Use of different subjective health indicators to assess health inequalities in an urban immigrant population in north-western Italy: a cross-sectional study

Data collection

Enrollment in this cross-sectional study took place between January and August 2012. To be eligible for inclusion, subjects had to be born abroad, not have Italian nationality, have a sufficient knowledge of Italian to fill in the questionnaire, be ≥ 18 years old and live in the city of Genoa.

Owing to the quantitative and qualitative heterogeneity of foreign communities [2] and the absence of a survey frame, snowball sampling was used. For this reason, the results cannot be generalized to the entire immigrant population. Initial seeds were recruited at parish centers, student hostels, cultural and social promotion associations, and during communities’ cultural events.

All participants were contacted personally and were informed as to the objectives of the survey and the voluntary nature of participation. Anonymous questionnaires were distributed only to those who had agreed to participate in the survey. There was no risk of any single participant being identified. The Ethics Committee of S. Martino Hospital (Genoa, Italy) approved the study protocol (N^o 20/2011).

Questionnaire

We used a questionnaire that consisted of the SF-12 Short Form Health Survey (SF-12), version 1 [53–55], integrated with items regarding socio-demographic characteristics and immigration background (gender, age, birth country, marital status, level of education, reason for migration, year of arrival in Italy), some common risk factors (current smoking status, alcohol intake, height, weight) and self-reported medical conditions.

The SF-12, which is the abbreviated version of the SF-36, is particularly suitable and advantageous for use in large-scale (n = 500+) health measurement and in situations in which there is a need to achieve summary information on physical and mental health status. The SF-12 comprises twelve five-point Likert-based questions, which cover eight health dimensions: physical and social functioning, role limitations due to physical health and role limitations due to emotional problems, mental health, vitality, bodily pain and general health. The SF-12 yields two composite scores, the Physical Component Summary (PCS-12) and the Mental Component Summary (MCS-12); a higher score corresponds to a better health status. The SF-12v1 has been validated for use in the Italian population [53–55]. The single Likert-based item on SRH was the question “Would you say your health is excellent, very good, good, fair, or poor?” [36–38].

All study participants were asked about their present smoking status (“Do you currently smoke?”). Questions on alcohol consumption concerned the frequency, quantity and type (beer, wine or spirits); intake was then converted into standard units (12.5 g of ethanol/drink) and dichotomized into non-drinking/light drinking (≤ 1 drink per day) and moderate/heavy drinking (> 1 drink per day) [56]. Participants were also asked to report any health problems and their height and weight, in order to calculate Body Mass Index (BMI).

Questionnaires were self-completed by participants. All questionnaires were checked on the basis of quality control; only fully completed questionnaires were considered, as missing responses to any of the twelve SF-12 items makes it impossible to calculate both summary scores.

Socio-demographic characteristics

Age was categorized according to the following age-classes: 18–24, 25–34, 35–44, 45–54, 55+. Marital status was dichotomized into (1) single never married, divorced/separated, widowed and (2) married/cohabiting. Educational level was defined as “middle school or lower” and “high school or higher”. Reasons for migration were classified as: study, work, family and religious reasons. Immigrants for religious reasons were defined as those whose first residence permit had been issued for religious reasons. Participants were divided into 2 groups on the basis of the time spent in Italy: recent (< 10 years) and long-term (≥ 10 years). This division was based on previous research on social determinants of health in an immigrant population [57].

Country of birth

As a pre-migration factor, the country of origin was classified in accordance with the Human Development Index (HDI) promoted by the United Nations (UN) [58]. The HDI is a summary composite index of human development that measures a country’s average achievements in three basic dimensions of human development: health (measured by life expectancy at birth), education (measured by mean years of schooling and expected years of schooling) and standard of living (measured by per capita Gross National Income [GNI]). The 2011 HDI classification uses a relative approach based on quartiles (from “very high” to “low”) [58].

Statistical analysis

SF-12 scoring was carried out by means of the algorithms described by Apolone et al. [53]; to obtain final PCS-12 and MCS-12 scores, physical and mental weights corresponding to a response choice were summed and then standardized by adding a constant (different for the two summary scores) to the sum of physical or mental weights [53].

Quantitative variables (SF-12 scores) were expressed as means and standard deviations (SD) or medians and interquartile ranges (IQRs), while qualitative variables were expressed as frequencies and percentages.

As the distribution of SF-12 scores was asymmetric, the Mann–Whitney U test was used to evaluate differences in quantitative variables between two groups (sex, marital status, schooling, duration of residence, current smoking, alcohol consumption, presence/absence of overweight/obesity and medical conditions), while the non-parametric Kruskal-Wallis test was used for three or more groups (age-class, reason for migration and HDI quartile of birth country). The chi-square test was used to compare categorical data, i.e. percentages of subjects who reported poor/fair SRH and those who reported good/very good/excellent SRH by sex, age-class, marital status, schooling, HDI of birth country, reason for migration, duration of residence, smoking, alcohol consumption, BMI, and presence/absence of medical conditions. Student’s t test was used to estimate differences in current age and age on arrival in Italy between genders. Spearman’s correlation coefficient was used to measure the correlation between SF-12 subscales, SRH and count of medical conditions.

Multivariate regression models (logistic for SRH and SF-12 scores and Poisson for medical conditions) were utilized to determine whether outcomes were associated with demographic and lifestyle variables. Those characteristics with a p-value lower than 0.15 on univariate analysis were considered in a multivariate model. A stepwise forward approach was used to select the variables for the final multivariate model, with a p-value of exclusion set to 0.10. The independent variables considered in regression models were: gender, age-class, time of residence, HDI of birth country, reason for migration, marital status, level of education, smoking status at the time of the survey, alcohol consumption in standard drinks per day and BMI. Measures of relative effect were expressed as odds ratios (ORs) for logistic regression and prevalence ratios (PRs) for Poisson regression, with 95% confidence intervals (CIs). The level of significance adopted was p < 0.05. All data were analyzed by means of the R stats package, version 2.15.1 [59].

Study population

Initially, 614 subjects were enrolled. Of these, 19 were unable to fill in the questionnaire because of language difficulties, 9 were aged less than 18 years, and 4 had been born in Italy. Another 58 subjects refused to participate because of lack of interest in the topic; the majority of these were females (64%) and subjects from Eastern Asia (42%). We also excluded 22 questionnaires from further analysis on the basis of quality control (not fully completed). Therefore, 502 questionnaires were analyzed.

Descriptive statistics

Multivariate regression analysis

Health disparities by gender

The results of our study highlight the presence of health inequalities within the immigrant population. First of all, we documented the presence of a gender difference, in that the health status of women proved to be poorer and displayed a higher prevalence of medical conditions than that of men, even after adjustment. This phenomenon has been amply demonstrated in both developed [64, 65] and developing countries [66, 67]. Similarly, studies on migrant health have found poorer health status [68] and a higher prevalence of chronic diseases [57] among female immigrants than among males. This seems paradoxical (“gender paradox”) as the life expectancy of men is shorter in almost all countries [69]. This gender divide is probably the result of the interplay between biological and social factors [18]. It has, for example, been suggested that women have a more acute perception of their medical problems, especially minor problems, than men do and, consequently, that they are more likely to report them [70]. However, the gender difference in self-assessed health indexes has been shown to be complex and to vary across age, social context and, obviously, morbidity background [71]. The gender divide may be particularly marked in migrant populations, as it is well known that immigrant women are particularly vulnerable to health problems and often have less access to prevention and healthcare [72].

Impact of reason for migration and duration of residence on subjective health indicators

In our study, subjects who had migrated for work reasons had a higher probability of reporting lower physical HRQoL, which is in contrast with the notion that those who migrate in search of work are generally young and healthy [73]. However, the health of labor migrants is often vulnerable as a result of various risk factors, such as lack of adequate health insurance, poverty and uncertain legal status [74]. Moreover, immigrants are often employed at the lower end of the labor market [75] even if they have a high level of education, a situation that may give rise to education-to-job mismatching. In this regard, it has been shown that overqualified immigrants, especially those of recent arrival, are at high risk of work-related injury [76]. This phenomenon could be particularly pronounced in Southern Europe, where there is a greater demand for unskilled and low-skilled migrants than for highly skilled ones. The most common problems associated with this fact include difficulties in social integration and uncertain long-term job prospects among low-skilled immigrants. In Italy, low- and unskilled jobs are often unregulated, which means that it is difficult for such immigrants to find regular employment [77]. Low-skilled jobs, in turn, often involve working in poor environmental conditions or without adequate safety [78]. Indeed, according to the Italian Workers Compensation Authority, immigrants are often engaged in particularly risky activities, work longer shifts and lack adequate vocational training, which results in a higher incidence of professional accidents [79]. Immigrant workers in Italy also display relatively high rates of work-related musculoskeletal diseases [80], which were the most frequently reported disorders in our study.

We found that migrants for religious reasons had a higher risk of poor physical HRQoL; by contrast, mental HRQoL scores were significantly higher in these subjects than in other groups. Indeed, research has often revealed a salutary effect of religious involvement, especially on mental health [81–83]. On the other hand, some research has underscored the possibility of a spurious association between health and religion [84]. Indeed, Sloan et al. [85] noted a weak and inconsistent association between health outcomes and religious practices. Moreover, Shmueli [86] documented an adverse effect of religiosity and spirituality on health, as measured by the SF-36 and visual analogue rating scale, thus confirming the ambiguity of the effect of religion on health. On the other hand, Franzini et al. [83] found a positive association between religiosity and a 5-point SRH scale, but no association between SF-12 scales and religious involvement among Mexicans in the United States.

Subjects who had migrated for family reasons showed higher MCS-12 scores on multivariate regression. This group comprised migrants for the purpose of family reunification and consisted of spouses, parents and children of immigrants as well as those seeking reunification with an Italian citizen. The better mental health of such subjects could be explained by the absence of exposure to pre-migration health risks, the relative ease of the migration process and social support from relatives who have already settled in the host country [87].

Long-term immigrants have been found to be at higher risk of chronic conditions. Dunn and Dyck [57] found a significant difference in reported health conditions according to the duration of residence, non-recent immigrants (10 or more years) being more likely to report at least one health condition (OR = 1.74, p < 0.001). Another Canadian study found that the time since immigration impacted differently on men’s and women’s health: long-term immigrant women were significantly more likely to report poor health than recent ones, while no significant difference was seen among immigrant men [88]. Similarly, we found a higher prevalence of medical conditions among both long-term immigrants and women.

Health and country of birth

Our study highlights the importance of pre-migration factors, particularly that of the immigrants’ birth countries; indeed, people from less developed societies were more inclined to report poor health status. Similar results also emerged from a Canadian study in which immigrants from countries with lower HDI values reported poor SRH more frequently than those from countries with higher HDI values [37]. Clearly, the HDI variable was used as a proxy measure. However, we believe that this broad indicator might be useful in migrant studies, since it is well documented that the distribution of material, capital, human and other key resources is an important predictor of a population’s average levels of health, income and education (which are the three dimensions of the HDI) [89]. It is therefore plausible that migrants bring their “health luggage” with them to a host country. For example, lower SF-12 scores have been reported in the general Iranian population (HDI = 0.707) [58, 90] than in the general populations of nine developed European countries (HDI ranges from 0.861 to 0.943) [54, 58]. However, this finding should be interpreted with caution for at least three reasons. Firstly, as the most important reason for migration is to improve quality of life [41], the health status of immigrants in a host country may not precisely reflect the baseline pre-migration level. Secondly, observed differences in SRH may be due to cross-cultural differences in health perception [91]. Finally, in accordance with the self-selection hypothesis, immigrants might initially be in better health than their compatriots who do not migrate, but subsequently their health might deteriorate over time, probably as a result of a combination of environmental and behavioral changes [5, 92].

Negative influence of overweight and obesity on SRH and physical HRQoL

Our results showed that being overweight or obese is a risk factor for poor health reporting as well as for lower scores of physical HRQoL; this is in line with previous studies [93–96]. In particular, Huisingh-Scheetz et al. [96] found an inverse independent association between PCS-12 and BMI and no association between MCS-12 and BMI among different racial and ethnic subgroups. Notably, the degree of this association among whites, blacks and Hispanics was quite similar, which indicates that programs to reduce obesity may yield an equal improvement in the physical dimension of HRQoL in all ethnic subgroups.

Different ability of subjective health indicators to predict health disparities

Finally, it is important to note that, while the different subjective health indicators are able to reveal inequalities within immigrant populations, they do so in different ways. Thus, morbidity and BMI were similarly associated with PCS-12 and SRH. Indeed, both PCS-12 and SRH displayed good ability to distinguish between immigrants with a specific health problem and those without that problem, as well as between overweight/obese and normal-weight subjects. Self-reported morbidity, in turn, can be used to predict quality of life and should be regarded as a reliable measure for risk adjustment in immigrant populations. On the other hand, only SRH was able to predict disparities by sex, HDI and BMI. By contrast, both SF-12 scores were associated with the reasons for immigration. Kaplan et al. [97] also found a moderate level of correlation between HRQoL and SRH; in their study, SRH predicted mortality better than HRQoL. These authors also found that, when there was overlapping between the two indicators, each of them provided new information. The results of recently published research conducted by Delpierre et al. [35] indicated that, in measuring inequalities, SRH may downplay socioeconomic differences, while HRQoL highlights them.

Study limitations

Our study is not without limitations. First of all, we used non-probabilistic sampling techniques, which are subject to some biases, selection bias in primis; the results cannot therefore be generalized [98]. However, even if the use of random techniques had been possible, the sampling frame would have been incomplete, as about 40,000 foreigners are granted Italian citizenship each year and disappear from official statistics. Specifically, citizenship can be applied for by long-term immigrants after 10 years of residence or after 2 years of marriage to an Italian citizen [99]. However, chain-referral methods can have some advantages; in particular, they can enable hard-to-involve populations to be reached through their social networks [100]. To mitigate the weaknesses of this sampling method, we started the snowball chain from people belonging to different social groups.

Second, the subjects enrolled in our study were sometimes inhomogeneous in terms of age and sex, since individual communities of immigrants display different patterns. For example, 85% of Ukrainian immigrants are women, while 82% of migrants from Senegal are men [2, 60]. Such inhomogeneity also applies to the numbers of participants born in each country; indeed, some foreign communities, such as Chinese, were underrepresented. On the other hand, the majority of immigrants to Genoa are from countries with high HDI, while those from very high HDI societies account for less than 10% [2, 58]. The use of HDI groupings can mitigate this limitation. However, our sample was age-representative, as the mean age of the study population reflected that of the whole foreign adult population of Genoa (37.6 years for males and 39.4 years for females) [60]. Similarly, the sample was representative of civil status [101].

Thirdly, we only enrolled subjects who had a sufficient knowledge of the Italian language, which could constitute a barrier for the most recent immigrants. Indeed, 3% of subjects were unable to fill in the questionnaire and were excluded. However, the German version of SF-12 has been successfully used among different ethnic groups in Germany [41].

Fourthly, our questionnaire did not contain some important items – such as housing, employment and working conditions, utilization of health services, etc. – that could affect health outcomes. The same applies to the issue of legal presence in Italy. A previous version of the questionnaire contained an item regarding the legality of residence in Italy. However, in our pilot study, several subjects (7 out of 30) omitted this question, even though they were assured that the questionnaire was anonymous. Therefore, in order to encourage openness and honesty in the answers, this question was deleted. Moreover, Pikhart et al. [36] did not find a statistically significant difference in terms of poor SRH reporting between legal and illegal immigrants in the Czech Republic.