A comparison of the health...

Social Science & Medicine 69 (2009) 370–378

Contents lists available at ScienceDirect

Social Science & Medicine journal homepage: www.elsevier.com/locate/socscimed

A comparison of the health status and health care utilization patterns between foreigners and the national population in Spain: New evidence from the Spanish National Health Surveyq Cristina Herna´ndez-Quevedo a, *, Dolores Jime´nez-Rubio b a b

European Observatory of Health Systems and Policies (WHO), LSE Health, Cowdray House, Houghton Street, London WC2A 2AE, United Kingdom Departamento de Economı´a Aplicada, University of Granada, Spain

a r t i c l e i n f o

a b s t r a c t

Article history: Available online 11 June 2009

The increasing proportion of immigrants in Spanish society is placing pressure on the National Health Care System to accommodate the needs of this population group while keeping costs under control. In the year 2000, a law was approved in Spain according to which all people, regardless of their nationality, are entitled to use health care services under the same conditions as Spanish citizens, provided that they are registered in the local population census. However, empirical evidence about differences in health status and health care utilization between the immigrant and the Spanish population is insufficient. This paper uses the 2003 and 2006 Spanish National Health Surveys to explore the existence of inequalities in health and in the access to health services for the immigrant population living in Spain, relative to that of Spaniards. Our results show that there are different patterns in the level of health and the medical care use between the national and the foreign population in Spain: while immigrants’ self-reported health relative to that of the Spanish population depends upon individual nationality, all immigrants, regardless of their nationality, seem to face barriers of entry to specialized care. Further research is needed to understand the nature of these barriers in order to design more effective health policies. Ó 2009 Elsevier Ltd. All rights reserved.

Keywords: Spain Self-assessed health Health care utilization Inequalities Immigrants

Introduction In Spain, immigration is a relatively new phenomenon with growing importance. In the period from 1998 to 2007, the proportion of foreigners registered in the census as a proportion of the total population has increased fivefold, with non-Spanish nationals representing 1.6% of the Spanish population in 1998 and 10% in 2007 (INE, 2008a). This has led Spain to become the main recipient of immigration flows in Europe (Eurostat, 2008).

The importance of the phenomenon of immigration for health services was manifested in the approval of the Law 4/2000 of 11th of January about rights and liberties of foreigners in Spain, according to which all individuals, regardless of their nationality, are entitled to use health care services under the same conditions as Spanish q The data used for this study has been provided by the Spanish National Statistics Institute (INE). Data collectors do not bear any responsibility for the analysis or interpretations presented here. The authors are grateful to David Epstein, Pilar Garcı´a, A´ngel Lo´pez Nicola´s and Alexandrina Stoyanova, participants at the XXVIII Jornadas de Economı´a de la Salud (Salamanca, May 2008) and the 7th European Conference on Health Economics (Rome, July 2008) and three anonymous referees for their comments on an earlier version of this work. * Corresponding author. Tel.: þ44 02079556342.

E-mail address: c.hernandez-quevedo@lse.ac.uk (C. Herna´ndez-Quevedo).

0277-9536/$ – see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.socscimed.2009.05.005

citizens (Dura´n, Lara, & Van Waveren, 2006). The only requisite for immigrants, whether legally accredited or not, to access health care services in the same way as Spaniards is that they should be registered in the local population census. Immigrants who are not registered in the population census are only covered by emergency services. Children and pregnant women have full coverage irrespective of their legal and administrative situation. In addition, the government has recently approved the ‘‘Citizenship and Integration Strategic Plan 2007–2010’’, which targets the whole population and intends to promote social cohesion through policies based on equality of opportunity and equality of rights and duties (Mladovsky, 2007). Also, the Regional Immigration Plans in most of the Autonomous Communities include as a priority the reduction of inequalities in health and in access to health services. However, these policies have been formulated without any sound scientific evidence corroborating the existence of such inequalities.

Antecedents Large-scale immigration began some decades ago in the US, Canada and in many Member States in the European Union. Their experience as immigration recipient countries has allowed them to collect a considerable amount of representative data about

C. Herna´ndez-Quevedo, D. Jime´nez-Rubio / Social Science & Medicine 69 (2009) 370–378

immigrant populations in their national health surveys. As a consequence, there is a great deal of literature on this issue for countries such as Canada and England. In Spain, however, immigration and especially work-related immigration is a relatively new phenomenon, and therefore studies about inequalities in health and equity in the utilization of the Spanish health care system for the immigrant population are rare.

International literature shows in many cases that although immigrants report better health when arriving to their country of destination (the so-called healthy immigrant effect) (McDonald & Kennedy, 2004), their level of health tends to decrease over time.

This is explained as a consequence of the deterioration of the previous healthy lifestyle of immigrants due to acculturation (diet, smoking, alcohol, fertility, social disconnection and isolation), together with the effect of socioeconomic status on morbidity (i.e., level of education achieved, employment status) (Go´mez, Kelsey, Glaser, Lee, & Sidney, 2004; Marmot, Adelstein, & Bulusu, 1984; Marmot & Syme, 1976) and genetics (Marmot et al., 1984).

For the Spanish case, several studies have explored the existence of inequalities in health and inequity in access to health care services for the Spanish population (e.g., Aba´solo, Manning, & Jones, 2001; Clavero & Gonza´lez, 2005; Garcı´a-Go´mez & Lo´pezNicola´s, 2004, 2007). However, so far, only a few studies have analyzed inequalities in health and unequal treatment for equal need for different nationality groups.

Using data from hospital admissions at the Hospital del Mar in Barcelona, Cots et al. (2002) find that immigrants have different needs than the Spanish population given their different age structure and their higher fertility rate. The analysis also shows that lowincome immigrants tend to access health care services primarily through the emergency department. A more recent study based on data from the same hospital suggests that immigrants tend to use hospital emergency services as a substitute for other health care services (Cots et al., 2007).

The study by Garcı´a-Go´mez (2007) based on the 2006 Catalan Health Survey shows that immigrants are less likely to report bad physical health status but are more likely to report bad mental health levels than Spaniards, other factors being equal. With respect to the use of health services, the results of this study suggest that immigrants have a lower probability of visiting a specialist doctor and a higher probability of visiting hospital emergency services than Spaniards. Since the differences in utilization are found to decrease with the immigrants’ number of years of residence in Catalonia, the study concludes that the different utilization patterns between the native and the immigrant population might be due to a limited knowledge of the functioning of the Spanish health care system by immigrants.

A national study based on the 2003 Spanish National Health Survey (Carrasco, Gil de Miguel, Herna´ndez, & Jime´nez, 2007) shows that, as compared to the Spanish population, immigrants present better lifestyle parameters than the national population, such as a lower consumption of alcohol and tobacco. As for the use of health care services, immigrants report high rates of hospitalization. However, the study does not find evidence of an excessive or inappropriate use of health services.

Finally, another national study based on the 2003 Spanish National Health Survey (Jime´nez-Rubio, 2008) corroborates previous results for Catalonia that suggest that, for immigrants, emergency care is an important mechanism of access to health services. The results of this study also show that the immigrant population has a higher probability of being hospitalized and a lower probability of visiting a specialist doctor than the Spanish population with the same health and socioeconomic characteristics.

The empirical literature referring to Spain presented in this section has two important limitations. Firstly, the studies using national data

371

are constrained by the small sample sizes for the immigrant population. Secondly, the studies with sufficiently large sample sizes are based on data for Catalonia. This paper aims to contribute to the research literature about the health status and health care utilization of immigrants in Spain using recently available nation-wide data. The data used is taken from the 2003 and 2006 editions of the Spanish National Health Survey (hereafter SNHS), which in 2003 started categorizing respondents according to their nationality. Pooling the 2003 and 2006 versions of the SNHS allows us to maximize the usable sample size for the immigrant population in Spain.

In particular, the objective of this paper is to use regression estimation techniques to explore how patterns of health and health care utilization compare between Spaniards and non-Spaniards. The next section provides an overview of the methodology that we have followed in the estimations, while Section 3 presents the data used.

Section 4 discusses the results, distinguishing between health and health care use specifications. Finally, Section 5 offers our conclusions.

Methods Empirical specifications For the purpose of our study, we apply different estimation techniques to our data, drawing a distinction between the self-reported health specification and the specification for health care utilization.

Specification for modeling self-assessed health For self-assessed health, we run an ordered logit model to explain our five-category measure of health status. We consider here a pooled specification for the years 2003 and 2006, applying the standard cross-section estimator. The log-likelihood used for the pooled model assumes that the observations are independent across versions of the data (2003 and 2006), and it uses the product of their marginal distributions (Jones, 2000).

Our ordered logit specification presents a measurement model, in which a latent variable (h*i ) is mapped to an observed variable (hi), through the thresholds s’s. The structural model is given by the following expression:

h*i ¼ a þ b ln inci þ

X

gk xk;i þ 3i ; with i ¼ 1; .; N

(1)

k

In (1), h*i represents a latent variable representation of the observed level of health limitations; ln inci is the logarithm of the equivalized net household income, xk,i is a vector of socioeconomic and demographic variables, and 3i reflects the individual error term. In our data, the latent outcome h*i is not observed. Instead, we observe a categorical measure of health (hi). The mechanism of observation (measurement model) is the following:

8 > 1ðvery badÞ; if N h*i < s1 > > > * > > < 2ðbadÞ; if s1 hi < s2 * hi ¼ 3ðfairÞ; if s2 hi < s3 > > * > > > 4ðgoodÞ; if s3 hi < s4 > : 5ðvery goodÞ; if s h* < N 4 i

(2)

Specification for modeling health care use Assuming a linear model, the utilization of health services can be explored by regressing medical care use (yi) on income, a vector of k medical need indicator variables (xk), and a set of p non-need variables (zp) using the following equation:

y*i ¼ a þ b ln ðinci Þþ

X k

gk xk;i þ

X p

dp zp;i þ 3i ; with i ¼ 1;.;N

(3)

C. Herna´ndez-Quevedo, D. Jime´nez-Rubio / Social Science & Medicine 69 (2009) 370–378

372

Need variables are those that ought to affect the use of health care, whereas non-need variables are those that ought not to affect current health care use. In spite of the substantial debate on the meaning of need and the value judgments involved in distinguishing between need and non-need variables (Gravelle, Morris, & Sutton, 2006), we follow the standard approach in the empirical literature and use morbidity variables (proxied by health status and health limitations) as need indicators, and variables, such as income, education, Autonomous Community (AC) of residence (as a proxy for availability of care), tenure of private insurance, and nationality, as non-need indicators.

For modeling a binary measure of health care utilization, we fitted pooled logit regressions collapsing the 2003 and 2006 SNHS data (Jones, 2000). Assuming that yi* in Eq. (3) above is a latent variable, the logit model can be written as

yi ¼

1 if y*i > 0 0; otherwise

(4)

We estimated the pooled logit regression models for our health and health care use variables using STATA 9.0. Individual weights (provided by the SNHS) were used in all computations in order to make the results representative of the Spanish population. Also, a year dummy for 2006 was included in the estimations to take into account the possibility that the 2003 and 2006 samples are independent (Wooldridge, 2006). For ease of interpretation, the estimates of the logit model are presented in odds ratios (see, e.g., Regidor, 2004).

The data Spanish National Health Survey In this study, we use the 2003 and 2006 editions of the SNHS (MSC, 2008). The SNHS is a representative survey of the Spanish population. It is disseminated every two years, coordinated by the Ministry of Health and Consumption and designed by the Spanish National Statistics Institute. The sampling of the data follows a threestage stratified design. The units for the first stage are the census sections. The units for the second stage are the main family households. Within each household, an adult (16 or older) is selected to fill all the questionnaires (INE, 2008b).

The SNHS includes a wide variety of information about health and socioeconomic conditions of Spanish residents, and it contains individualized samples for adults and children. For the purpose of this work, we restrict attention to the adult samples of the 2003 and 2006 versions of the SNHS. Previous editions of this survey do not allow us to identify the nationality or the country of birth of the respondent. A potential limitation of the SNHS for our research purposes is that the immigrant collective might be misrepresented.

One of the reasons for this is that the percentage of individuals unable to answer the questionnaire, although generally low, is higher among non-Spaniards, possibly due to language difficulties (INE, 2008b). In addition, the SNHS does not contain information about years since migration, a variable that would have been particularly useful to test whether any observed differences in health and utilization of health services between immigrants and Spaniards tend to converge or diverge over time.

Table 1 Sample sizes for 2003 and 2006 SNHS by nationality.

Nationality

Latin America European Union Africa Europe Asia North America Oceania Non-Spaniards Spaniards

Total 2003–2006

2003

2006

N

%

N

%

N

%

1250 742 446 150 76 24 8 2705 48,381

46 28 17 6 3 1 0.3 5 95

281 144 127 70 24 3 1 650 21,000

43 22 20 11 4 0.5 0.2 3 97

969 598 319 80 52 21 7 2055 27,381

47 29 16 4 3 1 0.3 7 93

foreigners and non-Spaniards. The variables employed in the estimations are described below.

Health status variables We use self-assessed health (SAH) as our dependent variable in the health status model. SAH is the most extensively used measure of health in the research literature, and it has been shown to be a strong predictor of subsequent use of health care services (Van Doorslaer et al., 2000) and mortality (Idler & Kasl, 1995). Our measure of an individual’s self-perceived health status is based on the question: ‘‘In the last twelve months, would you say that your health state has been: very good, good, fair, bad, or very bad?’’

For explaining health care utilization, we include self-assessed health and three other measures of health as a proxy for the need of health care use. The first is based on the question: ‘‘Did you have to reduce or limit your main activity during the last two weeks?’’ (no, yes). The second health status indicator employed in the study of health care utilization is an indicator of whether the individual suffered an accident of any kind during the twelve months previous to the survey. The last health-related question refers to long-term chronic conditions diagnosed by a health professional. On the basis of the number of chronic conditions reported by the survey respondents, we have created five categories ranging from no chronic conditions to four or more chronic conditions.

Health care utilization variables The use variables that we consider are different indicators of whether the individual has visited a GP, a specialist doctor, hospital services and hospital emergency services.

Measurement of the utilization of the general practitioner (GP) and medical specialist services is based on the question: ‘‘During the last two weeks (four weeks in the 2006 SNHS), about how many times have you visited: (a) a family doctor or general practitioner and (b) a medical specialist?’’ Hospital utilization is measured on the basis of the question: ‘‘How many times in the past 12 months have you (a) been a patient overnight in a hospital and (b) visited hospital emergency services?’’ The different recall periods for utilization of a GP and a specialist doctor in the 2003 and 2006 surveys imply that we will not be able to make predictions of use for each type of service. However, we can provide some insights from the sign of the estimated coefficients.

Sample and variables

Socioeconomic variables

Table 1 shows the sample size for our dataset split by nationality, our proxy for immigrant status. Throughout the text, nonSpanish nationals are referred to interchangeably as immigrants,

Several variables have been included in the econometric estimations to reflect both the demographic and socioeconomic characteristics of the individual.

C. Herna´ndez-Quevedo, D. Jime´nez-Rubio / Social Science & Medicine 69 (2009) 370–378

Age is captured by the following five dummy variables: 16–34, 35–44, 45–64, 65–74 and over 75 years. We allow for the interaction between age and sex variables, and we use 16 to 34 year-old males as the reference category.

In the SNHS, income is measured as a categorical variable with eight possible response categories that provide us with an estimate of the aggregate monthly income, after taxes and deductions, of all household members from all sources. Given the high proportion of missing values for income in the SNHS (25% in 2003 and 11% in 2006), we have imputed household income by regressing the lower and upper bounds of each income interval on a set of variables related to the household, such as region of residence, number of children and number of adults and the mean age of adults, together with several variables related to the main earner of the household, such as education, activity and socioeconomic position (A´lvarez, 2001; Jones, 2000). We have included non-response dummies in the estimations to allow for the possibility that items were not missing at random (Morris, Sutton, & Gravelle, 2005). Equivalent income has been computed by using the modified OECD equivalence scale (giving a weight of 1.0 to the first adult, 0.5 to the second and each subsequent adult, and 0.3 to each child in the household).

Other socioeconomic variables used in this study for both specifications of health status and use of health care are: AC of residence, job status, level of education and nationality of the respondents. We have included a dummy variable for each AC, except for the base category, Comunidad de Madrid, to allow for cultural and geographical differences in the distribution of health and use of health services among Spanish regions. Given their different status, Ceuta and Melilla have been excluded from the analysis, and instead, restricted attention has been devoted to the seventeen Spanish ACs. For education, we use four levels: no education, primary and secondary (first cycle) studies, secondary (second cycle) and post-secondary studies, and university studies (reference category). Job status is measured by six dummy variables that describe the activity status of the respondents: employed (base category), unemployed, retired, student, homemaker and other.

Nationality is captured by the following dummy variables: Spain (reference category), European Union, other European country, Canada or USA, Latin America, Asia, Africa and Oceania. Table 1 shows the number of individuals corresponding to the different categories of nationality included in the 2003 and the 2006 editions of the SNHS. After Spaniards, nationals from Central and South America are the most numerous, followed by European Union citizens, Africans and Europeans (from non European Union countries). Asian, Australasian and North American are the less represented nationalities in the survey. This is in line with data from the Spanish National Statistics Institute (INE, 2008a).

373

regularly (‘‘Have you consumed alcoholic beverages in the last twelve months?’’).

Given that the type of health insurance may have an important effect on the length of time an individual has to wait to receive health care assistance, we have included in the specifications for health care utilization a dummy variable taking the value one if the individual has private coverage for health care services, irrespectively of whether he has purchased the insurance himself, or if the state or a private company has contracted it on his behalf. The fact that the tenure of a private insurance in Spain is not always an individual’s choice but is based on the individual’s occupation (e.g., civil servants have the possibility to opt out between the National Health Service or private insurance companies (Dura´n et al., 2006)) implies that endogeneity in this context is less likely to be an issue.

Results Descriptive statistics Table 2 shows the mean statistics of the key (dependent) health status and health care use variables used in the estimations.

According to Table 2, non-Spaniards report a better level of health than Spaniards, a higher use of hospital emergency services, and a lower use of GP and specialized care as compared to the Spanish population. The next section explores whether these differences persist after controlling for all those factors that are known to affect health and health care use.

Sample means of the independent variables included in the regression models are presented in Table 1A of the Appendix. The sample shows that non-Spaniard individuals report higher levels of education compared to Spanish citizens. Furthermore, there are relatively more non-Spaniards employed, in the working age, and in the middle income categories. The socioeconomic characteristics of immigrants, in particular the similarity of their distribution by income level to the Spanish population and their high proportion in the upper income interval, suggest that the non-Spaniard sample might be misrepresented since the poorest immigrants might be excluded from surveys such as those of the SNHS. Regression results for econometric models of health status and health care use The estimated odds ratios of the logit models for the immigrant categories employed in the estimations are presented in Tables 3 and 4. Regression results for the remaining control variables used in the econometric estimations are presented in Tables 2A and 3A in the Appendix.

Other individual characteristics

Health status

In the self-reported health model, we use three indicators of lifestyle in the estimations: whether the individual smokes (‘‘Do you currently smoke?’’), an indicator of whether the individual practices physical exercise (‘‘Do you practice physical exercise in your spare time?’’) and whether the respondent consumes alcohol

Table 3 shows that, after controlling for a set of socioeconomic and demographic variables, there are no statistically significant differences in the self-perception of health status between the national and the immigrant population. However, the detailed results by nationality in Table 4 indicate that while European Union

Table 2 Sample means of key binary indicators of health status and health care use.

2003

Good or very good health GP visits Specialist visits Hospital visits Hospital emergency visits

2006

All

Spaniard

Non Spaniard

All

Spaniard

Non Spaniard

0.65 0.24 0.07 0.11 0.27

0.64 0.25 0.07 0.10 0.26

0.77 0.15 0.05 0.12 0.29

0.62 0.35 0.18 0.10 0.29

0.61 0.36 0.18 0.10 0.29

0.72 0.24 0.11 0.08 0.31

C. Herna´ndez-Quevedo, D. Jime´nez-Rubio / Social Science & Medicine 69 (2009) 370–378

374

Table 3 Results for the pooled logit analysis: odds ratios for non-Spaniards.

Nationality

Non-Spaniard

Self-assessed health

GP visits

Specialist visits

Hospital visits

Hospital emergency visits

Coef.

z

Coef.

z

Coef.

z

Coef.

z

Coef.

z

0.98

0.35

0.96

0.9

0.69***

5.9

1.23***

3.3

1.16***

3.5

Pseudo R2 0.1

0.2

0.1

0.1

0.2

55,663

24,291

15,860

13,971

24,372

49,802

49,123

49,123

49,123

49,123

Log-L N Note: The asterisks indicate significance at the 1% level (***) 5% level (**) and 10% level (*).

immigrants tend to report higher categories of self-assessed health status than Spanish individuals, those from Latin American countries tend to report worse levels of health than their Spanish counterparts, holding all other factors constant.

Regarding the socioeconomic variables, Table 3A shows that for both specifications of SAH there is a gradient for age, with individuals reporting worse levels of health as they get older. The size of the odds ratios for women is lower than for men, and the gradient is also steeper for women. Hence, female respondents report a worse level of health than male respondents, and the differences get worse with age. The results also show a gradient for the level of education, with more educated individuals reporting better levels of health than those with lower levels of education. These findings are consistent with the results found in the literature on the relationship between education and health (Grossman, 2000; Smith, 2004). With respect to activity status, retired and inactive people are more likely to report the lowest categories of SAH, while students are more likely to report the highest categories of SAH than the employed individuals. High levels of income are associated with better scores in self-assessed health, although the effect of income on perceived health status is lower than the effect of being retired or inactive. This is in line with the results obtained in the literature on the socioeconomic gradient and health (Adams, Hurd, McFadden, Merrill, & Ribeiro, 2003; Benzeval, Taylor, & Judge, 2000; Benzeval & Judge, 2001; Smith, 2004). In terms of the lifestyle variables, the results show that those individuals who smoke tend to report worse levels of health than non-smokers, while those who

Table 4 Results for the pooled logit analysis: odds ratios for non-Spaniard nationalities.

Nationality

Self-assessed health

GP visits

Specialist visits

Hospital visits

Hospital emergency visits

Latin America

0.82** ( 2.1) 1.29** (2.0) 1.06 (0.3) 0.91 ( 0.5) 0.85 ( 0.4) 1.08 (0.1) 0.43 ( 1.3) 0.1 55,645 49,802

0.99 ( 0.2) 0.97 ( 0.4) 1.20* (1.7) 0.42*** ( 3.2) 0.78 ( 1.1) 0.08** ( 2.3) 0.10 ( 1.5) 0.2 24,275 49,123

0.70*** ( 4.1) 0.56*** ( 4.9) 0.93 ( 0.5) 1.12 (0.5) 0.43** ( 2.3) 0.56 ( 0.9) 0.32 ( 0.8) 0.1 15,853 49,123

1.33*** (3.4) 0.89 ( 0.9) 1.75*** (4.3) 0.76 ( 0.8) 1.41 (1.2) 0.16 ( 1.2) 2.38 (0.9) 0.1 13,959 49,123

1.44*** (6.3) 0.85** ( 2.0) 1.26** (2.3) 0.66* ( 1.9) 0.92 ( 0.4) 0.51 ( 1.2) 0.40 ( 0.9) 0.2 24,350 49,123

European Union Africa Europe Asia North America Oceania Pseudo R2 Log-L N

Note: The asterisks indicate significance at the 1% level (***) 5% level (**) and 10% level (*).

consume alcohol tend to report better levels of health than nonconsumers. Practicing physical exercise is not found to be statistically significant in explaining health status in our model.

Health care utilization The regression results in Table 2A in the Appendix show that need is the most important determinant of health care use. Overall, the estimated coefficients on the need variables have the expected sign.

For instance, relative to being in very good health, being in very bad health increases the probability of using every type of health service considered in this study. In general, the coefficients for the variable self-assessed health also show the expected gradient. For hospital emergency services, the results suggest that for both males and females age reduces the probability of visiting emergency care. This can be justified on the grounds that younger people are more likely to be exposed to the practice of risky activities than older individuals (A´lvarez, 2001). Regarding the interactions of age and sex, an interesting result indicates that 16- to 34year-old females have a higher probability of contacting a GP, a specialist doctor, hospital emergency services and being hospitalized than their male counterparts, possibly indicating the use of maternity-related services by healthy women. In addition, another remarkable finding is that as women age, they tend to report a lower probability of visiting a specialist and using inpatient services than 16- to 34year-old males, whereas the opposite is true for men. While further research is necessary to understand this result, we believe that a possible explanation for this might be related to the fact that since we are using cross-sectional data, it is possible that healthrelated problems for older women have been diagnosed earlier than for men with the same health conditions, and their health care treatment may only require contact with a GP. A recent study funded by the European Union Public Health Programme has found that in Spain, women have a longer life expectancy at 50 years than men, but they are expected to live fewer healthy life years (Jagger et al., 2008).

Non-need factors were also found to be important determinants of health care utilization, including individual nationality. As found in previous research using Spanish data (Garcı´a-Go´mez & Lo´pez-Nicola´s, 2007), income is positively associated with the probability of contacting a specialist, while it is negatively associated with the probability of a GP visit. However, interestingly, our results suggest that higher income individuals have a higher probability of seeking emergency medical attention. On the other hand, our findings show that as the level of education increases, the probability of visiting a GP doctor and using hospital emergency services decreases, while the probability of visiting a specialist doctor increases. The tenure of private insurance increases the probability of visiting a specialist doctor and being hospitalized, and it reduces the probability of visiting a GP. These results are as expected given that people with private health insurance face shorter waiting lists and queues and

C. Herna´ndez-Quevedo, D. Jime´nez-Rubio / Social Science & Medicine 69 (2009) 370–378

can access specialist services without being referred by a GP (A´lvarez, 2001). For hospital emergency services, however, tenure of private health insurance does not exert a statistically significant influence in the utilization of these services. Also, while the effect of activity status on utilization varies according to the type of care, for emergency services occupational status does not appear to be an important determinant of use. The limited explanatory power of the non-need variables in the hospital emergency visits model might be partly explained by the unpredictable nature of the demand for this type of service, for which health-related variables are shown to be stronger predictors of use. Finally, the impact of the nationality of an individual on health care use varies across nationalities and types of health care and is described in more detail below.

GP visits For visits to a GP, the results show different patterns of utilization for different nationality categories. According to Table 4, African nationals report a higher probability of a GP visit than Spaniards, while nationals from Europe and North America report a lower probability of visiting a GP. Specialist visits With respect to specialist visits, the findings show robust evidence that immigrants have a lower probability of visiting a specialist physician as compared to Spanish individuals. By nationality, the analysis reveals that being a national from Latin American, Asia, and the European Union is associated with a lower probability of contacting a specialist.

Inpatient stays For non-national individuals, the results show robust evidence of a higher probability of spending a night in a hospital relative to Spanish citizens. Among non-Spaniards, the probability of being hospitalized is larger for Latin Americans and Africans and lower for North American citizens relative to Spaniards.

Hospital emergency services According to the results presented in Table 3, non-Spaniards have a higher probability of using hospital emergency services. In particular, the results show that Latin Americans and Africans have higher probabilities of an emergency visit as compared to Spaniards, while citizens from the European Union and Europe report lower probabilities of visiting hospital emergency services.

In sum, the results for health care use indicate that immigrants report higher probabilities of visiting a hospital and hospital emergency services and a lower probability of visiting a specialist doctor than Spanish nationals, other factors being equal.

Discussion This study aims to provide evidence on the patterns of health status and the access to health services for the immigrant population living in Spain relative to the national population, by using recent nation-wide data from the Spanish National Statistics Institute.

Ensuring that individuals in need of health care, regardless of their country of origin, have equal access to effective treatment that improves their health is a fundamental goal for the health system.

In order to analyze any differences in health outcomes for the national and immigrant populations, we focus on a measure of selfreported health that is widely used in the research literature. The main objective is to find whether, after controlling for a set of

375

socioeconomic and demographic variables, the level of perceived health differs substantially between the immigrant and Spanish individuals. Results show that richer individuals tend to report better self-perceived health scores, while individuals with low levels of education, who are retired or inactive, or who are old and female tend to report worse self-perceived health valuations. For the specific case of immigrants, we find that individuals from the European Union tend to report higher levels of health than the national population, while those from Latin American countries tend to report worse levels of health than the Spaniards, holding other factors constant.

In the analysis of the differences in health care utilization patterns by nationality groups, attention is drawn to whether, after having controlled for need variables (proxied by morbidity variables), utilization of a GP, a specialist doctor, inpatient services and hospital emergency services vary according to the nationality of the respondents. Other non-need variables included in the study are income, education, region of residence, tenure of private health insurance and activity status. The results indicate that need is the most important predictor of the probability of using any of the health care services analyzed in this study. However, other non-need factors were found to be statistically significant in predicting individual utilization of health services, including the nationality of the respondent.

The results for our health care utilization variables reveal that immigrants are more likely to be treated in a hospital than Spaniards are, and they are more likely to contact emergency medical services.

For specialist visits, the findings indicate that foreigners are less likely to contact a specialist doctor than national citizens. Since underutilization of specialist care services does not appear to be caused by reluctance to seek initial contact with a GP, these results may imply the existence of inequity in the access to specialist care with respect to nationality (Gravelle et al., 2006). The findings of this paper corroborate results from previous studies for Spain and suggest that emergency services may be an important mechanism of access to hospital services by immigrants, and that hospital emergency services are a potential substitute of specialized care for this population group (Cots et al., 2007; Garcı´a-Go´mez, 2007, Jime´nez-Rubio, 2008). Recent evidence for the UK also indicates that ethnic minorities have a lower use of secondary care than their non-minority counterparts despite having a higher use of primary care services (Morris et al., 2005; Smaje & Le Grand, 1997). The high reliance on hospital emergency services by non-Spaniards could imply that immigrants know the functioning of the health care system well and that they know that since access to this type of service is much more straightforward than access to ordinary care, they can use hospital emergency services to overcome, to some extent, the barriers that they face in accessing specialist services. Another possible interpretation of our results might be that immigrants are more likely than Spaniards to miss appointments once they are referred to a specialist doctor due to language difficulties, personal circumstances (working timetables, area of residence), or a limited knowledge about the functioning of the health system (Miralles & Dı´ez, 2008).

The results of this study should however be taken with caution for two main reasons. First, it is possible that immigrants might be misrepresented in the SNHS due for instance to language difficulties. In addition, our study does not account for the number of years of residence in Spain for immigrants, a variable that has been shown to be an important explanatory factor of health status and utilization of health services in other studies.

Nevertheless, contrary to previous research using Spanish data, our study uses both a sufficient sample for the immigrant population living in Spain and the most recent nationally representative micro dataset. In addition, the results of this study are generally consistent with earlier studies using Spanish data and the international literature on migration and health in showing that there are substantial variations in health and health care utilization

C. Herna´ndez-Quevedo, D. Jime´nez-Rubio / Social Science & Medicine 69 (2009) 370–378

376

patterns across immigrants relative to the national population of the recipient country. In particular, while the level of self-perceived health compared to Spaniards depends upon the nationality of the immigrants, non-Spaniards overall seem to face substantial barriers of entry to specialized health care. Our results suggest that health policies should focus on improving immigrants’ access to

secondary care by reducing legal, cultural and administrative barriers. Further understanding of the nature of these barriers (demand related: culture, language command, socioeconomic context or legal status; supply related: accessibility, staff attitudes) would help to target resources better to population needs and therefore ensure more effective health policies.

Appendix Table 1A. Sample means of key variables (average 2003–2006).

Income

All

Spaniards

Non-Spaniards

<360 euros 361–600 euros 601–900 euros 901–1200 euros 1201–1800 euros 1801–3600 euros 3601–6000 euros > 6000 euros

0.021 0.143 0.171 0.203 0.236 0.192 0.030 0.004

0.021 0.147 0.171 0.200 0.234 0.193 0.030 0.004

0.017 0.069 0.171 0.255 0.268 0.183 0.030 0.008

Self-reported health Very good Good Fair Bad Very bad

0.121 0.512 0.266 0.078 0.024

0.116 0.511 0.268 0.080 0.025

0.202 0.537 0.217 0.035 0.009

Limitations main activity (previous 2 weeks) Limited Non limited

0.147 0.853

0.148 0.852

0.134 0.866

Number of chronic conditions 0 1 2 3 4 or more Accident

0.421 0.243 0.152 0.088 0.097 0.102

0.410 0.244 0.155 0.090 0.101 0.103

0.625 0.222 0.089 0.038 0.026 0.089

Age and sex 16–34 year-old male 35–44 year-old male 45–64 year-old male 65–74 year-old male >75 year-old male 16–34 year-old female 35–44 year-old female 45–64 year-old female 65–74 year-old female >75 year-old female

0.107 0.092 0.122 0.054 0.046 0.127 0.109 0.170 0.088 0.086

0.101 0.090 0.125 0.057 0.048 0.118 0.108 0.173 0.092 0.090

0.213 0.124 0.077 0.013 0.007 0.286 0.133 0.121 0.014 0.013

Education None Primary and secondary (cycle 1) Secondary (cycle 2) and postsecondary University

0.143 0.483 0.229 0.146

0.147 0.490 0.220 0.143

0.078 0.358 0.380 0.184

Activity status Employed Retired Unemployed Student Homemaker Other

0.439 0.250 0.063 0.050 0.188 0.010

0.427 0.261 0.061 0.050 0.191 0.010

0.655 0.053 0.089 0.041 0.148 0.012

Health insurance Private health insurance Compulsory health insurance

0.145 0.997

0.146 0.998

0.124 0.970

Lifestyle variables Smoke No physical activity Alcohol consumption

0.468 0.410 0.615

0.469 0.409 0. 615

0.453 0.437 0.622

C. Herna´ndez-Quevedo, D. Jime´nez-Rubio / Social Science & Medicine 69 (2009) 370–378

377

Table 2A: Odds ratios for the logit specifications for health care utilization.

GP visits

Specialist visits

Hospital visits

Hospital emergency visits

(1)a

(2)b

(1)

(2)

(1)

(2)

(1)

(2)

Income (ln) Imputed income dummy

0.843*** 0.912***

0.848*** 0.911***

1.311*** 0.972

1.320*** 0.974

1.076 1.090**

1.093 1.092**

1.116** 0.954

1.132*** 0.959

Self-reported health Good Fair Bad Very bad Limitations main activity

1.382*** 2.100*** 2.149*** 1.811*** 2.965***

1.384*** 2.101*** 2.155*** 1.808*** 2.964***

1.487*** 2.590*** 4.191*** 4.701*** 2.204***

1.481*** 2.585*** 4.188*** 4.681*** 2.201***

1.274*** 2.780*** 6.197*** 8.428*** 1.639***

1.269*** 2.768*** 6.180*** 8.405*** 1.633***

1.354*** 2.652*** 5.231*** 5.704*** 1.901***

1.352*** 2.647*** 5.232*** 5.713*** 1.896***

Number of chronic conditions 1 2 3 4 or more Accident

1.654*** 2.062*** 2.321*** 3.048*** 1.222***

1.653*** 2.061*** 2.322*** 3.055*** 1.221***

1.268*** 1.436*** 1.531*** 1.470*** 1.511***

1.269*** 1.435*** 1.530*** 1.469*** 1.510***

0.979 1.039 1.031 1.199*** 1.700***

0.977 1.036 1.033 1.200*** 1.700***

1.234*** 1.455*** 1.589*** 2.024*** 10.739***

1.229*** 1.450*** 1.585*** 2.015*** 10.738***

Age and sex 35–44 year-old male 45–64 year-old male 65–74 year-old male >75 year-old male 16–34 year-old female 35–44 year-old female 45–64 year-old female 65–74 year-old female >75 year-old female

1.027 1.133*** 1.588*** 2.009*** 1.484*** 1.307*** 1.535*** 1.640*** 1.586***

1.027 1.137*** 1.603*** 2.024*** 1.487*** 1.311** 1.544*** 1.653*** 1.600***

1.146* 1.271*** 1.208** 1.203* 1.851*** 1.639*** 1.488*** 1.078 0.791**

1.149* 1.282*** 1.221** 1.219** 1.860*** 1.653*** 1.507*** 1.094 0.803**

1.061 1.319*** 1.503*** 1.648*** 2.106*** 1.203** 0.736*** 0.827* 1.004

1.060 1.334*** 1.532*** 1.679*** 2.116*** 1.212** 0.749*** 0.844* 1.027

0.776*** 0.575*** 0.615*** 0.736** 1.418*** 0.736*** 0.466*** 0.456*** 0.456***

0.777*** 0.578*** 0.618*** 0.739** 1.417*** 0.734*** 0.467*** 0.458*** 0.458***

Education None Primary and secondary (cycle 1) Secondary (cycle 2) and postsecondary

1.141** 1.117*** 1.112***

1.126** 1.109** 1.107**

0.788*** 0.867*** 0.978

0.783*** 0.866*** 0.983

1.009 0.984 0.989

0.988 0.973 0.988

1.065 1.109** 1.149***

1.060 1.103** 1.146***

Activity status Retired Unemployed Student Homemaker Other Private health insurance

1.277*** 1.042 0.820*** 1.135*** 1.189 0.775***

1.280*** 1.044 0.825*** 1.134*** 1.189 0.773***

1.548*** 1.185** 0.815*** 1.163*** 1.559*** 1.536***

1.549*** 1.187*** 0.818*** 1.157*** 1.558*** 1.538***

1.495*** 1.239*** 0.476*** 1.516*** 1.491*** 1.397***

1.503*** 1.247*** 0.477*** 1.510*** 1.494*** 1.400***

0.929* 0.990 0.989 0.972 0.920 0.958

0.938 0.996 0.991 0.978 0.928 0.960

Nationality Non Spaniard Latin America European Union Africa Europe Asia North America Oceania Year 2006 Pseudo R2 Log-L N

0.961

0.694***

1.681*** 0.1493 24,291

0.990 0.968 1.197* 0.416*** 0.778 0.077** 0.104 1.677*** 0.1499 24,275

49,123

49,123

1.233***

2.588*** 0.1210 15,860

0.704*** 0.556*** 0.928 1.122 0.425** 0.556 0.321 2.589*** 0.1214 15,853

49,123

49,123

1.163***

0.928** 0.1090 13,971

1.339*** 0.891 1.746*** 0.764 1.415 0.157*** 2.377 0.926** 0.1097 13,959

1.051** 0.1645 24,372

1.438*** 0.847* 1.260** 0.657* 0.918 0.513 0.397* 1.049** 0.1665 24,350

49,123

49,123

49,123

49,123

Note: The asterisks indicate significance at the 1% level (***) 5% level (**) and 10% level (*).

The reference category is a 16–34 year-old Spanish male with very good health, no chronic conditions and no limitations in main activity, and who has not suffered any type of accident during the twelve months previous to the survey, who has university studies and public health insurance, is employed and lives in Comunidad de Madrid in year 2003.

Estimates for regional dummies are not shown in the above table for brevity. a In specification (1), the proxy for immigrant status is an indicator of whether the individual is not Spanish. b In specification (2), immigrant status is captured by seven dummies indicating the nationality of the individual.

C. Herna´ndez-Quevedo, D. Jime´nez-Rubio / Social Science & Medicine 69 (2009) 370–378

378

Table 3A. Odds ratios for the ordered logit specification for SAH.a Self-assessed health (1)a

(2)b

Income (ln) Imputed income dummy Age and sex 35–44 year-old male 45–64 year-old male 65–74 year-old male >75 year-old male 16–34 year-old female 35–44 year-old female 45–64 year-old female 65–74 year-old female >75 year-old female

1.449*** 1.021

1.438*** 1.017

0.707*** 0.433*** 0.617*** 0.448*** 0.770*** 0.553*** 0.312*** 0.252*** 0.226***

0.707*** 0.432*** 0.619*** 0.449*** 0.772*** 0.556*** 0.312*** 0.252*** 0.227***

Education None Primary and secondary (cycle 1) Secondary (cycle 2) and postsecondary

0.419*** 0.629*** 0.829***

0.417*** 0.630*** 0.829***

Activity status Retired Unemployed Student Homemaker Other

0.415*** 0.768*** 1.480*** 0.754*** 0.363***

0.416*** 0.768*** 1.480*** 0.754*** 0.363***

Lifestyle variables Smoking No Physical Activity Consumes Alcohol

0.827*** 0.983 1.275***

0.822*** 0.982 1.279***

Nationality Non-Spaniard Latin America European Union Africa Europe Asia North America Oceania Year 2006

1.096***

0.827** 1.298** 1.063 0.911 0.859 1.088 0.435 1.094***

Cut-points Cut1 Cut2 Cut3 Cut4 Pseudo R2 Log-L

2.919 1.417 0.480 3.343 0.091 55,663

2.978 1.476 0.421 3.286 0.091 55,646

N

49,802

49,802

0.977

Note: The asterisks indicate significance at the 1% level (***) 5% level (**) and 10% level (*).

The reference category is a 16–34 year-old Spanish male who smokes, consumes alcohol regularly and does not practice physical exercise frequently, has university studies, is employed and lives in Comunidad de Madrid in year 2003.

Estimates for regional dummies are not shown in the above table for brevity. a In specification (1), the proxy for immigrant status is an indicator of whether the individual is not Spanish. b In specification (2), immigrant status is captured by seven dummies indicating the nationality of the individual.

References Aba´solo, I., Manning, R., & Jones, A. (2001). Equity in utilization of and access to public sector GPs in Spain. Applied Economics, 33, 349–364.

Adams, P., Hurd, M. D., McFadden, D., Merrill, A., & Ribeiro, T. (2003). Healthy, wealthy and wise? Tests for direct causal paths between health and socioeconomic status. Journal of Econometrics, 112, 3–56. ´Alvarez, B. (2001). La demanda atendida de consultas me´dicas y servicios urgentes ´micas, XXV(1), 93–138. ˜ a. Investigaciones Econo en Espan

Benzeval, M., Taylor, J., & Judge, K. (2000). Evidence on the relationship between low income and poor health: is the government doing enough? Fiscal Studies, 21(3), 375–399.

Benzeval, M., & Judge, K. (2001). Income and health: the time dimension. Social Science & Medicine, 52, 1371–1390.

Carrasco, G. P., Gil de Miguel, A., Herna´ndez, B. V., & Jime´nez, G. R. (2007). Health profiles, lifestyles and use of health resources by the immigrant population resident in Spain. European Journal of Public Health, 17(5), 503–507.

Clavero, A., & Gonza´lez, M. L. (2005). La demanda de asistencia sanitaria en ˜ a desde la perspectiva de la decisio´n del paciente. Estadı´stica Espan ˜ola, Espan 158, 55–87.

Cots, F., Castells, X., Garcı´a, O., Riu, M., Felipe, A., & Vall, O. (2007). Impact of immigration on the cost of emergency visits in Barcelona (Spain). BMC Health Services Research, 7, 9–17.

Cots, F., Castells, X., Olle´, C., Manzanera, R., Varela, J., & Vall, O. (2002). Perfil de la casuı´stica hospitalaria de la poblacio´n inmigrante de Barcelona. Gaceta Sanitaria, 16, 376–384.

Dura´n, A., Lara, J. L., & Van Waveren, M. (2006). Spain: health system review. Health Systems in Transition, 8(4), 1–208.

Eurostat. (2008). Net migrant flows in Europe. Available from. http://epp.eurostat. ec.europa.eu/tgm/table.do?tab¼table&init¼1&plugin¼0&language¼ en&pcode¼caa14608.

Garcı´a-Go´mez, P. (2007). Salud y utilizacio´n de los recursos sanitarios: Un ana´lisis de las diferencias y similitudes entre poblacio´n inmigrante y auto´ctona. Pre´ blico, 49(4), 67–85. supuesto y Gasto Pu Garcı´a-Go´mez, P., & Lo´pez-Nicola´s, A. (2004). The evolution of socio-economic health inequalities in Spain: 1987–2001. Working Paper No. 756. Universitat Pompeu Fabra.

Garcı´a-Go´mez, P., & Lo´pez-Nicola´s, A. (2007). The evolution of inequity in the access to health care in Spain: 1987–2001. Documento de Trabajo No. 10. Fundacio´n BBVA. Go´mez, S. L., Kelsey, J. L., Glaser, S. L., Lee, M. M., & Sidney, S. (2004). Immigration and acculturation in relation to health and health-related risk factors among specific Asian subgroups in a Health Maintenance Organisation. American Journal of Public Health, 94(11), 1977–1984.

Gravelle, H., Morris, S., & Sutton, M. (2006). Economic studies of equity in the consumption of health care. In A. Jones (Ed.), The Elgar companion to health economics. Cheltenham: Edward Elgar.

Grossman, M. (2000). The human capital model. In A. J. Culyer, & J. P. Newhouse (Eds.), Handbook of health economics (pp. 347–408). Amsterdam: Elsevier.

Idler, E. L., & Kasl, S. V. (1995). Self-ratings of health: do they also predict change in functional ability? Journal of Gerontology, 50B, S344–S353.

INE (National Statistics Institute). (2008a). Foreign population by nationality, autonomous communities, age and sex. Available from. http://www.ine.es/jaxi/ menu.do?type¼pcaxis&path¼/t20/e245/p05//a2007&file¼pcaxis.

INE (National Statistics Institute). (2008b). National Health Survey. Detailed methodology. Available from. http://www.ine.es/jaxi/menu.do?type¼pcaxis&path¼/t15/ p419&file¼inebase&L¼0.

Jagger, C., Gillies, C., Moscone, F., Cambois, Van Oyen, E., Nusselder, H., et al. (2008).

Inequalities in healthy life years in the 25 countries of the European Union in 2005: a cross-national meta-regression analysis. Lancet, 372, 2124–2131.

Jime´nez-Rubio, D. (2008). Equity in the use of health care services by immigrants in the Spanish National Health System. Estudios de Economı´a Aplicada, 26(3), 171–190.

Jones, A. M. (2000). Health econometrics. In A. J. Culyer, & J. P. Newhouse (Eds.), Handbook of health economics. Amsterdam: Elsevier North Holland.

McDonald, J. T., & Kennedy, S. (2004). Insights into the ‘‘healthy immigrant effect’’: health status and health service use of immigrants to Canada. Social Science & Medicine, 59, 1613–1627.

Marmot, M. G., Adelstein, A. M., & Bulusu, L. (1984). Lessons from the study of immigrant mortality. Lancet, 1(8392), 1455–1457.

Marmot, M. G., & Syme, S. L. (1976). Acculturation and coronary heart disease in Japanese-Americans. American Journal of Epidemiology, 104(3), 225–247.

MSC (Ministry of Health and Consumption). (2008). 2003 and 2006 editions of the National Health Survey. Available from. http://www.msc.es/estadEstudios/ estadisticas/solicitud.htm.

Miralles, M., & Dı´ez, E. (2008). Ana´lisis de la demanda asistencial de la poblacio´n inmigrante en la atencio´n neurolo´gica ambulatoria. Neurologı´a, 23(6), 361–366.

Mladovsky, P. (2007). Migration and health in the EU. Research note. Brussels: European Commission.

Morris, S., Sutton, M., & Gravelle, H. (2005). Inequity and inequality in the use of health care in England: an empirical investigation. Social Science & Medicine, 60, 1251–1266.

Regidor, E. (2004). Measures of health inequalities: part 1. Journal of Epidemiology and Community Health, 58, 858–861.

Smaje, C., & Le Grand, J. (1997). Ethnicity, equity and the use of health services in the British NHS. Social Science & Medicine, 45(3), 485–496.

Smith, J. P. (2004). Unravelling the SES health connection. Working Paper 04-02.

London: The Institute for Fiscal Studies.

Van Doorslaer, E., Wagstaff, A., Van der Burg, H., Christiansen, T., De Graeve, D., Duchesne, I., et al. (2000). Equity in the delivery of health care in Europe and the US. Journal of Health Economics, 19(5), 553–583.

Wooldridge, J. M. (2006). Introductory econometrics. Cincinnati, OH: South-Western College.