There has been considerable interest on the part of researchers and policy makers in understanding the social, economic, psychological, and behavioral factors that account for the many health disparities that are evident in the United States and the world [1-5]. Although socioeconomic factors are the most commonly examined drivers of disparities, social and interpersonal factors have also received considerable attention [3,6-9]. Two psychosocial constructs that have shown consistent associations with health outcomes are discrimination and perceived unfairness. The former has been extensively examined in both the United States [3,9-13] and globally [8,9,14], while studies of unfairness and health have been limited to a handful of studies in the United Kingdom and the Netherlands [15-17]. No studies have reported the joint effects of these 2 constructs.

In perhaps the first published study on unfairness, DeVogli et al [16] used the prospective data of 8298 individuals from The Whitehall II study, who were civil servants originally ages 35-55 at the study onset in 1985 through 1998. Baseline unfairness data were collected between 1991 and 1993 (during phase 3 of the study), and respondents were tracked for health outcomes on average 11 years through 2003 to 2004 (phase 7 of the study). The authors found that their single-item unfairness measure was associated with higher odds of clinically verified heart disease as well as worse physical and mental functioning, with the latter being assessed with the 36-Item Short Form Survey (SF-36) questionnaire [18,19] adjusted for sociodemographic and other risk factors. Unfairness was also positively related with being female, obese, and less physically active. Higher unfairness was associated with higher cigarette use but also higher rates of alcohol abstinence.

In a second study [15], again based on the Whitehall study, phase 3 data on unfairness were used to predict cardiovascular risk factors, over the subsequent 6 years through phase 5. The authors found that baseline unfairness was significantly associated with higher rates of low serum high-density lipoprotein, high serum triglycerides, hypertension, high fasting blood glucose, and elevated waist circumference.

A third study, conducted in the Netherlands, used a 9-item perception of unfairness scale, administered in 2008 [17]. Of those completing this unfairness questionnaire, 1282 adults also completed the SF-36 [18,19] in 2003, 2008, and 2010, which was used to classify respondents as experiencing either physical or mental health decline between 2003 and 2010 and between 2008 and 2010, with unfairness scores from 2008 being used as the predictor. In general, higher scores on unfairness were associated with significantly higher odds of both physical and mental health decline. Higher scores on the unfairness measure were also associated with lower socioeconomic position (eg, a composite of income and education).

The health effects of discrimination have been reported in hundreds of studies. One measure of discrimination that has been particularly popular is the Everyday Discrimination Scale developed by Williams and others [3,10,11]. The Everyday Discrimination Scale focuses on what are sometimes referred to as microaggressions, smaller acts of discrimination, racism, or prejudice; for example, the scale includes items such as “people act as if you are inferior,” “you are treated with less respect than others,” and “people act as if you are dishonest.” Everyday discrimination is distinguished from more major experiences of discrimination, such as being unfairly fired from a job, maltreated by the police, or denied a bank loan.

The everyday discrimination scale has been associated with a wide range of adverse mental and physical health outcomes including depression, anxiety, distress [8,9,12-14], and overall well-being [9,13], breast cancer among women under the age of 50 years [20], and high blood pressure, although the relationship with the latter is often conditional on other variables [9]. Some studies have used unfair treatment as the measure of discrimination [21], although the 2 constructs likely represent different psychological and social phenomena [22]. One distinction between the 2 constructs is that everyday discrimination measures the occurrence of events, whereas unfairness can be thought of as more related to the perception of such events. Additionally, while discrimination measures multiple potential events, unfairness in the Whitehall studies and in the current investigation was measured with a single global item.

No study has reported the relationship of both discrimination and unfairness with either mental or physical health outcomes. The current analyses from a recent cross-sectional survey conducted in a multiethnic sample of adults in the state of Michigan provide insight into the effects of discrimination and unfairness on medical, psychological, and behavioral outcomes.

A total of 2238 usable surveys were collected, of which two-thirds (n=1508) were completed by females. A little less than half (980/2228, 43.79%) had college or higher educational attainment, and about half were aged 45 or higher, married, and reported a household income above US $50,000 per year. Most, (1902/2230, 85.29%), were born in the United States. About half (1183/2232, 53.00%) were White, 23.52% (2232/2238) were African American, 18.06% (403/2232) were MENA, and 4.70% (105/2232) were multiracial.

As shown in Table 2, the mean value for the unfairness item was 2.62 (SD 1.61), with a range of 1-6; meanwhile, the mean for everyday discrimination was 1.87 (SD 1.06) with a range of 1-6. The 2 variables were moderately correlated (r=0.54; P<.001).

In bivariate analyses, higher unfairness values were significantly associated with lower educational attainment, lower age, lower household income, and being unmarried. Birthplace was unrelated to unfairness. For race, the highest values were observed for African American and multiracial respondents followed by MENA respondents. White respondents reported the lowest values, and they were significantly lower than African American, multiracial, and MENA respondents in pairwise comparisons.

As shown in Table 3, unfairness was significantly related to mental health symptoms, after adjustment for sociodemographic variables (Model 1), with higher unfairness values associated with worse (lower scores) mental health status. In addition to an overall significant effect, all pairwise contrasts were significant. The association, both overall and pairwise, remained significant after inclusion of everyday discrimination scores in the regression model (Model 2). The R² increased from 0.17 to 0.22 when everyday discrimination was added to the model.

As shown in Table 4, unfairness was significantly related to self-reported history of depression and high blood pressure, after adjustment for sociodemographic variables (Model 1). Higher unfairness values were significantly associated with higher odds of depression, when both the middle and high groups were compared to the lowest group, and for high blood pressure, when the highest group was compared to the lowest group. After inclusion of everyday discrimination scores in the model (Model 2), there was still an overall association with depression, and the pairwise contrast of highest to lowest remained significant. In Model 2, the effect of unfairness on high blood pressure disappeared. The addition of everyday discrimination to the model increased the R² value by 0.01% to 0.02%.

As shown in Table 5, unfairness was significantly related to 30-day marijuana use, 30-day cigarette use, and lifetime opiate use, after adjustment for sociodemographic variables (Model 1). There was no association with 30-day alcohol use. Specifically, higher unfairness values, when both the middle and highest groups were compared to the lowest group, were associated with significantly higher odds of cigarette and opiate use, whereas for marijuana, the contrast was significant only for the highest compared to the lowest group. When everyday discrimination scores were added to the model (model 2), the effect on marijuana was no longer significant and the effect on cigarettes was only significant for the comparison of the middle to the lowest group. In addition, the effect on alcohol use became significant. However, the effect was in the opposite direction than with the other substances. That is, higher unfairness was significantly associated with lower odds of 30-day alcohol use. The addition of everyday discrimination to the model increased the R² value by 0.01% to 0.02%.

Unfairness was significantly related to worse mental health functioning, net adjustment for sociodemographic variables, and everyday discrimination. Unfairness was related to self-reported history of depression and high blood pressure after an adjustment for sociodemographic variables was made although after including everyday discrimination in the model, only the association with depression remained significant.

Unfairness was also significantly related to 30-day marijuana use, 30-day cigarette use, and lifetime opiate use. Adding everyday discrimination scores to the model attenuated these effects, with the association with marijuana no longer being significant and the effect on cigarettes only being significant for the comparison of the middle group to the lowest group.

Our findings of a generally harmful effect of perceived unfairness on health are consistent with prior studies [15-17]. Perceived unfairness, although moderately correlated with everyday discrimination, appears to serve as a unique predictor of health status and health behavior. Alternatively, given that, in some instances, adding everyday discrimination to the model attenuated the impact of unfairness on outcomes, unfairness may also serve as a partial mediator of discrimination; that is, it may be one of the psychological pathways through which discrimination negatively impacts health: it may be a consequence of discrimination. On the other hand, adding discrimination to the model (except for mental health) generally only increased the R² value by 1% or 2%, suggesting that unfairness captures most of the same variance in health outcomes as does discrimination.

In our study, we found that higher unfairness was associated with lower educational attainment, lower age, lower household income, and being unmarried. The higher rates of unfairness observed for individuals with lower education and income is consistent with both UK [15,16] and Dutch studies [17]. Age was unrelated to unfairness in the Whitehall study [16], which differs from our findings. We found no differences by gender, which is inconsistent with the Whitehall study, where women had higher levels of reported unfairness [16].

With regard to race, we found the highest unfairness values were observed among African American and multiracial respondents, followed by MENA respondents. White respondents reported the lowest values, and they were significantly lower than African American, multiracial, and MENA respondents in pairwise comparisons. Prior studies did not report race effects, as their samples were more homogeneous than that of this study. The Whitehall study sample, for example, consisted of 90% White and 5% South Asian participants [27].

We found that the effect of unfairness on 30-day alcohol was in the opposite direction to the other substances we examined. That is, higher unfairness was associated with lower odds of 30-day alcohol use. The protective effect of unfairness on alcohol intake was not entirely unexpected. A few previous studies have found that higher levels of unfairness and discrimination were not associated with increased alcohol use [9,16,28]. In our study, this relationship was in part confounded by religiosity, as more religious respondents were both more likely to abstain from alcohol and to report higher levels of unfairness (data not shown). Controlling for religion reduced the protective effect of unfairness on alcohol use.

The study has several limitations. First, the data were all cross sectional, which limits our ability to determine causation; reverse causality cannot be excluded. For example, higher rates of illness or substance use could drive higher perceptions of unfairness or discrimination, rather than the inverse.

Secondly, unfairness may be caused by everyday discrimination. Experiencing discrimination could lead one to perceive life as unfair. Given the 2 variables were moderately correlated (r=0.54) and that the effects of unfairness on health outcomes generally remained after adjustment for discrimination, it appears that unfairness may function as a unique contributor to health outcomes, above that of discrimination. It should also be noted that while the 9-item everyday discrimination scale assesses various types of discriminatory behaviors, the single-item unfairness measure we used only assesses global perceptions. Perhaps more granular assessment of unfairness, more akin to how discrimination is typically assessed, would yield more robust findings.

Furthermore, illness outcomes were all self-reported, which likely increased the error of these measures. Given this, the analyses might have underestimated the association of unfairness on health, unless they share a common reporting bias, which then would have spuriously inflated their association. Finally, our sample was recruited in Michigan, so the generalizability of our findings to other populations is unclear.

Our findings have implications for both research and practice. With regard to research, our results indicate that unfairness should be included as a potential variable in studies of health and health disparities in addition to the standard set of socioeconomic and social indicators. Unfairness may also be more amenable to intervention than discrimination and other disparity drivers. Because unfairness relates more to the perception or perseveration of events rather than simply their occurrence, cognitive interventions that help individuals cope with these perceptions may merit investigation.