Content uploaded by Rocío García-Pina
Author content
All content in this area was uploaded by Rocío García-Pina on May 30, 2016
Content may be subject to copyright.
ORIGINAL PAPER
Absence of protective ethnic density effect on Ecuadorian
migrants’ mental health in a recent migration setting: a multilevel
analysis
Inma Jarrı
´n•Ana Garcı
´a-Fulgueiras •Vicente Iba
´n
˜ez-Rojo •De
´bora Alvarez •Rocı
´o Garcı
´a-Pina •
Alberto Ferna
´ndez-Liria •Visitacio
´n Garcı
´a-Ortu
´zar •Domingo Dı
´az •Maria A
´ngeles Rodrı
´guez-Arenas •
Lucı
´a Mazarrasa •Maria Victoria Zunzunegui •Alicia Lla
´cer •Julia Del Amo
Received: 2 February 2011 / Accepted: 17 May 2012
ÓSpringer-Verlag 2012
Abstract
Purpose We aimed to study the association between the
Ecuadorians’ ethnic density (EED) of the areas of resi-
dence (AR) with the mental health of Ecuadorians in Spain.
Methods Multilevel study of 568 Ecuadorian adults in 33
AR randomly selected from civil registries and interviewed
at home. Possible psychiatric case (PPC) was measured by
scoring C5 in General Health Questionnaire-28. Ecuadorians’
ethnic density was dichotomized in high and low EED
(\6 %). Multilevel logistic regression was used to estimate
odds ratios (OR) and 95 % confidence intervals (CI).
Results Prevalence of PPC, 24 % (95 %CI 20–28 %),
varied by area of residence. Ecuadorians’ ethnic density
varied by area of residence ranging from 0.9 to 19.5 %.
PPC prevalence in High Ecuadorians’ ethnic density AR was
29.5 and 20.4 % in low EED AR (p0.013). Ecuadorians from
High EED AR had higher odds of PPC than those from Low
EED AR (OR 1.65 95 %CI 1.01–2.72). Adjusting for indi-
vidual confounders (largely self-perceived discrimination),
OR decreased to 1.48 (95 %CI 0.87–2.55). The final model,
adjusted by area of residence and educational level, yielded an
OR 1.37 (95 %CI 0.78–2.40).
Conclusions No protective association between the Ecu-
adorians’ ethnic density of the Area of residence and
Ecuadorian migrants’ mental health was found. Mecha-
nisms underlying beneficial ethnic density effects may be
absent in recent migration settings.
Keywords Mental health Multilevel studies
Ethnic density Ecuadorians
I. Jarrı
´n(&)D. Alvarez A. Lla
´cer J. Del Amo
National Center of Epidemiology, Instituto de Salud Carlos III,
Monforte de Lemos 5, 28029 Madrid, Spain
e-mail: ijarrin@isciii.es
J. Del Amo
e-mail: jdamo@isciii.es
I. Jarrı
´nA. Garcı
´a-Fulgueiras D. Alvarez R. Garcı
´a-Pina
V. Garcı
´a-Ortu
´zar M. A
´. Rodrı
´guez-Arenas A. Lla
´cer
J. Del Amo
Ciber de Epidemiologia y Salud Pu
´blica (CIBERSP),
Parc de Recerca Biome
`dica de Barcelona, Doctor Aiguader 88,
8003 Barcelona, Spain
A. Garcı
´a-Fulgueiras R. Garcı
´a-Pina V. Garcı
´a-Ortu
´zar
Department of Epidemiology, Regional Health Council,
Ronda de Levante 11, 30008 Murcia, Spain
V. Iba
´n
˜ez-Rojo D. Dı
´az
Mental Health Department, CPE ‘‘Bola Azul’’,
Carretera de Ronda 226, 04009 Almeria, Spain
A. Ferna
´ndez-Liria
Mental Health Department, Hospital Universitario Prı
´ncipe de
Asturias, Carretera Alcala
´-Meco s/n. Alcala
´de Henares,
28805 Madrid, Spain
M. A
´. Rodrı
´guez-Arenas L. Mazarrasa
National School of Public Health, Instituto de Salud Carlos III,
Monforte de Lemos 5, 28029 Madrid, Spain
M. V. Zunzunegui
De
´partement de me
´decine sociale et pre
´ventive, Faculte
´de
me
´decine, Universite
´de Montre
´al, Montreal, QC H3C 3J7,
Canada
M. V. Zunzunegui
Axe Sante
´mondiale, CRCHUM, Montreal, QC, Canada
123
Soc Psychiatry Psychiatr Epidemiol
DOI 10.1007/s00127-012-0523-8
Introduction
The effect of contexts and places on individuals’ health is an
important research area in social epidemiology. In the last
decades, more attention has been placed on a particular type
of contextual effect known as ‘‘ethnic density effect’’ [1].
Multiple studies have reported beneficial ethnic density
effects on mental [2–5] and physical health [6,7] and health
behaviours [8]. A lower number of studies have reported no
associations [9,10], and fewer describe deleterious ones [3].
Economic deprivation of individuals and their neighbour-
hoods is associated with poorer health status too [11–15]and
the poverty of a neighbourhood may be associated with its
ethnic composition since migrants and ethnic minorities tend
to concentrate in areas with higher levels of poverty and
underinvestment in services [16]. The exact mechanisms
through which ethnic density may influence mental health in
ethnic minorities are not well established [17–21] but it has
been proposed that it may buffer the effects of everyday
racism and perceived risk of physical and psychological
intimidation by providing social support from local networks
and culturally specific services [17–21].
The differences in the direction of the association between
ethnic density and health outcomes could be partially
explained by the relative contribution of material and psy-
chosocial determinants of health, although important meth-
odological aspects have to be taken into account [17–20].
There is no consensus on how to model and categorise and
measure ethnic density nor onthe size of the area to be studied
[17]. Furthermore, some studies lump all ethnic minority
groups together and make no distinction as to whether
migrants areincluded nor to the relative mixof different ethnic
groups. Ethnic minorities include both established and
recently arrived members of the community but health status
and its determinants vary enormously between them [19,22].
Whereas first generation migrants tend to be a selected group
of people with good health, ‘‘healthy migrant effect’’, ethnic
minoritieshave poorer health than the rest of the population. In
spite of this good health on arrival, most migrant groups are
reported to have worse mental health indicators [19,22],
although it has been different in Canada for children [23].
Spain has experienced a recent and rapid increase in eco-
nomic migrants from the mid-90s onwards; Ecuadorians
account for one of the largest groups [24]. Ecuadorians have
settled in rural areas, which demanded agricultural work, and
in large cities which offered jobs in the building industry and
in the care of children and the elderly. Ecuadorians speak
Spanish; the majority are Catholics and most have ethnic
features which single them out [25]. Ethnic density effects on
health are very context specific and depend on the historical
and political processes which have determined the contact of
the different groups; such as cultural distance between com-
munities, velocity and duration of the settlements [17,20]. In
this paper, we expand on the literature of the positive and
negative associations of ethnic density with mental health in
the context of a South European country with a recent
migration process. Our hypothesis is that Ecuadorians living
in areas with a higher density of Ecuadorians will have better
mental health than those who do not. We aim to estimate the
association between the Ecuadorians’ ethnic density of the
Area of residence and the mental health of the Ecuadorians
living in 33 areas in Spain, accounting for material and psy-
chosocial individual and contextual confounders.
Methods
We designed a multilevel study which included 1,186
adults aged 18–55 clustered in 33 areas of residence (AR).
The 33 AR, 17 city neighbourhoods and 16 municipalities
(largely rural) within 4 regions in Spain (Alicante, Almerı
´a,
Madrid, Murcia) were chosen because the high influx of
migrants experienced over the last decade. We chose these
33 AR to reflect variability in immigration density allowing
for a minimum number of 200 Ecuadorians. A home survey
was conducted in a probabilistic sample obtained from the
civil registries allowing for an equal number of men and
women, Spaniards and Ecuadorians. A second sample was
drawn to account for invalid addresses, unavailable con-
tacts and refusals. Definition of Spaniards and Ecuadorians
was based on nationality. A ten-Euros token (phone card
for Ecuadorians and petrol voucher for Spaniards) was
given to participants. Ecuadorians were visited by trained
Latin-Americans interviewers, mostly women. A minimum
of two documented visits at different times were performed
before moving to the next candidate. The home survey was
conducted from September 2006 to January 2007, after a
piloting survey in January–February 2006. The overall
response rate (completed interviews/completed ?refusals)
was 61 %; 53 % for Spanish men and 57 % for Spanish
women, and 69 % for Ecuadorians. Median duration of the
interview was 20 min for Spaniards and 35 for Ecuadorians.
In this work, we will analyze data based on interviews of 568
Ecuadorians from the 33 AR. The collection of the contextual
level data was obtained during 2008. Detailed methodology of
this study has been previously published [26].
Individual-level variables
The outcome variable, possible psychiatric case (PPC), was
measured by scoring 5 or more in the Spanish version of
the General Health Questionnaire of 28 items (GHQ-28), a
mental health screening tool made up by four sub-scales
which capture recent changes in somatic symptoms, anxi-
ety, depression and social functioning [27,28]. The
response categories refer to the person’s experience in the
Soc Psychiatry Psychiatr Epidemiol
123
last 4 weeks compared to their ‘‘usual state’’ (better/same/
worse/much worse than usual). We used the coding scheme
that assigns values of 0,0,1,1 to these responses. We col-
lected information on socio-demographic characteristics
such as civil status, number of children, maximum educa-
tion attained. Social support was measured by the Duke
scale [29], social network diversity by asking about number
of friends, contact with neighbours and work colleagues
and participation in associations. Emotional support from
partner was explored through a five questions likert scale.
We also inquired for the presence of a confident (existence
of a person to talk about personal matters) and economic
confident (‘‘In case of need, do you have anybody from
whom to borrow 100 Euros?’’. Financial strain was
assessed by the question: how would you rate your diffi-
culty in making ends meet each month using your net
monthly income? Individual and family unit monthly
incomes were inquired as compared with the concurrent
national minimal wage (NMW). Subjects were asked about
their employment and type of contract, and about work
atmosphere through a 5-item likert scale. Time of arrival to
Spain, and whether subjects were still paying their migra-
tion debt were asked for. Perceived discrimination was
recorded through a 5 items scale inspired in the works of
Finch and Noh [30–32]. Detailed description of these
variables has been previously published [26].
Contextual level variables
Second-level data were obtained from all the secondary
sources available provided they were common to the 33 AR. A
detailed description of the sources and institutions providing
the data, as well as the size of the AR, has been previously
described [26]. The following second-level data were col-
lected for each of the 33 AR. The main exposure variable,
Ecuadorian’s ethnic density, EED (proportion of people with
Ecuadorian nationality among all subjects recorded in the
municipal council registry) was obtained from the Municipal
Registry 2006. We collected total ethnic density (proportion
of people lacking Spanish nationality among all subjects
recorded in the municipal council registry) and among the
various indicators of socio-economic level of the AR, we
chose to use the proportion of people with less than primary
education from the National Census 2001 as recommended by
Regidor et al. [33] and if the area was a neighbourhood (lar-
gely cities) or a municipality (largely villages).
As associations between ethnic density effects and
mental health cannot be assumed to be linear [20], we
assessed the shape of the relationship between EED with
the log odds of PPC. To do so, we categorised EED in
terciles, quartiles, quintiles and deciles and plotted it
against the log odds of PPC. The relationship was not
linear; the log odds of PPC in the three first quintiles were
similar and a higher odds of PPC was seen for 4th and 5th
quintiles. Therefore, we set the cut-off for high versus low
EED at the third tertile. This cut-off corresponded to 6 %
of Ecuadorians, which was, in fact, the mean EED in the
sample. Therefore, we categorised EED in two groups; one
defined as high EED and the other defined as low EED. The
other second-level variables: total ethnic density (% of
people whose country of origin was not Spain), and pro-
portion of people with less than primary education were
categorised in tertiles. Ethics committee’s approval was
obtained.
Statistical analyses
We used multilevel logistic regression models, with indi-
viduals at the first level and AR at the second level, to
estimate odds ratios (OR) and 95 % confidence intervals
(CI) accounting for the nesting of individuals within AR.
We performed four models. Model 1 was an empty model
(intercept-only model) that allowed us to calculate the
intraclass correlation coefficient (ICC) or the proportion of
the total variance in PPC that occurs at the AR level. The
ICC was estimated using the latent variable method [34].
Model 2 included EED to estimate the crude relation
between EED and PPC. Model 3 included EED and the
individual-level variables, i.e. sex, individual salary, eco-
nomic confident, attend associations, atmosphere at work
and discrimination, that were identified as confounders for
the relation between EED and PPC. That is, of all variables
collected in the survey summarised before and described in
previous publications [26], only those risk factors for PPC
who had a different distribution between areas with high
and low EED were included in the model. Finally, model 4
also included other area-level variables, such as the pro-
portion of people with less than primary education, as this
was the strongest area-level confounder of the association
between EED and PPC. Total ethnic density and urbanisation
of the area confounded the associationof interest in univariate
analyses but not when the proportion of people with less than
primary studies was included in the model. We tested possible
cross-level interactions between the second-level variable
EED and the first-level variables sex and discrimination. All
analyses were performed in Stata 10 [35].
Results
Overall, 568 Ecuadorians were analysed. The prevalence of
PPC was 24 % (95 %CI 20–28 %). There was variation in
this prevalence according to AR, which ranged from 0 % in
AR1 to 61 % in AR33 (Fig. 1). The distribution of EED
within the 33 AR ranged from 0.9 to 19.5 % with a median
of 4.7 % and a mean of 6.1 %. The prevalence of PPC in
Soc Psychiatry Psychiatr Epidemiol
123
AR with EED equal or over 6 %, from now onwards, high
EED, was 29.5 % and that of AR with less than 6 % EED,
from now onwards, low EED, was 20.4 %. This difference
was statistically significant (p=0.013).
The descriptive characteristics of subjects living in low
EED areas and high EED areas are summarised in Table 1.
Median time in Spain was 5 years (IQR 4–6) in low EED
AR and 5.5 (IQR 4–6) in high EED AR (p=0.13). There
were some small differences in the individual salary and
economic confident distributions between areas with low
and high EED. A larger proportion of people living in high
EED areas attended associations, reported worse atmo-
sphere at work and higher exposure to discrimination. Total
ethnic density was higher in areas with high EED, so was
the proportion of people with less than primary education
(a proxy for low socio-economic status). Up to 60 % of the
people living in areas with high EED were municipalities
compared to 43 % of those living in areas with low EED.
The upper part of Table 2describes the individual risk
factors for poor mental health. The prevalence of PPC was
higher in women and in those with lower salaries. Those with
lack of economic support, bad atmosphere at work and those
who participated in community associations were more likely
to be PPC. The probability of being a PPC increased with
increasing levels of self-perceived discrimination. The lower
part of Table 2describes the OR for PPC for area-level vari-
ables. Ecuadorians living in areas with lower levels of edu-
cation had higher odds of PPC (p=0.19), though these
differences were not statistically significant. Ecuadorians
living in municipal areas (that is, largely in villages) had a
61 % increase odd of PPC (p=0.06).
Table 3presents the multilevel modelling. Model 1 esti-
mates the ICC, that is, the proportion of the total variance in
PPC that occurs at the AR level, which was 6.9 %. Ecuadorians
living in areas with high EED had a 65 % increased odds of
PPC than those living in areas with low EED (OR 1.65 95 %CI
1.01–2.72) (model 2). After adjusting for individual con-
founders for poor mental health, the OR of interest decreases to
0
10
20
30
40
50
60
70
80
90
133
Area of residence (AR)
Prevalence of PPC (95% CI)
Fig. 1 Prevalence of possible psychiatric case (PPC) (95 % CI)
according to the area of residence (AR)
Table 1 Descriptive characteristics of Ecuadorians, globally and
according to the Ecuadorians Ethnic Density (EED) of the Area of
Residence (AR)
Low
EED
[N(%)]
High
EED
[N(%)]
All
[N(%)]
pvalue
a
Individual
characteristics
344 (61) 224 (39) 568 (100)
Sex 0.65
Men 174 (51) 109 (49) 283 (50)
Women 170 (49) 115 (51) 285 (50)
Age 0.62
B25 77 (22) 41 (18) 118 (21)
26–35 174 (51) 114 (51) 288 (51)
36–45 79 (23) 58 (26) 137 (24)
[45 14 (4) 11 (5) 25 (4)
Educational level 0.012
No formal education 10 (3) 16 (7) 26 (5)
Primary 104 (30) 85 (38) 189 (33)
Secondary 205 (60) 111 (50) 316 (56)
University 25 (7) 12 (5) 37 (7)
Marital status 0.33
Single 129 (37) 73 (33) 202 (36)
Married 189 (55) 128 (57) 317 (56)
Separated/widow/
divorced
26 (8) 23 (10) 49 (9)
Lives with partner 0.18
No 102 (30) 55 (25) 157 (28)
Yes 242 (70) 169 (75) 411 (72)
Emotional support
from partner
0.19
Low 81 (24) 49 (22) 130 (23)
Medium 61 (18) 52 (23) 113 (20)
High 137 (40) 93 (42) 230 (40)
Unknown 65 (19) 30 (13) 95 (17)
Has children 0.12
No 80 (23) 40 (18) 120 (21)
Yes 264 (77) 184 (82) 448 (79)
Has a confident 0.69
No 18 (5) 14 (6) 32 (6)
Yes, one 246 (72) 164 (73) 410 (72)
Yes, more than one 80 (23) 46 (21) 126 (22)
Attend associations 0.008
No 280 (81) 161 (72) 441 (78)
Yes 64 (19) 63 (28) 127 (22)
Contacts with
neighbours
0.013
No 138 (40) 67 (30) 205 (36)
Yes 206 (60) 157 (70) 363 (64)
Talks with work
colleagues
0.063
No 111 (32) 56 (25) 167 (29)
Soc Psychiatry Psychiatr Epidemiol
123
1.48 and confidence intervals include 1 (model 3). The single
largest individual confounder (which produced a 10 %
decrease in the OR of interest) was self-perceived discrimina-
tion; the adjusted OR was 1.49 (95 %CI 0.93–2.4). There were
other positive and negative confounders of the relationship
between EED and PPC which justified the inclusion of these in
model 3, though. Adjusting for atmosphere at work decreased
by 7 % the OR of interest and adjusting by individual salary and
economic confident increased the OR of interest by 9 and 5 %,
respectively. Model 4 adjusts for the only area-level con-
founder that remained in the multivariate analyses, the pro-
portion of people without primary education, and which further
decreased the OR of interest to 1.37. Although area urbanisa-
tion and total ethnic density behaved as confounders in the
univariate analyses, after adjusting by the proportion of people
without primary education (data not shown), they did not
produce any relevant change in the OR of interest. None of the
second-level interactions were statistically significant.
Table 1 continued
Low
EED
[N(%)]
High
EED
[N(%)]
All
[N(%)]
pvalue
a
Yes 233 (68) 168 (75) 401 (71)
Has friends 0.15
No 34 (10) 31 (14) 65 (11)
Yes 310 (90) 193 (86) 503 (89)
Social support 0.041
Low 162 (47) 113 (50) 275 (48)
Medium 118 (34) 56 (25) 174 (31)
High 62 (18) 55 (25) 117 (21)
Unknown 2 (1) 0 2 (1)
Employment 0.37
Working currently 300 (87) 203 (91) 203 (91)
Home 19 (6) 11 (5) 30 (5)
Student 4 (1) 1 (1) 5 (1)
Unemployed 21 (6) 8 (4) 29 (5)
Unknown 0 1 (1) 1 (1)
Type of contract 0.19
Independent worker 11 (3) 4 (2) 15 (3)
Civil servant/long-
term contract
92 (27) 63 (28) 155 (27)
Short-term contract 133 (39) 97 (43) 230 (40)
No contract 49 (14) 20 (9) 69 (12)
Does not know
duration
48 (14) 37 (17) 85 (15)
Unknown 11 (3) 3 (1) 14 (2)
Work dissatisfaction 0.52
No 270 (78) 175 (78) 445 (78)
Yes 30 (9) 25 (11) 55 (10)
Unknown 44 (13) 24 (11) 68 (12)
Atmosphere at work 0.010
Excellent/good 242 (70) 141 (63) 383 (67)
Regular/bad 55 (16) 59 (26) 114 (20)
Unknown 47 (14) 24 (11) 71 (13)
Individual salary 0.36
Higher than NMW 232 (67) 153 (68) 385 (68)
Similar to NMW 67 (19) 49 (22) 116 (20)
Inferior to NMW 26 (8) 9 (4) 35 (6)
Has no salary 4 (1) 5 (2) 9 (2)
Unknown 15 (4) 8 (4) 23 (4)
Economic difficulties 0.46
A lot 53 (15) 35 (16) 88 (15)
Some 125 (36) 74 (33) 199 (35)
Not much 84 (24) 62 (28) 146 (26)
Little/one 78 (23) 53 (24) 131 (23)
Unknown 4 (1) 0 4 (1)
Table 1 continued
Low
EED
[N(%)]
High
EED
[N(%)]
All
[N(%)]
pvalue
a
Economic confident 0.46
Yes 264 (77) 180 (80) 444 (78)
No 74 (22) 39 (17) 113 (20)
Unknown 6 (1) 5 (2) 11 (2)
Discrimination 0.001
Never 146 (42) 68 (30) 214 (38)
Sometimes 140 (41) 93 (42) 233 (41)
Always/almost
always
58 (17) 63 (28) 121 (21)
Area characteristics
Total ethnic density \0.001
1 (7–15 %) 175 (51) 18 (8) 193 (34)
2 (16–22 %) 151 (44) 36 (16) 187 (33)
3 (23–45 %) 18 (5) 170 (76) 188 (33)
Proportion of people
with less than
primary education
\0.001
1 (13–36 %) 145 (42) 36 (16) 181 (32)
2 (37–52 %) 72 (21) 116 (52) 188 (33)
3 (53–66 %) 127 (37) 72 (32) 199 (35)
Type \0.001
Neighbourhood 197 (57) 90 (40) 287 (51)
Municipality 147 (43) 134 (60) 281 (49)
NMW National minimal wage
a
pvalue for the comparison of individual and area characteristics
between low and high EED areas derived from the Chi-squared test
Soc Psychiatry Psychiatr Epidemiol
123
Discussion
We have found no protective association between the
Ecuadorians’ ethnic density of the area of residence and the
mental health of Ecuadorian economic migrants in a recent
migration setting in Southern Europe. Contrary to our
research hypothesis, the prevalence of poor mental health
in Ecuadorians living in areas with high EED was higher
than for those living in areas with low EED. This differ-
ence was statistically significant in univariate analyses and
became smaller and non significant after adjustment for
individual and contextual variables. The association
between EED and mental health was partially confounded
by self-perceived discrimination and by the socio-eco-
nomic status in the area of residence
A number of individual variables did partially confound
the crude association found between EED and poor mental
health. That is, part of the increased odds of poor mental
health in areas with high EED was explained by the
characteristics of the Ecuadorians who lived there. Ecu-
adorians from high EED areas perceived that their atmo-
sphere at work was worse and felt they were discriminated
against more frequently than those from low EED areas so,
adjusting for those negative confounders decreased the OR.
However, Ecuadorians living in areas with high EED had
higher earnings and a higher proportion had an economic
confident so adjusting for those increased the OR of
interest. Perceived discrimination was the strongest con-
founder. Our group has previously published a paper stat-
ing the individual risk factors associated to poor mental
health [31,36] and, as reported by many authors [37–39],
perceived discrimination was a very important risk factor
for poor mental health in the Ecuadorians [31].
In our study, the areas with high EED had a higher
overall proportion of migrants from all geographical ori-
gins (total ethnic density), were more rural and the average
education of their census population (a proxy of socio-
economic status) was lower than in the areas with low
EED. The negative effect on mental health associated with
a high EED was also partially confounded by socio-eco-
nomic deprivation of the area. Indeed, the proportion of
registered people with less than primary education was one
of the largest single confounders suggesting that much of
the effect associated to EED could be explained by the
poverty of the area. This effect has been found by many
other authors and favour material versus psychosocial
causes of ill health [11–13]. After adjusting for contextual
and individual confounders of PPC, the odds of having
poor mental health was still 37 % higher in Ecuadorians
Table 2 Prevalence and odds ratios for possible psychiatric case
(PPC) in Ecuadorians, according to selected covariates
PPC [N(%)] OR (95 % CI) p
Individual characteristics
Sex
Men 38 (13) 1.00 \0.001
Women 98 (34) 3.57 (2.31–5.51)
Age
B25 21 (18) 1.00 0.148
26–35 78 (27) 1.83 (1.04–3.22)
36–45 33 (24) 1.46 (0.77–2.76)
[45 4 (16) 0.94 (0.28–3.15)
Individual salary
Higher than NMW 78 (20) 1.00 \0.001
Similar to NMW 35 (30) 1.82 (1.11–2.99)
Inferior to NMW 18 (51) 5.16 (2.38–11.16)
Has no salary/
Unknown
5 (16) 0.75 (0.27-2.11)
Economic confident
Yes 92 (21) 1.00 0.002
No 41 (36) 2.37 (1.47–3.83)
Unknown 3 (27) 1.72 (0.41–7.23)
Attend associations
No 94 (21) 1.00 0.015
Yes 42 (33) 1.78 (1.12–2.84)
Atmosphere at work
Excellent/good 74 (19) 1.00 \0.001
Regular/bad 43 (38) 1.73 (1.09–2.74)
Unknown 19 (27) –
Discrimination
Never 35 (16) 1.00 0.003
Sometimes 59 (25) 1.72 (1.06–2.78)
Always/Almost
always
42 (35) 2.54 (1.47–4.38)
Area characteristics
Total ethnic density
1 (7–15 %) 41 (21) 1.00 0.42
2 (16–22 %) 42 (22) 1.09 (0.58–2.04)
3 (23–45 %) 53 (28) 1.48 (0.79–2.74)
Proportion of people with less than primary education
1 (13–36 %) 33 (18) 1.00 0.19
2 (37–52 %) 51 (27) 1.73 (0.92– 3.26)
3 (53–66 %) 52 (26) 1.61 (0.86–3.01)
Urbanisation
Neighbourhood 57 (20) 1.00 0.06
Municipality 79 (28) 1.61 (0.98–2.65)
NMW National minimal wage
Soc Psychiatry Psychiatr Epidemiol
123
who lived in areas with more than 6 % of Ecuadorians com-
pared to fewer than 6 %. We acknowledge that this difference
is not statistically significant and speculate that there may be
other unmeasured contextual effects. Nevertheless, the mag-
nitude of the association between these unmeasured con-
founders and PPC should have to be extremely high to reverse
the direction of the OR, and this is unlikely.
Our results join the pool of reports that have found no
protective effect of ethnic density for mental health in
ethnic minority members [9,10], but, as far as we know,
are the first to show these effects in a Southern European
setting of recently arrived economic migrants. We decided
to study Ecuadorians’ Ethnic Density rather than total
ethnic density as we were interested in co-ethnics support.
Table 3 Odds ratios for the
association between
Ecuadorians Ethnic Density
(EED) and Possible Psychiatric
Case (PPC) adjusting for
individual and contextual
variables
NMW National minimal wage,
SE Standard error, ICC
Intraclass correlation coefficient
Model 1,
Empty
model
Model 2,
EED
included
Model 3, EED
and individual-
level variables
Model 4, EED,
individual variables,
proportion of people
with less than
primary
education
Fixed effects
AR-level variables
EED
Low (\6 %) 1.00 1.00 1.00
High (C6 %) 1.65 (1.01–2.72) 1.48 (0.87–2.55) 1.37 (0.78–2.40)
Proportion of people with less than 1 education
1 (13–36 %) 1.00
2 (37–52 %) 1.34 (0.67–2.69)
3 (53– 66 %) 1.48 (0.77–2.84)
Individual-level variables
Sex
Men 1.00 1.00
Women 3.67 (2.21–6.09) 3.69 (2.22–6.12)
Individual salary
Higher than NMW 1.00 1.00
Similar to NMW 0.92 (0.52––1.64) 0.90 (0.51–1.61)
Inferior to NMW 2.95 (1.28–6.80) 2.95 (1.28–6.79)
Has no salary/
Unknown
0.56 (0.18––1.70) 0.55 (0.18–1.67)
Economic confident
No 1.00 1.00
Yes 0.51 (0.30–0.85) 0.52 (0.31–0.87)
Unknown 1.07 (0.22–5.20) 1.07 (0.22––5.19)
Attend associations
No 1.00 1.00
Yes 2.14 (1.26–3.61) 2.19 (1.29–3.70)
Atmosphere at work
Excellent/good 1.00 1.00
Regular/bad 1.91 (1.14–3.21) 1.85 (1.10–3.11)
Unknown 0.95 (0.46––1.92) 0.92 (0.45–1.88)
Discrimination
Never 1.00 1.00
Sometimes 1.32 (0.78–2.24) 1.32 (0.78–2.23)
Always/almost always 2.34 (1.28–4.31) 2.38 (1.30–4.38)
Random effects
Between AR variance
(SE)
0.246 (0.154) 0.179 (0.136) 0.187 (0.160) 0.157 (0.154)
ICC (%) 6.95
Soc Psychiatry Psychiatr Epidemiol
123
As mentioned in the introduction, we anticipated ethnic
density effects to be context specific. Our findings have to
be explained within the recent migration processes into
Spain that may have niether permitted the establishment of
the ethnic networks and community links nor the creation
of Ecuadorian institutions, newspapers or the establishment
of local businesses, thought to underlie the protective effect
for health outcomes. Indeed, publications reporting posi-
tive ethnic density effects come from Anglo-Saxon and
Northern European countries with established ethnic
minority communities [2–8], a different scenario to the
Spanish one, where Ecuadorian migrants started settling in
during the mid-90s. The median residence time in Spain of
the Ecuadorians in this study was 5 years. Building up a
community requires time and permanence in a given site
suggesting that the mechanisms through which ethnic
density effects may benefit established minorities may not
be present in recent migration settings.
Pickett and Wilkinson have described some of the
challenges and limitations of assessing ethnic density
effects on health [20]; the shape and the cut-off point for
ethnic density is an important one. We have taken special
care in deciding which cut-off point to choose after
exploring the shape of the relationship between EED and
PPC, which was far from linear. We repeated the analyses
with different cut-off points for EED and the main con-
clusion remained unaltered (data not shown). We are aware
that our sample size is small but, by conducting an ad-hoc
survey rather that using general surveys, we have been able
to collect with great detail a high number of individual
variables which have allowed us to perform fine adjust-
ments by first-level factors. However, collecting other area
level variables such as unemployment rates, housing prices
and measures of Ecuadorians’ social capital was an
impossible task as those data were not available at the area
level we were studying [26]. We have used the educational
level attained by the people of the AR as from the Census
2001 as a proxy of socio-economic status. This has been
used by Regidor et al. [33] in Spain. Attending an associ-
ation was associated with an increased odds of poor mental
health. Given the transversal designs of our study it is not
possible to establish directionality and this result may
suggest that people who have poor mental health or risk
factors to develop mental health problems seek help by
attending associations.
These are the first analyses exploring ethnic density
effects on mental health in a recent migratory context and
the first in a Southern European country. We have found no
protective association between the Ecuadorians’ ethnic
density of the area of residence and the mental health of the
Ecuadorians. In this study, we have identified that Ecu-
adorians living in areas with a higher density of Ecuado-
rians have poorer mental health than those who live in
areas with lower density and that, as well as individual risk
factors such as exposure to discrimination, part of that
effect is explained by the low socio-economic status of the
area of residence. These results are relevant for policy
interventions as the areas with high EED are largely rural,
have also higher total ethnic density and have little formal
education, mimicking the patterns described in other
countries. Efforts to fight against racism and to reduce
discrimination are needed. Besides, our results call for
further research into how ethnic density effects operate in
different contexts.
Acknowledgments This work was supported by the Spanish
Research Fund (FIS PI041026). DA was employed by CIBERSP
(Ciber of Epidemiology and Public Health). IJ is employed with funds
from RIS (Spanish HIV Research Network for excellence), RD06/
006. None of the funding bodies played a part in the research pro-
tocol, data analyses, data interpretation, or writing of the report.
References
1. Rabkin JG (1979) Ethnic density and psychiatric hospitalization:
hazards of minority status. Am J Psychiatry 136:1562–1566
2. Boydell J, van Os J, McKenzie K, Allardyce J, Goel R,
McCreadie RG et al (2001) Incidence of schizophrenia in ethnic
minorities in London: ecological study into interactions with
environment. BMJ 323:1336–1338
3. Halpern D, Nazroo J (2000) The ethnic density effect: results
from a national community survey of England and Wales. Int J
Soc Psychiatry 46:34–46
4. Veling W, Susser E,van Os J, Mackenbach JP, Selten JP, Hoek HW
(2008) Ethnic density of neighborhoods and incidence of psychotic
disorders among immigrants. Am J Psychiatry 165:66–73
5. Das-Munshi J, Becares L, Dewey ME, Stansfeld SA, Prince MJ
(2010) Understanding the effect of ethnic density on mental
health: multi-level investigation of survey data from England.
BMJ 341:c5367
6. Collins JW Jr, David RJ, Rankin KM, Desireddi JR (2009)
Transgenerational effect of neighborhood poverty on low birth
weight among African Americans in Cook County, Illinois. Am J
Epidemiol 169:712–717
7. Elo IT, Culhane JF, Kohler IV, O’Campo P, Burke JG, Messer
LC et al (2009) Neighbourhood deprivation and small-for-
gestational-age term births in the United States. Paediatr Perinat
Epidemiol 23:87–96
8. Be
´cares L, Nazroo J, Stafford M (2011) The ethnic density effect
on alcohol use among ethnic minority people in the UK. J Epi-
demiol Community Health 65:20–25
9. Cochrane R, Bal SS (1988) Ethnic density is unrelated to inci-
dence of schizophrenia. Br J Psychiatry 153:363–366
10. Henderson C, Diez Roux AV, Jr Jacobs DR, Kiefe CI, West D,
WilliamsDR (2005) Neighbourhoodcharacteristics,individual level
socioeconomic factors, and depressive symptoms in young adults:
the CARDIA study. J Epidemiol Community Health 59:322–328
11. Marmot M, Wilkinson RG (2001) Psychosocial and material
pathways in the relation between income and health: a response
to Lynch et al. BMJ 322:1233–1236
12. Lynch JW, Smith GD, Kaplan GA, House JS (2000) Income
inequality and mortality: importance to health of individual
income, psychosocial environment, or material conditions. BMJ
320:1200–1204
Soc Psychiatry Psychiatr Epidemiol
123
13. Silver E, Mulvey EP, Swanson JW (2002) Neighborhood struc-
tural characteristics and mental disorder: Faris and Dunham
revisited. Soc Sci Med 55:1457–1470
15. Hudson DL, Neighbors HW, Geronimus AT, Jackson JS (2011)
The relationship between socioeconomic position and depression
among a US nationally representative sample of African Amer-
icans. Soc Psychiatry Psychiatr Epidemiol (Epub ahead of print)
16. Karlsen S, Nazroo JY, Stephenson R (2002) Ethnicity, environ-
ment and health: putting ethnic inequalities in health in their
place. Soc Sci Med 55:1647–1661
17. Halpern D (1999) Minorities and mental health. Soc Sci Med
36:597–607
18. Becares L, Nazroo J, Stafford M (2009) The buffering effects of
ethnic density on experienced racism and health. Health Place
15:670–678
19. Bhugra D (2004) Migration and mental health. Acta Psychiatr
Scand 109:243–258
20. Pickett KE, Wilkinson RG (2008) People like us: ethnic group
density effects on health. Ethn Health 13:321–334
21. Whitley R, Prince M, McKenzie K, Stewart R (2006) Exploring
the ethnic density effect: a qualitative study of a London electoral
ward. Int J Soc Psychiatry 52:376–391
22. Razum O, Zeeb H, Rohrmann S (2000) The ‘healthy migrant
effect’—not merely a fallacy of inaccurate denominator figures.
Int J Epidemiol 29:191–192
23. Beiser M, Hou F, Hyman I, Tousignant M (2002) Poverty, family
process, and the mental health of immigrant children in Canada.
Am J Public Health 9:220–227
24. Instituto Nacional de Estadı
´stica (INE) Web site. Estadı
´stica de
Poblacio
´n. Padro
´n continuo de Poblacio
´n 2006. http://www.
ine.es/.http://www.ine.es/jaxi/menu.do?type=pcaxis&path=%2Ft
20%2Fe260&file=inebase&L=. Accessed 2010 Nov
25. Go
´mez Ciriano E, Tornos Cubillo A, Colectivo Ioe
´(2007) Ecu-
atorianos en Espan
˜a. Una aproximacio
´n sociolo
´gica. Documentos
del Observatorio Permanente de la Inmigracio
´n: nu
´mero 15.
Madrid: Ministerio de Trabajo e Inmigracio
´n (MTIN), p 262
26. Alvarez-Del Arco D, Lla
´cer A, Valero JA, Garcia-Fulgueiras A,
Garcia-Pina R, Garcia-Ortuzar V (2009) Methodology and
fieldwork logistics of a multilevel research study on the influence
of neighbourhood’s characteristics on natives and Ecuadorian’s
mental health in Spain. Rev Esp Salud Publ 83:493–508
27. Goldberg DP, Hillier VF (1979) A scaled version of the General
Health Questionnaire. Psychol Med 9:139–145
28. Lobo A, Perez-Echeverria MJ, Artal J (1986) Validity of the
scaled version of the General Health Questionnaire (GHQ-28) in
a Spanish population. Psychol Med 16:135–140
29. Broadhead WE, Gehlbach SH, Degruy FV, Kaplan BH (1988)
The Duke-UNK functional social support questionnaire: mea-
surement of social support in family medicine patients. Med Care
26:709–723
30. Finch BK, Kolody B, Vega WA (2000) Perceived discrimination
and depression among Mexican-origin adults in California.
J Health Soc Behav 41:295–310
31. Lla
´cer A, Amo JD, Garcia-Fulgueiras A, Ibanez-Rojo V, Garcia-
Pino R, Jarrin I et al (2009) Discrimination and mental health in
Ecuadorian immigrants in Spain. J Epidemiol Community Health
63:766–772
32. Noh S, Kaspar V (2003) Perceived discrimination and depression:
moderating effects of coping, acculturation, and ethnic support.
Am J Public Health 93:232–238
33. Regidor E, Pascual C, de la Fuente L, Santos JM, Astasio P,
Ortega P (2011) Socio-economic position, family demands and
reported health in working men and women. Eur J Public Health
21:109–115
34. Snijders TAB, Bosker RJ (1999) Multilevel analysis: An intro-
duction to basic and advanced multilevel modeling. Sage Publi-
cations, Thousand Oaks, CA
35. StataCorp (2001) Stata Statistical Software, release 7.0. Stata
Corporation, College Station
36. Del Amo J, Jarrı
´n I, Garcı
´a-Fulgueiras A, Iba
´n
˜ez-Rojo V, A
´lvarez
D, Rodrı
´guez-Arenas MA et al (2011) Mental health in Ecu-
adorian migrants from a population-based survey: the importance
of social determinants and gender roles. Soc Psychiatry Psychiatr
Epidemiol 46:1143–1152
37. Krieger N (1999) Embodying inequality: a review of concepts,
measures, and methods for studying health consequences of
discrimination. Int J Health Serv 29:295–352
38. Paradies Y (2006) A systematic review of empirical research on
self-reported racism and health. Int J Epidemiol 35:888–901
39. Chakraborty AT, McKenzie KJ, Hajat S, Stansfeld SA (2010)
Racism, mental illness and social support in the UK. Soc Psy-
chiatry Psychiatr Epidemiol 45:1115–1124
Soc Psychiatry Psychiatr Epidemiol
123