Disparities in Smoking and Heavy Drinking Behaviors by Disability Status and Age of Disability Onset: Secondary Analysis of National Health Interview Survey Data

Abstract

Objectives:

People with childhood-onset disabilities are living into adulthood, and the prevalence of smoking and illicit drug use among adults with disabilities is high. We evaluated the relationship between disability status and age of disability onset, current cigarette smoking status, and heavy alcohol drinking.

Methods:

We conducted a secondary data analysis of the National Health Interview Survey (NHIS), a US survey on illness and disability. Among 2020 NHIS participants aged 22-80 years (n = 28 225), we compared self-reported prevalence of current cigarette smoking and heavy alcohol drinking among those with and without disabilities and among those with childhood- versus adult-onset disabilities. We used adjusted logistic regression analysis to calculate the adjusted odds ratios (AORs) of current smoking and heavy alcohol drinking based on disability status and age of disability onset.

Results:

Compared with adults without disabilities, adults with disabilities were significantly more likely to report current smoking (23.5% vs 11.2%; P < .001) and significantly less likely to report heavy alcohol drinking (5.3% vs 7.4%; P = .001). The prevalence of these behaviors did not vary significantly by age of disability onset. In adjusted logistic regression models, adults with disabilities had significantly higher odds of current smoking (AOR = 1.76; 95% CI, 1.53-2.03) and similar odds of heavy alcohol drinking (AOR = 0.82; 95% CI, 0.65-1.04) compared with adults without disabilities. The odds of these health behaviors did not vary significantly by age of disability onset.

Conclusions:

Adults with disabilities overall may be at high risk for these unhealthy behaviors, particularly smoking, regardless of age of disability onset. Routine screening and cessation counseling related to smoking and unhealthy alcohol use are important for all people with disabilities.

Approximately 85.3 million people living in the United States, or 27.2% of the population, have a disability. ¹ A growing proportion of people with disabilities includes individuals with childhood-onset disabilities who are living into and thriving in adulthood. ² However, young adults with disabilities who are transitioning from the pediatric to the adult health care system often lack age-appropriate health care services. ³ This period is a particularly vulnerable time for health outcomes that are affected by decreased access and adherence to appropriate treatments and services. ⁴ Long-term smoking and heavy alcohol drinking also often develop during adolescence and young adulthood and are associated with poor health outcomes.^5,6 Some studies have shown that, compared with adults without disabilities, adults with disabilities have a similar or higher prevalence of cigarette smoking and illicit drug use but a lower prevalence of heavy alcohol drinking.^7-9 However, few studies have focused on substance use among adults with childhood-onset disabilities specifically. Findings from these studies are mixed and often limited in that they are diagnosis specific (eg, spina bifida) or focused solely on intellectual and developmental disabilities.^8,10-13 Overall, little data are available on whether age of disability onset (ie, childhood- vs adult-onset disability) is associated with these adverse health behaviors.

People with disabilities overall have lower rates of recommended preventive screenings, including substance use screening, than people without disabilities do.^13,14 Adults with childhood-onset disabilities may be at risk for substance use, yet, during adolescence and young adulthood, may not receive age-appropriate screening for these types of health behaviors. This lack of screening could be due to the often pervasive paternalistic attitudes and frequent infantilization of people with childhood-onset disabilities, which may cause health care providers to underestimate the prevalence of health behaviors such as smoking and drinking.^15-19 However, no strong evidence is available to suggest that people with childhood-onset disabilities are at overall lower risk for these adverse health behaviors than their nondisabled peers or people with adult-onset disabilities. This discrepancy may result in a missed opportunity for preventive or cessation substance use counseling. Because adults with disabilities are generally at higher risk of many substance use behaviors than adults without disabilities, improved understanding of the effect of age of disability onset may help guide or support screening and preventive strategies.

The 2020 National Health Interview Survey (NHIS), a nationally representative survey of community-dwelling adults in the United States that collects information on disability, allows for the unique opportunity to evaluate the relationship between age of disability onset and adverse health behaviors such as cigarette smoking and alcohol drinking. ²⁰ The objective of our study was to evaluate the relationship between both disability status and age of disability onset, current cigarette smoking, and alcohol drinking among a population-based sample of adults. We hypothesized that adults with disabilities would have a higher prevalence of current smoking and heavy alcohol drinking than adults without disabilities. In addition, we hypothesized that adults with childhood-onset disabilities and adults with adult-onset disabilities would have a similar prevalence of these adverse health behaviors. An exploration of these relationships could help to inform screening guidelines for people with disabilities from adolescence through adulthood.

Methods

Data Source

We completed a secondary analysis of public release data from the 2020 NHIS, which collects information on disease, health behaviors, and disability among respondents. Each year, the NHIS includes survey questions on functioning and disability originating from the Washington Group on Disability Statistics Extended Set on Functioning, which is a set of questions intentionally designed for census and national surveys. ²¹ Additional information related to the NHIS and its procedures is available elsewhere. ²⁰ Within the NHIS, the Sample Adult Questionnaire ²² collects information on 1 randomly selected adult in each household. Responses are either self-reported or, when the sample adult is physically or mentally unable to self-report responses, provided by a knowledgeable proxy. In 2020, the survey was administered to 31 568 adults (response rate, 48.9%), including 409 (1.3%) proxy responses. ²⁰ Both self-report responses and proxy responses were included in aggregate in our analyses.

Study Population

Our study sample included NHIS respondents aged 22-80 years at the time of the survey (n = 28 225). Based on the NHIS definition of disability, respondents were classified as having a disability if they self-reported having “a lot of difficulty” or “cannot do at all” to questions about walking or climbing stairs, communicating, remembering and/or concentrating, self-care, or doing errands alone (n = 3442). We used 22 years as the minimum age of our sample because the NHIS question about age of disability onset was administered to those aged ≥22 years.^20,22 We used a maximum age of 80 years to limit the presence of functional impairments related to frailty among older adults, particularly those aged >80 years, as frailty confers an increased risk for loss of function. ²³

Variables

For the 2020 NHIS, the Administration for Community Living sponsored a question related to age of disability onset: “Did this difficulty (any of these difficulties) begin before age 22 years?” This question was administered to all respondents aged ≥22 years who were classified as having a disability as defined previously. For our analyses, we designated age of disability onset based on responses to this question, which was our independent variable. Childhood-onset disabilities were categorized as impairment(s) that began before age 22 years and adult-onset disabilities as those beginning at age ≥22 years. This classification is consistent with the federal designation of developmental disability, the Individuals With Disabilities Education Act, and previous studies of health-related differences based on age of disability onset.^24-26

The primary outcomes of interest were prevalence of self-reported current cigarette smoking and heavy alcohol drinking. The NHIS has a standard definition of current smoking, which has been used in previous studies.^7,27-29 Respondents are first asked if they have smoked ≥100 cigarettes in their entire life; those who respond yes to the first question were then asked, “Do you NOW smoke cigarettes every day, some days, or not at all?” People who respond “every day” and “some days” are categorized as current smokers. The NHIS uses responses from several questions related to alcohol drinking and the respondent’s sex to classify a participant’s drinking status into several categories. Heavy drinking is defined as, on average, >14 drinks per week in the past year for men and >7 drinks per week in the past year for women, which has also been used by previous studies.^30-32

We also explored secondary outcomes related to smoking or drinking cessation counseling. NHIS respondents who have seen a health professional in the past year and who are current cigarette smokers or former smokers who have recently quit (within the preceding 12 months) are asked, “In the past 12 months, has a doctor, dentist, or other health professional ADVISED you about ways to stop smoking or prescribed medication to help you quit?” Similarly, respondents who have seen a health professional in the past year and who have had ≥1 drink in the past 12 months are asked, “In the past 12 months, has a doctor, dentist, or other health professional ADVISED you to stop or cut down on your drinking?”

We chose covariates a priori based on clinical reasoning and availability within the NHIS dataset. Participant characteristics included age, sex (male/female), race and ethnicity (Hispanic, non-Hispanic Black, non-Hispanic White, and non-Hispanic Other [non-Hispanic Asian, non-Hispanic American Indian/Alaska Native, other single race, and other >1 race]), marital/relationship status (coupled [married or living with a partner] or uncoupled), annual household income as a proportion of the federal poverty level (FPL)^33-36, and health insurance status (private, public, uninsured, other). The FPL in 2020 for a family/household of 4 was $26 200. ³⁷ All data for covariates were also obtained from self-reported NHIS questions.

Statistical Methods

We calculated descriptive statistics for demographic and smoking/drinking characteristics by disability status for the 28 225 participants included in our analyses and by age of disability onset (ie, childhood vs adult onset) among people with disabilities (n = 3442). US population-level estimates can be calculated from the NHIS data by using sampling weights. We applied these NHIS-provided weights and strata information to all analyses in accordance with NHIS-recommended procedures. NHIS describes detailed weighting procedures and bias assessment, including handling of missing data, elsewhere. ³⁸ Results are reported as unweighted or weighted frequencies and weighted percentages, unless otherwise specified.

We used unadjusted and adjusted logistic regression analysis to evaluate the relationship between disability status and current cigarette smoking, current heavy alcohol drinking, and receipt of cessation counseling for both smoking and drinking. Similarly, we used unadjusted and adjusted logistic regression analysis to determine the relationship between childhood- and adult-onset of disability and each of these outcomes. We adjusted the multivariable regression models for age, sex, race and ethnicity, marital/relationship status, annual household income, and health insurance status. For the multivariable regression models based on age of disability onset, we used the same covariates except for age, which was excluded in these models because of collinearity with age of disability onset (our primary explanatory variable). For analyses evaluating receipt of cessation counseling for smoking or drinking, only the subset of respondents who were asked about receiving cessation advice were included in these models. Significance was set at α = .05. We performed all analyses using SAS version 9.4 (SAS Institute, Inc). The Colorado Multiple Institutional Review Board determined this study was not human subjects research because it used publicly available data without personal identifiers (IRB no. 23-2129).

Results

Of the 28 225 respondents aged 22-80 years at the time of the 2020 NHIS, 12.3% (n = 3442) reported having a disability (Table 1). Of these, 444 reported a childhood-onset disability (Table 2).

Table 1.

Demographic and clinical characteristics of respondents aged 22-80 years at the time of the survey (N = 28 225), by disability status, 2020 National Health Interview Survey, United States ^a

Characteristic b	Weighted %		P value c
	Disability (weighted n = 3442)	No disability (weighted n = 24 783)
Unweighted no.	109 643	782 668	—
Age, median (IQR), y d	63.4 (54.1-71.3)	50.3 (36.4-63.2)	<.001
Female sex	62.2	52.6	<.001
Relationship status, coupled	40.8	60.0	<.001
Race and ethnicity			<.001
Hispanic	9.0	10.4
Non-Hispanic Black	13.4	8.9
Non-Hispanic White	72.3	73.4
Other e	5.3	7.3
Annual household income			<.001
<FPL	22.5	6.3
1-1.99 × FPL	29.6	12.7
2-3.99 × FPL	28.9	29.4
≥4 × FPL	19.0	51.5
Health insurance status			<.001
Private	35.9	70.3
Public	46.0	16.8
Uninsured	3.6	8.0
Other	14.6	4.9
Current cigarette smoking	23.5	11.2	<.001
Age when started smoking, y			<.001
<15	24.2	14.1
15 to <18	34.8	33.9
18 to <25	32.9	43.9
≥25	8.1	8.0
Advised to stop smoking by health professional in preceding 12 mo	66.3	49.4	<.001
Current heavy alcohol drinking	5.3	7.4	.001
Advised to stop/cut down on drinking by health professional in preceding 12 mo	8.2	3.5
			<.001

Abbreviations: —, does not apply; FPL, federal poverty level.

^aData source: 2020 National Health Interview Survey. ²²

^bWeighted percentage provided for each cell unless otherwise stated; data are weighted according to methods provided in the documentation for the National Health Interview Survey.

^cThe Rao–Scott χ² test of proportions was used to compare unadjusted differences between those with childhood- versus adult-onset disabilities; Student t test was used for continuous variables. P < .05 was considered significant.

^dMedian and IQR were used as measures of central tendency because of the complexity of estimating the SD for a finite population.

^eOther includes non-Hispanic Asian, non-Hispanic American Indian and Alaska Native, other single race, and >1 race.

Table 2.

Demographic and clinical characteristics of respondents aged 22-80 years at the time of the survey with a disability (n = 3442), by age of disability onset, 2020 National Health Interview Survey, United States ^a

Characteristic b	Weighted %		P value c
	Childhood-onset disability (weighted n = 444)	Adult-onset disability (weighted n = 2998)
Unweighted no.	14 957	94 686	—
Age, median (IQR), y d	45.2 (30.8-58.6)	64.9 (57.1-72.0)	<.001
Sex, female	56.1	63.1	.04
Relationship status, coupled	39.1	41.0	.58
Race and ethnicity			.01
Hispanic	7.1	9.3
Non-Hispanic White	9.5	14.0
Non-Hispanic Black	75.1	71.8
Other e	8.3	4.9
Annual household income			.01
<FPL	30.1	21.3
1-1.99 × FPL	26.2	30.1
2-3.99 × FPL	28.9	28.9
≥4 × FPL	14.9	19.7
Health insurance status			.24
Private	40.6	35.1
Public	44.4	46.2
Uninsured	3.6	3.6
Other	11.4	15.1
Current cigarette smoking	27.6	22.9	.12
Age when started smoking, y			.68
<15	26.6	23.9
15 to <18	37.8	34.4
18 to <25	29.1	33.4
≥25	6.5	8.3
Advised to stop smoking by health professional in preceding 12 mo	53.3	68.8	.02
Current heavy alcohol drinking	5.9	5.2	.68
Advised to stop/cut down on drinking by health professional in preceding 12 mo	10.3	7.8	.38

Abbreviations: —, does not apply; FPL, federal poverty level.

^aData source: 2020 National Health Interview Survey. ²²

^bWeighted percentage provided for each cell unless otherwise stated; data are weighted according to methods provided in the documentation for the National Health Interview Survey.

^cThe Rao–Scott χ² test of proportions was used to compare unadjusted differences between those with childhood- versus adult-onset disabilities; Student t test was used for continuous variables. P < .05 was considered significant.

^dMedian and IQR were used as measures of central tendency because of the complexity of estimating the SD for a finite population.

^eOther includes non-Hispanic Asian, non-Hispanic American Indian and Alaska Native, other single race, and >1 race.

Disability Status

Compared with adults without self-reported disabilities, adults with disabilities were older (median age, 63.4 vs 50.3 y), more frequently female (62.2% vs 52.6%), and less frequently part of a coupled relationship (40.8% vs 60.0%; Table 1). A much higher proportion of people with disabilities than without disabilities had public health insurance (46.0% vs 16.8%) and reported their annual household income as below the FPL (22.5% vs 6.3%). Compared with adults without disabilities, adults with disabilities more frequently reported current cigarette smoking (23.5% vs 11.2%) but less frequently reported heavy alcohol drinking (5.3% vs 7.4%). Among those who were current smokers, a higher proportion of those with disabilities than without disabilities reported starting smoking before age 15 years (24.2% vs 14.1%). Compared with their peers without disabilities, people with disabilities more frequently reported being advised by a health professional to stop smoking (66.3% vs 49.4%) and to stop or cut down on drinking (8.2% vs 3.5%) within the preceding 12 months.

In the logistic regression analysis, adults with disabilities had 76% higher adjusted odds of current cigarette smoking than adults without disabilities (Figure 1). However, the adjusted odds of current heavy alcohol drinking did not differ significantly by disability status. We also found that adults with disabilities had significantly higher odds of reporting receipt of advice on smoking cessation (AOR = 1.56) and drinking cessation (AOR = 1.91) from a health professional within the preceding 12 months than adults without disabilities.

Figure 1.

Unadjusted and adjusted odds of current smoking and heavy alcohol drinking among adults with a disability compared with adults without a disability, 2020 National Health Interview Survey, United States. Disability refers to self-reported difficulty walking or climbing stairs, communicating, remembering or concentrating, self-care, or doing errands alone. The reference group is no disability; logistic regression models adjusted for age, sex, relationship status, race and ethnicity, annual household income, and health insurance status. Data source: 2020 National Health Interview Survey. ²²

Childhood- Versus Adult-Onset Disability

Adults with childhood-onset disabilities were younger than those with adult-onset disabilities (median age, 45.2 vs 64.9 y), a smaller proportion were female (56.1% vs 63.1%), and a greater proportion had an annual household income below the FPL (30.1% vs 21.3%) (Table 2). The proportion of adults reporting current cigarette smoking and heavy alcohol drinking did not differ significantly between people with childhood- versus adult-onset disabilities. The age at which current smokers started smoking did not differ significantly based on age of disability onset. While the frequency of receiving alcohol cessation advice from a health professional within the preceding 12 months did not differ significantly by age of disability onset, those with childhood-onset disabilities less frequently reported receiving smoking cessation advice than those with adult-onset disabilities did (53.3% vs 68.8%).

In the logistic regression analysis, adults with childhood- and adult-onset disabilities had similar unadjusted and adjusted odds of current smoking and heavy drinking and similar odds of having received drinking cessation advice within the preceding 12 months (Figure 2). However, those with childhood-onset disabilities had 53% lower adjusted odds of receiving smoking cessation advice than those with adult-onset disabilities.

Figure 2.

Unadjusted and adjusted odds of current smoking and heavy alcohol drinking among adults with childhood-onset disability compared with adults with adult-onset disability, 2020 National Health Interview Survey, United States. Disability refers to self-reported difficulty walking or climbing stairs, communicating, remembering or concentrating, self-care, or doing errands alone. The reference group is adult-onset disability; logistic regression models adjusted for sex, relationship status, race and ethnicity, annual household income, and health insurance status. Age was colinear with age of onset, so it was not included in regression models. Data source: 2020 National Health Interview Survey. ²²

Discussion

We examined population-level data to test 2 hypotheses: (1) that adults with disabilities would have a higher prevalence of self-reported current smoking and heavy drinking than adults without disabilities and (2) that adults with childhood- versus adult-onset disabilities would have a similar prevalence of these adverse health behaviors. We found that adults with disabilities had higher odds of current smoking but similar odds of current heavy drinking compared with adults without disabilities, and the odds did not vary by age of disability onset. Our study is the first to use nationally representative data to explore differences in smoking and drinking behaviors based on age of disability onset. Our results suggest that adults with disabilities overall may be at high risk for these unhealthy behaviors, particularly smoking, regardless of age of disability onset.

Our finding of higher odds of smoking among adults with disabilities than among adults without disabilities is consistent with other studies, although some studies found a similar prevalence of cigarette smoking in adults with and without disabilities.^7-9,13 Few studies have examined the effect of age of disability onset on smoking and have shown mixed results: some demonstrated a lower likelihood of tobacco use in adults with developmental disabilities compared with the general population, and another found similar rates of cigarette smoking among adults without disabilities, with developmental disabilities, and with nondevelopmental disabilities.^11-13 However, we found that the odds of smoking did not differ based on age of disability onset. Respondents included in our study had similar adjusted odds of heavy drinking regardless of disability status or age of disability onset. Few prior studies have evaluated unhealthy drinking in these populations and had mixed results: 1 study found a lower prevalence of heavy alcohol drinking among adults with disabilities than among adults without disabilities ⁸ and another identified a higher incidence of alcohol-related disorders among adults with 2 childhood-onset disabilities (cerebral palsy and spina bifida) than among adults without these disabilities. ¹⁰ Our study advances prior evidence through analysis of data from a single national database inclusive of adults with any childhood-onset disability (ie, not limited to developmental or diagnosis-specific disabilities).

Adolescence may be a particularly crucial time for screening and counseling on unhealthy behaviors such as smoking and heavy drinking, as these behaviors often begin during adolescence and early adulthood.^5,6 The few prior studies that focused on adolescents rather than adults with disabilities and smoking and drinking had varied results and largely focused on adolescents with intellectual disabilities.^39-42 One study that evaluated students with a broad spectrum of disabilities found that disability status was a significant predictor of early-onset alcohol use (first drink at age ≤14 y). ⁴³ While the NHIS did not include questions on age of first alcohol use, our study found that age of smoking onset differed significantly between adults with disabilities and adults without disabilities, including a higher percentage of people with disabilities who started smoking before age 15 years. When we examined age of smoking onset between adults with childhood- versus adult-onset disabilities, we found no significant differences, suggesting that adolescence may be a particularly important time for screening adolescents with disabilities about these risky behaviors. Despite this important period, adolescents and young adults with disabilities often do not receive age-appropriate health care services as they transition from pediatric- to adult-based care. ³ These factors make adolescence and early adulthood a vulnerable yet critical time for appropriate preventive screening among young people with disabilities.

Compared with people without disabilities, people with disabilities generally face substantial disparities in health care access and health outcomes, including lower rates of preventive health care services, higher rates of chronic medical conditions, and greater disparities in social determinants of health.^44-46 Despite recommendations for screening for tobacco and alcohol use among all adults and adolescents,^47-49 previous research has shown that adults with disabilities have lower rates of primary care screening for smoking and alcohol use than adults without disabilities. ¹⁴ It is unknown what proportion of adolescents with disabilities are screened for tobacco and alcohol use. Our study found that people with disabilities who reported current smoking or heavy drinking had higher odds of receiving cessation advice by a health professional in the preceding 12 months than those without disabilities. While receipt of cessation advice is reassuring, our study could not explore rates of preventive screening. In addition, adults with childhood-onset disabilities had significantly lower odds of receiving smoking cessation advice than those with adult-onset disabilities, despite similar odds of current smoking. Although the reasons for this discrepancy are unknown, systemic ableist biases including paternalistic attitudes and the infantilization of adults with intellectual, developmental, and other childhood-onset disabilities may contribute to discrepancies in screening and counseling for these health behaviors.^15-19 Respondents with childhood-onset disabilities were significantly younger than those with adult-onset disabilities; thus, these 2 subpopulations may have developed smoking habits at different points in the antismoking movement, which strengthened in the 1980s to 1990s. ⁵⁰ However, the effects of this age difference are unclear, as both respondents with childhood- and adult-onset disabilities were equally likely to begin smoking in early adolescence (ie, age <15 y) and did not differ by current smoking behaviors despite this age difference. The results of our study underscore the importance of providing routine tobacco use screening to all people with disabilities beginning in early adolescence.

Our study had several implications for practice. First, our results indicate that adults with disabilities may be at high risk for these unhealthy behaviors, particularly smoking, and yet prior studies show they often have lower rates of preventive screening than adults without disabilities. Increased attention to ensure consistent preventive screening for people with disabilities, as recommended for the general population, is therefore needed. Second, adolescence may be a crucial time for screening and counseling on unhealthy behaviors such as smoking and drinking, including for adolescents with disabilities. Screening should begin in early adolescence and continue throughout the life course, and cessation counseling should be provided to all those with unhealthy behaviors, including people with childhood-onset disabilities.

Limitations

Our study had several limitations. First, because it was a secondary analysis of NHIS data, our findings were subject to the same limitations inherent in the NHIS design, including nonresponse bias, self-reported responses, and its cross-sectional design, which precludes causal inferences. NHIS surveys only the noninstitutionalized US population and, thus, does not provide information on those with disabilities residing in institutions. It also does not provide information on diagnoses or causes of disability, and, thus, we could not explore how smoking and drinking may differ between specific etiologies of disability. Second, NHIS provides a weighting strategy for use in analyses, which we incorporated; however, our study design included only a subset of all NHIS respondents; therefore, the weights provided by the NHIS were not adjusted for our study population. Third, the COVID-19 pandemic may have affected the results of the 2020 NHIS; however, this was the first year the age-of-disability-onset question was included, and questions related to alcohol use are not asked every year (ie, they were not asked in 2021). Future studies should reexamine NHIS data in subsequent years should the age-of-disability-onset question continue to be asked.

Conclusions

Our findings describe some of the first population-based estimates of associations among disability status, age of disability onset, and adverse health behaviors. Routine screening and cessation counseling related to smoking and unhealthy alcohol use are important for all people with disabilities regardless of age of onset, and adolescence may be a particularly vulnerable but crucial time. Future research should include primary data collection on substance use and other adverse health behaviors in people with disabilities of varying ages of onset, exploration of unique factors or risks associated with these behaviors for people with disabilities, and implementation of effective counseling and cessation strategies.