Intended for healthcare professionals

Primary Care

The health of children in refuges for women victims of domestic violence: cross sectional descriptive survey

BMJ 2001; 323 doi: (Published 28 July 2001) Cite this as: BMJ 2001;323:210
  1. Elspeth Webb, senior lecturera (webbev{at},
  2. Judith Shankleman, research health visitorc,
  3. Meirion R Evans, senior lecturerb,
  4. Rachel Brooks, clinical lecturera
  1. a Department of Child Health, University of Wales College of Medicine, Cardiff CF14 4XW
  2. b Department of Epidemiology and Public Health, University of Wales College of Medicine
  3. c Sure Start, Primary Care Directorate, Cardiff and Vale NHS Trust, Lansdowne Hospital, Cardiff CF11 8UL
  1. Correspondence to: E Webb
  • Accepted 14 May 2001


Objectives: To describe the health and developmental status of children living in refuges for women victims of domestic violence and to investigate their access to primary healthcare services.

Design: Cross sectional survey.

Setting: Women's refuges in Cardiff.

Participants: 148 resident children aged under 16 years and their mothers.

Main outcome measures: Completeness of records on the child health system (register of all children that includes data on the child's health) for named health visitor, named general practitioner, and immunisation uptake; satisfactory completion of child health surveillance; Denver test results for developmental status; Rutter test scores for behavioural and emotional problems; reports of maternal concerns.

Results: 148/257 (58%) children living in refuges between April 1999 and January 2000 were assessed. Child health system data were incorrect (general practitioner and/or address) or unavailable for 85/148 (57%) children. Uptake of all assessments and immunisations was low. 13/68 (19%) children aged <5 years had delayed or questionable development on the Denver test, and 49/101 (49%) children aged 3-15 years had a Rutter score of >10 (indicating probable mental health problems). Concerns were expressed by mothers of 113/148 (76%) children. After leaving the refuge, 22 children were untraceable and 36 returned home to the perpetrator from whom the families had fled.

Conclusions: The children had a high level of need, as well as poor access to services. Time spent in a refuge provides a window of opportunity to review health and developmental status. Specialist health visitors could facilitate and provide support, liaison, and follow up.

What is already known on this topic

What is already known on this topic A pilot study showed poor uptake of immunisations and surveillance among children who live in refuges for women victims of domestic violence

Qualitative studies suggest that these children are at risk of psychological ill health

What this study adds

What this study adds Baseline health and demographic data show that children in refuges have a high level of unmet health need, particularly in relation to mental health difficulties

Their families have poor access to health services

Editorial by Lynch and Gough


Children in refuges for women victims of domestic violence constitute one of several marginalised groups with poor access to services.1 They are a largely unstudied population, although a pilot study in Wales identified poor uptake of immunisation and surveillance.2 They have the added experience of growing up with violence; nearly 75% have witnessed violence, including 10% witnessing sexual assault of their mother.3 Between 20% and 70% of children exposed to domestic violence are themselves physically abused.4 Qualitative studies show that they may be at risk of psychological ill health. 5 6

Domestic violence is a common and serious problem.7 Nearly 35 000 children in England and Wales pass through refuges annually, with a similar number referred on to other safe accommodation. Women seek refuge after an average of 28 assaults (Women's Aid and Welsh Women's Aid, personal communication, 1995).

We studied children passing through all Cardiff refuges to (a) describe their health and developmental status; (b) assess the prevalence of serious emotional and behavioural disorders; (c) document the completeness of records in the child health system (a register of all children that includes data on the child's health); and (d) measure uptake of surveillance and immunisation.


We conducted the study from April 1999 to January 2000 in all five women's refuges in Cardiff, which are affiliated to Welsh Women's Aid; three are run by Cardiff Women's Aid and two (catering for women from ethnic minority communities) by the Black Association of Women Step Out (BAWSO).

A research health visitor (JS) agreed protocols with refuge staff. She also liaised with health and social services professionals to agree referral protocols and sharing of information so that identified needs could be addressed. Refuge staff informed her when a new family arrived. She then met the mother, explained the purpose of the study, and obtained oral consent to assess the woman's children.

The assessment included a general health questionnaire and history administered orally to the mother; developmental screening of children aged 0-4 years with the Denver screen8; and estimation of behavioural and emotional problems with a semistructured interview and the revised Rutter parents' scales.9 The mother was present for all assessments.

The questionnaire asked about the child's primary care team and included open ended questions to elicit the mothers' current concerns about their children. The Denver results were categorised as pass, questionable, or fail. The Rutter scales are based on symptom scores for behaviours present during the previous year. Symptoms reported in the month before assessment were not scored. A Rutter score of over 10 (on a scale of 0-50) shows probable mental health problems; such children may benefit from fuller psychological assessment.10

We reviewed data from the child health system for completeness. The system is a population register of all children that is generated from birth notification; it includes data on the child's primary health team, immunisations, and surveillance.

The researcher contacted health staff participating in the care of the child to gather further information on surveillance and immunisation and reviewed records held by health visitors and parents where available.

We had no sampling strategy; this was an unstudied and unknown population. We tried to see as many children as possible. Data entry and analysis ran concurrently with the study. Data collected after six months did not affect the overall outcomes, so data collection was stopped after 10 months.

The study was approved by the local research ethics committee.


During the study 257 children spent at least one night in a refuge. Assessments were completed for 148/257 (58%) children. Of the rest, 91 (35% of the total) children were resident for a very short time and the mothers of 18 (7% of the total) children failed to attend the appointment. No mothers refused to take part, although non-attendance could indicate refusal.

Twenty four children of 14 families resident at the start of the survey had been resident for between four and 361 days before assessment. (Refuge statistics for 1999 showed that 61% of the women entering a refuge run by BAWSO stayed more that one month, compared with 23% for Cardiff Women's Aid refuges; 49% of women entering Cardiff Women's Aid stayed fewer than seven nights. Statistics on short term residents were not kept by BAWSO.)

Table 1 shows the sex and age of the 257 children identified as living in refuges and their last address before coming into the refuge. Just under half were of preschool age. Most lived within 50 miles of Cardiff. The distribution of data on ages, sex, and last recorded address did not differ for the children who were assessed and those who were not assessed. Those not assessed were more likely to have been resident in a Cardiff Women's Aid refuge, suggesting a higher proportion of white Europeans among the those who were not assessed (assuming that most non-white women used the refuges run by BAWSO, which cater for ethic minority families).

Table 1

Personal characteristics, source, and destination of the 257 children passing through refuges for victims of domestic violence, Cardiff, April 1999 to January 2000. Values are numbers (percentages) of children

View this table:

Review of routine health records

Data from the child health system were available for 124 (84%) of the assessed children. Of these, 33 (27%) had either an incorrect address or no last address recorded, 80 (65%) had an incorrect or no general practitioner recorded, and 40 (32%) had both the incorrect address and incorrect general practitioner recorded. For those with a correct last address recorded (89), 48 mothers said they would not receive mail in the refuge as for their own safety they had not divulged their whereabouts. Consequently, 83 of 124 (67%) children would not receive any appointments sent by post.

Each refuge had a named health visitor who might visit once a fortnight and see any families who were present. These visits were ad hoc and unstructured. As no formal records were kept, no information was obtained from this source. The resources required for this work had never been estimated. All the named health visitors were attached to practices recognised to be under-resourced for health visiting, even without the refuge workload.

Table 2 summarises data on the uptake and outcome of health surveillance. The preschool and school age modules of the child health system were not added until 1988 and 1991 respectively. The totals reflect, therefore, those children who should have had each assessment by the time of the survey and whose age meant that they should have a record of that assessment on the child health system. In all, data on the birth visit was absent for 74/130 (57%) children and on the 18 month assessment for 40/105 (38%) children. Similarly, 37 of the 124 (30%) children for whom child health system data were available had failed to complete immunisations that were due by the date of assessment.

Table 2

Outcome of child health surveillance as recorded on child health system in all 148 children in study. Values are numbers (percentages) of children

View this table:

Developmental screening

Sixty eight of the 73 children aged 3-4 years were tested with the Denver developmental screen. Six (9%) of these children failed the screen, and a further seven (10%) had questionable results requiring review. These 13 children with problems were referred to appropriate agencies: three moved on to unknown destinations, three remained in a refuge (of which two were outside Cardiff), two returned home, and five were rehoused (though mostly not in their home area).

Behavioural assessment

Table 3 shows the results of the Rutter parents' scale administered to 101 children aged 3-15 years. Forty nine (49%) of the children scored >10. These results are presented in more detail elsewhere.11

Table 3

Rutter scores by age, sex, and ethnic group for children aged 3-15*

View this table:

Maternal concerns and referrals to health and social services

The mothers of 113 (76%) children identified at least one concern, and some identified several (table 4). Overall, 237 concerns were expressed, of which 67 related to the child's physical wellbeing, 129 to emotional and behavioural wellbeing, and 41 to management of the child (feeding, sleeping, and toileting).

Table 4

Concerns of 113 mothers about their children

View this table:

Referrals were made only with consent (excepting child protection referrals). Seventy three referrals were made for 65 children. Thirty six children were referred to a health visitor; nine to a paediatrician; eight to social services; six to a school nurse; four to a family therapist; three each to child protection, audiology, and enuresis services; and one to a dietitian.

After leaving refuge

Only 37 of 148 (25%) assessed children were in permanent accommodation at the end of the study. Of the remaining 111 children, 37 returned home (36 to the perpetrator), 24 remained in a refuge in Cardiff, 12 moved to refuges outside Cardiff, 7 were in “second stage” refuge accommodation (half-way house), 9 were staying with friends or extended family, and 22 were untraceable (the mothers of 6 had been evicted for behaviour that put other residents at risk—for example, allowing a partner into the refuge).


We have shown a high level of health needs in refuge populations, as well as poor access to child health services. Child health surveillance should be accessible to all children. Evidence of benefit is clearly set out in Health for All;12 indeed the Royal College of Paediatrics and Child Health recommend uptake of surveillance and uptake of immunisation as outcome measures.13

Failings in the system

The child health system is designed to facilitate implementation of the immunisation and developmental surveillance programmes, as well as other health services for children—for example, school health services and services for children in need. In our study many children in refuges were not identifiable or accessible via the child health system. In effect, these children form a largely invisible population, outside the health system and poorly served by it. This may be blamed on the “chaotic” lifestyle of these families or on a current system that fails these and other highly mobile vulnerable families. We did not explore the reasons for the poor quality of data from the child health system. The poor quality may result from system failure (failure of transfer of registration details from the general practitioner to the health authority or from the health authority to the child health system) or failure by families either to register with a general practitioner or to provide information on change of status and address. Fear and experience of discrimination, as well as mobility, may be important factors. Although mothers are encouraged to register with local general practitioners, many reported being told by practices to return and register only if their child needed an appointment. Other mothers opted to remain with their previous practices but not to inform them of their change of address.

Children in refuges are at risk from preventable infections and the late diagnosis of remediable developmental and other abnormalities. Although our figures do not suggest an excess of children with developmental problems, as do American surveys,8 access to services for assessment and therapy is poor. Many mothers request, for their safety, that previous records are not sought. Families drift into other temporary accommodation, move to other refuges, or are rehoused far from their original address. We are concerned that, without dedicated services to which these families can be referred, many fall through the system. Nearly half of the assessed children required referral to another agency or health professional. In the absence of this project, few of these would have been identified.

Mental health

The prevalence of mental health difficulties in this population was 48% compared with estimates of 10-26% in the general school age population. 14 15 Although our findings may not accurately reflect the impact of domestic violence on the psychological wellbeing of children generally, they support the argument that domestic violence is harmful and thus should be seen as a form of child abuse.7 Indeed a 1999 report by the Department of Health encouraged Area Child Protection Committees to work with local domestic violence groups.16

New strategies needed

Child and adolescent mental health services need to work with refuge organisations to develop strategies to meet the needs of children whose mothers are victims of domestic violence. Currently families are supported by non-specialist refuge workers with extremely limited access to advice and referral. A tier 3 level of service—as defined in the Department of Health's report Together We Stand17—which provides training and support for staff in direct contact with mothers and children may offer a solution. Supported refuge staff may then be better able to provide tier 1 services to families in crisis. They are well placed to follow through referrals to tier 2 services if necessary, when children are settled into permanent accommodation.

The English primary care trusts and their equivalents in Wales and Scotland have a duty of care to families in their local refuges. The time that children spend in refuges provides a window of opportunity to review their health, identify problems, and begin the process of investigation, treatment, and care. The numbers passing through refuges are about half of those in care, on whom a large amount of money is currently being spent in England and Wales. 18 19 Children in refuges also deserve investment and targeted services. Appropriately trained and skilled specialist health visitors could provide support and advocacy and facilitate access to mainstream services.


We thank the refuge staff, especially Margaret Smith (BAWSO) and Cathy Thomas (Cardiff Women's Aid); Alan Davies and Julie Allaway in the Cardiff Directorate of Child and Adolescent Health; the participating families; and John Clark for help with data analysis.

Contributors: EW initiated the research, participated in the study design and data analysis, jointly coordinated the study, and wrote the paper. JS developed partnerships with the collaborating voluntary agencies, participated in the study design and data analysis, collected and entered the data, and wrote the paper. MRE participated in the study design and analysis and edited the paper. RB established partnerships with the collaborating voluntary agencies, participated in the study design and data analysis, jointly coordinated the study, and edited the paper. EW will act as guarantor for the paper.


  • Funding The Wales Office for Research and Development in Health and Social Care (study number R/98/027) and the Sanoffi-Winthrop Foundation.

  • Competing interests None declared.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
View Abstract