-45.2)	39.9 (39.5-40.4)	46.6 (45.8-47.4)	43.5 (42.8-44.2)
15 000-24 999	55.3 (54.8-55.7)	60.1 (59.6-60.5)	53.4 (52.6-54.2)	56.5 (55.8-57.2)
Tobacco control spending as percentage of CDC’s best practices, mean (SE)^g	18.9 (4.0)	15.6 (5.1)	18.6 (8.6)	15.7 (8.7)
Indoor air policy at private worksites^h
No ban on smoking^f	9.8 (9.7-10.0)	43.6 (43.2-44.1)	14.3 (13.9-14.7)	41.0 (40.3-41.7)
Partial ban on smoking	29.1 (28.6-29.5)	22.3 (22.0-22.6)	19.5 (18.7-20.2)	24.4 (23.9-25.0)
Ban on smoking	61.1 (60.7-61.5)	34.0 (33.7-34.4)	66.2 (65.6-67.0)	34.6 (33.9-35.3)
Current smoker	29.2 (28.8-29.5)	33.3 (32.9-33.7)	100.0 (100.0-100.0)	100.0 (100.0-100.0)
Good or better health^f	69.6 (69.2-70.0)	68.5 (68.1-68.9)	63.2 (62.5-64.0)	61.5 (60.8-62.2)
Did not receive health care because of cost	29.3 (28.9-29.7)	36.9 (36.4-37.3)	34.5 (33.8-35.3)	43.2 (42.5-43.9)
Any health insurance^f	69.1 (68.6-69.5)	58.7 (58.3-59.2)	69.0 (68.3-69.8)	55.6 (54.9-56.3)

Abbreviations: CDC, Centers for Disease Control and Prevention; MSA, metropolitan statistical area.

^aData source: Behavioral Risk Factor Surveillance System (BRFSS)—annual survey data.²¹

^bLow income was defined as an annual household income of <$25 000 to best align with the Affordable Care Act state Medicaid eligibility of an income ≤138% of the federal poverty level.

^cCurrent smokers were adults who responded “current smoker—now smokes every day” or “current smoker—now smokes some days” to the smoking behavior question on the BRFSS survey.

^dExpansion states included Washington, DC, and 26 states: Arizona, Arkansas, California, Colorado, Connecticut, Delaware, Hawaii, Illinois, Iowa, Kentucky, Maryland, Massachusetts, Michigan, Minnesota, Nevada, New Hampshire, New Jersey, New Mexico, New York, North Dakota, Ohio, Oregon, Rhode Island, Vermont, Washington, and West Virginia. Nonexpansion states included 24 states: Alabama, Alaska, Florida, Georgia, Idaho, Indiana, Kansas, Louisiana, Maine, Mississippi, Missouri, Montana, Nebraska, North Carolina, Oklahoma, Pennsylvania, South Carolina, South Dakota, Tennessee, Texas, Utah, Virginia, Wisconsin, and Wyoming.

^eValues are percentages weighted to account for the complex survey design of the BRFSS survey, with the exception of tobacco control spending as a percentage of CDC best practices.

^fIndicates the reference category for a covariate variable used in regressions.

^gValues are mean (SE) of total state and federal tobacco control appropriations through 2010 as a percentage of CDC’s recommendations in 2007, calculated elsewhere.⁴⁰ Quartile ranking was used as the variable in the regression, with quartile 1 as the reference category. Quartiles were 0.0%-7.2%, 7.3%-17.9%, 18.0%-36.0%, and >36.0%.

^hState-level ban in 2014 on smoking in the private worksite, categorized by type. No provision = no ban on smoking; banned = ban on smoking; designated areas, separate ventilated areas, and designated smoking areas = partial ban on smoking.⁴¹

The interaction coefficients for the insurance status outcome variables were significant both in the total low-income population and in the low-income smoker population (Table 3). Low-income adults residing in expansion states had a 4.7 percentage-point (95% confidence interval [CI], 3.5-5.8; P < .001) increased probability of having insurance and a 1.4 percentage-point (95% CI, –2.5 to –0.3; P = .015) reduction in cost-related barriers to care compared with low-income adults in nonexpansion states after Medicaid expansion. Current smokers from expansion states had a 7.6 percentage-point (95% CI, 5.7-9.6; P < .001) increased probability of having insurance and a 2.0 percentage-point (95% CI, –3.9 to –0.1; P = .044) reduction in cost-related barriers to care compared with current smokers in nonexpansion states. Low-income adults residing in expansion states had a 1.6 percentage-point (95% CI, 0.6-2.7; P = .003) increased probability of reporting good or better health compared with low-income adults in nonexpansion states, whereas low-income smokers residing in expansion states had a 3.2 percentage-point (95% CI, 1.3-5.2; P = .001) increased probability of reporting good or better health compared with low-income smokers residing in nonexpansion states.

Table 3.

Unadjusted outcome rates and adjusted difference-in-difference coefficients among low-income adults and low-income current smokers aged 18-64 in expansion and nonexpansion states before and after implementation of the Affordable Care Act on January 1, 2014, Behavioral Risk Factor Surveillance System, United States, 2011-2016^a

Outcome Variable	Unadjusted Outcome		Adjusted Difference-in- Difference Coefficient^b (95% CI)	P Value^c
Outcome Variable	Before^d Implementation, %	After^d Implementation, %	Adjusted Difference-in- Difference Coefficient^b (95% CI)	P Value^c
Model for low-income adults^e
Any health care insurance
Expansion states^f	61.9	77.5	4.65 (3.50 to 5.79)	<.001^b
Nonexpansion states	53.5	64.7	0 [Reference]
Current smokers
Expansion states	30.3	27.9	–0.11 (–1.15 to 0.94)	.84
Nonexpansion states	34.5	32.0	0 [Reference]
Good or better health
Expansion states	69.5	69.8	1.61 (0.55 to 2.67)	.003
Nonexpansion states	69.0	67.9	0 [Reference]
Did not receive health care because of cost
Expansion states	33.2	24.8	–1.38 (–2.48 to –0.27)	.015
Nonexpansion states	39.5	33.9	0 [Reference]
Model for low-income current smokers^e,g
Any health care insurance
Expansion states	60.8	79.7	7.63 (5.69 to 9.57)	<.001
Nonexpansion states	50.3	62.0	0 [Reference]
Everyday smokers
Expansion states	71.1	69.8	–0.17 (–2.10 to 1.75)	.86
Nonexpansion states	71.6	70.6	0 [Reference]
Good or better health
Expansion states	63.2	63.4	3.23 (1.30 to 5.17)	.001
Nonexpansion states	62.7	60.0	0 [Reference]
Did not receive health care because of cost
Expansion states	39.1	28.7	–1.97 (–3.88 to –0.05)	.04
Nonexpansion states	745.7	40.1	0 [Reference]

^aData source: Behavioral Risk Factor Surveillance System—annual survey data.²¹

^bWe multiplied difference-in-difference coefficients by 100 to provide percentage-point changes in outcome variables in expansion compared with nonexpansion states.

^cSignificance was determined by using the t test from the interaction term of each difference-in-difference analysis. P < .05 was considered significant.

^eLow income was defined as an annual household income of $25 000 to best align with the Affordable Care Act state Medicaid eligibility of an income ≤138% of the federal poverty level.

^fExpansion states included Washington, DC, and 26 states: Arizona, Arkansas, California, Colorado, Connecticut, Delaware, Hawaii, Illinois, Iowa, Kentucky, Maryland, Massachusetts, Michigan, Minnesota, Nevada, New Hampshire, New Jersey, New Mexico, New York, North Dakota, Ohio, Oregon, Rhode Island, Vermont, Washington, and West Virginia. Nonexpansion states included 24 states: Alabama, Alaska, Florida, Georgia, Idaho, Indiana, Kansas, Louisiana, Maine, Mississippi, Missouri, Montana, Nebraska, North Carolina, Oklahoma, Pennsylvania, South Carolina, South Dakota, Tennessee, Texas, Utah, Virginia, Wisconsin, and Wyoming.

^gCurrent smokers were adults who responded “current smoker—now smokes every day” or “current smoker—now smokes some days” to the smoking behavior question on the Behavioral Risk Factor Surveillance System survey.

The gains in health insurance coverage and in reported health were larger for low-income smokers than for the total low-income population. Although the percentage of insured nonelderly, low-income adults increased more rapidly in expansion states than in nonexpansion states after the expansion date, this improvement was more distinct among the current smoker subpopulation (F = 12.60, P < .001) (Figure). Additionally, we found greater improvement in the health of current smokers in expansion states than in nonexpansion states (F = 3.97, P = .046). Although the baseline rates for each measure between groups differed in the 3 years before Medicaid expansion, the differences were not significant. We found no significant differences in the interaction term between expansion status and a yearly time variable before Medicaid expansion.

Figure.

Unadjusted percentages of outcome variables of low-income adults aged 18-64, by state Medicaid expansion status, Behavioral Risk Factor Surveillance System (BRFSS),²¹ United States, 2011-2016. Current smokers were adults who responded “current smoker—now smokes every day” or “current smoker—now smokes some days” to the smoking behavior question in the BRFSS survey. The vertical black line in each graph indicates the Affordable Care Act Medicaid expansion date of January 1, 2014. Expansion states included Washington, DC, and 26 states: Arizona, Arkansas, California, Colorado, Connecticut, Delaware, Hawaii, Illinois, Iowa, Kentucky, Maryland, Massachusetts, Michigan, Minnesota, Nevada, New Hampshire, New Jersey, New Mexico, New York, North Dakota, Ohio, Oregon, Rhode Island, Vermont, Washington, and West Virginia. Nonexpansion states included 24 states: Alabama, Alaska, Florida, Georgia, Idaho, Indiana, Kansas, Louisiana, Maine, Mississippi, Missouri, Montana, Nebraska, North Carolina, Oklahoma, Pennsylvania, South Carolina, South Dakota, Tennessee, Texas, Utah, Virginia, Wisconsin, and Wyoming. Percentages are weighted according to the BRFSS methodology.

Discussion

We found that state Medicaid expansion was associated with improvements in health insurance coverage, health status, and access to health care services. The findings from this study confirm previous evaluations that found positive effects of Medicaid expansion on health status and health insurance coverage.^13,44 Although insurance rates and health status improved during the study period for the entire low-income population, the subset of low-income smokers in expansion states had relatively greater improvements in health insurance coverage and in reported health status compared with the total low-income population.

Despite improvements in access to health care and in overall health status, smoking behavior did not improve. It is possible that individuals may have had improved access to smoking cessation resources through gains in insurance; however, the difficulty to quit smoking and the relatively low uptake of insurance-covered tobacco cessation resources may have contributed to the lack of improvement in smoking behavior.^13,45
-47 The lack of an obvious impact of Medicaid expansion on cigarette smoking behavior aligns with the conclusions of a previous study.²⁰ Our study adds to this literature by including state-level covariates for tobacco control policies and by providing a separate analysis on insurance coverage, health status, and everyday smoking behavior among low-income current smokers. Continuing evaluation of potential mechanisms that could affect smoking and other health behaviors of current smokers could provide policy makers and public health officials with information to create programs that can most effectively reduce poor health outcomes and high health care costs among this population.

Although our findings suggest no improvement in smoking behavior among those in expansion states compared with nonexpansion states, the greater improvements in health and access to care among low-income smokers in expansion states should lead to additional research. Long-term evaluations of Medicaid expansion in Oregon, which occurred 6 years before state Medicaid expansion under the Affordable Care Act, found decreases in smoking rates among newly insured Medicaid beneficiaries,¹⁸ which suggests the potential for future improvements in smoking behaviors for newly insured people under Affordable Care Act state Medicaid expansions.

Limitations

This study had several limitations. First, changes in the health care environment that could affect smoking and insurance rates outside of the direct expansion of Medicaid could have influenced the outcome variables. For example, states differ in mandatory Medicaid coverage of various smoking cessation treatments, which could affect smoking behaviors.⁴⁸ Second, data from the BRFSS, a patient-reported survey, may not accurately represent smoking behavior, because consumers have an incentive to underreport use based on the Affordable Care Act allowing discrimination based on smoking status. This discrimination is the same in expansion and nonexpansion states, which could limit the impact of the provision on reported smoking status. Additionally, use of the BRFSS did not allow for exact specification of eligibility for Medicaid, which may have biased results. Relatedly, a more fine-grained indication of income level may allow for additional analyses, such as the differential impacts of insurance gains among people who may be eligible for federal insurance premium assistance. Third, individuals may not have been able to change their smoking behavior in such a short time period after Medicaid reform.^13,45 Evaluating smoking habits over time may provide a better indication of changes in risk preferences from increased access to health services. Finally, our study examined rates of insurance status, good or better health, and smoking behaviors aggregated at the state level. State-level aggregated data may not capture the effect of the change in insurance status on an individual’s propensity to smoke some days or every day.

Conclusions

Our study found improved insurance, access to health care services, and health status among low-income smokers in Medicaid expansion states compared with the overall low-income adult population. The differential gain in insurance among low-income smokers indicates the potential for future health benefits from increased access to smoking cessation resources through Medicaid expansion. Improved access to health services and health that could influence smoking behavior warrants continued monitoring.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

References

DeNavas-Walt

Proctor

Smith

. Income, Poverty, and Health Insurance Coverage in the United States: 2009. Current Population Reports P60-238. Washington, DC: US Government Printing Office; 2010. http://www.census.gov/prod/2010pubs/p60-238.pdf. Accessed August 21, 2017.

Pub L No. 111-148, 42 USC §300GG (2010).

Price

Eibner

. For states that opt out of Medicaid expansion: 3.6 million fewer insured and $8.4 billion less in federal payments. Health Aff (Millwood). 2013;32(6):1030–1036.

National Federation of Independent Businesses v. Sebelius , 567 US 519 (2012).

Kaiser Family Foundation. Status of state action on the Medicaid expansion decision. http://www.kff.org/health-reform/state-indicator/state-activity-around-expanding-medicaid-under-the-affordable-care-act/?currentTimeframe=0&sortModel=%7B%22colId%22:%22Location%22,%22sort%22:%22asc%22%7D. Published 2017. Accessed August 21, 2017.

Uberoi

Finegold

Gee

. Health insurance coverage and the Affordable Care Act, 2010-2016. ASPE Issue Brief. http://garnerhealth.com/wp-content/uploads/2014/02/ACA2010-2016.pdf. Published 2016. Accessed August 21, 2017.

Sommers

Gunja

Finegold

Musco

. Changes in self-reported insurance coverage, access to care, and health under the Affordable Care Act. JAMA. 2015;314(4):366–374.

Buchmueller

Levinson

Levy

Wolfe

. Effect of the Affordable Care Act on racial and ethnic disparities in health insurance coverage. Am J Public Health. 2016;106(8):1416–1421.

McMorrow

Long

Kenney

Anderson

. Uninsurance disparities have narrowed for black and Hispanic adults under the Affordable Care Act. Health Aff (Millwood). 2015;34(10):1774–1778.

10.

Angier

Hoopes

Gold

. An early look at rates of uninsured safety net clinic visits after the Affordable Care Act. Ann Fam Med. 2015;13(1):10–16.

11.

Hoopes

Angier

Gold

. Utilization of community health centers in Medicaid expansion and nonexpansion states, 2013-2014. J Ambul Care Manage. 2016;39(4):290–298.

12.

Miller

Wherry

. Health and access to care during the first 2 years of the ACA Medicaid expansions. N Engl J Med. 2017;376(10):947–956.

13.

Sommers

Blendon

Orav

Epstein

. Changes in utilization and health among low-income adults after Medicaid expansion or expanded private insurance. JAMA Intern Med. 2016;176(10):1501–1509.

14.

Tsai

Rosenheck

Culhane

Artiga

. Medicaid expansion: chronically homeless adults will need targeted enrollment and access to a broad range of services. Health Aff (Millwood). 2013;32(9):1552–1559.

15.

Seiler

Malcarney

Horton

Dafflitto

. Coverage of clinical preventive services under the Affordable Care Act: from law to access. Public Health Rep. 2014;129(6):526–532.

16.

Ayanian

Weissman

Schneider

Ginsburg

Zaslavsky

. Unmet health needs of uninsured adults in the United States. JAMA. 2000;284(16):2061–2069.

17.

Hiscock

Bauld

Amos

Fidler

Munafò

. Socioeconomic status and smoking: a review. Ann N Y Acad Sci. 2012;1248(1):107–123.

18.

Bailey

Hoopes

Marino

. Effect of gaining insurance coverage on smoking cessation in community health centers: a cohort study. J Gen Intern Med. 2016;31(10):1198–1205.

19.

US Department of Health and Human Services. The Health Consequences of Smoking—50 Years of Progress: A Report of the Surgeon General. Atlanta, GA: US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health; 2014.

20.

Simon

Soni

Cawley

. The impact of health insurance on preventive care and health behaviors: evidence from the first two years of the ACA Medicaid expansions. J Policy Anal Manage. 2017;36(2):390–417.

21.

Centers for Disease Control and Prevention. Behavioral Risk Factor Surveillance System: annual survey data. https://www.cdc.gov/brfss/annual_data/annual_data.htm. Accessed August 21, 2017.

22.

Mukhtar

Murphy

Mitchell

. Use of data from the Behavioral Risk Factor Surveillance System optional diabetes module by states. J Public Health Manag Pract. 2003;suppl:S52–S55.

23.

Remington

Smith

Williamson

Anda

Gentry

Hogelin

. Design, characteristics, and usefulness of state-based Behavioral Risk Factor Surveillance: 1981-87. Public Health Rep. 1988;103(4):366–375.

24.

Pierannunzi

Town

Garvin

Shaw

Balluz

. Methodologic changes in the Behavioral Risk Factor Surveillance System in 2011 and potential effects on prevalence estimates. MMWR Morb Mortal Wkly Rep. 2012;61(22):410–413.

25.

Lipton

Decker

. The effect of health insurance coverage on medical care utilization and health outcomes: evidence from Medicaid adult vision benefits. J Health Econ. 2015;44:320–332.

26.

Pande

Ross-Degnan

Zaslavsky

Salomon

. Effects of healthcare reforms on coverage, access, and disparities: quasi-experimental analysis of evidence from Massachusetts. Am J Prev Med. 2011;41(1):1–8.

27.

Weissman

Zaslavsky

Wolf

Ayanian

. State Medicaid coverage and access to care for low-income adults. J Health Care Poor Underserved. 2008;19(1):307–319.

28.

Akinyemiju

Jha

Moore

Pisu

. Disparities in the prevalence of comorbidities among US adults by state Medicaid expansion status. Prev Med. 2016;88:196–202.

29.

Benitez

Creel

Jennings

. Kentucky’s Medicaid expansion showing early promise on coverage and access to care. Health Aff (Millwood). 2016;35(3):528–534.

30.

Gonzales

Henning-Smith

. The Affordable Care Act and health insurance coverage for lesbian, gay, and bisexual adults: analysis of the Behavioral Risk Factor Surveillance System. LGBT Health. 2017;4(1):62–67.

31.

Griffith

Evans

Bor

. The Affordable Care Act reduced socioeconomic disparities in health care access. Health Aff (Millwood). 2017;36(8):1503–1510.

32.

Courtemanche

Marton

Ukert

Yelowitz

Zapata

. Early impacts of the Affordable Care Act on health insurance coverage in Medicaid expansion and non-expansion states. J Policy Anal Manage. 2016;36(1):178–210.

33.

Kaestner

Garrett

Gangopadhyaya

Fleming

. Effects of ACA Medicaid expansions on health insurance coverage and labor supply. J Policy Anal Manage. 2017;36(3):608–642.

34.

Wallace

Sommers

. Health insurance effects on preventive care and health: a methodologic review. Am J Prev Med. 2016;50(5 suppl 1):S27–S33.

35.

Yurekli

Zhang

. The impact of clean indoor-air laws and cigarette smuggling on demand for cigarettes: an empirical model. Health Econ. 2000;9(2):159–170.

36.

Siegel

. The effectiveness of state-level tobacco control interventions: a review of program implementation and behavioral outcomes. Annu Rev Public Health. 2002;23(1):45–71.

37.

Agaku

King

Dube

. Current cigarette smoking among adults—United States, 2005-2012. MMWR Morb Mortal Wkly Rep. 2014;63(2):29–34.

38.

Barbeau

Krieger

Soobader

. Working class matters: socioeconomic disadvantage, race/ethnicity, gender, and smoking in NHIS 2000. Am J Public Health. 2004;94(2):269–278.

39.

Karaca-Mandic

Norton

Dowd

. Interaction terms in nonlinear models. Health Serv Res. 2012;47(1 pt 1):255–274.

40.

Kaufmann

Francis

Tynan

Pechacek

. State tobacco revenues compared with tobacco control appropriations—United States, 1998-2010. MMWR Morb Mortal Wkly Rep. 2012;61(20):370–374.

41.

Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health. CDC STATE System tobacco legislation—smokefree indoor air. https://chronicdata.cdc.gov/Legislation/CDC-STATE-System-Tobacco-Legislation-Smokefree-Ind/32fd-hyzc. Accessed August 21, 2017.

42.

Ryan

Burgess

Jr Dimick

. Why we should not be indifferent to specification choices for difference-in-differences. Health Serv Res. 2015;50(4):1211–1235.

43.

StataCorp. STATA [computer program]. Version 12. College Station, TX: StataCorp; 2012.

44.

Blumenthal

Collins

. Health care coverage under the Affordable Care Act—a progress report. N Engl J Med. 2014;371(3):275–281.

45.

Courtemanche

. Rising cigarette prices and rising obesity: coincidence or unintended consequence? J Health Econ. 2009;28(4):781–798.

46.

Cokkinides

Ward

Jemal

Thun

. Under-use of smoking-cessation treatments: results from the National Health Interview Survey, 2000. Am J Prev Med. 2005;28(1):119–122.

47.

Boyle

Solberg

Magnan

Davidson

Alesci

. Does insurance coverage for drug therapy affect smoking cessation? Health Aff (Millwood). 2002;21(6):162–168.

48.

Singleterry

Jump

Lancet

Babb

MacNeil

Zhang

. State Medicaid coverage for tobacco cessation treatments and barriers to coverage—United States, 2008-2014 [published erratum appears in MMWR Morb Mortal Wkly Rep. 2014;63(29):640]. MMWR Morb Mortal Wkly Rep. 2014;63(12):264–269.