Introduction
Women of Pakistani and Indian origin living in the UK have a higher prevalence of common mental disorder than other ethnic groups, but the reasons for this are not clear (Williams & Hunt, Reference Williams and Hunt1997; Weich et al. Reference Weich, Nazroo, Sproston, McManus, Blanchard, Erens, Karlsen, King, Lloyd, Stansfeld and Tyrer2004). Recently we found that few years of education, marked social difficulties and poor social support were associated with depressive disorder, suggesting that some aspects of Brown & Harris's social model of depression hold for this group although some of the vulnerability factors might be different (Brown & Harris, Reference Brown and Harris1978; Gater et al. Reference Gater, Tomenson, Percival, Chaudhry, Waheed, Dunn, Macfarlane and Creed2008). It has also been reported that distress or depression in this population is associated with lack of adapting to UK culture, racism and difficulty speaking English (Husain et al. Reference Husain, Creed and Tomenson1997; Williams & Hunt, Reference Williams and Hunt1997; Weich et al. Reference Weich, Nazroo, Sproston, McManus, Blanchard, Erens, Karlsen, King, Lloyd, Stansfeld and Tyrer2004). Lack of acculturation to the host country has been documented as a risk factor for depression in immigrant populations in different countries, but the results are conflicting (Miller & Chandler, Reference Miller and Chandler2002; Bhui et al. Reference Bhui, Lawrence, Klineberg, Woodley-Jones, Taylor, Stansfield, Viner and Booy2005; Parker et al. Reference Parker, Chan, Tully and Eisenbruch2005). The most recent studies have suggested that stressful life events may mediate the association between depression and lack of acculturation (Hwang & Myers, Reference Hwang and Myers2007; Hwang & Ting, Reference Hwang and Ting2008). The current study assessed whether lack of acculturation and social difficulties were independent correlates of distress, prevalent and persistent depressive disorder in a population-based sample of women of Pakistani origin in the UK. We hypothesized that depressed women would be less acculturated than non-depressed women, and that lack of acculturation would be an independent predictor of persistent depressive disorder.
Method
The study design was a cross-sectional two-phase population-based survey to determine correlates of depression, followed by a prospective cohort study of depressed subjects to identify risk factors for persistent depressive disorder.
Sample
The study sample of 487 (66% response rate) was drawn from four general practitioner (GP) practices in central Manchester, located in areas with a high population of British Pakistani Muslims. Practice registers were scrutinized to identify all 18- to 65-year-old female patients with recognizable Pakistani Muslim names. One subject was randomly selected from each household using a random number generator. The explanatory letters and the Self Reporting Questionnaire (SRQ), both in the English and Urdu language, were posted along with a return envelope to the potential participants selected from the GP lists. The letter also stated that if a completed SRQ or a refusal to participate letter was not received within a few weeks, a researcher would visit the woman's home. If a completed SRQ was not returned within 3 weeks, bilingual researchers visited the potential participants at home. During these visits the researchers again explained the study and answered any questions. Potential participants were requested to complete the SRQ at the time or it was collected later at a mutually agreed time.
If the potential participant was not at home repeatedly over five visits at different times of the day, the researcher tried to establish residency by enquiring from a neighbour. If, in this way, the person was found not to live at the address they were considered non-resident. Those who did not return the SRQ were considered as refusals. Participants gave written informed consent for inclusion in the study, and ethical approval was obtained from Central Manchester Ethical committee (no. CEN/00/122).
Measurement of psychiatric morbidity
The first phase used the SRQ, which is a widely used psychiatric screening questionnaire (WHO, 1994). It consists of 20 items to which the respondent answers ‘yes’ or ‘no’ according to whether they have experienced the symptom in the previous 30 days. The SRQ takes only a few minutes to complete. The SRQ has been successfully used in Pakistani populations and has been shown to have good sensitivity and specificity. It was validated earlier against the Psychiatric Assessment Schedule (PAS) (Husain et al. Reference Husain, Creed and Tomenson2000; Reference Husain, Gater, Tomenson and Creed2006). In the second phase of the study all high scorers (score of 7 or more on the SRQ) and a 1 in 4 sample of low scorers were contacted for interview using the Schedule for Clinical Assessment in Neuropsychiatry (SCAN) to assess whether subjects met the criteria for either International Classification of Diseases, tenth revision (ICD-10) or Diagnostic and Statistical Manual of Mental Disorders, 4th edition (DSM-IV) depressive disorders (WHO, 1992; Gater et al. Reference Gater, Tomenson, Percival, Chaudhry, Waheed, Dunn, Macfarlane and Creed2008). A total of 202 women (69% response rate) completed the SCAN interview. On the basis of these assessments the participants were divided into three groups: (1) depressive disorder, according to either ICD-10 or DSM-IV; (2) distressed, i.e. a high SRQ score (7 or more) but not depressed on the SCAN interview; and (3) not depressed, i.e. low SRQ group (<7). The reason for the three groups was, first, to examine whether an exposure–response relationship occurred between potential risk factors and depressed, distressed and non-depressed groups, and, second, whether the depression–acculturation association differed according to depression group.
Measurement of social stress
The Life Events and Difficulties Schedule (LEDS) was used to measure social stress (Brown & Harris, Reference Brown and Harris1978). This is a semi-structured interview, which covers a number of domains such as family, work, health and relationships to establish any life events and difficulties that may have occurred or been present in the 12-month period prior to the date of the interview. The interviewers in this study were trained in the use of this instrument at Bedford College, London. They were supervised by senior researchers experienced in the use of LEDS (F.C. and N.H.), and all events and difficulties were presented for rating to a group who were blind to whether the participant was depressed or not. This work has led to a supplement to the Bedford College LEDS dictionary because some of the events and difficulties are unique to the Pakistani population. The proportion of each group who had experienced a marked social difficulty, namely one that carried a marked degree of threat to the individual (1–3 on a scale from 1–7) (Brown & Harris, Reference Brown and Harris1978), is quoted. Marked difficulties were also rated according to the dimensions of ‘humiliation’ or ‘entrapment’ in an ongoing difficult situation. In order to assess change in the level of social difficulties between baseline and follow-up, a social difficulties score was derived for each of the following: physical health of the respondent or close relative, marital or other close relationship difficulties, housing and financial difficulties (Brown et al. Reference Brown, Adler and Bifulco1998).
Measurement of assimilation into British society
An acculturation questionnaire was derived from the Cultural Questionnaire (Stopes-Roe & Cochrane, Reference Stopes-Roe and Cochrane1990). The original questionnaire comprises 116 questions covering the areas of employment, friendships, language, cooking, movies and media, religious ceremonies and festivals, clothes, marriages and family pattern, attitude towards country of migration, and discrimination and prejudice. Since many of the questions overlapped (e.g. ‘do you speak any English at home?’ and ‘which language do you use with the family?’), the questionnaire was initially reduced to 38 items, and from these 38 only 16 questions, which covered a broad range of aspects of daily life (items displayed in Table 3), were included.
Each item was scored as 0=less or 1=more acculturated, with a range of 0–16 (mean 7.5, s.d.=3.4) and higher values indicating a higher level of acculturation. This was then dichotomized using a cut-off of 8/9; women scoring 9 or above (more than half the 16 items indicating good acculturation) were considered to be more acculturated, whereas those scoring 8 or less were considered to be less acculturated (Chaudhry, Reference Chaudhry2007). A factor analysis was carried out on the 16 items, and subsequent analyses used the factor analysis regression scores for each of the first five factors, which had eigenvalues >1, and which together explained 63.3% of the total variation.
Other variables
Information was collected on age, marital status and duration of full-time education (none, 1–7 years, 8–12 years or more than 12 years). Employment status was categorized as one of seven options: working full-time, working part-time, domestic work full-time in the house, not working due to ill health/disability, unemployed, student or retired.
Socio-economic status was rated using the Goldthorpe and Hope system (Goldthorpe & Hope, Reference Goldthorpe and Hope1974) according to the participant's most recent job (Escobar & Vega, Reference Escobar and Vega2000). If the participant had never worked, then her partner's current or last position was used.
Satisfaction with social support was determined by asking about social networks, the kind of social support received and whether the respondent was satisfied with the support received. The response was categorized on a four-point scale. The number who expressed dissatisfaction with social support is quoted.
Social isolation was determined by enquiring about the participant's daily activities, how often she went out of the house, autonomy, and contact with friends and acquaintances. This was rated on a four-point scale (1=marked, 2=moderate, 3=some, 4=little or none) and analyses compared the 72 (39%) women who were rated as experiencing any degree of social isolation with the remainder.
Intimate confiding (Brown & Harris, Reference Brown and Harris1978) was determined by asking the respondent with whom she could discuss problems. If it became clear that the woman had no one in whom she could confide then this was rated as no confidant. Other variables that were considered as possible vulnerability factors for depression were loss of mother before the age of 11 years, having three or more children under the age of 14 years, living in extended families, not having children and having only female offspring.
At 6 months after the baseline interview, the depressed and distressed participants were reinterviewed using the SCAN and LEDS, to identify the predictors for persistent depression. The non-depressed participants were not reinterviewed, as the sample was not large enough to examine predictors of new onset depression.
Statistical analysis
Data were analysed using SPSS (version 15; SPSS, Inc., USA). The first set of univariable analyses compared the three groups, depressive disorder, distressed, and not depressed with respect to sociodemographic factors, vulnerability factors, life events and difficulties, and acculturation items, using Pearson's χ2 test with test for linear trend across the three groups. Fisher's exact test was used to compare proportions of women with and without each of the potential risk factors who were classified as being acculturated.
Factor analysis of the Acculturation Questionnaire
Principal components analysis was used with Kaiser–Meyer–Olkin (KMO) and Bartlett's test for sphericity, and factors with eigenvalues >1 were identified. Factors were rotated to improve their interpretability. Both orthogonal (equamax) and oblique rotations (oblimin) were applied. The relationship between acculturation and depression group was examined using one-way analysis of variance on both the total acculturation score and the five factor scores derived from the factor analysis.
A hierarchical logistic regression analysis was performed to identify those variables independently associated with depressive disorder at baseline compared with non-depressed and distressed groups combined. Adjusted odds ratios (ORs) with 95% confidence intervals (CIs) were calculated. Independent variables included were those found to be significant in the univariate analyses, but not including any which may cause problems with multicollinearity. In detail, the independent variables were age and total acculturation score (model 1), plus social isolation (model 2), plus presence/absence of a marked health difficulty, marital and/or close relationship difficulty, and one or more independent severe life events (model 3). All variance inflation factors in the final model were checked, since any greater than 2 may indicate problems with multicollinearity.
For those women with depressive disorder at baseline who were reinterviewed 6 months later, univariable analyses used Fisher's exact test to identify baseline dichotomous variables which were associated with persistent v. resolved depression and t tests for continuous scores such as total SRQ score and acculturation scores. A hierarchical logistic regression analysis was used to determine factors relating to depression at follow-up (i.e. comparing persistent with resolved depression). The independent variables included were age >50 years and social isolation (model 1), plus baseline total SRQ score, baseline score for marital, partner and other close relationship difficulties, and the first three acculturation factor scores (model 2). In the third model, a change score (between baseline and follow-up) for marked difficulties was included.
Results
A total of 738 women of Pakistani origin were considered to be eligible for the study, and 487 completed the screening instrument SRQ (66% response rate). Of these 487, 227 (46.6%) scored 7 or more on the SRQ and all of these participants were invited to participate in the second-phase interview together with 66 randomly selected low scorers. A total of 202 completed the SCAN interview (158 high scorers, 70% response rate, and 44 low scorers, 67% response rate), and 193 of these completed the LEDS. Of these 193 participants, 175 completed the Acculturation Questionnaire.
Of the 202 participants, 96 had a depressive disorder, 65 had high SRQ scores but did not have depressive disorder (the distressed group), and 41 had low SRQ scores (not depressed). The adjusted prevalence of depressive disorder using inverse probability sampling weights was 31.1% (95% CI 24.1–38.0) (Gater et al. Reference Gater, Tomenson, Percival, Chaudhry, Waheed, Dunn, Macfarlane and Creed2008).
Correlates of depression at baseline (univariable analyses)
Women with depressive disorder were more likely to be aged over 50 years than the other two groups; they were less likely to have social support, but did not differ significantly with respect to marital status, education, employment or language used to complete the SRQ (Table 1). There was no significant difference between the three groups in terms of separation from mother before the age of 11 years, having three or more children, living in an extended family, and not having any children at all or any male children (Table 1). After Bonferroni adjustments to account for the fact that 14 variables were analysed in Table 1, only age and social isolation remained significant at p<0.05.
Table 1. Comparison of depressed, distressed and non-depressed Pakistani women with respect to sociodemographic variables
df, Degrees of freedom; SRQ, Self Reporting Questionnaire; LEDS, Life Events and Difficulties Schedule.
Data are given as number of subjects (percentage).
a Social variables were asked at the LEDS interview, and therefore the maximum group sizes are 92, 62 and 39, respectively.
b Missing data for one subject in the distressed group.
c Missing data for one subject in each group.
d Missing data for nine subjects in the depressed group and three subjects in the distressed group.
e Missing data for two subjects in the depressed group and one subject in the distressed group.
* Significant trend across the three depression groups (p<0.05).
Severe threatening life events and marked difficulties were experienced more often by the depressed than the other two groups. There was an exposure–response relationship for health difficulties, but for other types of difficulty and severe life events, the depressive disorder group was clearly different from the other two groups (Table 2), and was significant in all cases, except for housing difficulties where there were no significant differences between the three groups. After Bonferroni adjustments to account for the fact that eight variables were analysed in Table 2, all the significant variables except for severe independent life events and money-related difficulties remained significant at p<0.05.
Table 2. Comparison of depressed, distressed and non-depressed women and marked difficulties and severe events at baseline
df, Degrees of freedom.
Data are given as number of subjects (percentage).
Acculturation
Of the 175 women who completed the Acculturation Questionnaire, 64 (36.6%) scored 9 or more, and were considered to be more acculturated than the rest. For example, the majority of these more acculturated women felt that UK is their home country (70% v. 26% of less acculturated women); they speak fluent English (80% v. 16%) and eat either Pakistani or English food (91% v. 45%).
A total of 55 (73.1%) of the 78 depressed women, 33/60 (55%) of the distressed and 21/37 (56.8%) of the non-depressed women were less acculturated (p=0.058). The depressed group differed significantly from the remainder (54/97, 55.7%) (p=0.019).
The sociodemographic features of older age (11% acculturated v. 42% for younger women, Fisher's exact p=0.002), first-generation immigrant status (21% v. 79%, p<0.001), no formal education (5% v. 40%, p=0.002), social isolation (16% v. 50%, p<0.001) and absence of a confidant (10% v. 42%, p<0.001) were associated with less acculturation. On the other hand, the experience of individual types of marked difficulties were not significantly associated with acculturation: housing (37% v. 37%), money/possessions (14% v. 38%), health (29% v. 39%) and marital, partner or other relationships (25% v. 41%).
Factor analysis of the Acculturation Questionnaire
The principal components analysis resulted in a KMO of 0.77 and Bartlett's test for sphericity of χ2=888.0 (degrees of freedom=120; p<0.001). This indicates the sampling adequacy that factor analysis will give distinct and reliable factors.
The first five factors had eigenvalues >1. The cumulative variance explained by these five factors was 63.3% (Table 3).
Table 3. Factor loadings from principal components analysis of 16 acculturation items after equamax rotation of factors with eigenvalues >1Footnote a
a Loadings <0.4 are omitted for clarity. In further analyses, the item ‘speaks fluent English’ was loaded onto factor 1 because its loading was greater for factor 1 than for factor 2. The item ‘unacceptable for son/daughter to marry an English person’ was included in factor 1, since its largest loading, 0.39, was on that factor. Oblimin rotation yielded very similar groupings of items onto these five factors.
The mean acculturation scores for the depressed, distressed and non-depressed groups were: 6.8 (s.d.=3.3), 8.0 (s.d.=3.3) and 8.0 (s.d.=3.4) respectively (one-way analysis of variance, p=0.054). The only factor score significantly associated with depression group was factor 2 (friends and clothes): −0.17 (s.d.=0.9), 0.002 (s.d.=1.0) and 0.36 (s.d.=1.1) (p=0.025), indicating least acculturation in the depressed group and greatest acculturation in terms of friends and clothes being in the non-depressed group.
In logistic regression analysis with depressive disorder group versus the remainder as the dependent variable, low acculturation was not associated with depressive disorder when age was added to the regression equation (model 1). Older age, social isolation and marked difficulties concerning health and marital or other close relationships were independent correlates of depressive disorder at baseline (model 3, Table 4).
Table 4. Logistic regression analysis with depression at baseline as the dependent variable
Data are given as odds ratio (95% confidence interval).
a Model 1 includes age and acculturation score.
b Model 2 includes also social isolation.
c Model 3 includes also marked health and marital/other close relationship difficulties and independent severe life events. All variance inflation factors in the final model are less than 1.2, indicating no problems with multicollinearity.
Correlates of persistent depression (6 months after baseline)
A total of 89 of the 96 depressed women at baseline were followed up: two women refused, two could not be located and three could not be contacted. Of the 89 women with depressive disorder, 55 remained depressed at follow-up and 34 were no longer depressed.
Of the sociodemographic variables listed in Table 1, the following were significantly associated with depression at follow-up in univariable analyses: aged over 50 years (p=0.003), lack of education (p=0.023), lack of support (p=0.018), having no children (p=0.024), no male children (p=0.034), baseline SRQ score (p<0.001), and acculturation score (p=0.016). Only age >50 years and baseline SRQ score would remain significant after Bonferroni adjustments for multiple variable tests. The only type of marked difficulty at baseline which was associated with persistent depression was that of marital/partner or other relationships: 31 of 55 women (56%) with persistent depression experienced such a marked difficulty at baseline, and all except two of these still had the same marked difficulty at follow-up. By contrast, of the 34 women whose depression resolved, 11 had such a marked difficulty at baseline but in only five did this continue at follow-up.
Of the depressed women at baseline who completed the Acculturation Questionnaire, 50 remained depressed and in 25 it resolved. The respective acculturation scores for these groups were 6.0 (s.d.=3.0) and 8.0 (s.d.=3.4) (p=0.016), showing that the less acculturated women were more likely to have persistent depression; 41/56 (73%) less well-acculturated women were still depressed at follow-up, compared with nine out of 19 (47%) well-acculturated women (OR 3.0, 95% CI 1.0–8.9).
The only factor score associated with persistent depression was factor 1 (use of English); respective scores were −0.44 (s.d.=0.9) and 0.35 (s.d.=1.0) (p=0.001, for persistent versus resolving depression), indicating that good use of the English language was associated with recovery.
In logistic regression analysis, after adjusting for baseline SRQ score, persistent depression was associated with baseline and increased scores for marital and close relationship difficulties and factor 1 and 3 scores of acculturation (use of the English language and feels that Pakistan is home) (Table 5). The results were similar when oblimin factors were used instead of orthogonal factors. The result for ‘Pakistan is home’ was slightly more significant (OR 0.35, 95% CI 0.14–0.88, p=0.026).
Table 5. Logistic regression analysis with depression at follow-up as the dependent variable
SRQ, Self Reporting Questionnaire.
Data are given as odds ratio (95% confidence interval).
a Model 1 includes age and social isolation.
b Model 2 includes also SRQ total score, social difficulties score for marital and other close relationship difficulties, and acculturation factors 1–3 .
c Model 3 also includes change in social difficulties score for marital and other close relationship difficulties. All variance inflation factors in the final model are less than 1.4, indicating no problems with multicollinearity.
Discussion
The findings of this study confirm that the high prevalence of depression in women of Pakistani origin living in the UK is associated with older age, social isolation and marked social difficulties (Weich et al. Reference Weich, Nazroo, Sproston, McManus, Blanchard, Erens, Karlsen, King, Lloyd, Stansfeld and Tyrer2004; Gater et al. Reference Gater, Tomenson, Percival, Chaudhry, Waheed, Dunn, Macfarlane and Creed2008). What is new about these findings is the identification of certain aspects of acculturation as independent predictors of persistent depressive disorder in this group of women. Both our hypotheses were confirmed. Our data emphasize the very high rate of marked difficulties in close personal relationships experienced by these women. We did not find that the vulnerability factors for depression identified by Brown & Harris (Reference Brown and Harris1978), such as loss of mother before the age of 11 years; having three or more children under the age of 14 years; and lack of employment were relevant in this population, except lack of a confidant.
The literature concerning the relationship between acculturation and depression in immigrants is somewhat conflicting. On the one hand, there is some evidence that greater acculturation is associated with increased chance of depression (Hwang & Myers, Reference Hwang and Myers2007; Hwang & Ting, Reference Hwang and Ting2008). This may occur because of the gradual loss of protective cultural factors associated with greater acculturation and increased exposure to stressors that immigrants may not be equipped to handle (Escobar & Vega, Reference Escobar and Vega2000). On the other hand, it has been reported that acculturation is associated with lower rates of depression; this was found among immigrants who reported greater English usage and more integrated social relationships (Escobar & Vega, Reference Escobar and Vega2000; Miller & Chandler, Reference Miller and Chandler2002). These authors have suggested that working and living near English speakers and exposure to English cultural events or media enhances psychological well-being. It appears that the process of acculturation is specific to particular immigrant groups in different cultures and may depend on the age at which the immigration takes place (Oh et al. Reference Oh, Koeske and Sales2002; Hwang et al. Reference Hwang, Chun, Takeuchi, Myers and Prabha2005). Psychiatric disorder was reported to be inversely related to command of English in Greek Cypriots in London (Mavreas & Bebbington, Reference Mavreas and Bebbington1990). The authors also interestingly reported a gender effect, whereby the disorder was most prevalent in the highly acculturated males but in the least acculturated females. Other studies have reported effects between individual acculturation and area characteristics on psychiatric morbidity (Ghubash et al. Reference Ghubash, Hamdi and Bebbington1994).
In the prospective part of our study we found that two aspects of acculturation were predictive of persistent depression. This finding suggests a degree of specificity in the association between these aspects of acculturation and depressive disorder, as the total acculturation score was not a predictor. Limited use of the English language may predict persistent depressive disorder as it restricts a British Pakistani woman's opportunity to gain social support or become involved in activities outside of the home. This concurs with previous work that indicated that fluency in English was associated with less distress. Regarding Pakistan as home may represent a strong ethnic identity and traditionalism; these dimensions of acculturation could be regarded as a protective cultural factor (Escobar & Vega, Reference Escobar and Vega2000).
Prospective studies in this area of research are very few, but illustrate two different scenarios. Among Mexican American women living in the USA, prenatal depression was more prevalent in those who did not speak English readily but who lived in a household where English was the predominant language. Use of English did not predict post-natal depression; this was predicted by difficulties in close family relationships (Martinez-Schallmoser et al. Reference Martinez-Schallmoser, Telleen and MacMullen2003). In the other prospective study of Chinese Americans, depression was more likely to develop in those exposed to negative life events and social conflicts. Exposure to such stressors was associated with a greater degree of acculturation, which appeared to exacerbate the effect of negative life events (Hwang & Myers, Reference Hwang and Myers2007). Our study is more similar to the former study, as greater acculturation was associated with an increased chance that the depression would resolve. Having marked difficulties in marital and close relationships was also a predictor.
Prospective studies are clearly needed in this area of research, as we found that there was a different pattern of association in the cross-sectional and prospective part of our study. In the cross-sectional part, older age mediated the association between depression and acculturation. In the prospective part this was not so, and there were elements of both less and greater degrees of acculturation associated with persistent depression. Thus both increased fluency of English and perhaps protective cultural factors (seeing one's future in Pakistan) appear to be associated with resolving depression.
There also appears to be a difference according to the measure of depression. A study of Chinese people in Australia found that a greater degree of acculturation was associated with persistent depressive episodes, whereas current distress was associated with less acculturation (Parker et al. Reference Parker, Chan, Tully and Eisenbruch2005). In the cross-sectional part of our study we found that lack of acculturation was associated with depressive disorder, but those with high levels of distress below the depression cut-off (distressed group) showed no acculturation difference from those with neither.
Some strengths of the study
An important strength of this study lies in the use of recognized instruments administered by bilingual researchers who were from the same ethnic, gender and cultural background as the British Pakistani women interviewed. This meant that no participants were excluded on the basis of language, which has been a limitation of many previous studies. The rating of severity of life events and chronic social difficulties was performed with a full knowledge of the cultural context by raters who remained blind as to whether the woman was depressed or not. We had a good response rate at follow-up. We followed the suggestion of previous researchers to identify particular aspects of acculturation and not just use a single construct (Escobar & Vega, Reference Escobar and Vega2000; Hwang & Ting, Reference Hwang and Ting2008).
Some limitations of the study
Our study also had a number of limitations that must be recognized. The response rate of 66% for the initial phase of the study was slightly lower than the usual accepted rate but comparable with other studies in this population (van der Wurff et al. Reference van der Wurff, Beekman, Dijkshoorn, Spijker, Smits, Stek and Verhoeff2004; Weich et al. Reference Weich, Nazroo, Sproston, McManus, Blanchard, Erens, Karlsen, King, Lloyd, Stansfeld and Tyrer2004; Thapa & Hauff Reference Thapa and Hauff2005; Tinghog et al. Reference Tinghog, Hemmingsson and Lundberg2007). The sampling strategy could have missed some of the more acculturated women who had acquired British surnames by marriage, although these numbers if at all would be very small considering the Pakistani Muslim culture and the selection of GP practices with very high numbers of British Pakistanis. We followed up only the distressed and depressed women (response rate 93%), and present here only data for the latter because of small numbers in the former. There were too few onsets of depression in the distressed group to analyse risk factors for onset. However, we have shown elsewhere that while the rate of onset of depression in people of Pakistani origin living in UK is similar to that in white Europeans, it is the rate of persistence that is higher in people of Pakistani origin (Gater et al. Reference Gater, Tomenson, Percival, Chaudhry, Waheed, Dunn, Macfarlane and Creed2008). Therefore our findings regarding acculturation relate to persistent depression, not new onsets. Since it is clear that the process of acculturation differs greatly between immigrant groups and between countries, our results cannot be extrapolated to other immigrant groups or countries (Bhui et al. Reference Bhui, Lawrence, Klineberg, Woodley-Jones, Taylor, Stansfield, Viner and Booy2005).
Conclusion
The implication of this work is that women of Pakistani origin with depression should be encouraged to receive help in the use of English as one part of treatment that can prevent relapse. Our experience of working with such women is that this may be hard to achieve – it may be appropriate for English classes to be made available for them through a local Pakistani community centre or equivalent. In some centres a voluntary mental health service for people of ethnic minorities offers English classes. These initiatives could help bring depressed women of Pakistani origin together, which might offer social support in addition to providing English classes.
Acknowledgements
This study was funded by the UK Medical Research Council. We thank the participants and the GPs who took part in the study, the Health of Pakistanis and White Europeans (HOPE) research team and the Pakistani Community Centre, Longsight, Manchester, UK.
Declaration of Interest
None.
