Risk factors associated with primary care-reported domestic violence for women involved in Family Law care proceedings

This study provides information on risk factors associated with GP-recorded DVA


Background
Detrimental impacts of domestic violence and abuse (DVA) are felt by those directly exposed, as well as other family members; the World Health Organisation has classed the effect of exposure to DVA on children's health and wellbeing as a major public health concern 2 .
DVA is commonly underreported and prevalence estimates are wide ranging due to varying methods of measurement and data sources.As an example, the Office for National Statistics (ONS) Crime Survey for England and Wales estimated a one year prevalence (for the period ending March 2020) of 7.3%, i.e. 7.3% of all women surveyed aged 16 to 74 reported DVA within the previous year, and 27.6% reported DVA since the age of 16 3 .Other examples include a survey reporting that 17% of women attending their GP reported DVA; and studies using administrative GP data reported that less than 0.5% of women had records to indicate DVA 4 .More at risk groups such as mothers involved in family court proceedings experience higher DVA prevalence, with two studies reporting that 51% 5 and 65% 6 respectively of women reported DVA.
Where a child is identified as having suffered, or is at risk of suffering, significant harm (including bearing witness, or exposure to DVA) at the hands of a parent or caregiver, a local authority can initiate Family Court proceedings under section 31 (s.31) of the Children Act (1989).The Family Court appoints experts at Cafcass/Cafcass Cymru (Children and Family Court Advisory and Support Service) to ensure court decisions are made in a child's best interest.In Wales, administrative data collected by Cafcass Cymru includes details of court proceedings, and of individuals involved in proceedings; this data is made available in anonymous format in the SAIL Databank (Secure Anonymised Information Linkage Databank).SAIL is a privacyprotecting research environment holding administrative data sources relating to the population of Wales.
The National Institute for Health and Care Excellence (NICE) publish public health DVA guidelines, and report on established risk factors 7 .However there is a lack of evidence relating to DVA risk factors for more at risk groups, such as those involved in the family justice system.As such, this study investigated the prevalence of, and risk factors for, primary care recorded DVA for mothers involved in public law court proceedings across Wales.

What we did
This research linked anonymised Cafcass Cymru, health, and demographic data in the SAIL Databank.Our study cohort included mothers (with children aged under 18) involved in public law family court proceedings between 2011 and 2019 (n=4,695), as well as a group of comparison mothers from the general population not involved in public law family court proceedings, matched on key demographics (age and deprivation) (n=233,171).All individuals had GP (general practitioner) data available in the SAIL Databank.
We used published clinical Read codes 8 (a vocabulary used by clinicians to record patient findings) to identify mothers with exposure to DVA, as documented in their primary care (GP) records.We are mindful that the data we have used only captures DVA disclosed by mothers and recorded by a GP, and the scale of the problem may therefore be larger.The DVA outcome and risk factors were measured across a 2-year period up to the 'index' date (initial court proceedings date for cohort mothers).
There were two stages to the research.Initial analysis included all mothers (cohort mothers and comparison mothers) and examined six NICE risk factors as well as an additional risk factor describing whether a mother was involved in family court proceedings or not.The NICE risk factors included were: • aged under 25 years • intellectual disability • mental health conditions • number of house-moves • recent pregnancy or childbirth • substance misuse conditions.
The secondary analysis was restricted to cohort mothers only, and further risk factors were selected (chosen based on academic literature and data availability) and included: • living in a sparsely populated area • assault related Emergency Department (ED) attendances • having more than one child • having a child aged under 5 years old.
Inclusion of ED assault attendances as a risk factor may seem circular in nature (i.e., there is a link between assaults and DVA).However, "assault-related ED attendances" do not specify a perpetrator or a location (i.e., it is unknown whether the assault was due to DVA), and GPs do not have access to ED data, therefore the associations are not circular.
For both sub-analyses, univariable and multivariable logistic regression analyses were used to examine risk factors associated with GP reported DVA.We reported unadjusted odds ratios (OR) and adjusted odds ratios (AOR).

What we found
The findings revealed that: Just under six percent (5.7%; n=269) of cohort mothers, and 0.3% (n=626) of the comparison group mothers had GP data indicating exposure to DVA.In the final model we analysed a selection (based on unadjusted results) of the risk factors together (adjusted results) to account for the combined associations between multiple risk factors and DVA.We examined relationships, or associations between risk factors (such as mental health, age of the mother, number of children) and the GP-reported DVA.Initially we analysed each risk factor separately (unadjusted results) to understand the strength of the relationships and to inform which risk factors should be included in a final statistical model.In the final model we analysed all these risk factors together (adjusted results) to account for the combined associations between multiple risk factors and DVA.Unadjusted and adjusted results (shown in Figure 1) are reported as Odds Ratios (OR), which are a measure of association between the risk factors and GP-reported DVA.The risk factor with the largest adjusted association with GP-reported DVA was 'living in sparsely populated areas' with an adjusted OR of 3.9, or close to a fourfold increase in risk after accounting for other risk factors.Assault-related emergency department attendances, and mental health conditions were both associated with a close to two-fold increased risk, with adjusted odds ratios of 2.2 and 1.7 respectively.
The first stage of analysis included all mothers, and the unadjusted and adjusted results (taking all other risk factors into account) are shown in Table 1.Following adjustment, mothers with family court proceedings involvement were eight times more likely to have had DVA documented in primary care records.Four of the included NICE risk factors were statistically significant (aged under 25 years, mental health conditions, house-moves, and substance misuse conditions).

Why it matters
This study provides information on risk factors associated with GP-recorded DVA for mothers involved in public family law proceedings.This is compared with a general population comparison group and highlights an 8-fold increased risk.
Evidence of the association between having a mental health condition and experiencing DVA should help to further corroborate the necessity for continued and additional resources targeted at specialist mental health and DVA support services.The evidence that living in sparsely populated areas and assault-related emergency department attendances are associated with increased risk of DVA could be used to inform policy and practice interventions targeting prevention.It could also help tailor support services for those with exposure to DVA living in rural areas.
Parents may appear in public law family court proceedings under s.31 orders where the court concludes that a child is suffering or likely to suffer significant harm.Proving significant harm is complex, and decisions may be made based on many factors.One of these includes a child witnessing DVA between adults at home, but this is not the only reason a parent could appear in court under such proceedings.It is therefore of no surprise that findings show that mothers involved in family court proceedings have higher rates of GP reported DVA than mothers in the general population.However, research evidence in this area is scarce.This is the first study at large scale to examine whether DVA is recorded in primary care records especially for those involved in public law family court proceedings, and to use administrative data to investigate risk factors.Moreover, this evidence perhaps indicates or reinforces the importance of the role of primary care in responding to DVA as well as capturing incidents in data, which is essential for understanding the scope and scale of the problem.

What next?
To further understand the scale of the problem, other sources of DVA could be examined, such as that recorded in secondary health care, family, police, and criminal justice records.Further, DVA experienced by fathers, or other household members, also needs to be examined.

Limitations
Caution should be exercised when interpreting results, as it is important to consider underlying data creation mechanisms that may affect the results.There is widely acknowledged DVA underreporting, which is attributed largely to the sensitive and secretive nature of the issue and a hesitancy of victims to disclose the situation to outside authorities.The Crime Survey for England and Wales (CSEW) estimated less than 20% of women experiencing DVA in 2017-18 reported it to the police 9 ; data linkage studies, or those using GP data report higher underreporting compared with CSEW 8,10 .The 2-year GP-reported DVA prevalence findings reported here are therefore considered an underestimate of the true prevalence of lived experience DVA.
This study recreated NICE risk factors using administrative data sources and as such there is not a perfect alignment of definitions between them.Mapping decisions were made based on careful consideration of administrative data records and coding descriptions and a pragmatism with regards to data availability.Future work could seek to validate these methods and to develop a robust taxonomy of administrative data sources and definitions aligned to the risk factors as described by NICE.
Readers are encouraged to refer to the main study for additional detail on methods and limitations 1 .

Figure 1 -
Figure 1 -Unadjusted and adjusted odds ratios and 95% CIs for risk factors tested for associations with GP reported DVA for mothers involved in family court proceedings (study cohort mothers).

Table 1 -
Unadjusted and adjusted odds ratios (and 95% CIs) for seven risk factors (Family Court Involvement, and six NICE guideline risk factors) for the risk of GP-recorded DVA; includes all mothers (cohort and comparator mothers combined).