Research - Swedish study on mothers

Centre for Epidemiology, National Board of Health and Welfare, 106 30
Stockholm, Sweden (G Ringbäck Weitoft BA, B Haglund DMSc, M Rosén DMSc); and Department of Public Health and Clinical Medicine, Umeå University, Umeå, Sweden (G Ringbäck Weitoft, M Rosén)

Summary
Background An increasing number of lone mothers are experiencing financial and health disadvantages. Our aim was to assess mortality among lone mothers compared with mothers with partners.
Methods In this population-based study, overall and cause-specific
mortality, between 1991 and 1995, was estimated for 90 111 lone mothers and 622 368 mothers with partners from data collected in the Swedish Population and Housing Census 1990. We estimated relative risks by Poisson regression, adjusted for socioeconomic status, and, to handle health-selection effects, we adjusted for previous inpatient history from 1987 to 1990. Findings Lone mothers showed an almost 70% higher premature risk of death than mothers with partners. The excess risk remained significantly increased (relative risk 1•2 [95% CI 1•11•4]) after adjustments for socioeconomic status and previous severe somatic and psychiatric inpatient history. Increased mortality was especially pronounced for suicides (2•2 [1•53•1]),violence (3•0 [0•910•6]), and alcohol-related mortality (2•4 [1•44•1]) among mothers who were without a partner in 1985 and in 1990.
Interpretation
The increase of lone mothers in society shows financial, social, and health disadvantages. Nevertheless, the increased mortality risk of lone mothers seems to be partly independent of socioeconomic status and
health selection into lone motherhood. For long-term lone mothers the risks may be underestimated when adjusting for selection bias by taking hospital discharge history into account, since these events may be part of the consequences of the stress of lone motherhood. Lancet 2000; 355: 121519

Introduction
During the past few decades the proportion of lone-parent families has
increased substantially in Western countries. In Sweden, lone parents now constitute about 20% of all families with children.1 Most lone parents are women. Although studies show that Swedish lone mothers have an economically more favourable situation than their counterparts in other countries, many findings show that their situation is disadvantageous.14 Relying on only one income, lone-parent families are to a larger extent dependent on public subsidy. Social welfare and housing allowance are far more common for lone-parent families than for other families.1 The proportion of lone mothers who had difficulties in paying their household expenses increased from about 30% in 1979, to more than 50% in 1995. For couples, proportions were 12% in 1979, and 20% in 1995.3 There is a well-established link between socioeconomic status and health.59 Furthermore, the relation between marital status, parenthood, and health is well known.10,11 White-collar workers and high-income groups have better health than blue-collar workers, unemployed people, and those in low-income strata. Married people live longer and report better health conditions than unmarried and divorced people. Given the circumstances of lone mothers, it is not surprising that studies have found a high proportion of lone parents with poor health.4,1216 In a comparison of self-perceived health among lone
mothers and mothers with partners, in Great Britain and Sweden,17 the
increased relative risk for lone mothers was of the same size in both
countries despite a more favourable social policy in Sweden. There are few studies on mortality, although in some analyses of the effect of
occupational, marital, and parental roles on mortality, lone mothers were
found to have an increased risk of premature death.18,19

The explanations offered for these health disadvantages are mainly divided between theories of social causation and health selection. Social causation suggests that the psychological stress, stigma, and financial difficulties associated with lone parenthood have adverse health consequences. For previously married people the stress suffered during marital breakdown also contributes. The health-selection argument suggests that the "healthy state of marriage" is due to a lower probability for unhealthy people to get married, stay married, or remarry. Previous studies have mainly focused on elucidating the processes of social causation.14 Studies that examine health-selection mechanisms are rare, probably because of the scarcity of the data required for this kind of research.

We analysed the mortality of lone mothers in Sweden compared with mothers with partners on the assumption that the poor socioeconomic status of lone mothers contributes to higher mortality. We adjusted for previous inpatient history to handle health-selection factors. This was done for overall and cause-specific mortality. In addition, we assessed the modifying effects of social position, social welfare, and number of children on the association between mortality and lone motherhood.

Participants and methods
The population was generated from the Swedish Population and Housing Census 1990. In Sweden, it is mandatory for every household to answer and return a census questionnaire (from 1960 to 1990, the census was every 5 years). The drop-out rate in the 1990 census was 2•5%.20 We selected women aged 2954 years, living alone or with a partner, and with children aged 015 years, by using a variable about household type. We combined this variable with information about marriage or consensual union. For each individual in the census there is information about the oldest member of the household. To check our classification and to count the number of children in a family, we linked the members of each household and excluded all women classified as mothers who had no matching children. Information on country of birth, age, socioeconomic status, housing situation, and geography was also obtained from the 1990 census. Socioeconomic status was defined according to a
classification used by Statistics Sweden, which is based on occupation but also takes educational level of occupation, type of production (wares or services), and position at work into account.21

Information on social welfare and unemployment benefits for each participant was obtained from the Total Enumeration Income Survey 1990 (Statistics Sweden), which comprises information about the taxes and incomes for the entire population. We had access to the annual totals, and women who received any amount of allowance were classified as receiving either social welfare or unemployment benefit. The linkage procedure generated 1010 drop-outs (0•14%). We also linked data to the 1985 census to get data on family situation from an earlier period.
Information on mortality for 199195 was obtained from the National Cause of Death Registry. To control for a history of morbidity (psychiatric and somatic) we linked data to the Swedish Hospital Discharge Register for 198790, which comprises all discharges from Swedish hospitals (National Board of Health and Welfare, Centre for Epidemiology).

Analyses were made on total mortality and the following broad cause-of-death groups, as defined by the International Classification of Diseases, ninth revision (ICD-9): ischaemic heart disease (410414, 429.2), cancer
(140209), respiratory diseases (460519), and injuries and poisoning
(E800E999). External causes of death were further divided into the
following categories: suicide (E950E959 and E980E989); violence, assault, and homicide (E960E969). Alcohol-related mortality was defined as ICD 291, 303, 305.0, 357.5, 425.5, 535.3, 571.0571.3, E860, E980+980, on the basis of either underlying or contributing causes of death. Former illness was defined and divided into two categories: women who had been discharged from hospital between 1987 and 1990 with a psychiatric diagnosis (291319); and women who had been discharged during this period with a somatic diagnosis--ie, with a diagnosis other than a psychiatric or maternity-care (630679) diagnosis.
We assessed person-years at risk by adding up the years the individuals were alive and living in Sweden during the follow-up period (199195). The year of death contributed 6 months for that specific calendar year. We obtained an indicator of the duration of lone or couple motherhood by taking into account the civil status 5 years previously from the Population and Housing Census 1985.

Statistical analysis
Multivariate analyses were done by Poisson regression with mortality as the dependent variable. Age was entered as a continuous independent variable and was present in all models. Adjustments were also made for different groups of variables: those mainly reflecting the health status of the individuals before entering the study, and those mainly reflecting the social situation of the mothers. Health-status variables were used to control for possible confounding from selection of physically and psychologically vulnerable persons into lone motherhood, and social-situation variables served as explanatory variables of a possible excess mortality rate. We used SAS version 6.12 for statistical analyses.
After preliminary analyses, the categories country of birth, living in a big
city, and receipt of unemployment benefit were excluded from all models
because they did not seem to be confounders.

Results
We studied 90111 lone mothers and 622368 mothers who had partners. The socioeconomic status of lone mothers in comparison to mothers with partners is shown in table 1. A slightly higher proportion of lone mothers than mothers with partners were manual workers or high and medium non-manual workers, and were more likely to lack information about occupation. We also analysed the level of education (data not shown), and found a lower degree of higher education (ie, >12 years in school) among lone mothers (22687 of 90111 [25%]) than among mothers with partners (189940 of 622368 [31%]). Lone mothers were six times more likely to receive social welfare and two times more likely to receive unemployment benefits in 1990 than were mothers with partners. A higher proportion of lone mothers lived in one of the three largest cities in Sweden (Stockholm, Gothenburg, and Malmoe), and most rented their homes; more than 80% of mothers with partners owned their homes. Lone mothers had fewer children on average than mothers with
partners. Lone mothers were more likely than mothers with partners to have had a psychiatric diagnosis, whereas the relative differences in somatic diagnoses were small. 25•4% of lone mothers and 22•8% of mothers with partners had been hospital inpatients during 1987/90.

The crude mortality rate (number of deaths/person-years at risk) indicated an overall excess mortality risk for lone mothers irrespective of
stratification (table 1). For lone mothers, the risk of dying within a
5-year period was almost 70% higher than for mothers with partners when adjustments were made for age (table 2). However, the risks were reduced when hospital data indicating previous severe somatic or psychiatric morbidity were entered into the model. When socioeconomic conditions (socioeconomic status, social welfare, number of children, type of dwelling) were entered into the model, there was a greater decrease in the mortality risk (table 2). There was an excess risk for ischaemic heart disease among lone mothers when we adjusted only for age. However, when we adjusted for socioeconomic factors and previous psychiatric disease, which were highly correlated to smoking, the significant effects diminished. The greatest risks were found among external causes (E800E999), including suicide, with an almost 60% increased risk for lone mothers after adjustment for both hospital inpatient care and socioeconomic conditions.

By grouping the lone mothers and mothers with partners in 1990 according to their family situation in 1985, we took the duration of lone and couple motherhood into account (table 3). Women with children and who were without a partner at both times had the highest total mortality risk, both when age was controlled for, and when somatic and psychiatric inpatient history from 1985 to 1990 and socioeconomic variables (socioeconomic group, social benefits) were entered into the analysis (table 3). Having a partner on both occasions or changing from the single to the couple status was associated with lower mortality. There was no significant difference between lone mothers and mothers with partners for ischaemic heart disease, probably due to small numbers. For cancer mortality only mothers who had been separated
from their partner during the past 5 years had a significantly increased
risk after adjustment for age. The risk of death from external causes,
injuries, and poisoning doubled for mothers who were alone in 1985 and 1990, compared with mothers who had a partner in 1985 and 1990, after controlling for hospital and socioeconomic variables (table 3).
Women who were with partners on both occasions had the smallest risk of dying from suicide, violence, assault, homicide, or alcohol-related causes (table 3). When age was taken into account, mothers without partners in 1985 and 1990 had an almost four-fold risk of committing suicide and a five-fold risk of being a victim of violence (although the absolute numbers are quite small) or dying of alcohol-related causes. After controlling for hospital and socioeconomic variables the excess risks were lowered. An expected dose-response relation was apparent for many causes of death. Mothers who were living with a partner in both 1985 and 1990 had the lowest risks, followed by women who were alone in 1985 and cohabitant in 1990. Cohabitant women who separated from their partner during 198590 had higher risks, and the highest risks were found among women who were lone mothers in both 1985 and 1990.
Analyses were also done to assess whether the relation between lone
motherhood and mortality varied by socioeconomic group, receipt of social welfare, and number of children. The effects of lone motherhood on all-cause mortality were similar in all socioeconomic categories and among those who received social welfare as well as those who did not. We found a slight increase of the relative risks with increasing number of children (one child, 1•21 [95% CI 1•071•37]; two or three children, 1•29 [1•091•53]; at least four children, 1•49 [0•683•29]). We could not exclude that this effect modification was random, since the confidence limits around the interaction-term were wide and included 1•0.

Discussion
Our study of more than 90 000 lone mothers showed an increased premature death risk for lone mothers compared with mothers living with partners. The highest risks were for suicide, violence, and alcohol-related mortality, although absolute numbers of violent deaths were small.
Some of the mortality differences in the multivariate analyses were
accounted for by differences in socioeconomic status. To some extent we have controlled for health selection by inclusion of previous psychiatric and somatic morbidity in the models. However, after these adjustments there was still a significant increase in all-cause mortality, suicide, inflicted violence, and alcohol-related mortality.

Use of the variable household by type alone in a definition of lone mothers and mothers with partners results in the misclassification of 8% of lone mothers and 2% of mothers with partners, according to the Population and Housing Census 1990.20 However, we combined different variables, linked the members of each household and excluded all women classified as mothers but been minimised by this procedure and does not
imply systematic bias.

The likelihood that a woman is married or living with a partner may be
influenced by the health status or earlier illness of the woman. To control
for this potential selection bias, we analysed whether the woman was
admitted to hospital at any time in the 4 years preceding the study period; however, this period may have been too short. We had information about family situation for only the years 1985 and 1990 (census data) and no information on the exact timepoint for change of civil status. For long-term lone mothers, hospital admissions might have resulted from the stresses of lone motherhood. In that case, adjustment for hospital discharge will give an underestimation of the effect of lone motherhood on mortality.

Furthermore, we had no information about health problems that do not need medical attention in hospital. The limitations of the data available did not allow us to extend collection of morbidity data.

The sociodemographic variables were measured at one timepoint and merely served as indicators of socioeconomic situation. Although a broad range of variables was available, many aspects of a life situation cannot be elucidated from register data. For instance, we had no information about existing, or lack of, confiding/intimate relationships, or any experienced stress and stigma, all of which are factors suggested by Benzeval14 as related to lone mothers' health.

The use of multivariate models and a longitudinal approach gives an
understanding of the relative importance of different factors that may
explain the health differences between lone mothers and mothers with
partners. The interpretation needs to be made with care so that the
substantial public-policy problem of the poor health of lone mothers is not thought unimportant, just because it can be explained away by other
factors.14 However, from a policy perspective it is important to ascertain
whether the increased mortality of lone parents is primarily a result of
their being disadvantaged or whether there is something else in lone
motherhood that is damaging to health and leads to an increased risk of
premature death.

Studies among lone mothers with dependent children have shown that mothers in full-time employment have poor psychosocial health.12,22 In Sweden, an important cornerstone of family policies has been the goal of full employment for both men and women and the possibility of combining
parenthood and salaried work. Lone mothers in Sweden are more likely to work full time than are married mothers, and on average they work more hours per week.1 Although full-time employment gives the opportunity to increase the family income, and increases confidence and self-esteem, it may also lead to role overload,17 because of the strain of combining work and parental roles without the emotional or financial support of a partner.

Gähler2 found that divorced mothers had lower access to social support than other women. Although this finding was based on small numbers, another Swedish study reported that lone mothers, compared with mothers with partners, were more likely to be daily smokers, took less exercise, had a lower quality of their social network, and were more likely to work full time (52•2% vs 35•6%).23 There are economic reasons for the longer working hours of lone mothers, since although shorter hours are sanctioned by Swedish law for parents of small children, the mother may still have to work longer hours to support herself and her children. The investigators suggested that lone mothers should be given the financial incentives to work less.23 Whitehead and colleagues have also suggested that lone mothers suffer from time poverty.18 The expectations to work, despite a relatively weak position in the labour market, often in poor-quality jobs, and lower access to social support are factors that contribute to the poorer health of lone mothers.

However, an intact marriage is no guarantee of an emotionally healthy,
well-supervised home environment. Thus, lone parenthood is for some people an escape from a different set of problems and becoming a lone parent may actually improve their health chances. For example, in the UK, 20% of lone parents indicate violence and 15% give alcoholism and drug abuse as major factors in the breakdown of the relationship with their partner.24 So there might even be a positive health selection linked with lone motherhood, because it takes strength and power of initiative to leave a marriage that is not functioning. A divorce leading to lone motherhood might not necessarily be an entirely negative experience.

The strengths of our study were the completeness of the study population, the longitudinal approach, and the wide set of individual background characteristics, which enabled a multivariate approach. Other studies have also shown increased morbidity and mortality for lone mothers, but our study lends further support to the hypothesis that lone motherhood or separation are stressful events that lead to premature mortality. The increased mortality risks seem to be partly independent of a previous history of severe illness, socioeconomic status, &c. The analysis of mortality and changes in civil status also gives further support to this hypothesis.

Contributors
Gunilla Ringbäck Weitoft was responsible for study design, analyses, and
writing of the paper. Bengt Haglund contributed to discussion about study design and data analyses. Måns Rosén contributed to data analyses and writing of the paper.

References
1 Nyman H. An economic analysis of lone motherhood in Sweden (doctoral
thesis). Göteborg: Department of Economics, Göteborg University, 1998.
2 Gähler M. Life after divorce. Economical and psychological well-being
among Swedish adults and children following family dissolution. Dissertation
series 32 (doctoral thesis). Stockholm: Swedish Institute for Social
Research, 1998.
3 Vogel J. Economic problems. Living conditions and inequality 19751995 (in
Swedish with English summary). In: Levnadsförhållanden, rapport no 91.
Stockholm: Statistics Sweden, 1997.
4 Burström B, Diderichesen F, Shouls S, Whitehead M. Lone mothers in Sweden:
trends in health and socio-economic circumstances, 19791995. J Epidemiology
Community Health 1999; 53: 75056.
[ PubMed Community+Health&volume=53&year=1999&page=750&display=abstract> ]
5 Marmot MG, Shipley MJ, Rose G. Inequalities in death--specific
explanations of a general pattern? Lancet 1984; i: 100306.
6 Vågerö D, Leon D. Effect of social class in childhood and adulthood on
adult mortality. Lancet 1994; 343: 122425. [ Pub Med
7 Power C, Matthews S. Origins of health inequalities in a national
population sample. Lancet 1997; 350: 158489. [ Text
_research.10462.1> ]
8 Report of the Independent Inquiry into Inequalities in Health. Chaired by
Sir Donald Acheson. London: Stationery Office, 1998.
9 Lynch JW, Kaplan GA, Shema SJ. Cumulative impact of sustained economic
hardship on physical, cognitive, psychological, and social functioning. N
Engl J Med 1993; 337: 188995. [ PubMed
volume=337&year=1993&page=1889&display=abstract> ]
10 Macintyre S. The effects of family position and status on health. Soc
Sci Med 1992; 35: 45364. 11 Elstad JI. Inequalities in health related to women's marital, parental, and employment status: a comparison between the early 70s and the late 80s,
Norway. Soc Sci Med 1996; 42: 7589olume=42&year=1996&page=75&display=abstract> ]
12 Macran S, Clarke L, Heather J. Women's health: dimensions and
differentials. Soc Sci Med 1996; 42: 120316. 13 Berkman PL. Spouseless motherhood, psychological stress, and physical
morbidity. J Health Soc Behav 1969; 10: 32334.
14 Benzeval M. The self-reported health status of lone parents. Soc Sci Med
1998; 46: 133753.
15 Popey J, Jones G. Patterns of health and illness amongst lone parents. J
Soc Policy 1990; 19: 499534. volume=19&year=1990&page=499&display=abstract> ]
16 Shouls S, Whitehead M, Burström B, Diderichsen F. The health and
socio-economic circumstances of British lone mothers over then last two
decades. Popul Trends 1999; Spring (95): 4146.
17 Kotler P, Wingard DL. The effect of occupational, marital and parental
roles on mortality: the Alameda Count Study. Am J Public Health 1989; 79:
60712.
18 Whitehead M, Burström B, Diderichsen F. Social policies and the pathways
to inequalities in health: a comparative analysis of lone mothers in Britain
and Sweden. Soc Sci Med 2000; 50: 25570.
19 Martikainen P. Women's employment, marriage, motherhood and mortality: a
test of the multiple role and role accumulation hypotheses. Soc Sci Med
1995; 40: 199212.
20 Statistics Sweden. Evaluation of the Population and Housing Census 1990.
Household data and dwelling data (in Swedish with an English summary). In:
Statistiska meddelanden Be43 SM9201. Stockholm: Statistics Sweden, 1992.
21 Statistics Sweden: Swedish socio-economic classification (in Swedish with
an English summary). In: Meddelande i samordnings frågor 1982: 4. Stockholm:
Statistics Sweden, 1995.
22 Søgaard AJ, Kritz-Silverstein D, Wingard DL. Finnmark Heart Study:
employment status and parenthood as predictors of psychological health in
women, 2049 years. Int J Epidemiol 1994; 23: 8290.
23 Forssén A, Janlert U. Ensamstående mödrar har ökad risk för ohälsa
Allmänmedicin 1991; 12: 7276.
24 Bradshaw J, Millar J. Lone parent families in the UK. Department of
Social Security research report no. 6. London: HM Stationery Office, 1991.

Commentary

Suicide in women
One of the most consistent findings about suicide is the lower rate among
women than men. In England and Wales, for example, the male to female ratio
is around 3:1 but higher in the younger age-groups. Similar sex differences
are reported from many countries in the developing and developed worlds.1 A
notable exception is China (where the sex difference is reversed),2 and in
several countries rates are high among immigrant women born in the Indian
subcontinent.3

A male preponderance is found in other kinds of unnatural mortality.
Homicide is an offence usually committed by men, with men as their victims;
in England and Wales only a quarter of homicide victims and a tenth of
perpetrators are female.4 Those who die in accidents too are most likely to
be male, and in England and Wales only 6% of accident deaths occur in women
aged under 35. But it is the suicide figures that most need to be explained
as suicide becomes an international health priority, and two questions
arise. Why are suicide rates lower in women? And what are the factors that
put certain women at risk?

Two papers in today's Lancet address the risk of unnatural mortality,
particularly suicide, in women in two diverse cultural settings--Bangladesh
and Sweden. In Bangladesh half the unnatural deaths in women identified by a
painstaking case-finding exercise were by suicide, most by poisoning. The
largest numbers were in women aged under 30, and there was an association
with being married. The investigators conclude that their findings are
consistent with the view that economic adversity and marital and family
discord are important causes of suicide in Bangladesh. The Swedish study, a
population-based examination of deaths among single mothers, also highlights
the importance of psychosocial adversity in suicide and other causes of
premature death among women. However, the higher rates of mortality overall
and of suicide specifically that were found among lone mothers persisted
after the analysis was adjusted for social and economic factors and (as a
marker of previous health) for hospital admission for psychiatric and
somatic illness. The vulnerability of lone mothers could not be attributed
to these factors alone.

The disparity between men and women in suicide rates has been most apparent
in the past three decades, during which suicide rates among young men but
not young women have increased substantially in many countries. In England
and Wales in 1970 the population suicide rate was just over 10 per 100 000,
similar to recent figures. But in 1970 the rate for men was 12•2 and that
for women was 8•7, a male to female ratio of 1•4:1. Now the rate for men is
over 15 and that for women has fallen to under 5, so the excess of deaths
among men has doubled. Three broad explanations are possible. The causes of
the increased suicide rate among men may be less common among women, or they
may be experienced differently, or they may be counterbalanced by other
factors.

These possibilities are not mutually exclusive, and examples of each are
apparent in the links between suicide and unemployment, divorce, and misuse
of alcohol and drugs. Increases in the prevalence of these factors may have
contributed to the rise in suicide rates among men in recent years and the
widening gap between the sexes.5 Alcohol and drug misuse has also risen
among women, although rates remain lower than among men. By definition
divorce happens as commonly among men as among women, but the experience is
often different. In particular, women are more likely to retain
responsibility for the care of young children, a factor that seems to be
protective against suicide.6 Although population unemployment rates may be
associated with suicide in both sexes,7 the effect of rising unemployment
rates among men in the 1980s on the socioeconomic circumstances of women,
and therefore their risk of suicide, may have been offset by the increasing
entry of women into the job market, another possible protective factor.8
What, then, is different about those women who do take their own lives?
Inevitably they have high rates of these key risk factors. They also have
high rates of mental illness. Furthermore, the impact of severe mental
illness, such as schizophrenia, on suicide rates may be greater in women
than in men.9 This point may be one reason for the higher rate of recent
contact with psychiatric services among women who have committed
suicide--29% compared with 19% among men4--although similar findings on
contact with primary care10 suggest that such figures also reflect a greater
readiness to recognise and accept the need for mental-health care.
Suicide rates can also be influenced by the lethality of available methods
of self-harm. In general this fact protects women because they are less
likely than men to use violent methods,4 more likely to take non-fatal
overdoses of drugs,11 and more likely to survive their first episode of
self-harm.12 However, violent methods are a feature of suicide by women with
severe mental illness; the proportion dying by hanging or jumping from a
height is greater among women with severe mental illness than among the
female population as a whole.4 Certain violent methods are also associated
with suicide in particular cultures: in Asian women in England and Wales
burning is disproportionately common.13 In Bangladesh, women who commit
suicide do so mainly by poisoning, and they are thought not to use
psychotropic drugs or analgesics, as in most western countries, but to take
pesticides, which are widespread in a rural economy.

Suicide among women, as in every other group, has neither a single cause nor
a single solution, but the components of a prevention strategy are
universal. They include the alleviation of social hardship, reduction in
access to the main methods of suicide, and specific measures for high-risk
groups. Suicide rates in women overall may be comparatively low but concern
over rates in men should not disguise the high risk that some women also
face.
Louis Appleby
_____

School of Psychiatry and Behavioural Sciences, University of Manchester,
Withington Hospital, West Didsbury, Manchester M20 8LR, UK
1 La Vecchia C, Lucchini F, Levi F. Worldwide trends in suicide mortality.
Acta Psychiatrica Scand 1994; 90: 5364.
2 Pritchard C. Suicide in the People's Republic of China categorized by age
and gender: evidence of the influence of culture on suicide. Acta
Psychiatrica Scand 1996; 93: 36267
3 Patel SP, Gaw AC. Suicide among immigrants from the Indian subcontinent:a
review. Psychiatry Services 1996; 47: 51721.
4 Appleby L, Shaw J, Amos T, et al. Safer services. Report of the National
Confidential Inquiry into Suicide and Homicide by People with Mental
Illness. London: Department of Health,1999.
5 Hawton K. Why has suicide increased in young males. Crisis 1998; 19:
11924.
6 Appleby L. Suicide during pregnancy and in the first postnatal year. BMJ
1991; 302: 13740.
7 Gunnell D, Lopatatzidis A, Dorling D, Wehner H, Southall H, Frankel S.
Suicide and unemployment in young people: analysis of trends in England and
Wales, 19211995. Br J Psychiatry 1999; 175: 26370.
8 Burr JA, McCall PL, Powell Griner E. Female labor force participation and
suicide. Soc Sci Med 1997; 44: 184759.
9 Allebeck P, Wistedt B. Mortality in schizophrenia: a ten-year follow-up
based on the Stockholm County inpatient register. Arch Gen Psychiatry 1986;
43: 65053.
10 Haste F, Charlton J, Jenkins R. Potential for suicide prevention in
primary care? An analysis of factors associated with suicide. Br J Gen
Pract 1998; 48: 175963.
11 Hawton K, Fagg J, Simkin S, Bale F, Bond A. Trends in deliberate
self-harm in Oxford, 19851995: implications for clinical services and the
prevention of suicide. Br J Psychiatry 1997; 171: 55660.
12 Isometsä ET, Lönnqvist JK. Suicide attempts preceding completed suicide.
Br J Psychiatry 1998; 173: 53135
13 Soni Raleigh V, Bulusu L, Balarajan R. Suicides among immigrants from the
Indian subcontinent. Br J Psychiatry 1990; 156: 4650.