Methods

The data were taken from the “Neighbourhood characteristics, immigration and mental health” survey carried out between May and December 2006. Sampling was conducted in four regions (Alicante, Almeria, Madrid and Murcia) selected due to the high influx of immigrants to these regions in recent years and the large concentration of immigrants from Ecuador, employed primarily in the service and agriculture sectors. In Madrid, 128 neighbourhoods (barrios) in the city’s 21 districts were ranked in tertiles according to socioeconomic status (measured as the mean price of housing per square metre) and immigrant density. One neighbourhood was selected in each of the nine resulting strata. In both Alicante and Murcia, nine populations (neighbourhoods and cities) with high and low immigrant density were selected, and in Almeria, only six cities were selected. A list of residents was available from census rolls in each city with place of birth, age and sex. A sex-specific stratified random sample of Ecuadorian residents aged 18–55 years was selected to obtain nine men and nine women at each location. The final sample had 570 subjects from 33 areas stratified by mean housing prices and immigrant density (Madrid sample) or only by immigrant density (Almeria, Alicante and Murcia).

Assuming only two-thirds of addresses would actually be valid (estimated by a pilot study), a second sample of five men and five women was selected to replace those whom were not located during field work. A letter of invitation was sent to each potential participant followed by a home visit from trained interviewers. Response rates [completed interviews/(completed +refused interviews)] varied by location: 42.9% in Madrid, 79.5% in Alicante, 96.4% in Murcia and 89.6% in Almeria, giving an overall response rate of 69%. The questionnaire and the measurement scales used were tested in the pilot study.

The dependent variable was psychological distress measured by the General Health Questionnaire GHQ-28,23 a screening instrument consisting of four subscales (somatic symptoms, anxiety and insomnia, social dysfunction and depression). The response categories refer to the person’s experience in the last 4 weeks compared to their “usual state” (better/same/worse/much worse than usual). The coding scheme that assigns values of 0,0,1,1 to these responses was used and the summary score of the 28 items were calculated, the so-called GHQ score.24 A 4/5 cut-off point as a marker of “caseness” was adopted, meaning that those individuals with scores higher than four are considered as having a high probability of being a “possible psychiatric case” (PPC).25 The Spanish translation of the GHQ has been validated.26

To measure discrimination, we constructed a five-item scale based on the works of Finch and Noh.2728 Respondents indicated how often they had felt treated as described in each of five sentences due to their Latin American origin (table 1). Factor analysis using the data from the pilot study revealed a single factor. In the final sample, this factor explained 67% of the variance. Internal consistency measured with Cronbach’s alpha was 0.87. The total possible score on the discrimination scale ranged from 0 (reply of “never” to all five questions) to 20 (reply of “always” to all five questions). As a considerable proportion of respondents (37.7%) said they had never felt treated in any of these ways, a categorical variable was created to identify those who had never felt discriminated against for ethnic reasons, those whose cumulative score was in the upper tertile of the distribution, and those in intermediate situations. The categories were labelled: (0) never, (1) sometimes and (2) always or nearly always.

Financial strain was assessed by the following question: How would you rate your difficulty in making ends meet each month using your net monthly income? Responses were: a lot, quite a lot, neither a lot not a little, little and none.29 As responses in the two extreme categories were infrequent, they were collapsed with the adjacent categories. Categories for analysis were: a lot, some and little difficulty.

Two kinds of social support were assessed. Economic support was assessed by the following question: “In case of need, do you have anybody from whom to borrow 100 euros?” (yes/no). Emotional support was assessed by the following question: “How many friends do you have? I mean people who make you feel good, who you can talk to about personal matters and can ask for help when you need it?” For the analysis the answer was dichotomised to none and one or more.

Social integration was assessed by participation in community associations. A list of different groups was read by the interviewer: sports, neighbourhood, parents of school children, housewives, volunteers, immigrants, trade unions, dance groups, religious associations and others. As most participants did not participate in any activity, and those who did tended to belong to just one group, the variable was dichotomised.

Potential confounders were age, sex, area of residence, education, and self-rated health, measured by the general question: “How would you rate your health?”.

Data analysis

Discrimination was described according to age, sex, area of residence, education and self-rated health. Bivariate associations between caseness and discrimination were estimated using a χ² statistic. A multivariate logistic regression analysis of the association controlling for the possible confounding effects of age, sex and self-rated health was performed (p<0.05). The possible modifying effects of economic and social resources were evaluated on the association between discrimination and caseness by a descriptive stratified analysis and by testing the interaction terms (the product terms of financial strain, economic and emotional support, and social integration with discrimination) in the logistic regression. Lastly, 95% confidence intervals for the odds ratio of caseness for discrimination compared to the absence of discrimination were estimated. The statistical software used was SPSS version 16.0.

Results

Depending on the scale item considered, these Ecuadorian immigrants acknowledge ethnic discrimination as a relatively constant experience in their lives, and a sizeable proportion reported feeling discriminated against as a somewhat frequent experience (sometimes) (table 1).

Table 2 presents the distribution of the study participants by the main study variables. The mean age was 31.6 years. Some additional information may also be useful in characterising the sample. They primarily came from the “Sierra” region of Ecuador, which is home to nearly half the population. Most had migrated to have a better life and to improve their economic conditions and had lived in Spain for at least 5 years.

Almost one-third already had a stable legal situation, and barely 5% were in a legally precarious situation. One-third of them were acquiring their own home. This is a population with a relatively high educational level, most of whom (96%) were employed. The majority worked in manual occupations, with temporary work contracts.

The distribution of perceived discrimination was as follows: 37.7% did not report any discrimination, 41.1% had felt discriminated against on some occasions, and 21% reported feeling discriminated against always or almost always. Discrimination was not significantly associated with sex, age, educational level or self-reported health, but was higher among those residing in Murcia.

Some 28% of the participants met our definition of PPC. About 20% of those who reported no discrimination were PPCs, rising to 30% of those who sometimes felt discriminated against and 41% of those who continually perceived discrimination. Neither age nor education were significantly associated with classification as a PPC. Women and participants with fair or poor self-rated health were more likely to be PPCs (p<0.001). There were also differences by regions, with Almeria having the largest proportion of PPCs (36.8%).

Table 3 presents the stratified analysis by level of discrimination. The presence or absence of resources modified the effect of discrimination. The proportion of PPCs was highest when high financial strain was combined with continuous discrimination, with the result that 75% of the persons in this situation were PPCs. The lack of economic support combined with continuous discrimination resulted in a similar situation, with 62% in the category of PPC. Neither having friends nor belonging to associations modified the association between discrimination and PPC.

In the multivariate analysis, the magnitude of the OR for the association between discrimination and PPC after adjusting for the possible confounding variables (OR 3.0; 95% CI 1.7 to 5.2) varied little with respect to the bivariate analysis (OR 2.8; 95% CI 1.7 to 4.6) (table 4).

Table 5 shows the results of the final multivariate model, which includes the interaction terms for discrimination with available resources: financial strain, social support (emotional and economic) and social integration. The OR for continuous discrimination rose to 12 (95% CI 3.5 to 40.3) among those with high financial strain, and to 10 (2.4 to 41.7) when there was lack of economic support. Emotional support had an independent effect on PPC, with an OR of 1.8 (95% CI 1.0 to 3.6) for those who reported having no friends. Social integration through a community group or association was positively related to the probability of being a PPC, with an OR of 1.7 (95% CI 1.0 to 2.9).

Discussion

In this population of Ecuadorian immigrants in Spain, one-fifth reported being discriminated against continually. This unjust treatment for reasons of ethnic origin seems to have consequences for mental health beyond what could be expected by sociodemographic characteristics or health status. Moreover, economic resources and the ability to obtain economic support in case of need seemed to attenuate the effects of discrimination. Participation in some type of group or association and lack of emotional support from friends were independently associated with poor mental health.

Other studies have shown that most Latin American immigrants are the object of double discrimination in their host countries, both for being poor and due to their ethnic origin,30 although the health consequences of this unfair treatment are modulated by their integration process, which varies by ethnicity.3132

The first survey on discrimination against immigrants in Spain was not made until 2000; it showed that over 60% of the immigrant population had not perceived any discriminatory treatment. Latin American residents were one of the groups that generated the least feelings of rejection on the part of the Spanish population.33 Two years later, a more specific survey showed that Ecuadorians were the least integrated of the three major immigrant groups investigated (Morocco, Colombia, Ecuador), and the ones who most often (24%) reported suffering discrimination.

Perceived discrimination was more frequent in Murcia, the study area with the largest proportion of Ecuadorians in relation to the total population (3.7%) and to the foreign-born population (26%). For instance in Totana, a municipality of nearly 30 000 inhabitants, the proportion of Ecuadorians is 12%. It is possible that the local populations have difficulties in accepting large numbers of foreigners in their communities.

Consistent with our findings, perceived discrimination was not associated with sociodemographic variables, duration of residence in Spain, or administrative situation.10

In another study in an urban setting,34 some 50% of Ecuadorians – surpassed only by the Africans – said they had felt discriminated against at some time due to cultural differences, national origin or physical characteristics. Ecuadorians, primarily of Mestizo origin, have even reported experiencing discrimination among their own people.35

Answering the first research question, a clear and strong association was found between perceived discrimination and mental health, coinciding with Paradies’ recent review,6 which reported that the most consistent association found was between self-perceived racism and poorer mental health. Most of the studies included in the review were conducted in native ethnic minorities (primarily Afro–Americans) in the United States; those carried out in Latino populations referred mainly to first and second generation Mexican immigrants. Only 11% of the 138 studies included addressed specific aspects of the immigrant experience.

The second research question dealt with how the availability of economic resources and emotional support might modify the association between discrimination and mental health. The present results are consistent with Pearlin’s conceptual model. Financial stress has been identified as a source of depression,36 and this has also been seen in Hispanic elders in the USA,3738 but its interaction with discrimination was not examined in those studies. Economic support in times of crisis has been identified as protective for depression in a population of poor Afro–American mothers, but that study did not examine the joint effects of ethnic discrimination and lack of economic support.39 The present findings support the conclusions of a recent review that financial strain is one of the most powerful stressors, together with unemployment, that could explain the mechanisms by which ethnic discrimination influences health.17 However, because only 4% of the immigrants in the present sample reported being unemployed, it was not possible to study the effects of unemployment on mental health.

Having friends was independently associated with good mental health. Some 88% of the Ecuadorians in the present sample said they had friends in Spain, mostly Spaniards, (46%) and only 22% had friends exclusively from other countries. Probably because the present sample was a relatively well-established population of immigrants, it did not corroborate the findings of another study that Ecuadorian immigrants have difficulties making Spanish friends,10 despite their shared language.40

In agreement with the present findings, Finch reported in a study of Latinos of Mexican origin in California, that discrimination was associated with poorer self-rated health only among those who lacked social support.21 Other studies have found directs effects of discrimination on depression, independently of social support and income.41

It is difficult to interpret the negative effect of participation in community groups, such as parent–teacher and neighbourhood associations composed mainly of Spaniards. It is possible that such participation accentuates immigrants’ feelings of “otherness” in relation to natives.42 The present results are consistent with a study showing that participation in community activities increased psychological distress in Pakistani immigrants in Norway.43

The Ecuadorians interviewed in the present study participate very little in immigrant associations. Thus, they may not benefit from the ethnic support that has been reported to moderate depression, for example, in Korean refugees and immigrants in Canada.28

Study limitations

The choice of a cut-off point for the GHQ has been the subject of controversy. Goldberg showed that the cut-off to distinguish those who have a high probability of psychiatric disorder from those who do not varied between 3 and 7 in studies conducted across different countries.2544 The present choice of 4/5 corresponds to a validation study carried out in Greece,45 and has been used in Madrid, Andalucia46 and other studies.4748

The discrimination scale used describes the frequency and nature (insult, aggression, demeaning treatment) of the type of discrimination experienced, but not the place or sphere of activity where it occurred, without establishing a temporal framework. Given the lack of homogeneity in the measure of discrimination used in the studies consulted, a detailed comparison of the results is difficult.6

The method used to select the sample in this study most likely led to the identification of a well established subpopulation of Ecuadorian immigrants in Spain, as suggested by the facts that they had lived in Spain for a median of 5–6 years, that only 5% were in a precarious situation and only 4% were unemployed, that almost one-third were buying an apartment, and fully 47% had Spanish friends. Thus, the present results refer to a subgroup of immigrants who have been relatively fortunate in the integration process, and not to the whole immigrant population.

Concerning measurement bias, the strong associations between the factors associated with mental health, such as sex and poor perceived health, as well as the lack of association between experiences of discrimination and age, sex or education, support the validity of the measures of mental health and discrimination in the present work. The relatively high integration of the present sample may have influenced estimation of the extent of discrimination as perceived discrimination may depend on the length of residence in Spain and on cumulative experiences in different settings.49 Lastly, as this was a cross-sectional study, it cannot be ruled out that poorer mental health status leads to a greater perception of discrimination, although longitudinal studies seem to confirm the reverse.4150 Another possible explanation is that at least part of the association between discrimination and depression is the result of confounding factors not measured in the present study. For instance, personality factors, substance misuse problems and other factors which are known to increase the likelihood of mental disorder may also be associated with the likelihood of subjective experiences of discrimination.

Conclusion

Ethnic discrimination seems to be a relatively frequent and increasingly common phenomenon in Spain and other European countries,51 and the economic situation of immigrants is deteriorating due to rising unemployment.52 Discrimination appears to have mental health consequences among long-established immigrant populations, who are actively involved in the labour force and who often have Spanish friends. Discrimination seems particularly harmful among those with financial strain and lacking needed economic support. As the effects of discrimination decrease when the immigrant has access to help at times of crisis, the social services in Spain should seek to detect immigrants suffering a combination of ethnic discrimination and lack of economic resources as target populations for health and social policy interventions.