Article Text

Download PDFPDF

Rates and types of hospitalisations for children who have subsequent contact with the child protection system: a population based case-control study
  1. M O'Donnell1,
  2. N Nassar1,
  3. H Leonard1,
  4. P Jacoby1,
  5. R Mathews2,
  6. Y Patterson3,
  7. F Stanley1
  1. 1Telethon Institute for Child Health Research, Centre for Child Health Research, University of Western Australia, Perth, Western Australia
  2. 2Disability Services Commission, Western Australian Government, Perth, Western Australia, Australia
  3. 3Department for Education, Western Australian Government, Perth, Western Australia, Australia
  1. Correspondence to Melissa O'Donnell, Telethon Institute for Child Health Research, 100 Roberts Road, Subiaco, Perth, Western Australia 6008, Australia; melissao{at}ichr.uwa.edu.au

Abstract

Objectives To determine whether children who have child maltreatment allegation or substantiation have a higher rate of general hospital admissions and injury related admissions when compared to other children and to investigate other types of admissions, such as mental health, infections and admissions due to external causes.

Study design A prospective matched case-control study of children born in Western Australia between 1990 and 2005 using de-identified record linked Child Protection and Hospital Morbidity data. Rates of prior hospital admissions for cases versus controls were calculated, and conditional logistic regression was used to estimate the effect of hospital admission rate on the risk of child maltreatment allegation and substantiated allegation.

Results Children with child maltreatment allegations and substantiations had higher mean prior admission rates compared to controls. Higher rates of general admissions and admissions for injuries, infections, mental and behavioural disorders, and external causes of morbidity, were associated with a markedly increased risk of child maltreatment allegations and substantiation.

Conclusions The hospital system plays not only an important role both in the surveillance of maltreatment-related injuries and conditions but also in the role of prevention in the referral of families who may need support and assistance in ensuring the health and safety of their children. This research highlights the importance of moving to electronic patient records in identifying children who have high rates of admissions and the types of conditions they have previously presented with, particularly for injuries, mental and behavioural disorders and external causes of admissions.

  • Hospital morbidity
  • child abuse
  • child health
  • children
  • injury
  • violence RB

Statistics from Altmetric.com

With the increasing attention being placed on interagency involvement in the identification, intervention and prevention of child abuse and neglect, it is important to obtain a clear picture of the prior contact that children who are maltreated have with different agencies. The health system is integral in the identification of child maltreatment cases as a large proportion of substantiated child maltreatment occur in children aged 0–4 years, prior to them entering the education system.1 In addition, injuries are the leading cause of death and the third leading cause of hospitalisation in children; therefore, health professionals have the task of identifying maltreatment-related injuries and conditions and play an important role in child maltreatment surveillance.2 Therefore, it is important that patterns of healthcare utilisation are investigated to assist with this surveillance and intervention role.3

There is a small but growing number of studies which have linked health and child protection data to investigate patterns of prior healthcare use.4–6 Previous studies have found higher emergency department (ED) use prior to abuse diagnosis,5 and in a more recent study by Spivey et al,6 children with two injury related ED visits in a year were twice as likely to have a child maltreatment report and a substantiated report. These studies have provided important findings but were limited either by small samples4 or short duration of follow-up data.6

Through Western Australian government total population data linkage, it is possible to further investigate the prior hospital morbidity for all children who have contact with child protection compared with all those who have not. We hypothesised that children who have child maltreatment allegation or substantiation would have a higher rate of general hospital admissions and injury-related admissions. We also wanted to investigate other types of admissions, such as those for mental health, infections and admissions due to external causes.

Method

We conducted a prospective matched case-control study of children born in Western Australia from 1990 to 2005 using de-identified routinely collected information linked across datasets. The health data included Midwives Notifications, Birth Registrations, Death Registrations, Hospital Morbidity and Mental Health. The Midwives Notifications and Birth Registrations include information on maternal characteristics, pregnancy conditions, delivery details and infant outcomes for all births from 1990 to 2005. The Death register contains information on all children who have died in Western Australia from 1990 to 2005. The Hospital Morbidity data collection contains information on all hospital admissions (public and private) with corresponding diagnostic and procedural information using the International Classification of Diseases (ICD) coding system recorded for each episode of care for children from 1990 to 2005. All ICD codes were recoded to ICD-10 (Australian Modification). The Mental Health data contains information on mental health in-patient admissions for children from 1990 to 2005.

The Child Protection data collection includes information on Child Maltreatment Allegations and Substantiations from 1990 to 2005. Child Maltreatment Allegations consist of reports made to the Department for Child Protection regarding allegations of child abuse, neglect or harm to a child. An allegation is substantiated when following investigation “there is reasonable cause to believe the child has been, is being, or is likely to be abused or neglected or otherwise harmed”.1 Therefore, a child would receive an allegation prior to receiving a substantiated allegation resulting in overlap between the two groups of cases.

The Index of Relative Social Disadvantage, as defined by the Australian Bureau of Statistics, was used to determine the level of socioeconomic disadvantage in the census collection district (approximately 250 households) relating to the mother's residence at the time of the infant's birth.7 The Index of Relative Social Disadvantage, sex, maternal age, marital status and Aboriginality were obtained from the Birth Registrations and Midwives Notifications supplemented by information from the Child Protection demographic data for unknown Aboriginal status.

Each of the datasets is linked by the Western Australian Data Linkage Unit by matching the identifiers common to the sets of records such as name, address, sex and other available data.8 Only a unique project number and the individual's clinical information are provided to the researcher, and all identifying information, such as name and address information, are removed.

Study design

The population cohort comprised all children born in Western Australia between 1990 and 2005. Two study groups were selected, each with a set of cases and associated controls selected from the same population. The cases for the first study group were children who had child maltreatment allegation as reported by the Department for Child Protection from 1990 to 2005. The cases for the second group were children who had substantiated child maltreatment allegation from 1990 to 2005. Four controls were selected for each case using random sampling with replacement. Controls were matched to cases on birth year, and only controls that had not died nor had a child maltreatment allegation or substantiation during the follow-up time were eligible. Follow-up time was calculated from the 1st January of the child's birth year to the case child's maltreatment date. The number of hospitalisations, both overall and for specific admission types, prior to the date of allegation or substantiation was recorded for each case and its matched controls (excluding birth admissions).

Analyses

Demographics of cases and control participants were described using frequencies and percentages. Admission rates were calculated as number of admissions divided by follow-up time per year. ICD chapter groupings were used to investigate specific groups of admissions for mental health, external causes and injuries.

Conditional logistic regression analysis was used to estimate the effect of hospital admission rate on the risk of child maltreatment allegation after adjusting for sex, Aboriginality, maternal age, marital status and socioeconomic disadvantage at the time of birth. This was also done for the second study group for risk of child maltreatment substantiations. These results were presented using ORs and 95% CIs. For a prospective case-control design, the OR provides an estimate of the incidence rate ratio or RR and can be interpreted as the relative increase in risk of a child maltreatment allegation/substantiation for every one admission per year of life.9 SAS release V. 9.1 (SAS Institute) was used to conduct all analyses.

Ethics

This study has ethics approval from the University of Western Australia Human Ethics Committee, the Confidentiality of Health Information Committee and the Western Australian Aboriginal Human Information and Ethics Committee.

Results

There were 397 346 live births in Western Australia between 1990 and 2005. Of these, the first study group contained 13 648 children who received child maltreatment allegation, and the second study group contained 6486 children who received substantiated allegation. In both groups, compared with controls, cases were more likely to be Aboriginal and borne to mothers aged <20 years, of single marital status and living in areas of greater socioeconomic disadvantage (table 1).

Table 1

Demographics of case and control participants for the two study groups

Almost two thirds (61%) of cases and 40% of controls had at least one admission, with cases having a higher mean admission rate (0.7/year) compared to controls (0.2/year; table 2). In addition, 36% of children with an allegation and 40% of children with a substantiated allegation had two or more admissions, while 18% of controls had two or more admissions.

Table 2

Mean admission rates for cases and controls for both study groups

After adjusting for potential risk factors, the risk of a child maltreatment allegation increased by a factor of 1.49 (CI: 1.44 to 1.53) for every one admission per year. Similarly in the second study group, the risk for a substantiated allegation increased by a factor of 1.74 (CI: 1.65 to 1.83) for every one admission per year (table 3).

Table 3

Multivariate ORs (95% CIs) for admission rates

Higher rates of admissions for injuries, infections, mental and behavioural disorders and external causes of morbidity were associated with a marked increase in risk of child maltreatment allegations and substantiation (table 3). For infectious and parasitic disease admissions, for every unit increase in admission rate, there was a 2.6-fold increase in risk of child maltreatment allegation and a threefold increase in substantiated allegation. For admissions related to mental and behavioural disorder, there was a 10-fold increase in risk of child maltreatment allegation and a 26-fold increase in substantiated allegation. Similarly for injuries, with every unit increase in admission rate, there was a 10-fold increase in risk of allegation and a 21-fold increase in substantiated allegations, with external causes of admission having a eightfold increase in child maltreatment allegation and an 18-fold increase in substantiated allegation. Demographic characteristics associated with both child maltreatment allegations and substantiated allegations were being women, Aboriginal and children borne to mothers of single marital status, younger maternal age and of greater socioeconomic disadvantage (table 3).

Admissions for infections were predominantly for bacteria and viral agents, intestinal infectious diseases and pediculosis, acariasis and other infestations, while admissions for mental and behavioural disorders were mostly related to behavioural and emotional disorders with onset usually occurring in childhood and adolescence, disorders of psychological development and neurotic, stress-related and somatoform disorders. The most common types of injuries resulting in an admission were for head and neck, shoulder and arm, hip and leg, multiple body regions and other unspecified effects of external causes. External causes of admission included assault and other external causes of accidental injury.

Discussion

The main finding from this study is that children who have higher overall rates of hospital admissions prior to contact with child protection are at greater risk of child maltreatment allegations and substantiated allegations. In addition, risks of allegations and substantiations were also elevated for higher admission rates related to infections, injuries, external causes of morbidity and mental and behavioural disorders.

Our longitudinal study confirms and extends upon findings from the recent 2-year surveillance study,6 which found that children with a higher rate of injury-related ED visits were more likely to be reported and to have a substantiated maltreatment report. Our results also confirm, as expected, that children who have a higher rate of assault-related admissions will have a high risk of substantiated maltreatment. Our study also shows that children who have maltreatment allegations and substantiations have higher rates of infection admissions. There could be two potential reasons for this: first, that children who are later identified as maltreated become ill due to the home environment (eg, diet, sanitation, stress), or second, that children who are ill require more attention causing stress in homes and increasing vulnerability to child maltreatment.10

Children who received maltreatment allegations and substantiations also had higher rates of admissions for behavioural and emotional disorders with onset usually occurring in childhood and adolescence, and disorders of psychological development. There is a wealth of research that indicates that children who experience child maltreatment are at risk of mental health and adjustment problems.11 However, our findings suggest that psychological indicators of potential family problems were likely to be evident prior to any allegation of child maltreatment. Recent research has indicated that for children who have experienced physical abuse their mental health outcomes are strongly linked with the general family context in which the abuse occurred.12 Therefore, the elevated levels of mental health problems prior to any allegations may be an early indication that maltreatment was occurring or as a result of the social and family context of the child.

The strength of this study is that we used de-identified population level databases, linked across health and child welfare, to investigate the relationship between prior hospital morbidity and child protection outcomes without bias of participant participation and recall, and with acceptable statistical power. Researching issues as sensitive as hospital admissions for injuries and conditions and child protection involvement is a challenge, but the use of de-identified population level data allows us to investigate these events while protecting the privacy of research participants.

Using a matched case-control study has ensured that we take into account the age of children and equal follow-up time for the cases and controls in regard to their hospital admissions. Another strength is that we investigated hospital admissions in relation to all reported as well as substantiated maltreatment allegations. Spivey and colleagues6 argue that substantiated reports may not be valid or reliable as not all allegations are able to be investigated. They also argue that children with substantiated and unsubstantiated allegations have similar outcomes; therefore, “substantiation may be a distinction without a difference” (p2126). However, we also believe that if we are trying to prevent maltreatment, it is important to investigate the prior hospital morbidity for children who are suspected of being maltreated or where there are concerns for children's welfare; therefore, allegations and substantiations were included in this study.

While using linked administrative databases has many advantages, there are some limitations. Potential sources for under-ascertainment of types of admissions may arise with the use of diagnostic codes reported on the morbidity dataset to identify co-morbidities.13 Another is that child maltreatment is only identified by notification to and substantiation by child protection agencies, and there may be children who have been maltreated but not reported to child protection services, particularly as Western Australia did not have mandatory reporting by health professionals during the study period. It is also likely that certain groups are over-represented in reporting of suspected child maltreatment, specifically Aboriginal children.14

Another limitation is that administrative databases are limited to those individuals who have accessed the hospital system which we recognise is a small proportion of the potential health contacts for children (eg, emergency department and general practitioner contacts). In addition, the administrative data only have area-based estimates of socioeconomic disadvantage (which equates to approximately 200 households), which is less representative than individual-based measures, although in population-based research, these have been shown to be reliable but may potentially understate disadvantage.15 As we only had birth year, the children could not be matched on month or date of birth; however, this was consistent across cases and controls. Finally, while we ensured that children in our study groups did not die before their follow-up date, it is possible that some children may have left the state following birth, and this may have contributed to the reduced rate of admissions for control children. However, migration analyses in Western Australia show that this could potentially only account for 2% of the study group.16 Future research should investigate other health contacts such as general practitioner or health clinic contacts, prior to contact with child protection.

Despite these limitations, our research provides strong evidence that children who are maltreated experienced higher rates of prior hospital admissions as well as admissions for specific conditions such as infections, injuries, external causes of injuries and mental health related admissions. This has important practical implications for the prevention and intervention of child maltreatment. We argue that electronic medical records could have the potential to assist hospitals in identifying children who have high rates of admissions, particularly for injuries and external causes.6 This could assist not only in recognising children who are at risk of supervisory neglect and maltreatment but also, as Spivey et al6 have suggested, in providing discharge supervision guidelines for parents appropriate for children's developmental age to reduce the incidence of accidental injuries. It also may warrant hospital social worker contact to determine if families need ongoing assistance with referral to support agencies or, in the case of concerns of maltreatment, reports made to child protection agencies.

In conclusion, children who have higher rates of general admissions are at higher risk of receiving child maltreatment allegations and substantiated allegations, and in particular, children who have higher rates of admissions for infections, injuries, external causes of morbidity and mental and behavioural disorders are at a much greater risk of allegations and substantiated maltreatment. The hospital system plays an important role both in the surveillance of maltreatment-related injuries and conditions and in the role of prevention in the referral of families who may need support and assistance in ensuring the health and safety of their children (ie, secondary prevention3). All countries who are signatories to the “United Nations Convention on the Rights of the Child” have agreed to protect children's rights through “assisting parents in the performance of their child rearing responsibilities”. This highlights the need to improve our understanding of areas where we can identify and assist at-risk families to prevent child abuse and neglect.

What is already known on this subject

Previous research has found higher emergency department (ED) use prior to diagnosed child abuse and that children with a higher number of injury-related ED visits were more likely to have a substantiated child maltreatment report. These studies have provided important findings but were limited either by small samples or short duration of follow-up data.

What this study adds

This research used longitudinal population-level-linked health and child protection data to investigate prior hospital admission rates on the risk of child maltreatment allegation and substantiated allegation. This paper found that higher rates of overall admissions and admissions for injuries, infections, mental and behavioural disorders and external causes of morbidity were associated with a markedly increased risk of child maltreatment allegations and substantiation. This highlights the important role hospitals play in both the surveillance of maltreatment-related injuries and conditions and in the role of prevention, in the referral of families who may need support and assistance in ensuring the health and safety of their children.

Acknowledgments

The authors acknowledge the partnership of the Western Australian government Departments of Health, Child Protection, Education, Disability Services, Corrective Services and Attorney General who provided support as well as data for this project. This paper does not necessarily reflect the views of the government departments involved in this research. We would also like to acknowledge the Western Australian Data Linkage Unit who linked this data. Melissa O'Donnell acknowledges the support of an Australian Postgraduate Award Industry Scholarship, provided through an Australian Research Council Linkage Project Grant (LP0455417). Dr Nassar is supported by a Public Health Fellowship (404198) from the National Health and Medical Research Council of Australia.

References

Footnotes

  • Funding Australian Research CouncilWestern Australian Government. Melissa O'Donnell was supported by an Australian Postgraduate Award Industry Scholarship, provided through an Australian Research Council Linkage Project Grant (LP0455417). The Western Australian Government Departments of Health, Child Protection, Education, Disability Services, Corrective Services and Attorney General who provided support as well as data for this project. Dr Nassar is supported by a Public Health Fellowship (404198) from the National Health and Medical Research Council of Australia.

  • Competing interests None.

  • Ethics approval This study was conducted with the approval of the University of Western Australia Human Ethics Committee Confidentiality of Health Information Committee Western Australian Aboriginal Human Information and Ethics Committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.