A striking finding emerging from the burgeoning field of international psychiatric epidemiology is the wide variation in the prevalence rates of mental disorder recorded across countries and regions. Reference Demyttenaere, Bruffaerts, Posada-Villa, Gasquet, Kovess and Lepine1,2 For example, several national surveys have yielded substantially lower rates of common mental disorders in North East and South East Asia compared with English-speaking countries of the West. Reference Demyttenaere, Bruffaerts, Posada-Villa, Gasquet, Kovess and Lepine1,Reference Hwu, Yeh and Chang3–Reference Shen, Zhang, Huang, He, Liu and Cheng5 An important question, therefore, is whether the most commonly used measure, the Composite International Diagnostic Instrument (CIDI), underestimates the prevalence of mental disorder in some Asian cultures. Reference Shen, Zhang, Huang, He, Liu and Cheng5 We report the first epidemiological study comparing an indigenously derived measure with the CIDI among an Asian population, the Vietnamese, living in Vietnam and as immigrants in Australia. We examine whether being resident in a Western country increases the endorsement of CIDI-based mental disorders among Vietnamese people. The data are compared with those obtained from an Australian-born sample.
Method
Sampling and procedure
Mekong Delta survey
The Vietnam-based sample was obtained from a survey in the Mekong Delta region. The study was conducted between November 2004 and March 2005, covering the regions of
\batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{C}{\grave{\mathrm{a}}}\mathrm{n}\) \end{document}
City and 
Giang province.
\batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{C}{\grave{\mathrm{a}}}\mathrm{n}\) \end{document}
City, located 160 km south of Ho Chi Minh City on the banks of the Giang river, is the fifth largest city in Vietnam and the main urban centre of the Mekong Delta region (population of 1 121 141). The province of Giang lies adjacent to
\batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{C}{\grave{\mathrm{a}}}\mathrm{n}\) \end{document}
City and covers a rural population of 772 239.
The survey applied a multistage probabilistic cluster sampling frame with the commune or hamlet (the smallest geographic area for which census information is available) specified as the primary sampling unit. The first stage applied probability proportional to size sampling to identify 15 of a pool of 478 hamlets in Giang province and 16 from a pool of 503 hamlets in
\batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{C}{\grave{\mathrm{a}}}\mathrm{n}\) \end{document}
City. In each hamlet, 100 consecutive households were selected using a random commencement point. Within each household, a single respondent aged 18 years or older was selected using a Kish grid without replacement for non-response. Overall, 3039 people living in 3100 households agreed to participate, forming the Mekong Delta Vietnamese sample (response rate 98%).
Twenty-five local personnel (5 mental health physicians, 16 mental health nurses and 4 general staff) conducted the interviews. All interviewers received 40 h of initial training in the administration of the research instruments, as well as ongoing supervision.
Australian Vietnamese survey
The Australian Vietnamese survey was conducted between June 1999 and May 2000. A full description of survey methods has been provided elsewhere. Reference Steel, Silove, Chey, Bauman, Phan and Phan6 Sampling of private dwellings was carried out across five local government areas in New South Wales, Australia, housing 41 487 people or 75% of the total adult Vietnamese population resident in that state. A probability proportional to size cluster-sampling method was applied to select 44 census collection districts in which all identified households were approached. The occupants were contacted by an introductory letter and this was followed by a face-to-face visit. A Vietnamese-speaking bilingual interviewer then ascertained whether the household was occupied by persons of Vietnamese origin. Of the 6224 private dwellings approached, 1413 housed at least one Vietnamese person. A single respondent aged 18 years or older was randomly selected for interview. Interviews were conducted by trained bilingual Vietnamese-speaking lay workers. The sample included 1161 individuals, a response rate of 82%. Participants completed the interviews in Vietnamese (98.6%) or English (1.4%).
Australian Bureau of Statistics survey
The Australian Bureau of Statistics survey, described elsewhere, 7 was conducted between May 1997 and August 1997, consisting of a nationwide stratified multistage probability sample of 13 624 private dwellings. Interviews were conducted with 10 641 individuals randomly selected from each household (response rate of 78%). Responses of the 7961 participants born in Australia (Australians) were extracted for analysis.
Survey instruments
Diagnostic assessment
All three surveys applied the Composite International Diagnostic Interview (CIDI) 2.0, a lay administered instrument yielding DSM–IV diagnoses. Reference Andrews and Peters8 The measure has been widely employed in international psychiatric epidemiological studies. The CIDI identified the 12-month prevalence of DSM–IV anxiety disorders (panic disorder, agoraphobia, social phobia, generalised anxiety disorder, obsessive–compulsive disorder and post-traumatic stress disorder), mood disorders (depression, dysthymia, mania, hypomania and bipolar disorder) and substance use disorders (alcohol and substance harmful use/misuse and dependence). The ICD–10 9 diagnosis of neurasthenia was included since there is no equivalent category in DSM–IV.
The two Vietnamese studies also included the Phan Vietnamese Psychiatric Scale (PVPS), a questionnaire designed to identify culturally relevant idioms and expressions of psychological distress in that ethnic group. Reference Phan, Steel and Silove10 A full description of the development of the measure has been published previously Reference Phan, Steel and Silove10 and only a summary of the procedure will be provided herein. The measure was developed in sequential stages. First, items describing a wide range of emotional states were derived from a comprehensive review of both the medical and general Vietnamese-language literature. Next, a series of ethnographic studies involving members of the Vietnamese community yielded further items for inclusion in the pool. Items were then subjected to psychometric testing based on responses of Vietnamese samples attending psychiatric and general health clinics. From these analyses, three broad symptom constellations emerged, broadly recognisable as the domains of anxiety, depression and somatisation. The final measure included a 26-item depression scale, a 13-item anxiety scale and a 14-item somatisation scale. Psychometric tests revealed sound internal consistency (r=0.87–0.95) and test–retest reliability (r=0.81–0.89) for the scales. Criterion validity was assessed by comparing the anxiety, depression and somatisation sub-scales with diagnoses made by naturalist healers (κ=0.45–0.71), psychiatrists (κ=0.49–0.62) and by structured diagnostic instruments (κ=0.61–0.69) demonstrating satisfactory diagnostic agreement. To maintain the cultural foundations of the measure, threshold scores were adopted from diagnoses made by naturalist healers. Reference Phan, Steel and Silove10 In the present study, PVPS symptoms were recorded for the preceding 12 months to ensure consistency with the CIDI.
Disability
All surveys included two measures of disability: the Medical Outcomes Study Short Form-12 (MOS–SF–12) generates a physical (PCS) and a mental (MCS) health component functioning score, with lower scores indicating higher levels of disability. Reference Steel, Silove, Chey, Bauman, Phan and Phan6 The PCS includes items assessing physical functioning, including difficulties in role performance attributed to physical symptoms, bodily pain and general health. The MCS includes items assessing vitality, social functioning, role performance difficulties attributed to emotional problems and mental health. The PCS and MCS were categorised according to two levels of disability: none or mild (40 or above) and moderate to severe (below 40). Reference Steel, Silove, Chey, Bauman, Phan and Phan6
The number of disability days was based on two questions 7 assessing the number of days respondents were unable to work or to carry out normal activities, and the number of days respondents had to significantly reduce their activity because of ill health. The combined index was stratified according to three levels: no days of disability; 1–5 days of disability; and 5+ days of disability. Reference Steel, Silove, Chey, Bauman, Phan and Phan6
Service utilisation
Contact with service providers for all health and mental health problems in the previous 12 months was recorded. 7 Service providers included primary care physicians; a mental health practitioner; and, in the Vietnamese samples, a traditional healer (Chinese doctor, acupuncturist, herbalist, folk healer/witchcraft practitioner and fortune teller/cosmologist). Reference Steel, Silove, Chey, Bauman, Phan and Phan6
Translation–back-translation
The CIDI and MOS-SF-12 were translated in Australia using established methods
Reference Steel, Silove, Chey, Bauman, Phan and Phan6,Reference Bracken and Barona11
with minor discrepancies being reconciled by a Vietnamese mental health professional and a panel of seven bilingual healthcare interpreters. Minor modifications relevant to local language usage were made for the Mekong Delta survey by five senior mental health staff from
\batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{C}{\grave{\mathrm{a}}}\mathrm{n}\) \end{document}
health service.
Across all sites, interviewers administered the CIDI and other measures using prompt cards and a computerised data-entry system.
Statistical analysis
All three samples were stratified according to the age and gender distribution of the target population. For the Mekong Delta Vietnamese, the age and gender distribution of the general population was derived from the 1999 Vietnam census of housing and population. For Australian Vietnamese, weighting was based on the New South Wales Vietnamese population derived from the 2001 Australian census. The Australian sample was weighted according to the age and gender distribution of the 1996 Australian population. For the two Vietnamese samples, standard errors for prevalence estimates were calculated using the SAS software package 9.1.3 for Windows, adjusting for clustering effects. For Australians, a jackknife method using 30 replicate weights was applied to calculate the standard errors of the prevalence estimates in accordance with the guidelines provided by the Australian Bureau of Statistics. 7
Demographic characteristics of the three populations surveyed are presented as weighted counts and prevalence estimates (%). Chi-squared tests were used to compare categorical variables. Cohen's kappa (κ) and the area under the receiver operating characteristic curve (AUC) were used to assess the level of diagnostic agreement between the CIDI and PVPS. The κ-statistic ranges from 0.0 to 1.0, providing a measure of agreement corrected for chance. It is sensitive to population prevalence rates, however, even when sensitivity and specificity estimates remain constant across samples. In effect, this means that the κ is likely to be higher in clinic samples than in population-based studies. Reference Kessler, Abelson, Demler, Escobar, Gibbon and Guyer12 The AUC addresses this limitation because it is not sensitive to prevalence rates. Reference Kessler, Abelson, Demler, Escobar, Gibbon and Guyer12,Reference Haro, Arbabzadeh-Bouchez, Brugha, de Girolamo, Guyer and Jin13 Estimates for AUC range from 0.5 to 1.0. A score of 0.9 or greater indicates near perfect agreement; 0.8–0.9, substantial agreement; 0.7–0.8, moderate agreement; 0.6–0.7 fair agreement; and less than 0.6, slight agreement. Reference Haro, Arbabzadeh-Bouchez, Brugha, de Girolamo, Guyer and Jin13,Reference Landis and Koch14
Results
The average period of residency for Australian Vietnamese in Australia was 11.3 years (s.d.=5.9). Table 1 presents weighted age and gender characteristics for the three samples. There were no gender differences (χ2 (2)=1.90, P>0.05) but the Vietnamese samples were younger than Australians (χ2 (10)=405, P<0.0001). More Mekong Delta Vietnamese were married and fewer were divorced (χ2 (6)=327.3, P<0.0001). Australian Vietnamese had higher levels of education compared with Mekong Delta Vietnamese (χ2 (4)=825, P<0.0001). Australians had higher levels of tertiary education than either of the Vietnamese groups. Workforce participation was lower among Mekong Delta Vietnamese (49.9; χ2 (4)=687, P<0.0001).
Weighted socio-demographic characteristics of Mekong Delta Vietnamese (n=3039), Australian Vietnamese (n=1161) and Australians (n=7961)a
| Mekong Delta Vietnamese | Australian Vietnamese | Australian | ||||
|---|---|---|---|---|---|---|
| n | Estimated prevalence, % (s.e.) | n | Estimated prevalence, % (s.e.) | n | Estimated prevalence, % (s.e.) | |
| Gender | ||||||
| Male | 1431 | 47.1 (1.2) | 572 | 49.3 (2.0) | 3839 | 48.2 (0.6) |
| Female | 1608 | 52.9 (1.2) | 589 | 50.7 (2.0) | 4122 | 51.8 (0.6) |
| Age, years | ||||||
| 18-24 | 757 | 24.9 (1.1) | 192 | 16.5 (2.3) | 1193 | 15.0 (0.7) |
| 25-34 | 849 | 27.9 (1.3) | 326 | 28.1 (2.1) | 1775 | 22.3 (0.8) |
| 35-44 | 616 | 20.3 (0.6) | 346 | 29.8 (1.8) | 1654 | 20.8 (0.8) |
| 45-54 | 355 | 11.7 (0.7) | 157 | 13.5 (1.2) | 1268 | 15.9 (0.7) |
| 55-64 | 187 | 6.2 (0.6) | 52 | 4.5 (0.6) | 857 | 10.8 (0.6) |
| 65+ | 274 | 9.0 (0.7) | 88 | 7.6 (1.0) | 1215 | 15.3 (0.7) |
| Marital status | ||||||
| Never married | 818 | 26.9 (1.1) | 313 | 26.9 (2.5) | 1834 | 23.0 (0.9) |
| Married | 2153 | 70.8 (1.1) | 739 | 63.7 (2.4) | 5038 | 63.3 (1.0) |
| Separated/divorced | 19 | 0.6 (0.1) | 74 | 6.4 (0.8) | 619 | 7.8 (0.6) |
| Widowed | 49 | 1.6 (0.3) | 33 | 2.9 (0.6) | 471 | 5.9 (0.5) |
| Highest qualification/trade | ||||||
| None/at school/< 12 years | 1900 | 61.5 (2.8) | 689 | 59.4 (2.0) | 2842 | 35.7 (0.5) |
| Secondary or vocational training | 739 | 23.9 (1.5) | 358 | 30.8 (2.2) | 4008 | 50.3 (0.6) |
| Tertiary | 380 | 12.3 (2.2) | 114 | 9.8 (1.5) | 1111 | 14.0 (0.4) |
| Employment status | ||||||
| Employed | 1201 | 39.5 (1.7) | 495 | 42.7 (2.0) | 5135 | 64.5 (0.6) |
| Unpaid labour | 1522 | 50.1 (1.9) | 574 | 49.5 (1.9) | 2503 | 31.4 (0.6) |
| Unemployed/not in workforce | 315 | 10.4 (1.0) | 91 | 7.9 (1.0) | 323 | 4.1 (0.3) |
| Housing | ||||||
| Owner | 3001 | 98.8 (0.3) | 375 | 32.3 (2.6) | 5037 | 63.3 (0.6) |
| Renting privately/from government | 21 | 0.7 (0.2) | 644 | 55.5 (3.1) | 1968 | 24.7 (0.6) |
| Other | 16 | 0.5 (0.2) | 141 | 12.2 (1.7) | 956 | 12.0 (0.4) |
Prevalence of CIDI mental disorders
Table 2 presents the weighted prevalence estimates with design corrected standard errors for CIDI mood, anxiety and substance use disorders as well as ICD–10 neurasthenia. The prevalence of all CIDI disorders for the Mekong Delta Vietnamese was 1.9% compared with 6.7% for Australian Vietnamese and 17.1% for Australians.
12-month prevalence of Composite International Diagnostic Interview (CIDI) and Phan Vietnamese Psychiatric Scale (PVPS) mental disorders among Mekong Delta Vietnamese (n=3039), Australian Vietnamese (n=1161) and Australians (n=7961)a
| Mekong Delta Vietnamese | Australian Vietnamese | Australian | ||||
|---|---|---|---|---|---|---|
| n | Estimated prevalence, % (s.e.) | n | Estimated prevalence, % (s.e.) | n | Estimated prevalence, % (s.e.) | |
| DSM–IV disorders | ||||||
| Total DSM–IV disorders | 58 | 1.9 (0.3) | 77 | 6.7 (1.0) | 1365 | 17.1 (0.4) |
| DSM–IV anxiety disorders | 12 | 0.4 (0.1) | 35 | 3.1 (0.8) | 467 | 5.9 (0.3) |
| Panic/agoraphobia | 5 | 0.1 (0.1) | 6 | 0.6 (0.4) | 129 | 1.6 (0.1) |
| Social phobia | 1 | 0.0 (0.0) | 4 | 0.3 (0.3) | 123 | 1.5 (0.1) |
| Generalised anxiety disorder | 3 | 0.1 (0.1) | 8 | 0.7 (0.2) | 217 | 2.7 (0.2) |
| Obsessive–compulsive disorder | 3 | 0.1 (0.1) | 6 | 0.5 (0.2) | 56 | 0.7 (0.1) |
| Post-traumatic stress disorder | 2 | 0.1 (0.0) | 17 | 1.5 (0.4) | 114 | 1.4 (0.1) |
| DSM–IV mood disorders | 14 | 0.4 (0.2) | 30 | 2.6 (0.5) | 563 | 7.1 (0.3) |
| Major depression | 9 | 0.3 (0.2) | 21 | 1.8 (0.4) | 533 | 6.7 (0.3) |
| Dysthymia | 3 | 0.1 (0.1) | 11 | 1.0 (0.3) | 38 | 0.5 (0.1) |
| DSM–IV substance disorders | 34 | 1.1 (0.3) | 19 | 1.6 (0.6) | 699 | 8.8 (0.3) |
| Alcohol use disorders | 33 | 1.1 (0.3) | 13 | 1.1 (0.5) | 533 | 6.7 (0.3) |
| Drug use disorders | 1 | 0.0 (0.0) | 6 | 0.5 (0.3) | 258 | 3.2 (0.2) |
| ICD–10 neurasthenia | 3 | 0.1 (0.1) | 12 | 1.0 (0.3) | 134 | 1.7 (0.1) |
| PVPS disorders | ||||||
| Total PVPS disorders | 224 | 7.4 (0.8) | 98 | 8.4 (0.7) | ||
| Depression | 37 | 1.2 (0.7) | 27 | 2.4 (0.5) | ||
| Anxiety | 122 | 4.0 (0.2) | 42 | 3.6 (0.5) | ||
| Somatisation | 162 | 5.3 (0.6) | 82 | 7.1 (0.6) | ||
The pattern was consistent when anxiety and depressive disorders were examined separately. The rates of substance use disorders were similarly low for the two Vietnamese samples (1.1% for Mekong Delta Vietnamese and 1.6% for Australian Vietnamese), about a fifth of the prevalence for Australians (8.8%). Both Vietnamese samples had low rates of neurasthenia compared with Australians.
Prevalence of PVPS mental disorders
Table 2 presents prevalence rates yielded by the PVPS. Mekong Delta Vietnamese (7.4%) and Australian Vietnamese (8.4%) had similar overall prevalence rates, with the same pattern emerging for the individual domains of anxiety, depression and somatisation. Somatisation was the most commonly assigned diagnosis in both samples, followed by anxiety and then depression.
Comparing CIDI and PVPS cases
Table 3 compares data from the CIDI and the PVPS for the Vietnamese groups. The aggregated prevalence of CIDI and PVPS disorders was 8.8% for the Mekong Delta Vietnamese, with the PVPS identifying 84.1% and the CIDI 21.9% of all participants identified by both systems. The PVPS identified 208 unique cases, the CIDI 42 unique cases and both systems 16 cases, indicating minimum diagnostic overlap (AUC=0.59). For Australian Vietnamese, the two measures yielded a combined prevalence of 11.7%. The PVPS identified 72.2% of all cases and the CIDI 57%. The PVPS identified 58 unique cases, the CIDI 38 cases, and both systems 40 cases, representing a moderate degree of diagnostic agreement (AUC=0.77).
Diagnostic agreement between Composite International Diagnostic Interview (CIDI) and Phan Vietnamese Psychiatric Scale (PVPS) mental disorders among Mekong Delta Vietnamese (n=3039) and Australian Vietnamese (n=1161)
| Mekong Delta Vietnamese | Australian Vietnamese | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| CIDI & PVPS combined prevalence, %a | CIDI coverage of cases, % | PVPS coverage of cases, % | Kappa | AUC (95% CI) | CIDI & PVPS combined prevalence, %a | CIDI coverage of cases, % | PVPS coverage of cases, % | Kappa | AUC (95% CI) | |
| Agreement within diagnostic subdomains | ||||||||||
| PVPS total disorders & any CIDI disorder | 8.8 | 21.9 | 84.1 | 0.086 | 0.59 (0.55-0.63) | 11.7 | 47.0 | 72.2 | 0.408 | 0.77 (0.72-0.82) |
| PVPS anxiety & any DSM–IV anxiety disorder | 4.2 | 9.1 | 94.4 | 0.061 | 0.64 (0.59-0.70) | 5.3 | 57.4 | 68.5 | 0.391 | 0.71 (0.64-0.79) |
| PVPS depression & any DSM–IV mood disorder | 1.5 | 28.9 | 79.4 | 0.148 | 0.85 (0.56-0.73) | 3.9 | 65.9 | 60.5 | 0.404 | 0.73 (0.64-0.81) |
| PVPS somatisation & ICD–10 neurasthenia | 5.4 | 2.0 | 98.5 | 0.009 | 0.64 (0.59-0.68) | 7.4 | 13.9 | 95.1 | 0.150 | 0.83 (0.78-0.89) |
| Agreement across diagnostic subdomains | ||||||||||
| PVPS anxiety & any CIDI disorder | – | 34.7 | 72.6 | 0.113 | 0.53 (0.49-0.58) | – | 80.7 | 44.2 | 0.369 | 0.74 (0.67-0.79) |
| PVPS depression & any CIDI disorder | – | 64.8 | 41.4 | 0.102 | 0.56 (0.49-0.62) | – | 92.0 | 32.6 | 0.374 | 0.83 (0.78-0.89) |
| PVPS somatisation & any CIDI disorder | – | 28.5 | 79.1 | 0.116 | 0.53 (0.49-0.57) | – | 61.9 | 65.8 | 0.392 | 0.68 (0.83-0.74) |
Table 3 indicates that in both Vietnamese populations, depression was the domain with the greatest level of concordance across the diagnostic systems (Mekong Delta Vietnamese, 64.8%, Australian Vietnamese, 92%) followed by PVPS anxiety (Mekong Delta Vietnamese, 34.7%; Australian Vietnamese, 80.7%). The PVPS somatisation scale showed a low level of overlap with CIDI neurasthenia (Mekong Delta Vietnamese, 28.5%; Australian Vietnamese, 61.9%). Virtually all those with neurasthenia also had PVPS somatisation, but a larger number of Vietnamese fell into the latter category alone.
Functional impairment
In the first set of analyses, individuals identified by either diagnostic measure were examined. In all three samples, being assigned a mental disorder was associated with substantial functional impairment (Table 4). A greater percentage of participants from the Mekong Delta Vietnamese (48.3%) and Australian Vietnamese (48.3%) groups reported five or more disability days compared with the Australian group (28.1%) (χ2 (2)=23.00, P≤0.0001). Findings from the MOS–SF–12 PCS scale indicated that Australian Vietnamese (52.2%) with a mental disorder reported the greatest level of impairment in physical functioning, followed by Mekong Delta Vietnamese (34.2%) and Australians (22.6%) (χ2 (2)=64.95, P≤0.001). Australian Vietnamese (37.9%) with a CIDI or PVPS disorder and Australians (35.2%) with a CIDI disorder reported similar levels of impairment in mental health functioning, whereas Mekong Delta Vietnamese reported lower levels of dysfunction on that index (20.8%) (χ2 (2)=22.52, P≤0.001).
Disability in the 4 weeks prior to interview for Composite International Diagnostic Interview (CIDI) and Phan Vietnamese Psychiatric Scale (PVPS) mental disorders among Mekong Delta Vietnamese (n=3039), Australian Vietnamese (n=1161) and Australians (n=7961)
| CIDI and PVPS mental disorders | No mental illness | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Mekong Delta Vietnamesea (n=267) | Australian Vietnamesea (n=l36) | Australianb (n=1365) | Mekong Delta Vietnamese (n=2772) | Australian Vietnamese (n=1025) | Australian (n=6596) | |||||||
| n | Estimated prevalence, % (s.e.) | n | Estimated prevalence, % (s.e.) | n | Estimated prevalence, % (s.e.) | n | Estimated prevalence, % (s.e.) | n | Estimated prevalence, % (s.e.) | n | Estimated prevalence, % (s.e.) | |
| Level of disability | ||||||||||||
| Mental health | ||||||||||||
| Mild to none (MCS ≥40) | 211 | 79.2 (3.5) | 84 | 62.1 (4.0) | 884 | 64.8 (1.3) | 2679 | 96.6 (0.6) | 957 | 93.4 (0.9) | 6161 | 93.4 (0.3) |
| Moderate to severe (MCS <40) | 55 | 20.8 (3.5) | 51 | 37.9 (4.0) | 480 | 35.2 (1.3) | 93 | 3.4 (0.6) | 68 | 6.6 (0.9) | 435 | 6.6 (0.3) |
| Physical health | ||||||||||||
| Mild to none (PCS ≥40 | 175 | 65.8 (3.1) | 65 | 47.8 (4.1) | 1057 | 77.5 (1.1) | 2482 | 89.5 (0.6) | 876 | 85.5 (1.7) | 5524 | 83.7 (0.5) |
| Moderate to severe (PCS <40) | 91 | 34.2 (3.1) | 71 | 52.2 (4.1) | 308 | 22.6 (1.1) | 290 | 10.5 (0.6) | 149 | 14.7 (1.7) | 1072 | 16.3 (0.5) |
| Total days of disability | ||||||||||||
| No days | 121 | 45.3 (3.0) | 43 | 31.7 (5.3) | 669 | 49.0 (1.4) | 2082 | 75.0 (1.6) | 814 | 79.4 (2.2) | 4645 | 70.4 (0.6) |
| 1-5 days | 17 | 6.4 (1.8) | 27 | 20.0 (3.2) | 313 | 23.0 (1.1) | 198 | 7.1 (1.0) | 123 | 12.0 (1.7) | 1163 | 17.6 (0.5) |
| >5 days | 129 | 48.3 (3.2) | 65 | 48.3 (4.3) | 383 | 28.1 (1.2) | 492 | 17.8 (1.2) | 88 | 8.6 (1.4) | 788 | 12.0 (0.4) |
Levels of functional impairment were then compared across the two diagnostic systems. For the Australian Vietnamese, those identified by the CIDI and PVPS did not differ across any of the three indices of functional impairment (MOS–SF–12 MCS: χ2 (1)=0.059, P=0.808; MOS–SF–12 PCS: χ2 (1)=1.552, P=0.213; disability days: χ2 (2)=5.02, P=0.081). For the Mekong Delta Vietnamese, the overall level of disability associated with the two diagnostic systems was similar (five or more disability days: CIDI 36.4% v. PVPS 51.1%, χ2 (1)=6.25, P=0.45; MOS–SF–12 MCS: CIDI 18% v. PVPS 22.2%, χ2 (1)=6.25, P=0.45). Nevertheless, for the MOS–SF–12, people identified by the PVPS reported greater impairment in physical functioning compared with the CIDI (MOS–SF–12 PCS: CIDI 20.3% v. PVPS 38%, χ2 (1)=6.25, P=0.012).
Health service utilisation
Table 5 shows that 21.5% of Mekong Delta Vietnamese with a mental disorder had a consultation with a primary care physician in the previous 12 months compared with over 83.5% of Australian Vietnamese and Australians. Primary care consultations specific to a mental health problem were low (1.9%) for Mekong Delta Vietnamese compared with Australian Vietnamese (24.4%) and Australians (30.9%). Australians with a disorder were more likely to consult a mental health specialist than affected Australian Vietnamese (24% v. 10.2%). Vietnamese with mental disorders rarely consulted traditional healers, with Mekong Delta Vietnamese having notably low rates.
Health service utilisation in the 12 months prior to interview for Composite International Diagnostic Interview (CIDI) and Phan Vietnamese Psychiatric Scale (PVPS) mental disorders among Mekong Delta Vietnamese (n=3039), Australian Vietnamese (n=1161) and Australians (n=7961)
| CIDI and PVPS mental disorders | No mental illness | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Mekong Delta Vietnamesea (n=267) | Australian Vietnamesea (n=l36) | Australianb (n=1365) | Mekong Delta Vietnamese (n=2772) | Australian Vietnamese (n=1025) | Australian (n=6635) | |||||||
| n | Estimated prevalence, % (s.e.) | n | Estimated prevalence, % (s.e.) | n | Estimated prevalence, % (s.e.) | n | Estimated prevalence, % (s.e.) | n | Estimated prevalence, % (s.e.) | n | Estimated prevalence, % (s.e.) | |
| Consultation with primary care physicians | ||||||||||||
| General consults | 57 | 21.5 (3.4) | 119 | 87.5 (2.8) | 1140 | 83.5 (1.0) | 451 | 16.3 (1.0) | 763 | 74.4 (2.5) | 5366 | 81.4 (0.5) |
| Mental health consults | 5 | 1.9 (1.0) | 33 | 24.4 (5.4) | 422 | 30.9 (1.3) | 12 | 0.4 (0.2) | 33 | 3.2 (0.8) | 264 | 4.0 (0.2) |
| Consultation with mental health professional | 5 | 2.0 (0.7) | 14 | 10.2 (3.6) | 328 | 24.0 (1.2) | 6 | 0.2 (0.1) | 2 | 0.2 (0.1) | 207 | 3.1 (0.2) |
| Consultation with traditional healers | ||||||||||||
| General consults | 9 | 3.3 (1.0) | 35 | 25.7 (4.0) | 54 | 2.0 (0.3) | 164 | 16.0 (1.7) | ||||
| Mental health consults | 1 | 0.5 (0.4) | 8 | 5.7 (1.8) | - | - - | 4 | 0.4 (0.3) | ||||
Discussion
Our study revealed a low prevalence of common CIDI-defined mental disorders for Vietnamese living in Australia and in Vietnam compared with an Australian-born sample and, more generally, with other studies undertaken in high-income countries of the West. Reference Demyttenaere, Bruffaerts, Posada-Villa, Gasquet, Kovess and Lepine1,2 The findings, therefore, add to the already substantial evidence indicating low rates of DSM-derived disorders among populations from East Asia. 2,Reference Shen, Zhang, Huang, He, Liu and Cheng5,Reference Kawakami, Shimizu, Haratani, Iwata and Kitamura15,Reference Lee, Tsang, Zhang, Huang, He and Liu16 That trend does not appear to be restricted to studies using the CIDI, with investigations applying other Western-based diagnostic instruments also yielding lower rates of mental disorders among North and South East Asian communities. Reference Lee, Kwak, Yamamoto, Rhee, Kim and Han4,Reference Compton, Helzer, Hwu, Yeh, McEvoy and Tipp17,Reference Parker, Chan and Hadzi-Pavlovic18
The key question addressed by the present study was whether the addition of a culturally derived diagnostic measure would yield a higher overall prevalence of mental disorder among the Vietnamese. The PVPS added a substantial number of new cases in relation to both Vietnamese populations. Among Mekong Delta Vietnamese the rates increased fourfold. One possibility was that measures such as the PVPS identify individuals with less severe mental disturbances compared with a structured diagnostic instrument. The disability data argue against that conclusion, with those identified by the PVPS being at least as disabled as those identified by the CIDI.
The PVPS somatisation scale accounted for a large percentage of additional cases of mental disorder among Vietnamese in both settings. The importance of the somatic focus among Vietnamese was emphasised further by the higher endorsement of items on the PCS scale of the MOS–SF–12 by Mekong Delta Vietnamese with a mental disorder. This adds to the growing body of literature documenting a preference for reporting distress in the somatic idiom among North and South East Asian populations. Reference Shen, Zhang, Huang, He, Liu and Cheng5,Reference Ryder, Yang, Zhu, Yao, Yi and Heine19,Reference Lin, Ihle and Tazuma20 A key implication is that the pre-eminence given to psychological rather than somatic symptoms in the hierarchical organisation of DSM–IV and ICD–10 may result in the under-enumeration of mental disorders in East Asian populations. Reference Shen, Zhang, Huang, He, Liu and Cheng5
The ICD–10-defined category of neurasthenia identified very few participants among the Vietnamese, lower than among the Australian-born. It is noteworthy that in its development, the ICD–10 category was modelled closely on the Western construct of chronic fatigue, Reference Paralikar, Sarmukaddam, Agashe and Weiss21 whereas the items included in the PVPS scale have a greater similarity to those defining the category of neurasthenia in the Chinese classification system. Reference Lee22
The combined prevalence of CIDI- and PVPS-identified individuals in Vietnam (8.8%) and among Vietnamese immigrants (11.7%) was lower than that yielded by the CIDI alone among Australians (17.1%). These findings suggest two broad explanations. First, there may be genuine differences in the vulnerability to common mental disorders across ethnic groups. Alternatively, there may be a threshold effect, with Vietnamese having to experience a greater level of disability than individuals from Western backgrounds before they endorse psychiatric symptoms. Such a tendency may be accounted for by a greater degree of cultural stoicism (suffering without complaint) combined with stigma or feelings of shame associated with revealing mental symptoms in Asian cultures. Reference Compton, Helzer, Hwu, Yeh, McEvoy and Tipp17,Reference Ryder, Yang, Zhu, Yao, Yi and Heine19,Reference Beiser and Fleming23 Inconsistent findings have emerged from the international literature in relation to these issues. Simon et al recorded higher levels of disability associated with major depressive disorder in primary care settings for those countries with low prevalence rates. Reference Simon, Goldberg, Von Korff and Ustun24 Patients from North East Asia in particular had both low prevalence rates and high levels of associated disability. In contrast, the World Mental Health Survey Reference Demyttenaere, Bruffaerts, Posada-Villa, Gasquet, Kovess and Lepine1 undertaken among general population samples failed to identify an association between lower prevalence rates and greater disability associated with mental disorder across countries.
The CIDI data obtained from Australian Vietnamese suggest that cultural adaptation to Western society may influence the affinity of immigrants for endorsing Western-derived symptoms on instruments such as the CIDI. The rates yielded by the PVPS were similar for the two Vietnamese samples, suggesting that immigrants also retain their culturally based mode of expressing distress, even after a prolonged period of residency in a Western country. The AUC data support these inferences by showing greater overlap of the CIDI and PVPS among Australian Vietnamese than Mekong Delta Vietnamese.
Australians and Australian Vietnamese with a mental disorder exhibited much higher consultation rates with primary healthcare physicians than Mekong Delta Vietnamese. The difference in the availability and accessibility of services across the two countries undoubtedly was a major contributor. Reference Wang, Aguilar-Gaxiola, Alonso, Angermeyer, Borges and Bromet25 The tendency for Australian Vietnamese to restrict their help-seeking to primary care physicians may reflect the stigma associated with mental disorder in that culture and/or the limited availability of mental health professionals from a Vietnamese background residing in Australia.
Limitations
Several limitations of the study need to be acknowledged. Mekong Delta Vietnamese had a higher response rate compared with the other two samples. Evidence is mixed as to whether people with psychiatric disorders are less likely to participate in surveys. Reference Kessler, Chiu, Demler, Merikangas, Walters and Merikangas26 If that trend pertained in the present study, it would have the effect of attenuating the differences found between the two Australian-based samples and the Mekong Delta Vietnamese group.
There was a difference in the timing of the investigations, with Mekong Delta Vietnamese being studied more recently. If a process of secular shift exerted any influence, it would have lessened differences between Australian Vietnamese and the Mekong Delta Vietnamese because of the recent acceleration of Westernisation in the home country. In keeping with other studies of this type, none of the surveys included the itinerant, those who were homeless or those living in institutions. Low-prevalence disorders such as psychosis and organic disorders were not assessed, nor were the impulse-control disorders included in some recent surveys. Reference Demyttenaere, Bruffaerts, Posada-Villa, Gasquet, Kovess and Lepine1
The Australian sample was derived from a national survey. To preserve anonymity of participants, the Australian Bureau of Statistics did not allow extraction of Australians from the specific localities where Australian Vietnamese lived. For logistic reasons, we sampled the Australian Vietnamese in the state of New South Wales, which has the largest concentration of Vietnamese immigrants in the country. We were unable to sample very low-density suburbs, possibly excluding more acculturated Vietnamese. The Mekong Delta was chosen because most Vietnamese in Australia are from the south of Vietnam and they originated from both urban and rural areas. Nevertheless, at the time of the study, the lifestyle in the delta may have been more traditional than in the major cities such as Ho Chi Minh where the process of modernisation has gained more ground in recent times.
The level of diagnostic agreement (as indicated by the kappa statistic in Table 4) between the PVPS and the CIDI among Australian Vietnamese was somewhat lower than recorded in the development of the PVPS among a sample of Vietnamese primary care and mental health patients. Reference Phan, Steel and Silove10 The findings are consistent with the general tendency for concordance estimates to be lower in general population samples than in clinical populations. Reference Andrews and Peters8 We note that the level of agreement demonstrated in the present study are comparable with concordance estimates for related diagnostic instruments obtained in other population studies. Reference Andrews and Peters8,Reference Kessler, Wittchen, Abelson, McGonagle, Schwartz and Kendler27 Our study did not undertake a clinical recalibration of the CIDI with other DSM–IV-based diagnostic instruments. Reference Haro, Arbabzadeh-Bouchez, Brugha, de Girolamo, Guyer and Jin13,Reference Lee, Tsang, Zhang, Huang, He and Liu16 However, studies undertaken among East Asian populations, including neighbouring China, have indicated satisfactory concordance between the CIDI and measures such as the Structured Clinical Interview for DSM–IV in settings where there are similarly low prevalence estimates. Reference Lee, Tsang, Zhang, Huang, He and Liu16 A consistent pattern of low prevalence rates of mental disorder among Vietnamese populations has also been recorded using a range of other measures apart from the CIDI. Reference Beiser and Hou29,Reference Hinton, Chen, Du, Tran, Lu and Miranda30
The cut-off thresholds applied to the sub-domains of the PVPS were derived from naturalist healer diagnoses in primary care and mental health clinical samples and may not generalise to a population setting. Reference Van Ommeren28 The disability data indicated, however, that people identified by the PVPS and CIDI exhibited the same level of impairment, suggesting that the thresholds applied did not result in the indigenous PVPS detecting less severe cases.
In summary, our findings throw potential light on the large variation in rates of common mental disorders that have emerged from modern epidemiological studies undertaken across nations and regions. Several factors may be relevant in explaining this pattern. Cultural variation in the disability threshold for reporting symptoms may be relevant but results are inconsistent. Reference Demyttenaere, Bruffaerts, Posada-Villa, Gasquet, Kovess and Lepine1,Reference Simon, Goldberg, Von Korff and Ustun24 Culture-specific ways of expressing mental symptoms appear to be of great importance, with the present study indicating that the addition of an indigenous measure to the CIDI identified many more cases of equal disability among Vietnamese in two settings. The data also suggest that immigration and acculturation to Western environments may alter patterns of symptom endorsement. Vietnamese immigrants retained their base rate of indigenously defined disorders but exhibited higher rates of Western-derived diagnoses. It seems likely that by becoming more familiar with a Western culture, ethnic groups such as the Vietnamese develop a greater affinity for the host society's idiom for expressing mental distress. As a consequence, responses to Western and indigenous measures are likely to show a greater overlap among immigrants. The rapidity of cultural change around the world suggests that the balance between indigenous and Western modes of expressing psychological distress may evolve in a dynamic manner, making it vital to monitor both domains of symptom expression in transcultural settings over time. Reference Kirmayer31
The outcomes of the research suggest that it is feasible to integrate universalistic (etic) and culturally specific (emic) approaches Reference Van Ommeren28,Reference Kirmayer31,Reference Patel32 in studying the mental health of ethnic groups such as the Vietnamese and, by extension, other populations from non-Western traditions. The data demonstrate that a sole reliance on diagnostic systems developed in the West runs the risk of underestimating mental health needs in regions such as South East Asia. Assuming that the rates of mental disorder are low in these settings may inadvertently retard the development of mental health services in the very settings where, as shown by the present study, the majority of people with mental disorders do not have access to appropriate care. Reference Saxena, Thornicroft, Knapp and Whiteford33
eLetters
No eLetters have been published for this article.