Darker skin color is associated with a lower likelihood of smoking cessation among males but not females

Abstract

Darker skin color is associated with discrimination and unfair treatment and may contribute to persisting health disparities. This study examined whether darker skin color was associated with smoking cessation and whether this association was moderated by sex and race. This study also explored whether biological and psychosocial factors, including nicotine and cotinine concentrations, discrimination, distrust, and neuroticism, mediated this association. The data for this study came from a prospective smoking cessation intervention that included 224 Black and 225 White adults from Kansas City, Missouri. Skin color was assessed using a DermaSpectrometer to measure melanin contained within the skin. Point prevalence smoking abstinence was biochemically-verified and assessed at weeks 4 and 26. Hierarchical logistic regression analyses were conducted to evaluate hypothesized relations between skin color and smoking cessation. Interactions between race and sex with skin color were also evaluated. While skin color was not associated with smoking cessation in the overall sample or among Blacks only, results indicated that sex moderated the effect of skin color on smoking cessation after adjusting for race and other covariates. Among males, darker skin color was associated with lower odds of achieving smoking abstinence at weeks 4 (OR = 0.60 [95% CI= 0.36, 0.99]) and 26 (OR = 0.52 [95% CI = 0.29, 0.91]). Skin color did not predict smoking cessation among females. Skin color was positively correlated with discrimination (r = 0.15, p = 0.02), cynicism/distrust (r = 0.14, p = 0.03) and neuroticism (r = .24, p < 0.01) among males only. However, these factors did not mediate the association between skin color and smoking cessation. Skin color may contribute to cessation-related health disparities among Black males, but more research is needed to understand the psychosocial and biological mechanisms through which skin color influences tobacco cessation.

INTRODUCTION

Blacks in the United States are disproportionately burdened with tobacco-related morbidity and mortality (Cunningham et al., 2017; Ho and Elo, 2013), even though the prevalence of smoking among Blacks is nearly identical to Whites (Jamal et al., 2018). Blacks also begin smoking at a later age and smoke less frequently than Whites (Freedman et al., 2011; Trinidad et al., 2011). Thus, high rates of tobacco-related morbidity and mortality among Blacks may be partially due to a lower likelihood of smoking cessation (Nollen et al., 2019; West et al., 2018). Understanding the mechanisms that contribute to this disparity in smoking cessation is critically important to improving treatment and reducing tobacco-related health disparities (NCI, 2017). Nollen et al. (2019) explored sociodemographic, treatment, and biological factors that might contribute to the Black-White disparities in smoking cessation outcomes. They found that compared to Whites, a disproportionate number of Black adults did not own a home, had a lower household income, and resided in neighborhoods with many systemic problems including high traffic density and pollution, inadequate infrastructure and transportation, and limited availability of local facilities and amenities. These factors fully explained the Black-White disparity in smoking cessation outcomes and highlighted the importance of identifying factors associated with race that made it more difficult for Blacks to quit smoking compared to Whites (Nollen et all, 2019).

Among Blacks in the United States, infrequent homeownership, lower-income, and a higher likelihood of living in impoverished and disadvantaged neighborhoods are partially the result of the historical marginalization of Blacks economically, politically, and socially through discriminatory laws, policies, and norms (Fiss, 2009). This discrimination was based not only on race but also skin color, such that Blacks with darker skin color were treated less favorably by Whites than Blacks with lighter skin (Dixon and Telles, 2017). Notably, Blacks themselves associate negative characteristics with darker skin color (e.g., criminality and aggressiveness), and associate positive characteristics with lighter skin color (e.g., attractiveness and intelligence; Maddox and Gray, 2002). Not surprisingly, studies have shown that Blacks with darker skin color, compared to Blacks with lighter skin, receive less education and longer prison sentences, have lower-incomes, and hold jobs with lower prestige (Monk, 2014; Viglione et al., 2011). Also, Blacks with darker skin color have a greater risk for all-cause mortality (Stewart et al., 2018), high blood pressure (Gravlee and Dressler, 2005; Laidley et al., 2019), and obesity (Wassink et al., 2017), and they report poorer overall health (Monk, 2015).

Skin color is a key dimension of race, and societies use both concepts to create hierarchies that place Whites (or the lightest skin color) at the top and Blacks (or the darkest skin color) at the bottom (Dixon and Telles, 2017). Skin color, however, transcends race in social importance when race and skin color cannot be readily distinguished, such as in Latin America where skin color is used to categorize people into ethno-racial groups (Telles and Paschel, 2014). Similarly, because Blacks share common African ancestry and have significant variation in skin color (Crawford et al., 2017), many Blacks will use color labels, such as “high yellow” and “blue-black,” to socially classify other Blacks (Wilder, 2010). In addition to race, the social classification and hierarchy of skin color may further influence the psychological and physiological well-being of Blacks and shape racial health disparities.

In social science research, skin color has been assessed primarily through self-report or self-identification and interviewer-rated measurements (Dixon and Telles, 2017). Self-reported assessments reflect personal perceptions about skin color, whereas, interviewer-rated assessments reflect the perceptions of generalized others (Cobb et al., 2016); both techniques estimate where an individual is positioned within the social hierarchy of skin color. A third option is to use reflectance spectrometers to measure the amount of light reflected by the skin and obtain a melanin index score (Khalid et al., 2017). Briefly, melanin, which is contained within the skin, hair, eyes, heart, lung, liver, and brain, is the primary determinant of skin color (Visscher, 2017). There are two types of melanin: constitutive and facultative melanin. Constitutive melanin is genetically determined, and facultative melanin refers to an increase in melanin due to physiological regulation (Yamaguchi et al., 2007), such as the body’s response to ultraviolet light exposure (e.g., tanning reaction). Melanin index scores are based on facultative and constitutive melanin (Blair et al., 2002).

Only a few studies have examined the association between skin color and smoking (King et al., 2018, 2009; Tamai et al., 2014), and these studies have primarily used spectrometers to measure skin color. Research has shown that nicotine accumulates in human tissues containing melanin, such as hair, skin, and gingiva (Yerger and Malone, 2006). King et al. (2009) were the first investigators to examine whether melanin levels were related to nicotine dependence among Black adults. King et al. (2009) assessed the amount of melanin contained in non-sun exposed skin (i.e., constitutive) and sun-exposed skin (i.e., facultative). They found that having higher levels of facultative melanin, as found in darker skin tones, was associated with greater accumulation of nicotine (i.e., higher cotinine levels) and tobacco dependence, but constitutive melanin was not related to these variables.

In a separate study, King et al. (2018) found that higher tanning capacity, which is the extent to which the skin can darken due to sun exposure or some other physiological reaction, was associated with higher odds of smoking cigarettes within the first 5 minutes and the first 30 minutes after waking, hallmarks of nicotine dependence. From these findings, they hypothesized that higher levels of melanin allow for a greater amount of nicotine to accumulate within the skin or other melanin-containing tissues, which reduces nicotine clearance and increases the amount of time an individual is exposed to nicotine. Specifically, un-metabolized nicotine then continuously moves from melanin-containing tissues through the bloodstream into the brain and activates nicotine receptors, which increases nicotine dependence (King et al., 2009; Yerger and Malone, 2006).

Psychosocial mechanisms may also link skin color to tobacco use and poor cessation outcomes. Blacks with darker skin color are predisposed to experience discrimination and unfair treatment (Uzogara and Jackson, 2016), and previous studies have shown that experiencing discrimination is associated with smoking (Chen and Yang, 2014; Parker et al., 2017), greater nicotine dependence (Kendzor et al., 2014b), and difficulty quitting smoking (Alexander et al., 2019; Kendzor et al., 2014a). Secondarily, because Blacks with darker skin color report high levels of discrimination, they may also be more likely to believe that people are intentionally harmful and inherently dishonest and believe that society is uncontrollable and harmful (Brondolo et al., 2011; McClendon et al., 2019). As such, Blacks with darker skin color may be more distrustful of others, and because they perceive society to be uncertain, uncontrollable, and harmful, they are more likely to respond to social interactions with negative feelings, such as distress, worry, and anxiety (i.e., neuroticism; Sutin et al., 2016). Recently, high levels of distrust have been shown to increase the risk of current smoking among Black adults (Sims et al., 2017), and neuroticism, or feelings distress, worry, and anxiety (i.e., neuroticism), is consistently associated with current smoking and relapse (Hakulinen et al., 2015; Zvolensky et al., 2015).

Nonetheless, Black females and males with darker skin color face a disparate set of psychosocial and socioeconomic challenges that may have important implications for smoking cessation. Researchers have shown that Black females with darker skin color, compared to lighter skin color, have lower marriage rates, are perceived to be less attractive, and have lower self-esteem (Fink et al., 2006; Hill, 2002). Conversely, Black males with darker skin color, compared to lighter skin color, have lower self-efficacy and occupy lower-status jobs (Hill, 2000; Monk, 2014). Black females and males with darker skin color face unique and difficult challenges, but recent research suggests that skin color may have a more significant health impact on Black females than males. Perreira, Wassink, and Harris (2019) found that darker skin color was associated with higher depressive symptoms and body mass index (BMI) and poorer overall health among females with darker skin color. Although similar associations (i.e., depressive symptoms and overall health) were found among males with darker skin color, these associations did not persist after adjusting for discrimination, stress, racial/ethnic self-identification, and socioeconomic status (Perreira et al., 2019). Importantly, discrimination may be an important mechanism that links skin color to poor smoking cessation outcomes among males because Black males, not females, may respond to discrimination by using tobacco and engaging in other poor health behaviors (Brodish et al., 2011). Likewise, because a recent study showed that darker skin color was associated with higher levels of perceived discrimination among males and not females (Assari and Caldwell, 2017), darker skin color may also be more strongly associated with distrust and neuroticism among males than females.

In summary, skin color may be an important factor that contributes to an explanation of why Blacks have difficulty quitting tobacco and experience high rates of tobacco-related morbidity and mortality (NCI, 2017). Nevertheless, the few studies that have been completed thus far on skin color have been cross-sectional and focused on tobacco use and dependence, and therefore additional research is needed to investigate the association between skin color and smoking cessation. Thus, the current study examined whether skin color was associated with short- and long-term smoking cessation among Blacks and Whites enrolled in a smoking cessation trial. Race and sex, which were balanced as per the parent study design (Nollen et al., 2016, 2019), were evaluated as moderators of the association between skin color and smoking cessation. It was hypothesized that darker skin color would be associated with lower odds of smoking cessation, and we evaluated whether this association would be stronger among Blacks and males than among Whites and females. Last, we explored whether biological and psychosocial factors, including nicotine and cotinine concentrations, discrimination, distrust, and neuroticism, mediated the association between skin color and smoking cessation. Findings will provide information about how skin color may contribute to tobacco-related health disparities through psychosocial and biological pathways.

METHODS

The present study is a secondary analysis of a prospective smoking cessation intervention. The design, protocol, and primary outcomes have been described in detail elsewhere (McClendon et al., 2019; Nollen et al., 2016). Briefly, participants were recruited and stratified based on self-reported race (Black, White), and within race, on age (<40, ≥40) and sex, and income level (≤ 400% Federal Poverty Level) was also restricted to ensure that Blacks and Whites were comparable on these important covariates related to smoking abstinence (Nollen et al., 2016). All participants were provided 12 weeks of pharmacotherapy (varenicline), six counseling sessions through Week 16, and follow-up through Week 26. The stratified design of the parent study created similar numbers of participants within race and sex cohorts and is a strength for this present study because analyses are focused on the moderating effects of race and sex.

Measures

Dependent variables.

The primary dependent variables were biochemically verified 7-day point-prevalence smoking abstinence, which was defined as a self-report of complete abstinence from cigarettes during the previous seven days at 4 and 26 weeks after the baseline visit. Self-reported smoking abstinence was biochemically verified by salivary cotinine levels of ≤ 15ng/ml (Kim, 2016).

Independent variable.

Skin color was measured using a DermaSpectrometer (DSM II Color Meter, Cortex Technology). Following standardized protocols (van der Wal et al., 2013), one measurement of constitutive melanin and two measurements of facultative melanin were taken. Constitutive melanin was measured by examining the left inner arm, one inch above the inside of the elbow, and facultative melanin was measured by examining the forehead, in between and ½ inch above the eyebrows; and the back of the left hand, one inch below the knuckle of the middle finger. Participants were asked to remove any makeup from these locations, and hair, freckles, and birthmarks were avoided when possible. Two measurements were taken at each site, and the scores on the melanin index were averaged. Higher scores on the melanin index indicate darker skin pigmentation.

Two additional variables were calculated to reflect overall skin color and tanning capacity. Overall skin color was created by averaging the constitutive and facultative melanin readings, and this score was then standardized and rounded to the nearest whole number to reflect substantial deviations in skin pigmentation from the average (Gravlee and Dressler, 2005). Tanning capacity was created by subtracting constitutive melanin (i.e., upper top arm) from facultative melanin (i.e., forehead) and was used as a covariate in the primary analyses to account for skin that darkens due to physiological regulation, such as ultraviolet light exposure (King et al., 2018).

Mediators and covariates.

Potential mediators included baseline nicotine and cotinine concentrations. Total nicotine equivalents (TNE) and cotinine concentrations were collected at baseline through urinalysis. TNE was quantified by examining the sum of nicotine plus six of the primary nicotine metabolites (NIC + COT + 3HC + NNO + CNO + NNIC + NCOT). TNE was divided by creatinine to derive the TNE/creatinine ratio, which has the strongest correlation with nicotine dose in lab-based studies and represents the current gold standard for estimating nicotine exposure (Benowitz et al., 2010).

Perceived discrimination, which was measured using the Everyday Discrimination Scale (Sternthal et al., 2011), neuroticism, which was measured using 4-items from the Neuroticism scale of the NEO Five-Factor Inventory (McCrae and Costa, 2004), and cynicism/distrust of others, which was measured using 8-items from the Cook-Medley Hostility Scale (Cook and Medley, 1954), were also explored as potential mediators. Covariates included self-reported race and ethnicity (Non-Hispanic Black and Non-Hispanic White), sex (female and male), years of age, years of education, homeownership (yes or no), annual household income per household member (household member/$1000), study visits completed (range 0-6), and total score on the neighborhood problems scale (range 10-30; Steptoe and Feldman, 2001). All variables were measured at baseline, which occurred one week before the quit date.

Plan of analysis

Descriptive statistics were computed for all variables included in the primary analysis. The primary analyses were completed using logistic regression and hierarchical logistic regression. Univariate and multivariable logistic regressions were conducted to explore the association between skin color and smoking cessation, and hierarchical logistic regression was conducted to determine whether the effect of skin color on smoking cessation was moderated by race and sex. Multiplicative interactions (skin color x race x sex, skin color x race, and skin color x sex, sex x race) were manually removed by comparing nested models according to information criterion (IC) scores. Interactions that were not significant (p > .05) or did not improve model fit were removed. Significant interactions were decomposed using simple slopes or spotlight analyses (Bauer and Curran, 2005) using the PROCESS macro in SAS 9.4. The primary analyses included covariates that were previously identified to be the best predictors of smoking cessation within this sample, which included homeownership, annual household income per household member, study visits completed, baseline cotinine concentration, and neighborhood problems (Nollen et al., 2019). Age, tanning capacity, and education were also included as covariates in primary analyses.

Potential mechanisms of the association between skin color and smoking cessation were investigated in exploratory analyses using a two-step process. First, partial correlations between skin color and biological and psychosocial variables were computed in the overall sample and separately by sex, and these correlations adjusted for race. Second, biological and psychosocial variables significantly correlated (p ≤ 0.05) with skin color were analyzed as potential mediators using the PROCESS macro in SAS 9.4 to investigate whether these variables explained the association between skin color and smoking cessation (Hayes, 2017). Mediation analyses controlled for race only. Across all analyses (i.e., primary and exploratory), participants with incomplete data to calculate smoking abstinence, which was 14.5% of participants at Week 4 and 17.4% at Week 26, were considered smoking (i.e., intent-to-treat). Less than 1% of data points were missing for explanatory variables in the analyses.

RESULTS

Sample characteristics

As shown in Table 1, study participants were 41.9 (SD = 11.7) years of age on average, and most participants had at least attended college or technical school (66.2%). Participants earned an average annual household income of $28,567 (SD = $19,841), most participants did not own a home (74.4%), and nearly three persons lived in each participants’ household on average (M = 2.8, SD = 1.7). At baseline, participants had average urine cotinine and TNE/creatine ratio of 289.5 ng/mL (SD = 150.5) and 63.4 ng/ml (SD = 39.5), respectively. About 22.7% of participants were biochemically confirmed abstinent 4-weeks after the baseline visit, and a similar proportion of participants were abstinent 26-weeks after the baseline visit (19.4%).

Table 1.

Sample characteristics (N = 449)

Variable	N (%) or M (SD)
Race
Non-Hispanic White	225 (50.1%)
Non-Hispanic Black	224 (49.9%)
Sex
Male	225 (50.1%)
Female	224 (49.9%)
Age	41.9 (SD = 11.7)
Education
Less than a high school degree	44 (9.8%)
High school graduate	108 (24.1%)
Some college or tech school	206 (45.9%)
College graduate or higher	91 (20.3%)
Annual household income	$28,565.8 (SD = $19,839.9)
Persons living in a household	2.8 (SD = 1.7)
Homeownership
No	334 (74.4%)
Yes	115 (25.6%)
Perceived neighborhood problems	14.5 (SD = 4.2)
Baseline urine cotinine concentration	289.5 ng/ml (SD = 150.5)
Total nicotine equivalents/creatinine	63.4 ng/ml (SD = 39.5)
Study visits completed	4.8 (SD = 1.7)
Overall melanin reading (e.g., skin color)a	64.5 nm (SD = 21.8)
Tanning capacity	6.5 nm (SD = 10.8)
7-day point prevalence smoking abstinence (4-weeks post-baseline visit)
No	282 (62.8%)
Yes	102 (22.7%)
Missing	65
7-day point prevalence smoking abstinence (26-weeks post-baseline visit)
No	284 (63.3%)
Yes	87 (19.4)
Missing	78 (17.4%)

Skin color and smoking abstinence

In unadjusted logistic regression analyses, each standard deviation (SD) increase in skin color decreased the odds of smoking abstinence at 4 weeks (OR = 0.62 [95% CI = 0.48, 0.79]) and 26 weeks (OR = 0.67 [95% CI = 0.52, 0.87]) post-baseline visit. However, skin color was not directly associated with smoking abstinence 4- (OR = 0.82 [95% CI = 0.55, 1.23]) or 26- (OR = 0.76 [95% CI = 0.49, 1.19]) weeks post-baseline visit, after adjusting for covariates, including race (non-Hispanic White vs. non-Hispanic Black), sex (male vs. female), age, education, annual household income per household member, homeownership, baseline cotinine concentration, neighborhood problems, the number of sessions completed, and overall tanning capacity.

Interaction terms were introduced into the model to determine whether the effect of skin color was moderated by sex and race. As shown in Table 2, there was no evidence of a three-way interaction between skin color, race, and sex (i.e., skin color x race x sex) or a two-way interaction between skin color and race (i.e., skin color x race). However, there was strong evidence in the unadjusted and adjusted analyses that the effect of skin color on smoking abstinence was moderated by sex (i.e., a skin color x sex interaction). In addition, the introduction of the skin color x sex interaction term into the model significantly improved the fit of the model predicting smoking abstinence at 4 weeks (χ² = 5.42, DF = 1, p = 0.02) and 26 weeks post-baseline visit (χ² = 7.30, DF= 1, p = 0.01). Among males, each SD increase in skin color significantly decreased the odds of smoking abstinence at 4 weeks (OR = 0.60 [95% CI = 0.36, 0.99]) and at 26 weeks post-baseline visit, (OR = 0.52 [95% CI = 0.29, 0.91]). Among females, skin color was not associated with smoking abstinence at weeks 4 (OR = 1.12 [95% CI = 0.70, 1.78]) or 26 (OR = 1.16 [95% CI = 0.69, 1.96]).

Table 2.

The interaction of skin color, race, and sex and the effect on short- and long-term smoking abstinence (0 = abstinent, 1 = smoking; N=449)

Variable	Week 4 smoking status		Week 26 smoking status
	OR (95% CI)	aOR (95% CI)	OR (95% CI)	aOR (95% CI)
	3-way interaction
Skin color x Race x Sex	0.93 (0.18, 4.74)	0.72 (0.13, 4.08)	1.37 (0.23, 8.15)	0.84 (0.11, 6.26)
Skin color x Race	0.83 (0.25, 2.74)	0.94 (0.27, 3.28)	0.74 (0.21, 2.65)	1.01 (0.24, 4.21)
Skin color x Sex	1.70 (0.46, 6.31)	2.08 (0.50, 8.61)	1.59 (0.36, 6.97)	2.66 (0.49, 14.51)
Race x Sex	1.12 (0.25, 5.17)	1.24 (0.24, 6.34)	1.01 (0.20, 5.15)	0.86 (0.13, 5.59)
	2-way interaction
Skin color x Race	0.79 (0.35, 1.76)	0.78 (0.33, 1.82)	0.88 (0.37, 2.12)	0.94 (0.35, 2.51)
Skin color x Sex	1.71 (1.02, 2.86)*	1.89 (1.10, 3.25)*	1.89 (1.10, 3.24)*	2.27 (1.24, 4.14)*

Note: Skin color was assessed at baseline using the overall score from multiple melanin readings from a reflectometer (i.e., top of left hand, top of left arm, and forehead). The overall skin color score was standardized. Past 7-day smoking status was self-reported and biochemically-confirmed (i.e., cotinine concentration). Missing data for smoking status was imputed using penalized imputation (i.e., missing = smoking). The adjusted analyses controlled for race (non-Hispanic White vs. non-Hispanic Black), age, education, annual household income per household member, home ownership, baseline cotinine concentration, neighborhood problems, the number of sessions completed, and tanning capacity.

^*p ≤ .05

^**p ≤ .10

The breakdown, by race, of the proportion of males and females with skin tones that were, 0, 1, 2, and 3 SD below or above the mean and the observed and predicted prevalence of smoking abstinence, are shown in Table 3. The majority of White participants’ skin complexion fell within the mean or 1 SD below the mean, whereas most Black participants typically had skin complexion above the mean. As shown in Table 3, the predicted probabilities of smoking abstinence decreased at weeks 4 and 26, respectively, as skin color became darker among males, but not females. For instance, a male whose skin color was one SD below the average (1% Black, 76.1% White) had a 19% chance of being abstinent at week 26, whereas a male whose skin color was one SD above the average had a 6% chance of being abstinent (43.8% Black, 1% White; see Figure 1). Comparatively, the predicted probabilities of smoking abstinence at week 26 for a female whose skin color was one SD below or above the average were very similar (9% vs. 12%, respectively).

Table 3.

The observed prevalence and predicted probabilities of short- and long-term smoking abstinence among males and females based on values of skin color

Skin Color	N		Smoking abstinence at 4 weeks		Smoking abstinence at 26 weeks
	Whites	Blacks	Observed prevalence	Predicted probabilitya	Observed prevalence	Predicted probabilitya
Females (N = 224)
− 2 SD	0	1	0%	23%	0%	8%
− 1 SD	86	7	26%	25%	18%	9%
0	26	46	24%	27%	21%	10%
+ 1 SD	0	43	19%	29%	16%	12%
+ 2 SD	0	13	15%	32%	15%	13%
+ 3 SD	0	2	0%	34%	0%	15%
Males (N = 225)
− 1 SD	86	1	36%	35%	32%	19%
0	26	34	23%	25%	20%	11%
+ 1 SD	1	49	8%	16%	8%	6%
+ 2 SD	0	24	8%	11%	8%	3%
+ 3 SD	0	4	0%	7%	0%	2%

Note: Skin color was assessed at baseline using the overall score from multiple melanin readings from a reflectometer (e.g., top of the left hand, top of left arm, and forehead). The overall skin color score was standardized. Past 7-day smoking status was self-reported and measured biochemically (i.e., cotinine concentration). Missing data for smoking status was imputed using penalized imputation (i.e., intent-to-treat).

^aThe predicted probabilities were based on model coefficients, which adjusted for race (non-Hispanic White vs. non-Hispanic Black), age, education, annual household income per household member, home ownership, baseline cotinine concentration, neighborhood problems, the number of sessions completed, and tanning capacity.

Figure 1.

The effect of skin color on the predicted probability of smoking at 4 and 26 weeks post-baseline visit stratified by sex.

Note: Skin color was assessed at baseline using the overall score from multiple melanin readings from a reflectometer (e.g., top of the left hand, top of left arm, and forehead). The overall skin color score was standardized, where zero represents the average skin color of the entire sample. Past 7-day smoking status was measured biochemically (i.e., cotinine concentration). Missing data for smoking status was imputed using penalized imputation (i.e., intent-to-treat). Predicted probabilities were based on model coefficients, which adjusted for race (non-Hispanic White vs. non-Hispanic Black), age, education, annual household income per household member, homeownership, baseline cotinine concentration, neighborhood problems, the number of sessions completed, and tanning capacity.

We conducted sensitivity analyses that examined the association between skin color and smoking cessation using standardized melanin index scores taken form each measurement location (face, inner arm, and back of the left hand). Findings from the sensitivity analyses were largely consistent with our original analyses supporting the robustness of the findings reported.

Identifying potential mechanisms

Adjusting for race, the relation between skin color and the number of discrimination events was significant among males (r = 0.15, p = 0.02) but not females. Skin color was also positively correlated with cynicism/distrust (r = 0.14, p = 0.03) and neuroticism (r = .24, p < 0.01) among males, and these factors were not correlated with skin color among females (see Table 4). Urine cotinine concentration was not correlated with skin color among males (r = 0.06, p = 0.35) or females (r = 0.05, p = 0.49). Likewise, the TNE/creatine ratio was not correlated with skin color among males (r = 0.04, p = 0.59) or females (r = 0.04, p = 0.52). Though skin color was significantly correlated with perceived discrimination, cynicism/distrust, and neuroticism among males, results indicated that these factors did not mediate the association between skin color and smoking cessation (see Table 5).

Table 4.

Inter-correlations of variables males and females

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	Full Sample (N = 449)
Skin Color (1)	-	-	-	-	-	-	-	-
Tanning Capacity (2)	.31*	-	-	-	-	-	-	-
Frequency of discrimination (3)	.11*	.03	-	-	-	-	-	-
No. of discrimination situations 4)	.07	.01	.83*	-	-	-	-	-
Cynicism/distrust (5)	.14*	.02	.26*	.18*	-	-	-	-
Neuroticism (6)	.14*	.08	.29*	.23*	.16*	-	-	-
Cotinine concentration (7)	.05	.10*	−.02	−.01	−.04	−.03	-	-
Nicotine concentration (8)	.01	.03	−.03	−.03	−.07	−.05	.48*	-
	Females (n = 224)
Skin Color (1)	-	-	-	-	-	-	-	-
Tanning Capacity (2)	.24*	-	-	-	-	-	-	-
Frequency of discrimination (3)	.02	−.02	-	-	-	-	-	-
No. of discrimination situations 4)	−.02	−.06	.84*	-	-	-	-	-
Cynicism/distrust (5)	.09	.01	.24*	.17*	-	-	-	-
Neuroticism (6)	.04	.02	.30*	.21*	.15*	-	-	-
Cotinine concentration (7)	.05	.15*	.03	.03	−.02	−.06	-	-
Nicotine concentration (8)	.04	.06	.06	.06	.03	−.07	.49*	-
	Males (n = 225)
Skin Color (1)	-	-	-	-	-	-	-	-
Tanning Capacity (2)	.32*	-	-	-	-	-	-	-
Frequency of discrimination (3)	.15*	.01	-	-	-	-	-	-
No. of discrimination situations (4)	.12	.02	.81*	-	-	-	-	-
Cynicism/distrust (5)	.14*	−.07	.24*	.16*	-	-	-	-
Neuroticism (6)	.24*	.11	.28*	.25*	.15*	-	-	-
Cotinine concentration (7)	.06	.05	−.05	−.05	−.08	−.01	-	-
Nicotine concentration (8)	.04	.11	−.10	−.12	−.11	.02	.54*

Note: Skin color was assessed at baseline using the overall score from multiple melanin readings from a reflectometer (e.g., top of left hand, top of left arm, and forehead). The overall skin color score was standardized. Analyses adjusted for race (Whites vs. Blacks).

^*p ≤.05

Table 5.

Mediation models linking skin color to smoking status among males (0 = smoking, 1 = abstinent; N = 225).

Mediator (M)	X→M (a path)		M→Y (b path)	X→Y (c’path/direct effect)	X→M→Y (ab path/indirect effect)
	B	SE	OR (95% CI)	OR (95% CI)	OR (95% CI)a
		Week 4 Smoking Abstinence
Frequency of everyday discrimination	1.14	0.49	0.97 (0.90, 1.04)	0.67 (0.37, 1.22)	0.96 (0.84, 1.05)
Cynicism/Distrust	0.34	0.16	0.85 (0.69, 1.04)	0.69 (0.38, 1.26)	0.95 (0.85, 1.01)
Neuroticism	2.39	0.65	1.01 (0.96, 1.06)	0.63 (0.35, 1.15)	1.02 (0.87, 1.21)
		Week 26 Smoking Abstinence
Frequency of everyday discrimination	1.14	0.49	0.97 (0.90, 1.04)	0.53 (0.28, 0.99)	0.96 (0.84, 1.05)
Cynicism/Distrust	0.34	0.16	0.87 (0.70, 1.07)	0.54 (0.28, 1.01)	0.95 (0.83, 1.02)
Neuroticism	2.39	0.65	1.02 (0.96, 1.07)	0.49 (0.26, 0.93)	1.04 (0.93, 1.21)

Note: X = independent variable (overall skin color at baseline), M = mediator (frequency of discrimination, cynicism/distrust, and neuroticism at baseline), and Y = dependent variable (smoking abstinence 4 and 26 weeks post-baseline visit). Analyses included race as a covariate. Adjusting for race but not adjusting for any potential mediator, the effect of race on week 4 and 26 verified abstinence (OR, 95% CI of lower limit-95% CI of upper limit) was 0.64 (0.36-1.16) and 0.51 (0.27-0.95), respectively. Bolded text indicates a significant association.

^aAnalyses included race as a covariate.

^bBias corrected bootstrapped confidence intervals (10000 bootstrap samples).

DISCUSSION

The current study examined the association between skin color, a characteristic closely associated with race and ethnicity, and smoking cessation among Black and White adults enrolled in a smoking cessation intervention study. Findings indicated that darker skin color, most common among Black males, was associated with a reduced likelihood of achieving abstinence among males, but not females, at 4- and 26-weeks after the baseline visit. Skin color was not correlated with nicotine or cotinine concentration overall, or among males and females specifically. Notably, darker skin color among males was also associated with experiencing more discrimination and more distrust of others, and perceiving the world as distressing, threatening, and unsafe (i.e., neuroticism); however, these factors did not explain the reduced likelihood of smoking cessation. Study findings are particularly relevant for Black males because their skin complexion is much darker compared to most other racial and ethnic groups (Alaluf et al., 2002). Skin color may contribute to cessation-related health disparities among Black males (Henley et al., 2016), but more research is needed to understand the biological and psychosocial mechanisms through which skin color influences tobacco cessation.

King et al. hypothesized that while Blacks begin smoking at a later age and smoke at a lower frequency than Whites (Freedman et al., 2011; Trinidad et al., 2011), low cessation rates among Blacks might be explained by greater nicotine dependence due to greater accumulation and slower release of nicotine from melanin-containing tissues (King et al., 2018, 2009). However, skin color was not significantly correlated with baseline (pre-quit) urinary cotinine or nicotine concentrations in the full sample or by sex, suggesting that higher melanin-related nicotine accumulation did not occur before quitting. It is still possible that melanin-related nicotine clearance was slower for those with darker skin color when they were reducing cigarette use or quitting, but that was beyond the scope of the parent study. Other research has shown that Blacks have slower metabolic clearance of nicotine to cotinine and also slower clearance of cotinine from the bloodstream than Whites (Benowitz et al., 2009; Pérez-Stable et al., 1998). More research is needed to determine whether nicotine and cotinine accumulation and clearance varies by skin color before and after a quit attempt.

Although current findings indicated that skin color was positively correlated with pre-quit reports of everyday discrimination frequency among males, discrimination frequency was not found to mediate the association between skin color and smoking cessation. This finding is surprising given previous research indicating that individuals smoke to cope with negative feelings and sensations caused by discrimination (Parker et al., 2016; Purnell et al., 2012), and everyday discrimination lowers confidence to resist smoking in high-risk situations (Alexander et al., 2019. Discrimination also reduces the willingness to trust others and increases cynicism and neuroticism (Sutin et al., 2016). One potential explanation for the unexpected finding is that this study only measured everyday discrimination (e.g., minor incidents often involving discourteous behavior), and Kendzor et al. (2014b) previously found that everyday discrimination was not associated with the likelihood of smoking cessation, whereas reporting one or more major discrimination events (e.g., acute and severe incidents with potentially long-lasting impact) decreased the likelihood of achieving longer-term abstinence. Future studies of skin color might consider assessing multiple types of discrimination (e.g., every day and major discrimination). Also, this study measured discrimination before the quit attempt, and future studies may want to measure major and everyday discrimination before and after the quit attempt to determine whether the timing and type of discrimination experiences are important. In addition to discrimination, researchers might consider other psychosocial stressors, such as financial strain, relationship distress, psychological work stress, perceived inequality, and childhood adversity. Discrimination is only one domain of stress, and research shows that Blacks have higher levels of stress than Whites across many domains (Hatch and Dohrenwend, 2007), and these psychosocial stressors increase their risk of smoking and smoking relapse (Slopen et al., 2013, 2012). Importantly, larger sample sizes may be needed to ensure adequate power to evaluate discrimination and other psychosocial stressors as mediators of the relation between skin color and cessation.

Findings also indicated that skin color did not influence smoking cessation among females, and more specifically, cessation rates were low among females regardless of skin color (see Figure 1). Previous research has shown that females have more difficulty quitting smoking than males (Smith et al., 2016). Further, relative to males, females have many unique challenges to quitting including smoking for weight control, smoking to cope with negative affect, greater dependence, and hormonal changes throughout the menstrual cycle that influence withdrawal symptoms (Nakajima and Al’Absi, 2012; Smith et al., 2016). Thus, darker skin color may not confer additional risk for smoking following a quit attempt for females. More notably, society seems to treat males and females with darker skin color differently. For instance, a recent study showed that darker skin color was associated with higher levels of perceived discrimination among males only (Assari and Caldwell, 2017). Also, compared to females, males have been shown to respond to discrimination by using tobacco and engaging in other poor health behaviors (Brodish et al., 2011), though this relationship was not confirmed in the current study. Overall, additional research is needed to gain a better understanding of sex differences in the impact of skin color.

Findings from this study may inform tobacco cessation treatment among males with darker skin color, which would be especially relevant to Black males. Clinicians who provide pharmacotherapy for smoking cessation might consider providing higher dosages of medication and extending the duration of therapy. Second, because males with darker skin color in the current study reported more discrimination than their females counterpart, and because discrimination has previously been linked with depression (Pascoe and Smart Richman, 2009), clinicians might consider implementing depression screening and the provision of pharmacological treatment for depression in conjunction with traditional smoking cessation treatments. Pharmacological treatment could be supplemented with specific psychosocial treatments, such as mindfulness-based therapy (Halland et al., 2015). Mindfulness-based therapy has been shown to mitigate the depressive symptoms and physical sensations caused by discrimination (Hofmann et al., 2010) and improve cessation outcomes (Heppner et al., 2016; Vidrine et al., 2016). Strategies designed to increase positive perceptions and challenge and reduce negative beliefs and expectations about those with darker skin color (e.g., media communications, public health campaigns educational approaches) are also worth considering.

This study has some limitations. First, 84% of Black participants and 57% of White participants could be considered low-income (i.e., earning ≤ 200% of the federal poverty level), and therefore study findings may be less generalizable to adult cigarette smokers of higher socioeconomic status. Relatedly, the present study also excluded Hispanics/Latinos, Asians, and individuals of other racial/ethnic backgrounds with darker skin color. Also, it is possible that there were other unmeasured genetic or biological factors that might confound the association between skin color and smoking cessation. Past research has shown that Blacks in the United States share some genetic ancestry with Europeans or Whites (Bryc et al., 2015), and Blacks with lighter skin may have a higher percentage of European genetic ancestry than Blacks with darker skin color (Parra et al., 2004). Therefore, darker skin color could be a proxy for other physiological characteristics that might increase the risk of smoking after a quit attempt.

Future research may benefit from several refinements. First, future studies on smoking and skin color might include self-report and interviewer-rated measures of skin color because some studies have shown that perceptions about skin color have stronger associations with health and health behavior than objective measurements (Dixon and Telles, 2017; Gravlee and Dressler, 2005). Also, self-reported and interviewer-rated measures of skin color characterize perceptions that might better reflect the social hierarchy associated with skin color than objective measures (Gravlee and Dressler, 2005). More broadly, the inclusion of skin color measurements within studies that address tobacco-related health disparities may result in the collection of more nuanced information regarding the impact of race and ethnicity on tobacco cessation and tobacco-related diseases (NCI, 2017). Second, future research should also include hypothesis-driven and optimally timed measurements of potential mechanisms linking skin color with smoking cessation, such as discrimination, negative affect/depression, self-efficacy, and nicotine metabolism. Finally, previous research on skin color and health has focused on Black and Latino populations (Dixon and Telles, 2017), while findings from this study support a future focus on samples with greater racial/ethnic diversity. Ultimately, it will be important to determine whether skin color influences the development of tobacco-related morbidity and mortality independent of race/ethnicity. Overall, skin color may play an essential role in the persistence of tobacco-related health disparities, and targeted strategies may be needed to increase cessation rates among those with darker skin color.

• Among males, darker skin color increases the odds of smoking after a quit attempt.

• Among females, darker skin color does not predict smoking after a quit attempt.

• Skin color among males only is linked to discrimination, distrust, and neuroticism.

• However, these factors failed to explain the sex difference in smoking cessation.