Exposure to racial insults among youth in Jujuy, Argentina, was examined as a factor associated with smoking behavior.
Youth sampled from eighth-grade classes in 27 randomly selected middle schools completed annual surveys in the ninth and tenth grades. Demographics, race/ethnicity (Indigenous/Amazonian, Indigenous/Andean, Indigenous unspecified group, Mixed European-Indigenous, European), cigarette smoking, and other attitudinal and behavioral factors were measured. Exposure to racial insults, measured in the ninth grade, was modeled to predict cigarette smoking in the previous 30 days (defined as current) in the tenth grade conditional on ninth grade smoking.
Of the 3,122 respondents, 35.5% reported exposure to racial insults and 33.8% were current smokers. Factors associated with racial insults were being male, indigenous language spoken at home, ever and current smoking, smoking in a ceremonial context, exposure to second-hand smoke at home, number of friends who smoke, having low expectations for the future, low identification with conforming role models, high identification with defiant role models, and depressive symptoms. Reported exposure to racial insults increased the risk of current smoking in the 10th grade among Indigenous Amazonian respondents (OR = 3.8; 95% CI 1.4–10.4) and among the Indigenous-unspecified group (OR = 1.8; 95% CI 1.1–2.8), but not among European or Indigenous Andean youth.
Exposure to racial insults is commonplace among youth in Jujuy. Evidence of a longitudinal effect of ninth-grade racial insults on tenth-grade smoking rates provides support for an association of racial insults with smoking behavior.
In South America race and ethnicity have been singled out as enduring determinants of opportunities. Indigenous peoples are the largest disadvantaged group and continue to suffer from a greater incidence of disease and discrimination [1, 2]. In northwest Argentina, the predominant population of Andean or mountain peoples have for millennia adapted to a sedentary, agriculture-based lifestyle. However, Amazonian peoples with a nomadic way of life based on the integrity of their forest environment have had greater difficulty in coping with the encroachment and demands of western development.
Racial prejudice and discrimination have emerged as important health risk factors in a growing number of studies that document an association with indicators of poor physical and mental health [3–7]. Definitions of racism refer to an organized system that assigns hierarchical status to specific groups and uses this ranking to preferentially allocate societal goods and resources to those who are regarded as inherently superior . In the US context, racism leads to the development of negative attitudes and beliefs toward, and different treatment of, specific groups .
Although associations between racial discrimination and health outcomes have been reported in US samples [4, 6, 8, 10], the mechanisms underlying these associations remain unclear. It has been hypothesized that discrimination may lead to negative health outcomes by promoting stress and coping through unhealthy behavior . Several studies have focused on health risk behavior, for example smoking, excessive alcohol consumption, or illicit drug use [12–15]. In cross-sectional studies racial discrimination has been associated with cigarette smoking among immigrant adolescents in the US [16–18], and among adolescents in Johannesburg, South Africa . Among adults, a longitudinal study showed that African Americans experiencing racial discrimination had higher risk of current tobacco use. Other studies using cross sectional data reported positive associations between racism and smoking among adult African Americans [12, 20], ethnic minorities, and Whites in the US , and among the Maori in New Zealand .
Prevalence of youth smoking in Latin America continues to be among the highest in the world, with about 20% of the youth in the cities of Chile, Colombia, Ecuador, and Argentina currently smoking , but data are limited on smoking behavior among indigenous youth . Furthermore, there are no published studies on the effects of racial/ethnic discrimination as determinants of smoking behavior in these populations. This study examines the relationship between exposure to racial insults and smoking behavior, in a longitudinal, multiethnic sample of youth from the Province of Jujuy, Argentina.
The study was conducted in the province of Jujuy, Argentina, a poor area in the northwest on the border with Bolivia and Chile. The population is diverse, and includes Indigenous populations of the mountain areas (Andean) and of the lowland forests (Amazonian), Europeans predominantly of Spanish background, and individuals of mixed Indigenous-European origin.
The study methods, which have been described elsewhere , are summarized below. A random sample of 27 middle schools was selected from within the three geographic areas of Jujuy: the mountain region; the provincial capital; and the agricultural lowlands. The baseline survey was self-administered in 8th-grade classrooms between June and August of 2004 and follow-up surveys were conducted in 2005 (9th grade) and 2006 (10th grade). Students who completed all three surveys are the focus of this analysis. In follow-up, respondents who dropped out of school or who changed school were interviewed in their homes. The UCSF Committee on Human Rights and an NIH-certified human subjects research board in Buenos Aires based at Centro de Educación Médica e Investigaciones Clínicas (CEMIC) approved the research protocol. Passive consent was requested from parents or caretakers and students signed an active consent form that also allowed follow-up contact.
The smoking behavior items were adapted and translated from US youth surveys  and new items were developed from qualitative research. Current smokers were defined as having smoked at least one puff in the previous 30 days. The subjects were also asked about exposure to second-hand smoke at home and about the number of friends who smoked. To address the use of tobacco in ceremonial contexts among respondents who reported ever smoking, we asked in what place or situation they had first tried smoking and where they felt like smoking the most. Responses that mentioned the Pachamama and Todos Santos celebrations were categorized as “ceremonial smoking”.
Racial discrimination was measured by assessing exposure to racial insults commonly used in the study site that denigrate the physical appearance or intellectual capacity of Indigenous individuals. These included items asking whether on the street or in school a respondent had ever been called “coya tonto”, “chaguanco”, or “indio cabeza dura”, or “boliviano/bolita”; and whether someone had used the terms “coya”, “chaguanco”, “indio”, or “boliviano/bolita” just to bother the respondent. Items were developed through qualitative work with the target population. Youth participating in semi-structured interviews provided information about the situations, locations, and wording used in racial insults. Exposure to racial insults was assessed over the lifetime and each item was coded as having occurred or not.
Respondents reported their sex, date of birth, age, and religion. The geographical location of the school was noted. Low socioeconomic status (SES) was defined as having parents who had not completed elementary school or who were welfare recipients compared with others with higher achievements. Ethnic self-identification questions were derived from previous qualitative work resulting in a list of five ethnic categories: Indigenous, mixed Indigenous and European, European, Arab and Other. If respondents had self-categorized as Indigenous or mixed, in a follow up question they were asked to select from a list of Andean and Amazonian Indigenous groups in the region. We combined youth’s responses to the ethnic self-identification questions to create the following race/ethnic categories: European, Indigenous-European mix, Indigenous of unspecified group, Indigenous Andean, Indigenous Amazonian. The Indigenous-unspecified group resulted when youth were not able to identify a specific group (Amazonian or Andean) in the follow up question even though they acknowledged their Indigenous origin. Respondents reported on whether an Indigenous language was spoken at home, even if only a few words, and whether theirs was a two-parent household.
Positive expectations for the future was measured with the following questions that provided answers with response categories of always, almost always, sometimes, almost never, and never: “I think that when I am older I will be able to have all that I need” and “I think that when I am older I will be able to work in what I like”. Questions with a similar response structure were used to measure respect for parents “I obey my parents or the people that care for me”, parent support “My parents or the people that take care of me understand me”, concern for family “I am concerned about the people in my family”, and social concern “I am saddened by the fact that there are people who do not have food”, “I am saddened by the fact that there are people who do not have a job”. Religiosity was measured with a scale ranging from 1 (not important) to 10 (very important). Respondents self identified with role models by choosing from a list of socially conforming and defiant role models. Identification with “athletes” was used as an indicator of a conforming role model. “Villeros” is a negative stereotype of shanty-town dweller youth constructed by mainstream society and was used as an indicator of a defiant role model. To assess concern for body image we asked whether respondents were trying to lose, gain, or maintain their weight, or were not concerned about their weight.
We ascertained the presence of depressive symptoms in the past year on the basis of previous results showing an association with current smoking [23, 25]. Similarly, we used items to measure thrill-seeking orientation [23, 26].
Standard errors and confidence intervals were estimated via the Taylor series method  to accommodate clustering of students within schools. We cross-tabulated racial insults measured in the 9th grade by current smoking measured in the 10th grade. We also cross-tabulated the racial insult measure by smoking behavior factors identified in our previous research , including demographic, and attitudinal variables. Explanatory variables assessing stable individual characteristics were assessed in the eighth grade: sex, religion, poverty, ethnicity, region, and Indigenous language spoken in the family. Other explanatory variables describing relatively mutable risk factors were assessed in the ninth grade.
Logistic regression models were fit to estimate the longitudinal effect of exposure to racial insults in the ninth grade on current smoking in the tenth grade, controlling for current smoking in the ninth grade and all covariates. We estimated adjusted odds ratios and 95% confidence intervals. The models tested for two-way interactions of discrimination with ethnicity, sex, region, and low SES. Within the models, any interaction effects with p-values greater than 0.05 were removed through backward elimination.
The 27 participating schools included a total of 4,276 registered eighth-grade students in 2004. Within each geographic stratum participation for the baseline sample was 81.5, 84.3, and 91.5%. For this analysis the sample only included students who were 13–15 years old at baseline and participated in the follow up surveys in which questions about racial insults were included. Ethnic groups with low frequency, Asians, Arabs, Indigenous Mapuche, and a mix of Andean/Amazonian Indigenous were excluded from this analysis leaving a final sample of 3,122. In 2005 and 2006, the ninth and tenth-grade sample sizes equaled 2,955 (94.7%) and 2,787 (89.3%), respectively.
The demographic characteristics of ninth-grade respondents were 53.7% girls, 63% age 13 or 14 years, and 24.7% who fit the low SES definition. The largest proportion self-identified as Andean (43.0%) followed by Indigenous of unspecified group (24.2%), Indigenous-European mix (21.1%), European (7.3%), and Amazonian (4.4%), and 33.6% reported an Indigenous language spoken in the family. The distribution of low SES varied by race/ethnicity with a greater percentage among Amazonian youth (31.8%), followed by Andeans (28.8%), Indigenous of unspecified group (23.2%), Indigenous-European mix (20.1%), and Europeans (10.6%) (p < 0.0001). The prevalence of current smokers was 33.8% and was higher among boys (37.2%) than girls (31.0%) (p = 0.004).
The Cronbach alpha for the racial insults items was 0.80 for the ninth-grade sample and 0.75 for the tenth-grade sample. Over one-third of respondents (35.5%) reported exposure to racial insults, which did not vary significantly by ethnicity but was significantly higher among respondents who reported an Indigenous language spoken at home (Table 1). The proportion of respondents who reported exposure to racial insults was also significantly higher among boys than girls and among current smokers. Several other variables were significantly associated with racial insults: older age, not having a two-parent household, smoking in ceremonial contexts, exposure to second hand smoke at home and having friends who smoke, having low positive expectations for the future, low levels of social concern, low identification with conforming role models, high identification with defiant role models, and having depressive symptoms.
Carbon monoxide (CO) was measured for a total of 427 participants in 8th grade, corresponding to 11.6% of the total sample. Among the 202 students who reported they were not smokers, only 1 (0.5%) had expired CO values of 10 ppm or more.
Several covariates significantly increased the risk of current smoking (Table 2). Boys and older youth were at higher risk, and likelihood of current smoking was almost twice for Andean respondents as for Europeans (adjusted OR 1.7; 95% CI 1.1–2.5). Although smoking in ceremonial contexts, identification with defiant role models, and having friends who smoked were risk factors, living in the mountain region, having parental support, and identification with conforming role models were significant protective factors. The main effect of racial insults on current smoking (p = 0.003) was significantly modified by ethnicity. Amazonian respondents who reported having ever experienced racial insults while in the ninth grade (2006) had almost four times the odds of current smoking in the tenth grade (2007), compared with respondents of this ethnic group who were not exposed to insults (adjusted OR = 3.8; 95% CI 1.4–10.4). Respondents of the Indigenous-unspecified group who experienced racial insults in the previous wave had relatively lower, but still significant increased, odds of current smoking in the tenth grade (adjusted OR 1.8; 95% CI 1.1–2.8), compared with those of the same group who did not report exposure to insults. A significant interaction effect was also found between living in the mountain region and experiencing discrimination (adjusted OR = 2.9; 1.8–4.9), versus living in the capital city or the lowland region.
This study suggested that racial/ethnic discrimination is commonplace among youth in this Latin American sample. This analysis provides evidence for a longitudinal effect of ninth-grade racial insults on tenth-grade cigarette smoking (conditional on ninth-grade smoking) among Amazonian and Indigenous-unspecified ethnic groups. Within these groups the effects persisted even after controlling for a range of factors known to be strongly associated with smoking in this population and others [23, 28–31]. Amazonian Indians were the group most affected by perceived discrimination, resulting in a fourfold increase in the risk of current smoking, by far the largest among the variables tested in this study. The effect was also significant for youth of the Indigenous-unidentified group but not among the largest group of Andeans. Amazonian youth also had the highest rates of low SES among all ethnic/race categories, although we did not find evidence of a relationship between low SES and smoking behavior.
These findings emphasize the hierarchical nature of Latin American society and provide an indication of its potential health impact . Indigenous peoples are positioned in contexts of unequal power relations that establish multiple levels of top-down discriminatory practices . Thus, achieving equity in social and health assets will involve overcoming intra and inter-group barriers. Social boundaries have been created, among other means, by rhetoric that devaluates the other persons . Amazonian peoples, in particular, are a minority in a society that places them at the bottom of the social stratification system and this may result in increased sensitivity to discrimination. Andeans as the largest ethnic group in Jujuy may actually be more resilient to discrimination and to its association with cigarette smoking. Youth who self-identified as Indigenous but did not recognized a specific group of reference (Andean or Amazonian) may have more difficulties dealing with identity issues because they may not have the support of such a group. Thus, they may be more emotionally vulnerable to racial discrimination. These results are similar to findings in the US showing that racial and ethnic discrimination constitutes a significant component of adolescents’ daily experiences [3, 35–37] and strengthens the evidence from cross sectional data supporting a significant association of perceived discrimination with smoking [16–19].
Our study is subject to several limitations. Our exposure variable, racial insults, was measured through self-report from respondents. Individual emotional and psychological factors may affect both recollection and the reporting of discrimination events that could lead to underestimation by youth with high self-esteem or overestimation by youth emotionally susceptible to aggressive behavior. Respondents with positive expectations for the future reported racial insults in a significantly lower proportion than youth with low expectations. Similarly, in bivariate analysis a higher percentage of youth with depressive symptoms reported exposure to racial insults but these variables were controlled for in regression models. Our data on smoking are based on self-reports from students who might under or over-report their behavior, but is the standard used in youth surveys and generally found to be valid and a predictor of future smoking . The CO tests conducted in our study indicate that the incidence of under-reporting of smoking was very low.
These findings are relevant in that they contribute to overcoming the dearth of data on the relationship between racism and health status in Latin America. However, this is an initial step toward expanding the research agenda. On one hand there is need to develop methodological tools. Our discrimination variable for example, measured only one domain, racial insults. New, more comprehensive and locally relevant measures should be developed to include the diverse domains of the racial discrimination experience. Other forms of discrimination, for example gender or class-based, should also be explored. In addition, the results are conducive to addressing a set of new research questions that may include the assessment of new risk factors and mediator variables, particularly in the cultural domains, examining differences by the characteristics of the receiver and the sender and the context of exposure.
This research was funded by grant no. TW05935 from the Tobacco Research Network Program, Fogarty International Center, National Cancer Institute, National Institute of Drug Abuse, and National Institutes of Health and by grant no. 001726-037 from Research on International Tobacco Control, International Development Research Center, Canada. We thank Dana Nickleach for invaluable help in data analysis, Constanza Almiron for critical support in survey development and data management, Marina Bejarano, Susana Durán y Carina Delgado who administered surveys and supported the research work in Jujuy, and Elvira Gomez, Cambria Garrell, and Cecilia Populus-Eudave for administrative and research support at UCSF.
Conflict of interest There is no conflict of interest for any of the authors.
Ethel Alderete, Consejo Nacional de Investigaciones Científicas y Técnicas, Instituto de Ciencia y Tecnología Regional, Universidad Nacional de Jujuy, Jujuy, Argentina.
Madalena Monteban, Department of Anthropology, University of Georgia, Athens, GA, USA.
Steve Gregorich, Division of General Internal Medicine, Department of Medicine, Medical Effectiveness Research Center for Diverse Populations, University of California San Francisco, 3333 California Street, San Francisco, CA 94143-0856, USA.
Celia P. Kaplan, Division of General Internal Medicine, Department of Medicine, Medical Effectiveness Research Center for Diverse Populations, University of California San Francisco, 3333 California Street, San Francisco, CA 94143-0856, USA.
Raúl Mejía, Universidad de Buenos Aires, Buenos Aires, Argentina.
Eliseo J. Pérez-Stable, Division of General Internal Medicine, Department of Medicine, Medical Effectiveness Research Center for Diverse Populations, University of California San Francisco, 3333 California Street, San Francisco, CA 94143-0856, USA.