Prevalence, nature, severity and correlates of postpartum depressive symptoms in Vietnam
Dr J. Fisher, Key Centre for Women's Health in Society, WHO Collaborating Centre in Women's Health, School of Population Health, University of Melbourne, Victoria, 3010 Australia.
Objective To examine depressive symptomatology in women after childbirth in Ho Chi Minh City, Vietnam.
Design A cross sectional survey.
Setting Hung Vuong Obstetrics and Gynaecology Hospital and the Maternal, Child Health and Family Planning Center of Ho Chi Minh City, Vietnam.
Population Mothers of infants aged ±six weeks attending well-baby clinics.
Method Participants were recruited consecutively in the postnatal wards and invited to take part in the study at the first clinic visit. Individual structured interviews about health and social circumstances, including the Edinburgh Postnatal Depression Scale (EPDS) were administered during clinic visits. The interview schedule was translated into Vietnamese, back translated for verification and piloted. Interviewers were specifically trained members of staff of the two centres.
Main outcome measures EPDS scores and responses to structured questions about specific and non-specific symptoms.
Results Of 506 women who participated, 166 (33%) had EPDS scores in the clinical range of >12 and 99 (19%) acknowledged suicidal ideation. In a forward stepwise logistic regression analysis, 77% of cases with EPDS scores >12 were correctly classified in a model which included unwelcome pregnancy, lack of a permanent job, <30 days complete rest after childbirth, an unsettled baby, not being given special foods, avoiding proscribed foods and being unable to confide in their husbands.
Conclusion Depressive symptomatology is more prevalent among parturient women in Ho Chi Minh City, Vietnam than reported rates in developed countries and is at present unrecognised.
Provide feedback or get help You are viewing our new enhanced HTML article.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
There is substantial evidence from the industrialised world that women are at heightened risk of developing clinically significant psychological distress in the year after giving birth.1–3 Up to 50% of mothers with young infants experience some depressive symptoms and 10% to 15% a major depressive episode, which in 3–5% is of at least moderate severity.1–4 Unrecognised and untreated maternal depression heightens risk of prolonged psychiatric illness, and is associated with impaired relationships and developmental delays and behaviour problems in offspring.2,5,6 In contrast, there is relatively little systematic evidence about psychological functioning in parturient women living in resource poor developing countries,7,8 but there is emerging evidence that there are high rates of depression.9,10 Although there is continuing debate about the aetiological contribution of biological factors, it is generally agreed that psychosocial factors are central to the development of depression after childbirth.1,2,11
The World Health Organisation (WHO) estimates that depressive disorders will be the second leading cause of global disease burden by 2020.12 Rates of depressive illness in women of reproductive age are at least twice those observed in men. The poor face a greater risk of mental illness because they have greater exposure to adverse life experiences and chronic stressors, but are less likely to have access to affordable mental health care or to live in countries with adequate mental health services.12 A recent WHO review concludes that the sex-linked discrepancy in prevalence reflects women's lower social position and greater likelihood of living in poverty and experiencing chronic adversity.13
Vietnam is the second most densely populated country in South East Asia, with a population of approximately 77 million people, 76% of whom live in rural areas. In the last century, Vietnam experienced three major wars, segregation from the non-communist world and more recently, the reforms of a free market economy, globalisation and structural adjustment programs. The Human Development Index (UNDP, 2001) is a composite score including average life expectancy, years of schooling and per capita income on which Vietnam was rated 108 out of 175 countries in 2001. The average annual per capita income in 2000 was US$1669. There are disparities in relative poverty levels between those living in rural and urban settings, but it is estimated that 37% of the population live in absolute poverty with an income less than US$1 per day.14
Two studies have examined the mental health of parturient women who were born in Vietnam and have immigrated to Australia. Rates of clinically significant depressive symptomatology are similar to those observed in locally born cohorts.15,16
We could find no published prevalence studies of depressive symptomatology in parturient women in Vietnam. Local clinicians working in obstetric and maternal and child health settings were of the opinion that postpartum distress did not exist because mothers enjoyed an honoured status and were given increased supportive care for a month post-delivery. Ethnographic studies have documented traditional practices and argued that observation of culturally prescribed rituals, mandated rest, increased practical assistance and social recognition of mothers of newborns are protective of women's health and wellbeing in resource poor settings, including in Southeast Asia.17–19 These authors argue further that women are placed at heightened risk of compromised health after immigration because they might be less able to observe culturally important practices.20,21 Although not in a resource poor setting, some evidence to support the protective benefits of ritualised postpartum care has been found in a comprehensive systematic study of parturient Chinese women living in Hong Kong. They were at higher risk of postpartum depression if they had not been able to observe peiyue or culturally prescribed customs and if they had a critical mother-in-law.22,23
Systematic national perinatal mortality data are unavailable in Vietnam. Nevertheless, in a detailed classification of cause of 2882 maternal deaths in pregnancy or up to 42 days postpartum in three provinces in Vietnam (1994–1995), the leading cause (29%) was external events including accidents, murder and suicide.24 Overall, 14% of these maternal deaths were by suicide. In contrast in the industrialised world, suicide in women after childbirth appears to be extremely rare (1–9/105) and is usually associated with severe mental illness.25,26 Responsibility for a dependent infant has been found to be protective against suicide, but rates are higher among very young women and in those in whom the pregnancy is unwanted.25,26 Confining maternal mortality assessments to the first six weeks postpartum probably leads to under-estimations of maternal mortality from suicide which may occur much later in the postpartum period.27 The British Confidential Enquiry into Maternal Deaths found that maternal deaths from psychological causes, most usually suicide, were at least as prevalent as deaths from hypertensive disorders of pregnancy when data collection was extended to 12 months postpartum.28 These were not only associated with severe mental illness, but were also related to domestic violence and the complications of substance abuse.
There is a debate about whether depressive symptomatology is expressed in universal ways or is culturally determined.29,30 It is suggested that in cultures in which discussion of emotions is proscribed or in which distress is associated with shame or stigma, it may manifest as non-specific somatic symptoms.31 There is related debate about appropriate methods of measurement. However, if the complexities of translation, literacy levels and familiarity with test taking are considered, it appears that the Edinburgh Postnatal Depression Scale (EPDS)32 may be used with caution cross-culturally as a screening instrument for emotional distress warranting additional professional assistance.30,33,34 It is also argued that there are benefits in transcultural settings to assessing symptoms rather than attempting to form diagnoses, because these are a more accurate reflection of individual experience and reduce the likelihood of over- or under-estimates of prevalence of a diagnostic entity that may not be culturally meaningful.35
This project had two aims. First, to examine the incidence, nature, severity and correlates of depressive symptomatology in a group of newly delivered women in Ho Chi Minh City, Vietnam. Second, to ascertain whether self-reports of non-specific and specific somatic symptoms are evident or appear to be associated with psychological distress in this population.
This was a collaborative investigation between the Key Center for Women's Health in Society, a WHO Collaborating Center in Women's Health at the University of Melbourne, Australia; the Hung Vuong Obstetrics and Gynaecology Hospital (a tertiary facility in which 16,000 women including those referred at high risk, give birth annually) and the Maternal and Child Health and Family Planning Center of Ho Chi Minh City, which provides limited birthing facilities for women with low risk pregnancies and has no provision for operative interventions. Its central service is the provision of child health clinics to monitor infant growth and development and immunisation. Most women in Ho Chi Minh City (92%) give birth in medical settings and attendance rates by mothers at infant health clinics are high.36
There were no existing studies in the country from which an estimate of prevalence of postpartum depression could be made and a sample size calculated. In the opinion of Vietnamese clinicians consulted in the development of this project, postpartum mental disorders were very unlikely to be observed (estimated at <1% of the parturient population) and therefore a large sample was necessary in order to detect any cases of probable depression. Participants in this project were women returning to the Hospital or attending the Centre for infant health clinics and/or for medical review six to eight weeks after giving birth.
Ethics approval for the study was obtained from the University of Melbourne's Human Research Ethics Committee and each study site.
A structured interview assessed marital, educational and occupational status, parity, childbirth experiences, observation of traditional post-childbirth practices and practical and emotional support. It also included two mechanisms for assessing psychological distress. The EPDS is a widely used, valid and reliable 10-item self-report screening instrument for depressive symptomatology in English-speaking populations in the postpartum year.35 It does not provide a diagnosis of a major depressive episode and may not predict major depression as accurately as the clinical assessment of a trained health professional.37 However, it has high specificity and sensitivity to depressive symptomatology. It has been widely translated into other languages and is the established standard screening instrument for postpartum mood disorders in women. It had been translated into Vietnamese for use in Australia with studies of immigrant populations.15,16 For the present study, this version of the EPDS was further modified to include appropriate local linguistic expressions. In particular, item 6 (Things have been getting on top of me) translated literally would indicate either that expected tasks exceeded intellectual capacity or in concrete terms that a flood or natural disaster had put things on top of the individual. It was therefore altered to say: Do you feel that you have too many tasks to manage? Similarly, to avoid misinterpretation, item 10 (I have had thoughts of harming myself) was altered to say Have you had thoughts that you do not want to live any more, and if so, how often?
In order to explore the possibility that it may be more common for Vietnamese women to express their emotional distress as non-specific somatic symptoms than as explicit psychological ones, a checklist of such symptoms was constructed in consultation with health professionals in Ho Chi Minh City and a Vietnamese–Australian psychiatrist. It also included common symptoms of postpartum depression, including nighttime waking apart from caring for the baby and general anxiety.
The whole questionnaire, including the symptom checklist, was translated from English to Vietnamese, back translated for verification and refined by face-to-face and e-mail review between the investigators, according to the principles described by Brislin.38
The questionnaire was designed as a self-report instrument. However, pilot testing with 15 informants revealed that this was an unfamiliar, time consuming and generally difficult task and it was therefore decided to apply it as an individual structured interview. This approach is recommended in these circumstances.35,39 Members of the medical and nursing staff at the centres, who had been trained by the investigators, conducted the interviews in private rooms. Women were approached consecutively in the postnatal wards and invited to take part in the study when returning for the first well-baby or postnatal clinic six to eight weeks postpartum.
Data were entered into a Statistical Package for the Social Sciences (SPSS) v10.0 spreadsheet. Descriptive statistics, univariate measures of association, including χ2, analysis of variance and correlation and multivariate forward stepwise logistic regression were used to analyse the data.
There was a high response rate to the invitation to participate in this study. Approximately 2% (<10) of those approached did not agree to take part, most usually because they were returning to stay with their families in a rural area and would not re-attend the hospital or clinic for the first infant health check. In all, 506 women completed the interview schedule (306 from the Hung Vuong Hospital and 200 from the MCH and FP Center) between 30 August 2000 and the 11 December 2000.
The average age of participants was 28 years (range 16–49 years). Most (99%, 502/506) were married, one was divorced and three were single. Vietnam's public policy recommends that families have no more than two children born at a well-spaced interval. Most of the women in this study (90%, 457/506) had given birth to either a first or second baby, and only 14/506 (3%) to a fourth child. The mean fertility rate was 1.6 (0.73) comparable to the national rate of 1.8 for women aged 15–49.40 The mean age of the babies was 5.2 (1.4) weeks (range 3.7–9.9 weeks).
Vietnam has high participation rates in primary education. In this group of women, 57% (288/506) had up to nine years of formal education, 35% (179/506) had completed secondary schooling and a small group (7.7%, 39/506,) post-secondary education, comparable to the average rate of female education of women aged 18–39 of 7.7 years.36 In contrast, 47% (211/504) of their partners had completed secondary education and 15% (76/504) had a post-secondary qualification. Many of these women (371/506, 73%) had employment to return to after having a baby. Women with at least six years of education were significantly more likely to have a permanent job (336/444, 91%) than those with five or fewer years (35 / 62, 57%) (χ21= 10.28, P < 0.002). Most of their partners (432/502, 86%) had permanent jobs.
Most participants (466/506, 92%) were of Vietnamese heritage; the others (40/506, 7.9%) were ethnically Chinese.
Of these women, 26% (132/506) reported experiencing at least one previous menstrual regulation procedure, or induced abortion (the 1997 national rate of having at least one of these procedures was 18%).40 They continue to be used as a method of family planning, despite the now wider availability of a range of contraceptive alternatives. The rate of previous miscarriage was 11% (53/506). Most (500/506, 99%) had conceived spontaneously. The pregnancy was welcomed by 71% (358/506); described as difficult, but welcome by 23% (114/506); and as completely unwelcome by the remaining 6.7%. Most women reported good health in pregnancy, but 15% (74/506) had pregnancy health problems and three had been hospitalised antenatally.
Most (375/506, 74%) had spontaneous vaginal births; for 78 (15.4%) instrumental assistance had been used and the caesarean section rate was 11% (53/506) (the 1997 national urban caesarean rate was 12%).40 Vietnam is unique in the region in that most births (77%) occur in a medical facility. Nationally, women having first births in urban areas are the group most likely to have medically supervised births and operative interventions in delivery. These patterns reflect the availability of specialist trained personnel and facilities. Most of the participants in this study were primiparous women and all were living in an urban area.40 At the time of interview, 50% of the babies were exclusively breastfed and a further 37% were having some breastfeeds, comparable to the national rate of 53% exclusive breastfeeding at one month.40 There had been no infant deaths in the cohort although previously born children of four women had died. Perinatal maternal and infant morbidities including haemorrhage, infection and prematurity were representative of the normal rates for these centres.
Although exact comparable population data are unavailable, the sample appears adequately representative in socio-demographic and obstetric terms of the urban parturient population. It may not be representative of the majority of the population who live in rural areas.
Traditional health beliefs in Vietnam draw heavily upon the Chinese humoral schema, emphasising the balance of ‘cold’ (yin or am) with ‘hot’ (yang or duong). Childbirth renders women ‘cold’ and physiologically vulnerable and the body must be warmed through a special diet, avoidance of wind (through all orifices), wearing additional clothing and lying over a charcoal fire or other heat source for a month postpartum. Fresh, raw or sour foods are regarded as ‘cold’ and to be avoided. The recommended ‘hot’ diet is rice and salty pork (which women can find monotonous) with special foods including chicken and ginger soup and herbal tonics. Women should have complete rest and, apart from breastfeeding, relinquish household tasks and infant care to others.41 Failing to follow these prescribed practices is believed to place future health and wellbeing in jeopardy. Almost all the participants (475/506, 96%) had observed at least some traditional postpartum practices: lying over heat (62%), not bathing (65%) and using cotton swabs in their ears to protect against cold (78%) and taking special herbal medicines (48%). Many (344/506, 70%) had been assisted to have a complete rest for at least 30 days after giving birth, but the remainder had only been assisted to rest for shorter periods, for 17% (85/506) only seven or fewer days. Many (350/506, 70%) had been given special foods and 71% (357/506) had avoided proscribed foods.
Non-specific somatic symptoms including difficulty swallowing (12/506, 2.4%), heart palpitations (35/506, 6.9%), breathing difficulties (24/506, 7.8%) or a heavy heart (42/506, 8.3%) were not prevalent. However, a persistent feeling that the body was cold was more common (197/506, 39%). Although the somatic symptoms of nausea (14/506, 2.8%) and flatulence (24/506, 4.7%) were rare, headaches (170/506, 34%) and gastrointestinal disturbance in the form of constipation or diarrhoea (228/506, 45%) were more common.
Specific questions about common symptoms of depression appeared to be meaningful to participants and were more prevalent than most non-specific somatic symptoms. These included: difficulty going to sleep (48%), waking in the night apart from caring for the baby (180/506, 36%), general worrying (210/506, 42%) and severe fatigue (231/506, 46%).
The average EPDS score was 9.49 (6.32) [range 0 to 26]. Scores of >12 on the EPDS indicate clinically significant depressive symptomatology. Population and cohort studies in industrialised countries consistently find 10–15% of women have scores in this range.1 In this study, 166/506 women (33%) had scores >12. Overall, 99 (19%) explicitly acknowledged suicidal ideation (item 10), 64 of whom reported having suicidal ideas quite frequently or often.
A number of non-specific symptoms were associated with EPDS scores in the clinical range, the strongest being a persistent feeling of somatic cold. Women with elevated scores were more likely to report gastrointestinal disturbance. It was striking that there were strong associations between established depressive symptoms and scores in the clinical range (see Table 1).
Table 1. Self-reported symptoms and EPDS scores.
|Difficulty swallowing||1.8% (6)||3.6% (6)||ns|
|Heavy heart||6.8% (23)||11% (19)||0.05|
|Breathing difficulty||3.2% (11)||7.8% (13)||0.02|
|Heart palpitations||4.7% (16)||11% (19)||0.005|
|Body feels cold||31% (105)||55% (92)||0.000|
|Nausea||2.6% (9)||3.0% (5)||ns|
|Headaches||35% (107)||38% (63)||ns|
|Flatulence||3.2% (1)||7.8% (13)||0.02|
|Constipation/diarrhoea||39% (132)||58% (96)||<0.0001|
|General worrying||29% (99)||45% (75)||<0.0001|
|Nighttime waking apart from infant care||25% (84)||58% (96)||<0.0001|
|Difficulty falling asleep||37% (124)||73% (121)||<0.0001|
|Severe fatigue||34% (114)||71% (117)||<0.0001|
The observation of most traditional practices did not confer a psychological benefit. The exceptions were those that required sustained active care from others. Women who had assistance to rest and someone to prepare special foods were less likely to be distressed. It has been suggested that the avoidance of proscribed foods may reflect a response to critical scrutiny or active enforcement from others, in particular, a mother-in-law and that this may be why it was associated with the higher rates of distress in the group who were observing this practice (see Table 2).
Table 2. Observation of traditional practices and EPDS scores.
|Not bathing||65% (221)||66% (109)||ns|
|Taking herbal medicines||48% (164)||48% (80)||ns|
|Lying over heat||60% (204)||66% (109)||ns|
|Cotton swabs in ears||75% (256)||82% (136)||0.06|
|Avoiding proscribed foods||68% (231)||76% (126)||0.04|
|Having special foods||35% (120)||22% (36)||0.001|
|≥30 days of complete rest||85% (253)||68% (108)||<0.001|
As summarised in Table 3, there were significant univariate associations between some social and reproductive factors and an EPDS score >12.
Table 3. Univariate differences between non-clinical and clinical EPDS score groups.
|Unwelcome pregnancy||20% (68)||48% (80)||<0.001|
|Maternal education <5 years||8.5% (29)||20% (33)||<0.001|
|No maternal permanent employment||21% (71)||39% (64)||<0.001|
|No paternal permanent employment||11% (37)||21% (34/164)||0.004|
|Caesarean delivery||8.2% (28)||15% (25)||0.02|
|Baby cries for episodes >10 minutes||27% (90)||47% (78)||<0.001|
|Baby gaining <1000 g per month||21% (72)||30% (49)||0.03|
|Insufficient breastmilk||28% (83/296)||37% (54/146)||0.04|
|Not being given special food||65% (220)||78% (130)||0.001|
|Avoiding proscribed foods||68% (231)||76% (126)||0.04|
|<30 days free of work after birth||29% (99)||45% (75)||0.001|
|Unable to confide in partner||62% (212)||83% (137)||<0.001|
Using a forward stepwise procedure, a logistic regression model including all variables with a significant univariate association was constructed to examine which constellation of factors best predicted scores in the EPDS clinical range. A model with 92% specificity, 48% sensitivity, a positive predictive value of 74%, a negative predictive value of 78% and in which 77% of cases were correctly classified was found. Having an unwelcome pregnancy, no permanent job to return to after having the baby, less than a month of complete rest after childbirth, caring for a crying unsettled baby, not being given special foods, avoiding proscribed foods and being unable to confide in their husbands were associated with clinically significant depressive symptomatology (see Table 4).
Table 4. Logistic regression model predicting EPDS scores ≥ 13.
|No permanent job||0.64||0.4||1.0|
|Unable to confide in partner||0.45||0.27||0.75|
|Not being given special foods||0.61||0.37||0.99|
|Avoiding proscribed foods||2.5||1.5||4.2|
|<30 days of complete rest after birth||1.9||1.1||3.2|
|Unsettled crying baby||1.9||1.2||3.0|
|Baby gaining <1 kg per month||0.61||0.37||1.0|
This study has used a valid, published screening instrument with strong psychometric properties and acceptable cultural relevance with a carefully constructed systematic structured interview that incorporated questions about pertinent psychosocial and obstetric factors. There was a high response rate; the sample appeared accurately representative of the population of urban parturient women in southern Vietnam and large enough to allow multivariate statistical analysis of the variables. There is internal validity in that the scores on the screening questionnaire, the EPDS are strongly associated with the presence of established depressive symptoms. We believe therefore that our findings about the prevalence of clinically significant depressive symptomatology can be regarded as reliable and able to be generalised with some confidence. It is a limitation of the study that neither sex of the infant, in a culture with preference for male offspring, nor exposure to intimate partner violence was assessed. This study used a research screening instrument and did not include a diagnostic interview for depression and therefore the possibility that the depressive symptoms found might be reflecting unhappiness as part of a process of adjustment to life change, rather than mental illness cannot be excluded.42 In assessing women, including a large group of primiparae early in the postpartum period, it may be more likely that adjustment phenomena were being observed and the persistence of this mood disturbance cannot be ascertained from this study. Associations observed in cross sectional studies cannot be regarded as causally linked; however, the logistic regression model is easily interpretable and fits closely with comparable international findings about risk factors for postnatal depression.
It appears that symptoms of depression are recognisable and meaningful to urban Vietnamese women and if asked about specifically, are acknowledged. Apart from a persistent sense of being ‘cold’, which in the humoral schema may reflect a general sense of unwellness or distress, depressive symptoms appeared not to be exclusively expressed as non-specific or specific somatic symptoms. Most of the existing research into childbirth and the postpartum in developing countries have been generated from descriptive anthropological studies of traditional beliefs and practices, which have been assumed to be psychologically protective for mothers. The evidence of this study suggests that, on the contrary, there are by international standards very high rates, between two and three times those reported among women living in developed countries of clinically significant depressive symptomatology in parturient women in Ho Chi Minh City. The rates are similar to those reported in the few comparable studies in other resource poor developing countries.9,10 In the developed world, while suicidal ideas are expressed by women with severe postnatal depression,43 acts of self-harm and completed suicide in women in the first 42 days after childbirth are relatively rare.25 However, when the period of ascertainment is extended to the postpartum year, suicide makes a major contribution to maternal mortality.28 The findings of this study in which suicidal ideas were acknowledged by an unexpectedly high proportion of this cohort within 42 days of giving birth with the findings that suicide is a common cause of maternal mortality in Vietnam suggest that depression is a very serious, but under-recognised, health problem among newly delivered women in Vietnam.
These findings also suggest that apart from increased practical and emotional support, the observation of traditional practices does little to protect women against the development of symptoms of depression. The predictors of clinically significant depressive symptomatology reflect closely those observed in international investigations of risk factors for postpartum depression. There is robust evidence that a poor marital relationship including the inability to confide in a partner, low support, here in terms of relief from household tasks and preparation of special foods,1,2 unwelcome pregnancy and no job to return to after maternity leave11 and an unsettled baby44,45 contribute to depression after childbirth in women in developed countries. Women are in a state of heightened vulnerability following childbirth and it is probable that criticism and coercion are especially harmful to them at this time, which may be why being monitored to ensure that proscribed foods are avoided, contributes strongly to distress.23
These findings provide further strong evidence of the importance of social and economic adversity and human rights to mental health. Rates of depression are high for women living in poverty13,46 and are worsened by low social position, low education, lack of employment and the double burden of paid and unpaid work. Other contributing factors include experiencing violence from an intimate or being trapped in humiliating circumstances,13 were not specifically assessed in this study but may have underpinned the findings.
We are seriously concerned by our findings, which suggest that women's mental health including in the postpartum period in Vietnam requires close and urgent consideration. The nature and extent of this mental health problem had not been identified by existing clinical services. Further studies to confirm through diagnostic clinical interviews the validity of the results of this screening survey and to elucidate whether the incidence of clinically significant depressive symptomatology is the same or different in rural areas should follow this prevalence study. The persistence of depressive symptomatology needs to be examined through longitudinal follow up as does the impact on infant development of maternal depression in this setting. Our findings support the WHO's call for greater awareness of mental health and its integration into primary health care systems, including those providing maternal and infant health care.
Supported by a grant (WP TLR/ICP/RPH/002 05.01.RG RB 00) from the WHO Regional Office of the Western Pacific and a University of Melbourne International Joint Research Project Award. Dr Vu Thi Nhung, Director of Hung Vuong Obstetrics and Gynaecology Hospital, Dr. Huynh Thi Trong, Director of HCM City MCH Center, Dr Nguyen Trong Hieu assisted with statistical analysis and Ms Le Thi Hanh, Dr Tran Huynh Kim and Dr Le Trung conducted the interviews and entered the data.
Accepted 15 February 2004