Subjective social status,self-rated health and tobacco smoking: Brazilian Longitudinal Study of Adult Health (ELSA-Brasil)

Abstract

Using baseline data from ELSA-Brasil (N = 15,105), we investigated whether subjective social status, measured using three 10-rung “ladders,” is associated with self-rated health and smoking, independently of objective indicators of social position and depression symptoms. Additionally, we explored whether the magnitude of these associations varies according to the reference group. Subjective social status was independently associated with poor self-rated health and weakly associated with former smoking. The references used for social comparison did not change these associations significantly. Subjective social status, education, and income represent distinct aspects of social inequities, and the impact of each of these indicators on health is different.

Keywords

ELSA-Brasil health inequities self-rated health subjective social status tobacco smoking

Introduction

Health inequities do not result exclusively from absolute income levels, education, or type of occupation but also from people’s relative position in social hierarchy (Marmot, 2004; Wilkinson, 1997). The subjective social status (SSS) measure aims to capture individuals’ perception of their place in social hierarchy (Davis, 1956) and their sense of belonging to a certain social stratum (Rosenberg, 1953). It intends to seize former and current socioeconomic situation, future perspectives, household resources, life opportunities, how people experience society, and how they perceive themselves in relation to others (Singh-Manoux et al., 2003). SSS probably reflects the socioeconomic conditions throughout life (Chen et al., 2012) and allows the investigator to apprehend a relevant dimension of social stratification, which cannot be measured by objective indicators (Adler et al., 2000).

There is a consistent association between SSS and self-rated health (SRH) (Demakakos et al., 2008; Hu et al., 2005; Operario et al., 2004; Singh-Manoux et al., 2003, 2005). Low SSS has also been associated with poor mental and physical health (Adler et al., 2008; Chen et al., 2012; Demakakos et al., 2008; Singh-Manoux et al., 2005; Subramanyam et al., 2012) as well as with some health-related behaviors, such as smoking (Finkelstein et al., 2006; Reitzel et al., 2011; Ritterman et al., 2009). In longitudinal studies, SSS predicted changes in the health status of adults (Singh-Manoux et al., 2005) and the elderly people (Chen et al., 2012) over and above conventional objective social status, including occupation (Singh-Manoux et al., 2005).

The MacArthur Scale is a tool often used to assess the SSS in epidemiologic studies (Adler et al., 2008; Demakakos et al., 2008; Singh-Manoux et al., 2005). The scale employs two SSS ladders with two distinct reference groups. The “society ladder” uses a distal reference group and measures people’s perception of their status when compared with others in general society, taking into account income, education, and occupation. The “community ladder” uses a proximal reference group and refers to people’s standing compared to others from their local community, considering their living standards (Adler and Stewart, 2007). Most studies that analyze associations between SSS and health indicators have adopted only distal references (Adler et al., 2000, 2008; Demakakos et al., 2008; Operario et al., 2004; Singh-Manoux et al., 2005).

The Brazilian Longitudinal Study of Adult Health (ELSA-Brasil) has included both MacArthur Scales of SSS adapted to Portuguese and an additional ladder, which uses the workplace as a reference group for comparison (work ladder). This work ladder was created because ELSA-Brasil is an occupational cohort, and we intended to assess whether job-related social status also contributes to social inequities in health (Giatti et al., 2012).

Inequality and poverty levels in Brazil have decreased throughout the last several years, but the Gini index remains among the highest in the world—0.50 (Instituto Brasileiro de Geografia e Estatística, 2011). So far, most studies on social inequities in health in the country are based on objective factors. However, a recent study has shown that SSS is also important in understanding the social gradient in health in Brazil (Macinko et al., 2012)

SRH is a strong predictor of hospitalization and mortality. It is also a very inclusive and comprehensive measure, as it incorporates elements of health-related behaviors, psychological aspects, and social well-being (Jylhä, 2009; Singh-Manoux et al., 2006). Tobacco smoking is the most important risk factor for chronic diseases. Its prevalence has decreased sharply in Brazil (34.8% in 1989 vs 14.8% in 2011) (Levy et al., 2012; Secretaria de Gestão Estratégica e Participativa, Secretaria de Vigilância em Saúde, Ministério da Saúde, 2012), but not evenly among all socioeconomic groups. Smoking is becoming a problem of the most socially unprivileged portion of the population. The association between SSS and SRH has been shown to be consistent in literature, but it has not yet been studied in the Brazilian context. As for the association between SSS and tobacco smoking, results from other studies are inconsistent.

This study has the following objectives: (1) to investigate whether SSS is associated with SRH and tobacco smoking, independently of objective social status and depression symptoms, among civil servants participating in ELSA-Brasil and (2) to explore whether the magnitude of these associations varies according to the reference group used by the SSS ladders.

Our main hypotheses were as follows: (1) lower levels of SSS would be associated with poor SRH and with current and former smoking, (2) these associations would be independent of objective social status, and (3) these results would not be attributed to depression symptoms. Additionally, we hypothesized that these associations would be stronger in the society ladder, as the social distribution of the participants in their local community and workplace tends to be more homogenous than in the society as a whole.

Methodology

This study used the baseline data from ELSA-Brasil 2008–2010 (Release 15, January 2013), which enrolled 15,105 civil servants, aged between 35 and 74 years, from five universities and one research institute. The baseline examination included detailed face to face interviews. The participants ranged from highly skilled professionals (lecturers and researchers) to unskilled manual workers with different degrees of schooling and ethnicities. This study population was chosen in order to ensure a satisfactory level of retention during the follow-up. ELSA-Brasil was approved by the National Research Ethics Commission (CONEP) (Aquino et al., 2012).

Measures

Health outcomes

SRH was assessed with the following question: In general, compared to other people your age, would you say your health is (very good, good, fair, poor, or very poor)? In order to allow comparisons with other studies, these options were grouped into two categories: good (very good and good) and poor (fair, poor, and very poor). The category “good” was used as reference.

Smoking was assessed by the following questions: “Are you or have you ever been a smoker, that is, have you smoked at least 100 cigarettes (five cigarette packs) throughout your life?” and “Do you currently smoke cigarettes?” (never-smokers, former smokers, and current smokers). Never-smokers were the reference group.

SSS

The MacArthur Scale of SSS (society and community ladders) and the work ladder, created by researchers from ELSA-Brasil, were used (Giatti et al., 2012). The scales were translated from English to Portuguese by two Brazilian researchers fluent in English and underwent testing and adjustments. When translating the community ladder to Portuguese, we noticed that the word community was often understood as “slum” by the Brazilian population. Therefore, in order to maintain the scale’s original meaning, the word community was replaced by neighborhood. (Giatti et al., 2012).

The study, which evaluated the reliability of the MacArthur Scales of SSS and the work ladder, in a subsample of ELSA-Brasil participants, showed that the ladders presented a substantial agreement. The intraclass correlation coefficient was 0.67 in the society ladder, 0.64 in the community ladder, and 0.75 in the work ladder (Giatti et al., 2012).

The MacArthur’s ladder, which has an illustrated format of a ladder with 10 rungs, was presented to each participant with a card and he or she was asked to locate himself or herself in relation to the reference population (Giatti et al., 2012). For instance, in the work ladder, they were asked,

Consider this ladder as representing where people stand in their workplace. People define work in different ways; please define it in whatever way is most meaningful to you. On the top of the ladder are people who have the most valued jobs, as the director or the president, for example. At the bottom of the ladder are people who hold the less valued jobs. Considering your work, where would you place yourself on this ladder?

In this analysis, the answers to the three SSS ladders were grouped into four categories: “Low” (rungs 1–4), “Medium” (rungs 5 and 6), “High” (rungs 7 and 8), and “Very high” (rungs 9 and 10). We adopted such categorization because the scores presented an asymmetric distribution, skewed to the right. Other studies have also used the SSS as a categorical variable, but no cutoff point has ever been established (Chen et al., 2012; Hu et al., 2005; Miyakawa et al., 2012; Singh-Manoux et al., 2003, 2005).

Objective indicators of social position

Education level was categorized as follows: postgraduation, university degree, high school, complete elementary school, and incomplete elementary school.¹ The per capita household income was calculated using the monthly net household income, divided by the total number of people living in the household, and was categorized into quintiles.

Depression

Participants who presented depressive disorder or mixed anxiety and depressive disorder according to their answers to the Clinical Interview Schedule–Revised (CIS-R) were classified as having depression symptoms.

Other sociodemographic variables

Three demographic characteristics were assessed as covariates: gender, age, and race/color (White, Brown, Black, Asian descent, and Brazilian indigenous).

Data analysis

Initially, we described the sociodemographic characteristics of the participants in the study and the SSS. The presence of correlation between the SSS ladders and education and per capita household income was assessed by Spearman’s correlation coefficient.

The analysis was performed using logistic regression for SRH and multinomial logistic regression for smoking. The results were presented as odds ratio (OR) and its 95 percent confidence intervals (CIs) and p-values obtained from the Wald test. We verified whether health outcomes were independently associated with SSS by adding, one by one, all the covariables considered in this study.

The multicollinearity between the explanatory variables was assessed by the variance inflation factor (Neter et al., 1996). The significance level used was 5 percent. The analyses used the Stata 12.0 software (Stata Corporation, College Station, TX, USA).

Results

Among the 15,105 participants from the ELSA-Brasil baseline, the majority were women (54.4%) and White (52.2%); 39.3% percent were aged between 45 and 54 years, 52.6 percent had a university degree or postgraduation, and the average per capita household income was 915 dollars (Supplementary Table).

The distribution of the SSS was slightly skewed to the right, especially for the community and work ladders (Supplementary Figure). In all ladders, approximately only 10 percent of the participants placed themselves between the rungs 1–4 (Supplementary Figure and Supplementary Table).

The SSS ladders were moderately and positively correlated. The society ladder also showed moderate positive correlations with education and was weakly positively correlated with per capita household income. The community and work ladders presented weak positive correlation with per capita household income and education (Table 1).

Table 1.

Correlations (Spearman) between subjective social status, per capita household income, and education—Brazilian Longitudinal Study of Adult Health (ELSA-Brasil), 2008–2010.

	1	2	3	4	5
1 Subjective social status—society	1.00
2 Subjective social status—community	0.60*	1.00
3 Subjective social status—work	0.57*	0.48*	1.00
4 Per capita household income	0.39*	0.34*	0.34*	1.00
5 Education	0.44*	0.38*	0.38*	0.61*	1.00

Correlation is significant at p < 0.001.

The prevalence of poor (fair, poor, and very poor) SRH was 19.9 percent, prevalence of current smoking was 13.1 percent, and the prevalence of former smoking was 30 percent. In the univariable analysis, the prevalence of poor SRH was lower among White individuals, increased with age and with the presence of depression symptoms, and decreased as per capita household income and education level increased. Gender was not associated with poor SRH (Table 2). Former smoking was positively associated with older age and was less frequent among women and Black and Brown individuals and more frequent among individuals with lower education (Table 2). Current smoking was positively associated with the age ranges 45–54 years and 55–64 years, Brazilian indigenous race/color, lower per capita household income, education level, and depression symptoms. It was negatively associated with the female gender (Table 2). The prevalence of poor SRH among the “Low” SSS category was more than two times higher than observed among the “Very high” SSS category (Table 2). The prevalence of current and former smoking was also higher among those who declared lower SSS, in all three ladders (Table 2).

Table 2.

Univariate analysis of the association between health outcomes and sociodemographic variables, subjective social status, and depression symptoms—Brazilian Longitudinal Study of Adult Health (ELSA-Brasil), 2008–2010.

Characteristics			Smoking
	Poor self-rated health		Former smoking		Current smoking
	Prevalence (%)	OR (95% CI)	Prevalence (%)	OR (95% CI)	Prevalence (%)	OR (95% CI)
Gender
Male	20.0	1.00	35.3	1.00	14.3	1.00
Female	19.9	0.99 (0.92–1.08)	25.5	0.58 (0.54–0.63)*	12.1	0.68 (0.62–0.75)*
Age in years
35–44	13.6	1.00	16.0	1.00	10.3	1.00
45–54	19.5	1.54 (1.37–1.74)*	30.6	2.64 (2.37–2.95)*	16.1	2.15 (1.88–2.46)*
55–64	23.2	1.94 (1.71–2.19)*	37.1	3.40 (3.04–3.81)*	12.7	1.80 (1.56–2.09)*
65–74	26.4	2.30 (1.98–2.67)*	38.5	3.34 (2.90–3.84)*	8.4	1.13 (0.91–1.40)
Race/color
White	15.0	1.00	31.5	1.00	12.3	1.00
Brown	23.8	1.77 (1.61–1.95)*	28.7	0.89 (0.82–0.97)*	13.6	1.08 (0.96–1.21)
Black	27.9	2.19 (1.97–2.44)*	27.2	0.83 (0.75–0.93)*	14.6	1.14 (1.00–1.31)
Asian descent	21.4	1.54 (1.20–1.99)*	29.4	0.87 (0.69–1.10)	10.2	0.77 (0.54–1.09)
Brazilian indigenous	33.3	2.84 (2.02–3.98)*	30.1	1.05 (0.74–1.52)	17.8	1.58 (1.02–2.45)*
Per capita household income in US$
5th percentile (highest)	12.90	1.00	31.55	1.00	10.99	1.00
4th percentile	13.49	1.05 (0.90–1.23)*	29.80	0.90 (0.80–1.01)	9.86	0.85 (0.72–1.02)
3rd percentile	16.90	1.37 (1.18–1.60)*	28.92	0.89 (0.79–1.00)	11.99	1.06 (0.89–1.26)
2nd percentile	23.73	2.10 (1.82–2.43)*	29.32	0.96 (0.85–1.08)	15.09	1.42 (1.20–1.68)*
1st percentile (lowest)	31.90	3.16 (2.75–3.64)*	31.05	1.09 (0.97–1.23)	17.32	1.75 (1.49–2.06)*
Education
Postgraduation	11.4	1.00	29.6	1.00	8.8	1.00
University degree	14.1	1.28 (1.11–1.48)*	26.1	0.85 (0.76–0.95)*	10.2	1.12 (0.95–1.33)
High school	23.8	2.43 (2.19–2.69)*	29.4	1.13 (1.03–1.23)*	16.4	2.12 (1.87–2.39)*
Complete elementary school	35.4	4.26 (3.66–4.96)*	37.9	1.82 (1.57–2.11)*	18.9	3.06 (2.52–3.72)*
Incomplete elementary school	48.0	7.17 (6.14–8.38)*	37.0	1.85 (1.59–2.17)*	21.6	3.65 (3.00–4.46)*
Subjective social status—society ladder^a
Very high	13.29	1.00	31.29	1.00	10.01	1.00
High	15.91	1.23 (1.04–1.46)*	29.31	0.94 (0.82–1.07)	12.06	1.21 (0.99–1.47)
Medium	22.06	1.85 (1.56–2.19)*	29.22	0.96 (0.84–1.10)	13.98	1.44 (1.18–1.76)*
Low	34.45	3.43 (2.84–4.14)*	34.43	1.32 (1.12–1.55)*	16.56	1.98 (1.57–2.49)*
Subjective social status—community ladder^a
Very high	12.77	1.00	30.05	1.00	10.91	1.00
High	16.85	1.38 (1.20–1.59)*	29.41	0.99 (0.89–1.11)	12.45	1.16 (0.99–1.35)
Medium	23.24	2.07 (1.79–2.38)*	29.80	1.04 (0.93–1.17)	14.17	1.37 (1.16–1.61)*
Low	35.18	3.71 (3.14–4.38)*	33.67	1.31 (1.13–1.53)*	16.03	1.72 (1.41–2.12)*
Subjective social status—work ladder^a
Very high	14.99	1.00	31.77	1.00	12.09	1.00
High	16.47	1.12 (0.98–1.27)	29.02	0.87 (0.78–0.96)	11.98	0.94 (0.81–1.09)
Medium	23.32	1.72 (1.50–1.98)*	29.39	0.91 (0.81–1.02)	13.78	1.12 (0.96–1.32)
Low	32.85	2.77 (2.38–3.23)*	32.89	1.17 (1.01–1.34)*	17.35	1.62 (1.34–1.95)*
Depression symptoms
No	17.24	1.00	30.29	1.00	12.49	1.00
Yes	32.96	2.36 (2.15–2.59)*	28.69	0.98 (0.89–1.08)	15.99	1.32 (1.17–1.50)*

OR: odds ratio; CI: confidence interval.

The categories represent the following: “Low” = rungs 1–4; “Medium” = rungs 5 and 6; “High” = rungs 7 and 8; “Very high” = rungs 9 and 10.

OR is significant at p < 0.05 in Wald test.

Table 3 shows the results of the multiple regression analysis. The associations between SSS and health outcomes were little affected by the adjustments for gender, age, and race/color. After adding per capita household income and education to the analyses, the associations between SSS and SRH remained significant, but their magnitudes decreased. These results did not change when depression was included in the models. With regard to former smoking, the association disappeared in the community ladder after adjusting for objective indicators, but it remained in the “Low” SSS category in the society and work ladders. In the society ladder, participants who declared “Low” SSS presented a 29 percent higher chance of being a former smoker; in the community ladder, the chances were 23 percent higher. These results did not change when depression symptoms were included in the models. In all ladders, the association between SSS and current smoking disappeared after considering the impact of per capita household income and education. None of the explanatory variables present in the final models exhibited multicollinearity. The associations between SSS and SRH were very similar in all ladders (Table 3).

Table 3.

Results from multiple regression analysis—Brazilian Longitudinal Study of Adult Health (ELSA-Brasil), 2008–2010.

Subjective social status	Poor self-rated health			Former smoking			Current smoking
	Model A: Ladders, age, and race/color	Model B: Model A plus education and per capita household income	Model B plus depression symptoms	Model A: Ladders, age, gender, and race/color	Model B: Model A plus education and per capita household income	Model B plus depression symptoms	Model A: Ladders, age, gender, and race/color	Model B: Model A plus education and per capita household income	Model B plus depression symptoms
	OR (95% CI)	OR (95% CI)	OR (95% CI)	OR (95% CI)	OR (95% CI)	OR (95% CI)	OR (95% CI)	OR (95% CI)	OR (95% CI)
Society ladder^a
Very high	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00
High	1.31 (1.10–1.56)*	1.27 (1.06–1.51)*	1.25 (1.05–1.50)*	1.09 (0.96–1.25)	1.08 (0.94–1.24)	1.08 (0.94–1.24)	1.25 (1.02–1.52)*	1.17 (0.95–1.44)	1.17 (0.95–1.43)
Medium	1.87 (1.57–2.23)*	1.41 (1.18–1.70)*	1.38 (1.14–1.65)*	1.20 (1.05–1.38)*	1.10 (0.95–1.37)	1.09 (0.94–1.26)	1.49 (1.21–1.82)*	1.10 (0.89–1.37)	1.09 (0.88–1.35)
Low	3.21 (2.64–3.89)*	1.89 (1.54–2.33)*	1.86 (1.51–2.29)*	1.53 (1.29–1.82)*	1.30 (1.09–1.27)*	1.29 (1.08–1.55)*	1.90 (1.50–2.40)*	1.15 (0.90–1.48)	1.14 (0.89–1.46)
Community ladder^a
Very high	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00
High	1.38 (1.19–1.59)*	1.26 (1.09–1.47)*	1.25 (1.08–1.45)*	1.07 (0.95–1.19)	1.03 (0.92–1.16)	1.03 (0.92–1.16)	1.17 (1.00–1.37)	1.06 (0.90–1.25)	1.05 (0.90–1.24)
Medium	1.96 (1.70–2.27)*	1.45 (1.25–1.69)*	1.44 (1.23–1.68)*	1.14 (1.02–1.29)*	1.03 (0.91–1.16)	1.03 (0.90–1.16)	1.38 (1.17–1.62)*	1.07 (0.85–1.20)	1.01 (0.85–1.20)
Low	3.25 (2.73–3.86)*	1.89 (1.57–2.27)*	1.86 (1.54–2.24)*	1.32 (1.12–1.54)*	1.09 (0.92–1.29)	1.08 (0.92–1.28)	1.62 (1.31–2.00)*	0.99 (0.79–1.24)	0.98 (0.78–1.22)
Work ladder^a
Very high	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00
High	1.20 (1.05–1.37)*	1.16 (1.01–1.33)*	1.15 (1.00–1.33)*	1.03 (0.92–1.15)	1.01 (0.91–1.14)	1.02 (0.91–1.13)	0.98 (0.84–1.14)	0.92 (0.79–1.08)	0.92 (0.79–1.08)
Medium	1.80 (1.56–2.08)*	1.37 (1.18–1.59)*	1.37 (1.18–1.59)*	1.18 (1.04–1.33)*	1.08 (0.95–1.22)	1.07 (0.95–1.22)	1.19 (1.01–1.41)*	0.90 (0.76–1.07)	0.90 (0.75–1.07)
Low	2.72 (2.32–3.19)*	1.68 (1.42–1.98)*	1.62 (1.37–1.92)*	1.44 (1.24–1.67)*	1.24 (1.06–1.45)*	1.23 (1.05–1.44)*	1.64 (1.35–1.99)*	1.06 (0.87–1.30)	1.05 (0.86–1.28)

OR: odds ratio; CI: confidence interval.

The categories represent the following: “Low” = rungs 1–4; “Medium” = rungs 5 and 6; “High” = rungs 7 and 8; “Very high” = rungs 9 and 10.

OR is significant at p < 0.05 in Wald test.

Discussion

This article presented the distribution of SSS, using three reference groups, in a cohort of Brazilian civil servants. Our results showed that the lower the SSS, the higher the chances of having poor SRH and that this association is only partially attributable to the effect of income and education. Moreover, we found that SSS is not independently associated with current smoking, although it is weakly associated with former smoking in the society and work ladders. These results suggest that education and income are more important than SSS to comprehend smoking in this population. Our results do not support the hypothesis that psychological state is a confounder in the associations between SSS and SRH. Finally, we found that the magnitude of these associations did not vary significantly from ladder to ladder, considering the three different reference groups used.

The distribution of SSS was asymmetric, showing a higher concentration above the medium point level, which probably reflects the high education and income levels of the studied population. Similar distribution was found in a study with White American women, who also had a high educational level (Adler et al., 2000). In more socially heterogeneous populations, such as in the English Longitudinal Study of Ageing (Demakakos et al., 2008), the SSS distribution tends to present a normal shape.

The distributions of the scores of the scales, which use proximal groups as reference, were, in our study as well as in others’, more rightly skewed than the society ladder distribution (Ghaed and Gallo, 2007; Subramanyam et al., 2012; Wolff et al., 2010a). This may be explained, in part, by the smaller sociocultural contrasts among individuals who live in the same community or work in the same place vis-a-vis what is found in the society in general. It is also possible that the former scales, which do not evoke the same objective indicators as the society ladder, capture additional aspects not apprehended by the society ladder, such as self-esteem and self-efficacy (Wolff et al., 2010b). In addition, Brazilian higher education and research professionals, as in many countries, are not very well paid. Thus, in the society ladder, they would tend to position themselves in the middle of the ladder. However, in the work ladder, their reference is the distribution of wages and schooling in their occupation, and therefore, the variation is much smaller and many are quite well in comparison with others.

Income and education were more strongly correlated with the society ladder than with the community and work ladders. This is consistent with the scale’s instructions and suggests that the society ladder captures better aspects of objective social status. This find, along with the higher right deviation of SSS scores in the ladders, which use proximal references, reinforces the conclusion that depending on the reference group, individuals have different perceptions of their social status.

The SSS scales are more correlated among themselves other than with education and income. According to Cundiff et al. (2013), this suggests that this indicator measures a construct different from the one captured by objective indicators. This conclusion is corroborated by the absence of multicollinearity between the explanatory variables in our final models. However, the validity of the SSS ladder construct remains uncertain because the real meanings of this rating are not yet completely clear (Adler and Stewart, 2007).

The association between SSS and SRH was, in our study as well as in others (Demakakos et al., 2008; Hu et al., 2005; Miyakawa et al., 2012; Operario et al., 2004; Singh-Manoux et al., 2003, 2005), only partially attributed to income and education. This association was little affected by the adjustments for depression symptoms, which decreases the probability of a psychological bias, corroborating other studies (Adler et al., 2000; Hu et al., 2005; Miyakawa et al., 2012; Operario et al., 2004; Singh-Manoux et al., 2003). On the contrary, in a study carried out by Operario et al. (2004), the correlation between SSS and poor SRH decreased when adjusted for negative affect. This decrease was not higher than the one observed, after adjusting for negative affect, in the model that used objective indicator as main explanatory variables to elucidate the variability of SRH. These results suggest that negative affect would not be a confounder in the association between SSS and SRH instead, a mediating variable between socioeconomic conditions (objective and subjective) and health (Operario et al., 2004). Using a different methodology, Cundiff et al. (2013) achieved similar results. By means of a mediation analysis, Cundiff et al. (2013) demonstrated that psychosocial factors (depression symptoms, optimism, neuroticism, and marital adjustment) would be significant mediators of the association between SSS and SRH. Thus, some authors believe that the adjustment for psychosocial factors in the analysis of the association between SSS and SRH may be an overadjustment (Adler et al., 2000; Miyakawa et al., 2012). Notwithstanding, our findings did not support this mediation relationship, since the magnitude of the association between SSS and SRH did not decrease as expected when we adjusted for mediating variables.

We found that people with “Low” SSS in the society and work ladders had slightly higher chances of classifying themselves as former smokers. This implies that people with lower SSS present greater odds of having been a smoker in the past, despite income and education differences. Once again, adjusting for depression symptoms did not change these results. SSS and the prevalence of current smoking were not independently associated, after considering the effect of income or education exclusively (data not shown), revealing the strong association of these two indicators with tobacco smoking in this population. Our findings are contrary to results among Mexican and American adolescents (Finkelstein et al., 2006; Ritterman et al., 2009). In adults, low SSS seems more related to the likelihood of smoking relapses after abstinence (Reitzel et al., 2007, 2010, 2011) than to the prevalence of smoking (Castro et al., 2010; Ghaed and Gallo, 2007; Manuck et al., 2010).

The above-mentioned results reveal that the association between SSS and health differs depending on the studied event. This indicates that SSS, education, and income present different aspects of social inequities. The impact of these variables on health and on health-related behaviors seems to be mediated by distinct mechanisms. This observation reinforces the idea that although correlated, these indicators are complementary.

Our results did not confirm the hypothesis that the magnitude of the associations would be greater when using more distal reference groups, since the differences between the magnitudes of these associations were not statistically significant. Our findings are consistent with the results from the Jackson Heart Study (Subramanyam et al., 2012). However, some studies have found stronger associations using distal groups (Wolff et al., 2010a) and others using proximal references (Ghaed and Gallo, 2007).

In this study, occupation was not included in the analyses. Occupation was the most strongly social factor associated with SSS among English workers (Singh-Manoux et al., 2003). However, it was not associated with SSS among African Americans (Adler et al., 2008). It is possible that in our analysis, the magnitude of the association between SSS and health would be reduced if occupation had been a confounding variable.

The ELSA-Brasil population is formed by workers who have a steady and formal job or a retirement plan. These characteristics may have led us to underestimate the association between SSS and health and between objective social status and health. Such characteristics of the cohort will possibly have an impact on the external validity of the study. However, our study was carried out in major cities, with large and relatively heterogeneous populations. This suggests that even if the validity of these results cannot be applied to the entire Brazilian population, it can be extended to urban centers with similar characteristics within the country (Aquino et al., 2012).

We believe that lower levels of SSS contribute to deteriorate health status in the long term, by, for instance, increasing the prevalence of health-risk behaviors. This association has been observed in some longitudinal studies (Chen et al., 2012; Singh-Manoux et al., 2005). On the contrary, it is also possible that having poor health lead to worse SSS. Nonetheless, in a dynamic process, as in longitudinal studies, the direct relationship seems to play a role.

This study provides empirical evidence about the importance of SSS as an indicator of socioeconomic position in the Brazilian context. Although, many questions are still to be addressed in future researches. To further understand this indicator of relative social standing, it is important to investigate the mechanisms responsible for the connection of this indicator with health and, with the aid of qualitative studies, seize which factors affect SSS evaluations.

Footnotes

Funding

The ELSA-Brasil baseline study was supported by the Brazilian Ministry of Health (Science and Technology Department) and the Brazilian Ministry of Science and Technology (Financiadora de Estudos e Projetos and CNPq National Research Council), grants 01 06 0010.00 RS, 01 06 0212.00BA, 01 06 0300.00 ES, 01 06 0278.00 MG, 01 06 0115.00SP, 01 06 0071.00 RJ.LVC received a research fellow from Fundação de Amparo à Pesquisa do Estado de Minas Gerais (FAPEMIG) and SMB is research fellow of the National Research Council (CNPq, grant nº 300159/99-4).

Notes

References

Adler

Stewart

(2007) The MacArthur Scale of Subjective Social Status. Available at: www.macses.ucsf.edu/research/psychosocial/subjective.php (accessed 28 August 2012).

Adler

Singh-Manoux

Schwartz

. (2008) Social status and health: A comparison of British civil servants in Whitehall II with European- and African-Americans in CARDIA. Social Science & Medicine 66(5): 1034–1045.

Adler

Epel

Castellazzo

. (2000) Relationship of subjective and objective social status with psychological and physiological functioning: Preliminary data in healthy, White women. Health Psychology 19(6): 586–592.

Aquino

Barreto

Bensenor

. (2012) ELSA-Brasil (Brazilian Longitudinal Study of Adult Health): Objectives and design. American Journal of Epidemiology 175(4): 315–324.

Castro

Gee

Takeuchi

(2010) Examining alternative measures of social disadvantage among Asian Americans: The relevance of economic opportunity, subjective social status, and financial strain. Journal of Immigrant and Minority Health 12(5): 659–671.

Chen

Covinsky

Stijacic Cenzer

. (2012) Subjective social status and functional decline in older adults. Journal of General Internal Medicine 27(6): 693–699.

Cundiff

Smith

Uchino

. (2013) Subjective social status: Construct validity and association with psychosocial vulnerability and self-rated health. International Journal of Behavioral Medicine 20(1): 148–158.

Davis

(1956) Status symbols and the measurement of status perception. Sociometry 19(3): 154–165.

Demakakos

Nazroo

Breeze

. (2008) Socioeconomic status and health: The role of subjective social status. Social Science & Medicine 67(2): 330–340.

10.

Finkelstein

Kubzansky

Goodman

(2006) Social status, stress, and adolescent smoking. Journal of Adolescent Health 39(5): 678–685.

11.

Ghaed

Gallo

(2007) Subjective social status, objective socioeconomic status, and cardiovascular risk in women. Health Psychology 26(6): 668–674.

12.

Giatti

Camelo

Rodrigues

JFC

. (2012) Reliability of the MacArthur Scale of Subjective Social Status. Brazilian Longitudinal Study of Adult Health (ELSA Brasil). BMC Public Health 12: 1096.

13.

Adler

Goldman

. (2005) Relationship between subjective social status and measures of health in older Taiwanese persons. Journal of the American Geriatrics Society 53(3): 483–488.

14.

Instituto Brasileiro de Geografia e Estatística (2011) Pesquisa Nacional por Amostra de Domicílios 2011. Rio de Janeiro, MG: Instituto Brasileiro de Geografia e Estatística.

15.

Jylhä

(2009) What is self-rated health and why does it predict mortality? Towards a unified conceptual model. Social Science & Medicine 69(3): 307–316.

16.

Levy

De Almeida

Szklo

(2012) The Brazil SimSmoke policy simulation model: The effect of strong tobacco control policies on smoking prevalence and smoking-attributable deaths in a middle income nation. PLoS Medicine 9(11): e1001336.

17.

Macinko

Mullachery

Proietti

. (2012) Who experiences discrimination in Brazil? Evidence from a large metropolitan region. International Journal for Equity in Health 11: 80.

18.

Manuck

Phillips

Gianaros

. (2010) Subjective socioeconomic status and presence of the metabolic syndrome in midlife community volunteers. Psychosomatic Medicine 72(1): 35–45.

19.

Marmot

(2004) Status Syndrome. London: Bloomsbury Publishing.

20.

Miyakawa

Hanson

LLM

Theorell

. (2012) Subjective social status: Its determinants and association with health in the Swedish working population (the SLOSH study). European Journal of Public Health 22(4): 593–597.

21.

Neter

Wasserman

Kutner

(1996) Applied Linear Statistical Models. Chicago, IL: Irwin.

22.

Operario

Adler

Williams

(2004) Subjective social status: Reliability and predictive utility for global health. Psychology & Health 19(2): 237–246.

23.

Reitzel

Businelle

Kendzor

. (2011) Subjective social status predicts long-term smoking abstinence. BMC Public Health 11: 135.

24.

Reitzel

Mazas

Cofta-Woerpel

. (2010) Subjective social status affects smoking abstinence during acute withdrawal through affective mediators. Addiction 105(5): 928–936.

25.

Reitzel

Vidrine

. (2007) The influence of subjective social status on vulnerability to postpartum smoking among young pregnant women. American Journal of Public Health 97(8): 1476–1482.

26.

Ritterman

Fernald

Ozer

. (2009) Objective and subjective social class gradients for substance use among Mexican adolescents. Social Science & Medicine 68(10): 1843–1851.

27.

Rosenberg

(1953) Perceptual obstacles to class consciousness. Social Forces 32: 22–27.

28.

Secretaria de Gestão Estratégica e Participativa, Secretaria de Vigilânciaem Saúde, Ministério da Saúde (2012) Vigitel Brasil 2011: Vigilância de Fatores de Risco e Proteção para Doenças Crônicas por Inquérito Telefônico. Brasília, DF: Ministério da Saúde.

29.

Singh-Manoux

Adler

Marmot

(2003) Subjective social status: Its determinants and its association with measures of ill-health in the Whitehall II study. Social Science & Medicine 56(6): 1321–1333.

30.

Singh-Manoux

Marmot

Adler

(2005) Does subjective social status predict health and change in health status better than objective status? Psychosomatic Medicine 67(6): 855–861.

31.

Singh-Manoux

Martikainen

Ferrie

(2006) What does self-rated health measure? Results from the British Whitehall II and French Gazel cohort studies. Journal of Epidemiology and Community Health 60(4): 364–372.

32.

Subramanyam

Diez-Roux

Hickson

. (2012) Subjective social status and psychosocial and metabolic risk factors for cardiovascular disease among African Americans in the Jackson Heart Study. Social Science & Medicine 74(8): 1146–1154.

33.

Wilkinson

(1997) Socioeconomic determinants of health: Health inequalities: Relative or absolute material standards? British Medical Journal 314(7080): 591–595.

34.

Wolff

Acevedo-Garcia

Subramanian

. (2010a) Subjective social status, a new measure in health disparities research: Do race/ethnicity and choice of referent group matter? Journal of Health Psychology 15(4): 560–574.

35.

Wolff

Subramanian

Acevedo-Garcia

. (2010b) Compared to whom? Subjective social status, self-rated health, and referent group sensitivity in a diverse US sample. Social Science & Medicine 70(12): 2019–2028.