Relationship Between Smoking and Obesity Among Women

Abstract

Objectives

To examine the relationship between smoking and weight status in adult women and whether this association differed by race.

Methods

The study sample consisted of 22,949 African American and 7831 white women enrolled in the Southern Community Cohort Study from 2002 to 2006.

Results

Both African American and white current smokers had decreased odds of being overweight or obese compared to normal-weight nonsmokers, and the inverse trends between current smoking and BMI held for both groups.

Conclusion

A strong relationship exists between smoking and weight status, with patterns nearly identical for African Americans and white women.

Obesity has reached epidemic proportions in the United States with over a third of the U.S. population being classified as obese (body mass index (BMI ≥30).^1,2 The prevalence of obesity has more than doubled between 1960 (13%) and 2004 (33%) among adults 20–74 years old. Obesity is more prevalent among women compared to men, and among African Americans compared to whites.^2–4 Obesity is a risk factor for several diseases including diabetes, hypertension, stroke, dyslipidemia, and breast and colon cancers.^5–7 The direct and indirect economic impact of obesity on the US health care system is estimated at over 50 billion dollars annually.⁸

Smoking is one of the leading causes of premature mortality worldwide and remains the leading cause of preventable death in the United States.⁹ Smoking is a major risk factor for several diseases including lung cancer, coronary heart disease, stroke, and respiratory diseases.^10,11 Apart from the burden of smoking to individuals, smoking costs society billions of dollars annually as measured by health care costs and mortality-related productivity losses.¹² An estimated 21% (approximately 45 million) of US adults currently smoke cigarettes.¹³ Smoking prevalence varies by racial group, age, gender, and education. American Indian/Alaska Native adults (32.9%) have the highest prevalence whereas Asian American adults (11.6%) have the lowest smoking prevalence rate. Smoking is also more prevalent among adults who are between the ages of 18 and 44, more prevalent among men than women, higher for adults with lower incomes and for people with lower educational attainment.^13,14

Although it has been established that on average smokers weigh less than non-smokers,^15–17 the relationship between smoking and weight is complicated. Some studies examining measures of adiposity such as waist-to-hip ratio (WHR) have found that smoking is positively correlated with WHR for both younger and older smokers.^18,49 Yet other studies show that smoking is associated with increased body weight for up to 2 years after initiation and heavier smoking has been associated with higher BMIs.^20,21 In addition, this relationship is further complicated by the fact that several demographic factors such as level of education, age, and socioeconomic status moderate the impact of smoking on weight status.^15,22 Few studies have assessed whether the smoking-obesity patterns may differ for African American women, among whom the prevalence of obesity is especially high.

The interest in the relationship between smoking and weight has been fueled in part by the reputed impact of smoking on suppressing appetite and increasing metabolic rate. Studies have demonstrated that smoking increases energy expenditure for a short duration.²³ However, the long-term effects of smoking on metabolic rate are unclear. Some studies have found no differences in resting metabolic rates after smoking cessation whereas others found that the resting metabolic rate decreases at least in women who had quit smoking for a month.²⁴

Despite the uncertainty of the impact of smoking on body weight, there has been considerable interest in understanding the perceptions and beliefs people have about this relationship. Several investigators suggest that smoking is used as a means of weight control, especially among women.^25,26 Adolescents frequently believe that smoking helps with weight loss and control, with this belief being more prevalent among white girls. Also, dieting has been associated with an increased risk for smoking in adolescents.²⁷ Among adults, it is unclear if dieting and weight concerns are associated with smoking although fear of weight gain is cited as a major barrier to smoking cessation.^28,29 There is some evidence that young adults, especially women smokers, report higher levels of weight concerns than do nonsmokers and are more likely to smoke if they are trying to lose weight.^30,31 However, a limitation of the majority of the literature on smoking and weight concerns is that the findings are from primarily white samples.

The purpose of this study was to examine the relationship between smoking and weight status in low-income African American and white women. The rationale for this study is that the majority of the prior literature on the relationship between smoking and weight has focused on white populations. This relationship in African Americans has not been well established. Furthermore, the differential impact of race on this relationship has not been adequately documented. Another rationale for this study is that weight has been categorized into 5 categories (Overweight, Obese, Obese Class I, Obese Class II, and Obese Class III). This provides interesting information about the dynamics of this relationship across different weight categories, findings that add to our understanding of this relationship. Finally, the study populations consist of white and African American women between 40 and 79 years of age and recruited to the Southern Community Cohort Study (SCCS). The SCCS is a National Cancer Institute-funded large-scale prospective cohort study designed to investigate the causes of cancer disparities in incidence and mortality between African Americans and whites, and urban and rural groups in the southeastern United States.³² This study’s population is unique in that participants are primarily low income and have a higher prevalence of obesity (African American = 58%; white = 50%) and smoking (African American = 34%; white = 39%) compared to the national rates.

METHODS

The recruitment strategy for the SCCS included enrolling participants from 48 community health centers (CHC) across 12 states in the southeastern region of the United States. CHCs serve primarily low-income and underserved populations by providing primary and preventive medical care.³³ These CHCs were located in Alabama, Arkansas, Florida, Georgia, Kentucky, Louisiana, Mississippi, North Carolina, South Carolina, Tennessee, Virginia, and West Virginia.

Study enrollment at each CHC was conducted by trained interviewers who approached and informed potential participants about the study, asked eligible people to participate, and obtained written informed consent. The interviewers used a computer-assisted personal interview (CAPI) system installed on laptop computers to assist with baseline interviews. Information collected from participants included demographic characteristics (age, race, income, education, marital status, employment status), health insurance coverage, smoking status, height and weight, weight history, dietary behaviors, food intake, and physical activity.

The eligibility criteria included being between 40 and 79 years of age, English speaking, and not having received treatment for cancer within the past year (with the exception of nonmelanoma skin cancer). For the present cross-sectional analysis, we considered only women enrolled at one of 48 CHCs between March 2002 and December 2006 who self-reported their race as either white (N=7831) or African American (N=22,949). This research protocol was approved by the Meharry Medical College and Vanderbilt University institutional review boards.

For this study of women in the SCCS, we excluded persons for whom data about their current height or weight were missing (N=318) or who were in the underweight category (BMI <18.5) (N=400). Underweight women were excluded because the small sample size prevented meaningful analysis of the detailed relationships examined in this study. The final sample size thus included 30,780 women. The primary outcomes of interest among the African American and white women were BMI (calculated as self-reported weight in kilograms divided by the square of self-reported height in meters) and smoking status. The 5 standard BMI categories used included normal weight (18.5 ≤BMI ≤24.9), overweight (25 ≤BMI ≤29.9), obese class I (30 ≤BMI ≤34.9), obese class II (35 ≤BMI ≤39.9), and obese class III (BMI ≥40).

Smoking status was categorized as former, current, or never smokers. Current smokers consisted of those participants who endorsed “Yes” on the question “Do you smoke cigarettes now?” Participants were categorized as never smokers if they responded “No” to the questions “Have you smoked at least 100 cigarettes (5 packs of cigarettes) in your entire life?” Former smokers were those who responded “Yes” to the question “Have you smoked at least 100 cigarettes (5 packs of cigarettes) in your entire life?” and “No” to the question “Do you smoke cigarettes now?” Other smoking variables of interest included age of smoking onset and smoking amount (number of cigarettes reportedly smoked).

Covariates assessed in the analysis included demographic variables (age, education, employment, household annual income, marital status, and health insurance), depression, and total physical activity. Depression was measured using a 10-item version of the Centers for Epidemiological Studies Depression (CES-D) scale. Cutoffs of 10, 15, and 20 define mild, moderate, and severe levels of depressive symptoms.³⁴ The CES-D has demonstrated measurement equivalency in samples with differential characteristics including race and gender.³⁵

Physical activity was measured by MET-hours/day [metabolic equivalent task-hours] per day). Metabolic equivalent, or MET, is a unit useful for describing the energy expenditure of a specific activity. A MET is the ratio of the rate of energy expended during an activity to the rate of energy expended at rest. Total physical activity was categorized into 4 groups based on quartiles. Quartile 1 corresponds to a sedentary level of daily physical activity (sleeping, watching television), Quartile 2 corresponds to a light level of daily physical activity (desk work, writing, typing), Quartile 3 corresponds to a light-plus level of physical activity (light housekeeping, bowling, fishing), and Quartile 4 corresponds to a moderate and vigorous level of physical activity (walking at a brisk pace, dancing, jogging, heavy housekeeping).

Data Analysis

Demographic, lifestyle, and health-related characteristics were computed for both African American and white women and chi-square tests were used to test differences between the groups. An ANOVA with Scheffé’s multiple comparisons tests was used to test for significant differences in BMI by race and smoking status. Several demographic and lifestyle variables were adjusted for in this analysis.

Multinomial logistic regression models were used to estimate odds ratios (and 95% confidence intervals) as the measures of association between BMI and smoking status for African American and white women. Established BMI categories used to classify normal weight, overweight, and obesity classes served as the response variable for these analyses. These analyses used normal-weight non-smokers as the reference group. In order to adjust for potential confounding factors, terms for demographic variables, depression, and total physical activity were included in the regression models. Linear trends across BMI categories were examined using the Mantel-Haenszel test (Table 3). All data analyses were conducted using SAS/STAT software, Version 9.1 of the SAS System for Windows (SAS Institute Inc, Cary, NC, USA).

Table 3.

Odds Ratios and 95% Confidence Intervals of Overweight and Obesity According to Cigarette Smoking Status Among African American and White Women

Variables	African American						White
	Normal	Overweight	Obese			Trend Test	Normal	Overweight	Obese			Trend Test
			Class I	Class II	Class III				Class I	Class II	Classm
		OR (95% CI)	OR (95% CI)	OR (95% CI)	OR (95% CI)			OR (95% CI)	OR (95% CI)	OR (95% CI)	OR (95% CI)
Never Smoker	1.00						1.00
Smoking Status
Former		0.92 (0.81 1.05)	1.03 (0.90 1.17)	1.01 (0.88 1.16)	1.04 (0.91 1.20)	0.08		1.06 (0.88 1.28)	1.14 (0.94 1.37)	0.98 (0.80 1.21)	1.18 (0.96 1.45)	0.18
Current		0.59 (0.54 0.65)	0.42 (0.38 0.46)	0.31 (0.27 0.34)	0.21 (0.19 0.24)	0.01		0.59 (0.50 0.70)	0.42 (0.36 0.50)	0.28 (0.23 0.34)	0.22 (0.18 0.27)	0.01
Age of Smoking Onset
< 18		0.54 (0.48 0.61)	0.38 (0.34 0.43)	0.30 (0.26 0.34)	0.19 (0.16 0.22)	0.01		0.58 (0.48 0.70)	0.42 (0.35 0.51)	0.27 (0.22 0.34)	0.21 (0.17 0.27)	0.01
≥18		0.65 (0.58 0.73)	0.45 (0.39 0.50)	0.32 (0.27 0.36)	0.24 (0.21 0.28)	0.01		0.63 (0.51 0.78)	0.41 (0.33 0.51)	0.28 (0.22 0.37)	0.25 (0.19 0.33)	0.01
Smoking Usage
0 < & ≤20		0.59 (0.53 0.65)	0.42 (0.38 0.46)	0.30 (0.27 0.34)	0.22 (0.19 0.24)	0.01		0.60 (0.50 0.71)	0.40 (0.33 0.48)	0.28 (0.23 0.35)	0.22 (0.18 0.28)	0.01
>20		0.67 (0.51 0.87)	0.33 (0.25 0.45)	0.34 (0.25 0.47)	0.17 (0.12 0.25)	0.01		0.60 (0.47 0.77)	0.47 (0.37 0.61)	0.27 (0.20 0.37)	0.23 (0.17 0.32)	0.01

Notes.

Adjusted variables: Age-group, education level, employment status, annual household income, marital status, health insurance, depression, and total physical activity.

RESULTS

Demographic and lifestyle factors for African American and white women from the baseline SCCS interview are presented in Table 1. African American women comprised 75% of the total study sample; and compared to white women, they were more likely to be under the age of 60 years (82% vs 76%), employed (39% vs 33%), have an annual household income below $25,000 (85% vs 79%), and have health insurance (61% vs 58%). For both groups of women, the majority of women had an education level of high school or less (65%). White women were more likely to be married or cohabitating compared to their African American counterparts (43% vs 26%), to report having depressive feelings (49% vs 42%), and to be more physically active (Mean METS: 53.90 vs 52.59; P<.05).

Table 1.

Demographic and Lifestyle Characteristics Among African American and White Women Enrolled in the SCCS

Participant Characteristics	African American (N=22,949)		White (N=7831)
	N	%	N	%
BMI
Normal weight (18.5 ≤BMI ≤ 24.9)	3742	16.31	1859	23.66
Overweight (25 ≤ BMI ≤ 29.9)	5911	25.76	2073	26.47
Obese class I (30 ≤ BMI ≤ 34.9)	5798	25.26	1787	22.82
Obese class II (35 ≤ BMI ≤ 39.9)	3798	16.55	1058	13.51.
Obese class III (BMI ≥ 40)	3700	16.12	1060	13.54
Age
40–49	11349	49.45	3281	41.90
50–59	7395	32.22	2642	33.74
60–69	3101	13.51	1440	18.39
70–79	1104	4.81.	468	5.98
Education
< 9 years	1759	7.67	729	9.31
9–11 years	5553	24.21	1615	20.63
High school	7716	33.64	2773	35.42
Vocational, technical, or business training	1302	5.68	404	5.16
Some college or junior college	4546	19.82	1489	19.02
College graduate and beyond	2062	8.99	818	10.45
Employment
Employed	8815	39.28	2502	32.54
Unemployed	13629	60.72	5186	67.46
House Annual Income
Under $15,000	13955	61.59	4528	58.51
$15,000 – $24,999	5395	23.81	1586	20.49
$25,000 – $49,999	2537	11.20	999	12.91
$50,000 or More	772	3.41	626	8.09
Marital Status
Married/cohabitate	6057	26.43	3349	42.81
Divorced/separated	7878	34.37	2764	35.33
Widowed	3322	14.49	1068	13.65
Single, never been married	5664	24.71	642	8.21
Health Insurance
Yes	13913	61.05	4468	57.50
No	8876	38.95	3302	42.50
Smoking Status
Never	10675	46.61	2801	35.85
Former	4456	19.46	1929	24.69
Current	7770	33.93	3084	39.47
Age of Smoking Onset (years old)
Never	10675	46.75	2801	35.89
<18	6166	27.00	3170	40.61
≥18	5992	26.24	1834	23.50
Smoking Amount (1 cigarette/day)
Never	10675	46.78	2801	35.96
0<&≤20	11098	48.63	3656	46.93
>20	1048	4.59	1333	17.11
Depression (based on CES-D)a
Yes	9660	42.44	3778	48.62
No	13102	57.56	3992	51.38
Total Physical Activity (METs-hours/day)b
0–31.30 (sedentary level)	5908	26.41	1882	24.49
31.31–47.43 (light level)	5383	24.07	2029	26.40
47.44–70.30 (light plus)	5631	25.18	1828	23.79
70.31+ (moderate and vigorous level)	5445	24.34	1946	25.32

Notes.

^aBased on the 10-question Center for Epidemiologic Studies Depression scale (Scale 0–30, depression categorized as No if score <10 and Yes if score ≥10)

^bSum of household and occupational work plus moderate and vigorous sports

Differences in BMI and Smoking Status by Race

African American women were more likely to be obese (58% vs 50%) compared to white women (P<.01), with 16% having BMI values of 40 and above (compared to 14% for whites). Fewer African American compared to white women were current smokers (34% vs 39%; P<01), and a much smaller percentage (5% vs 17%; P<.05) smoked more than one pack per day.

Racial differences in average BMI by smoking status were compared using an ANOVA. Several demographic variables, depression, and physical activity, as well as amount smoked and age of smoking onset were controlled for in these analyses. As shown in Table 2, African American women had higher average BMI compared to white women within each smoking category, with the differences highly significant (P<.0001) due to the large sample sizes. Average BMI differed significantly (P<.0001) among the 3 smoking categories, with pairwise comparisons using Scheffé’s tests revealing significantly lower average BMI among current smokers compared to former smokers and never smokers, but no significant differences between former and never smokers. There were no significant interactions between smoking status and race for average BMI. These results indicated that both groups of currents smokers had lower BMIs compared to former and never smokers.

Table 2.

Means, Standard Deviations, and Main Effects for BMI by Race and Smoking Status

	African American	White	Main Effect Effect of Race	Main Effect of Smoking Status	Interaction Betweeem Race and Smoking Status
	Mean ± SD	Mean ± SD	F, P value	F, P value	F, P value
For BMI
Smoking Status
Never	33.64 ± 7.91	32.26 ± 8.05	70.12, P<0.001	789.11, P<0.001	0.48, P= 0.62
Former	33.80 ± 7.74	32.67 ± 8.25
Current	30.22 ± 7.30	29.37 ± 7.25

Notes.

Adjusted variables: Age-group, education, employment, house annual income, marital status, health insurance, age of smoking onset, smoking amount (# of cigarettes smoked per day), depression, and total physical activity.

Odds Ratios for BMI According to Smoking Categories by Race

As shown in Table 3, a multinomial logistic regression model was conducted to investigate the relationships between smoking variables and established BMI categories by race.

Smoking Status

The odds ratios of being overweight or in the any of the 3 obese classes compared to being a normal-weight nonsmoker were all close to 1.0 among former smokers, with no significant difference in BMI between nonsmokers and former smokers for both African Americans and whites. In sharp contrast, the odds ratios of being overweight or in any of the 3 obese classes were significantly lower among current smokers relative to never smokers for both African American and white women. The odds ratios monotonically declined with increasing BMI, dropping to 0.59 (0.54 0.65), 0.42 (0.38 0.46), 0.31 (0.27 0.34), and 0.21 (0.19 0.24) with rising BMI for African American women. A nearly identical pattern was seen for white women, with odds ratios declining to 0.59 (0.50 0.70), 0.42 (0.36 0.50), 0.28 (0.23 0.34), and 0.22 (0.18 0.27) with increasing BMI category.

Age of Smoking Onset

The odds ratios of being overweight or in any of the 3 obese classes among current smokers who started smoking before vs after age 18 relative to never smokers are also shown in Table 3. The declines in odds ratios were evident in both age-at-start groups, with the same trend strength for before or after 18 years old to start smoking. However, the patterns were essentially the same for African American and white women.

Smoking Amount

Table 3 also shows odds ratios of being overweight or in the any of the 3 obese classes among current smokers classified by number of cigarettes smoked per day (from fewer than 1 cigarette up to and including one pack per day vs more than one pack per day). The odds ratios declined with increasing BMI in both smoking intensity categories, with little consistent difference in the trends between the 2 smoking intensity categories. Again, the patterns were similar for black and white smokers.

CONCLUSIONS

The population studied had higher rates of obesity (50–58%) compared to the national rate (37%) for women between the ages of 40 and 79 years.² In addition, smoking rates in this sample were also higher (34–39%) compared to the national rate (21%).^13,14 The higher prevalence of smoking and obesity needs to be understood in the context of the unique demographic characteristics of the SCCS population, which was drawn from persons primarily of low income and educational attainment, 2 factors that are correlated with smoking and obesity prevalence.^3,4,14

Although most studies have indicated that smokers on average have lower body weights,^15–17 a few have reported that smoking and body weight are positively correlated.^20–23 The results from our large study clearly show that smokers were less likely to be overweight or obese compared to people who had never smoked. The inverse association between smoking and weight is limited to current smokers, with mean BMI among former smokers being nearly the same as among lifelong nonsmokers. Furthermore, current smokers were increasingly less likely to be in the upper categories of obesity. Remarkably, the patterns between smoking and BMI were essentially identical among black women and white women. Indeed, we found odds ratios among current smokers for overweight and obesity classes I, II and III, respectively, of 0.6, 0.4, 0.3, and 0.2 (relative to 1.0 for normal-weight never smokers) for both blacks and whites. The declining trends among both groups were not due to socioeconomic or other confounding factors, which we could adjust for in our statistical models.

Racial differences regarding concerns about body weight have been well documented. Studies have consistently indicated that white women are more likely than their African American counterparts to be concerned about their weight and engage in dieting practices and smoking as weight-management strategies.^25–27 Our findings, however, did not reveal any racial differences in the relationship between smoking and weight status. For both racial groups of women, current smokers were less likely to be overweight and obese compared to normal-weight nonsmokers. Our data suggest that although white women may be more likely to have weight concerns and engage in dieting behaviors, among smokers, racial differences are minimal. Hence, a similar percentage of women in both groups may be smoking to control their weight. In addition, smoking may affect multiple biological processes associated with weight or weight gain. The underlying mechanisms would not a priori be expected to differ by race, a notion supported by the nearly identical smoking-obesity patterns we observed between black and white women.

The lack of racial differences in the relationship between BMI and smoking status may be partially explained by the socioeconomic status (SES) of participants. The SCCS sample was made up of primarily low-income participants. Prior literature has indicated that lower SES is related to higher rates of smoking and obesity;^36–39 and women of lower SES may be less concerned about weight gain, weight control, and restrictive food practices.^40–42 It may be that for the lower SES participants in this study, the lack of racial differences may not be driven by using smoking to control weight; rather, it may be driven by the shared metabolic and appetite suppression advantage that smoking offers.^43,44

Smoking amount has been reported to affect weight status in that the risk of being obese increased with the number of daily cigarettes smoked in some studies.^22,23 The results of our study are not consistent with these prior research findings in that both heavy smokers (>20 cigarettes a day) and light and moderate smokers (<20 cigarettes a day) were less likely to be obese, with little difference in the trends with increasing obesity between heavy and light smokers. The age of this sample may potentially explain these discrepant findings. Metabolic rate gradually slows with increasing age; hence, people naturally gain weight as they age. For older heavy smokers, gains in metabolic rate due to smoking may translate into a slower weight gain as the result of aging as compared to former or never smokers, who would not have this metabolic effect.

Study Limitations and Strengths

This study has several notable limitations. Smoking status was derived from self-report, and although there is evidence supporting the validity of self-reported smoking,⁴⁵ there may be biases due to under- or overreporting of smoking behaviors. As part of our validation studies for the SCCS questionnaire, we compared self-reported smoking using the smoking items with smoking status determined using cotinine level. The total sample size for this validation study was 716 randomly selected participants from the Southern Community Cohort Study database. The results indicated a 97–98% agreement between biochemical and self-report methods of classifying current smokers. Height and weight were also self-reported. Although self-reports of these variables have been generally shown to be accurate,⁴⁶ new research has suggested that especially in overweight and obese women, BMI is underestimated because of underreporting of weight and overreporting of height.^47,48 For this study, errors in estimating BMI are likely minimal because there was a high agreement (97%) between observed and self-reported height and weight on a subset of participants.³² Another limitation of this study is that the data are cross-sectional in nature, limiting causal inferences. This study also had several notable strengths, including the large sample size, information obtained in a consistent, structured manner from in-person interviews using a standardized questionnaire, and good representation of respondents from various demographic categories, including a large number of African Americans in the sample.

Future Directions

The results from our study clearly showed that smokers were less likely to be overweight or obese compared to people who had never smoked. The inverse association between smoking and weight was limited to current smokers, with mean BMIs among former smokers being nearly the same as those among lifelong nonsmokers. Furthermore, current smokers were increasingly less likely to be in the upper categories of obesity. Remarkably, the patterns between smoking and BMI were essentially identical among African American and white women. In addition, the inverse trends between current smoking and BMI held for heavy and light smokers and for initiation of smoking before or after the age of 18 years. Future studies should evaluate the utility of incorporating the impact of smoking on weight in existing or new smoking interventions.