The Effectiveness of Child Injury Database and Registry in Identifying Children Who are Potentially at Risk of Abuse: A Systematic Review

Introduction

A systematic review of existing official databases that are available internationally and locally on injuries (focusing mainly on children) and evaluating their effectiveness in identifying children at risk of neglect or abuse was conducted. This systematic review addressed the broad research question: What is the availability of child injury databases and what is their effectiveness in identifying children potentially at risk of neglect or abuse. The review aimed to identify, appraise, select and synthesize all high-quality research evidence relevant to the above question [1]. Therefore, the aims of this review were to:

• Systematically review trauma and injury databases and/or Child Protection Registries that include children ages 18 years and below that are available globally;
• Determine whether these databases and registries are effective in identifying children who are potentially at risk of abuse within the community.

The systematic review process employs literature review methods to select only those studies that meet specific criteria which reasonably confirm the rigour of the ‘evidence’ produced by a previously published study. The primary characteristic of a systematic review is that it uses a rigorous set of criteria by which to appraise the reliability and validity of previously published research [2].

According to the Merriam-Webster dictionary, for the purpose of this study, a database was defined as ''a large collection of data in a computer, organized so that it can be expanded, updated, and retrieved rapidly for various uses or known as an online database” [3]. Therefore, health databases can be in an electronic form which can comprise computer patient records (CPRs) or electronic medical records (EMRs) or disease registries or in can be in a paper-based form (hard copy).

These electronic records may be accessible in the organisation with multiple levels of user-dependent authentication [4]. As the system provides a seamless and integrated patient-specific medical history, it facilitates healthcare professionals to make better diagnosis and provide quality care services [5].

In Malaysia, there are currently 21 patient registries which covered important chronic diseases (such as diabetes, cardiovascular disease, cancer, etc.) [6]. Patient registries are considered to be important sources of information as they provide information important in the clinical setting such as; i) the natural history of a disease, ii) the clinical or cost effectiveness of specified health care products/services, iii) measuring or monitoring safety and harm of products and services; and iv) measuring the quality of care provided for the patient [6]. At least three- quarter of these registries are fully funded by the Government whereas only three are funded either by private foundations/organizations or through public or individual donations. As such, these registries are not hospital-based registries.

There is evidence to suggest that many childhood abuse-related deaths are not identified as abuse prior to their deaths, further stressing the importance of early detection [7, 8]. A registry should be an integral part of any modern surgical service. With appropriate patient privacy safeguards in place, these data can provide valuable guides for evaluation and treatment of patients at the individual level, and for program design and allocation of resources at the system level. As mentioned in the previous chapter, a registry can complement the standard indicator of child abuse and neglect (CAN) by raising the suspicion index and increasing detection rate in the clinical setting [9, 10].

The next sections describe the search strategy, selection criteria, findings and a comparison between the reviews.

Methods of Review

A total of 13,170 articles were yielded from PubMed (730), EMBASE (730), PsychInfo (11,633) and ProQuest (77) after applying the filters in the search. All articles were screened by title and abstract. A total of six articles from PubMed, three articles from EMBASE, and six articles from ProQuest were selected for evaluation and full texts were obtained and downloaded and saved in Zotero online (no articles were sourced from PsychInfo). These articles were combination of cumulative articles based on separate MesH headings used in the Search Strategy section.

Click to enlarge

The other articles were excluded as they were not related to the research question which addressed determining the effectiveness of child injury database/register in identifying children who are potentially at risk of abuse. At the conclusion of the selection process, 12 articles were chosen for review after excluding repeated articles. Figure 1 shows the process of extracting, reviewing and excluding the non-relevant literature.

Database used / Targeted Population/Country

No Journal title Author

Database: 1999-2002 1. Vital birth records (California Department of Health) 2. Child Protective Services data Population: Children up to the age of 5 n=293,441 Country: US A public health approach to child maltreatment surveillance: Evidence from a data linkage project in the United States Putnam- Hornstein E, Webster D, Needell B, Magruder J.

1

Result

Several themes emerged and were identified across all 12 articles through analysis of injury databases to determine their effectiveness in identifying children at risk of abuse. A summary of articles included in the systematic review is shown in Table 1 below.

Effectiveness of

database in identifying child Summary – Important points Themes

abuse

1. Risk factors of maltreatment by the age of five - Poverty (22% of children covered by Medi-Cal were reported) - Maternal education (~20% of children born to a mother with less than a high school degree was reported). - Children in single-parent families (about 1/3 of children with missing paternity establishment). - Children with a health risk at birth (Nearly 18% identified as possible victims of maltreatment).
2. Almost 1/3 children born to black mothers in California had been reported to child protective services during the first five years of life.
3. Information gained from these record linkages provides - Unavailable information about the characteristics of children reported to child protective services. - Identification of very high-risk group of children who could be targeted for services from the day of birth.

Effective. Children potentially at

risk of abuse could be identified based on personal, familial and Risk Factors

socio-demographic characteristics at birth.

1997, n = 4327; 2000, n = 4305; 2003, n = 4409; 2006, n = 4473; 2009, n = 4782 Country: US Western Australian emergency department presentations related to child maltreatment and intentional injury: population level study utilising linked health and child protection data.

Database:

2001 - 2005 Western

Australia Emergency Department data Collection (EDDC): Population:

All children aged 0-17

$$ n = 6 5 7 6 5 6 $$ O'Donnell M, Nassar N, Jacoby P, Stanley F.

9 Database: January 1, 2001, and December 31, 2002, Utah Division of Child and Family Services (DCFS), birth certificate, death certificate and emergency department state-wide databases Population:

Prediction of child abuse risk from emergency department use.

Guenther E, Knight S, Olson LM, Dean JM, Keenan HT.

10 physical abuse based on data from child protective services. 3. Challenge of using a single source of data to track a complex problem such as child physical abuse.

injuries in medical documentation of child abuse over time.

This might impede efforts in comparing the evolution of data throughout the years.

2.KIDs cover all pediatric conditions rather than only focus on child abuse.

1. A large proportion of females had ED presentations related to maltreatment and examination for assault.
2. Children aged less than 1 year and over 12 years made up a large proportion of maltreatment-related ED presentations as well as children from socio- economically disadvantaged backgrounds.
3. Large proportion of Aboriginal children had maltreatment (24%) and intentional injury (16%) presentations compared with the proportion of Aboriginal children in the WA population aged 0–18 years (approximately 6%).

Fairly effective. Several risk factors

were identified for abused children. However, local injury Presenta- tion at the Emergency Department codes were used instead of ICD-10. Thus, data could not be compared internationally.

1. Children with supported child abuse have a higher use of the emergency department, before diagnosis of abuse, when compared with the general pediatric population. 2. Neither the frequency of emergency department uses nor the pattern of diagnoses offers sufficient specificity to be useful markers of future abuse.

Not effective. Frequency of

Presenta- tion at the Emergency Department

emergency department or pattern of diagnosis unable to predict risk of abuse.

Children <13 years of age - Cases (n = 9795) - Control (n = 9795) n=884 Country: US Database: 2006 1. Two child protection registers in the Shropshire area 2. Computerized records of ED attendances Population: Children aged 0-12 years - Cases (n = 220) - Control (n = 1950) Country: UK Emergency department attendance by children at risk of abuse.

Leaman AM, Holt A, Ramakrishnan RG.

11 Database: 1998 - 2008 Pediatric injury registry of emergency department in Umea˚ University Hospital Population: Children aged 0–19 years, n=33,781 Country: Northern Sweden Injuries in children and adolescents— Analysis of 41,330 injury related visits to an emergency department in northern Sweden.

1. In the 2 years before being placed on a child protection register, the ED and outpatient attendances of at-risk children are no greater than those of an age- matched control group of children attending the ED for injury or illness. 2. Maltreated children had more frequent changes of healthcare provider.

Presenta- tion at the Emergency Department

Not effective. Frequency of ED visits are not an effective marker of child abuse.

1. The rate of injury related visits to the emergency department was 110/1000- person years.
2. Variations in type of injury, mechanism and activity at injury in different age group and sex reflected the hazards associated with different periods in growth.
3. 1039 visits resulted from interpersonal violence such as violence between peers, domestic violence and neglect, with crude rate of 28 (26–30)/10,000 per year.

Not effective. Many cases were

likely to be underreported due to Presenta- tion at the Emergency Department

reluctance to seek medical care and difference in description of event of injury at presentation.

A summary of the emerging themes from the review is shown in Figure 2.

Click to enlarge

Discussion

Clinical acumen, skill and diplomacy among healthcare professionals are needed in early recognition and management of child abuse [25, 26]. Failure in early detection of child abuse poses risks of continuous suffering to an innocent child that might lead to permanent disability or mortality [27]. Thus, comprehensive child injury databases are considered essential tools and resources to assist in detecting not only those who are abused or neglected, but to identify those at risk, and to inform the development of additional programs/resources to prevent child injury.

The first point of care for abused or neglected children is most often the clinical setting, specifically the Emergency Department (ED), in addition to general practitioners, the police and the Department of Social and Welfare (DSW). While it has been noted that cases of child abuse detected through ED could be underestimated as some would seek other healthcare services to evade detection by authorities, or hide their reason for visiting the ED [24, 28]. EDs are often the point where first-hand information is received either from the child or their parents/caretakers. However, many generic ED databases are minimum datasets and provide insufficient data to predict risk of abuse/neglect. Thus, information keyed into the database should be thorough and accurate, while acknowledging the intricacies of detecting child abuse/neglect solely from one visit. It is important for the attending doctor to obtain a good history from the child’s initial visit. Information such as risk factors, repeated admission, patterns of injuries, and concordance with other databases need to be taken into consideration.

The documentation of certain risk factors in databases or registries that should raise the index of suspicion for child abuse/neglect, and there are numerous international studies that have demonstrated that it is essential to examine and identify risk factors with the use of good data. For example, an initial examination of Child Protection Services (CPS) data showed that children of black ethnicity were more likely to present to CPS. This finding was is consistent with the findings of a prospective study in the United States (US), again, using inferior data [29]. With the use of more in-depth and robust data, and making appropriate adjustments for important factors such as socioeconomic deprivation, subsequent studies found that there were, in fact, no significant racial differences in reports of child maltreatment [30]. Poverty instead was a major risk factor in the causation of child abuse [31, 32, 33]. This could be explained by harsher parenting styles and also stress associated with socioeconomic deprivation or disadvantage [34, 35, 36, 37].

Similarly, single parenthood and having unwed mothers were identified as risk factors in children suspected of abuse [14]. This was correlated by other studies, but these studies also suggested that maternal employment may compromise quality care-giving environment, thus putting children at risk [34, 38]. Perinatal depression was also linked to increased rates of child abuse, supported by a Turkish study which found that the most common maternal psychiatric disorder leading to child abuse was depression (5.6%) [39, 40]. The relationship between Munchausen syndrome by proxy in mothers with perinatal depression and child abuse should also be noted [41]. With the availability of this maternal information when children present with an injury, children at risk could be identified and targeted for intervention [40]. Another risk factor is parental alcohol use, as there is a propensity for violence to occur [25, 33, 42, 43]. This is evidenced by the involvement of a majority (54.7%) of maternal alcohol abusers with the CPS, with 42 percent of children removed from their care [44]. Substance abuse is also another risk factor, where one or both parents use illegal substance. Children whose parents abuse substances are almost three times more likely to be abused and four times more likely to be neglected than other children [37, 45, 46].

In addition, non-nationality is associated with increased risk of child abuse. This is explained by the tensions due to differences in cultural orientation between parent and child, leading to intergenerational- intercultural conflict [47]. Moreover, the differences in punishment methods for example, corporal punishment in immigrant families has been shown to contribute to the higher prevalence of child physical abuse [48]. However, the paucity of available data regarding the child’s immigration status warrants further research to understand the associations between these factors and risk of abuse/neglect [47].

Apart from the above risk factors, repeated admission is another predictor of potential child abuse. Chronic child abuse and recurrence of abuse are not uncommon. Many children re-enter the CPS after the first incidence of abuse within two to three years. A National Child Abuse and Neglect Data System (NCANDS) study estimated that 22 percent of abused children were re-reported within 24 months [49]. Another study also reported that the recurrence rate of abuse was 47 percent at the end of 36 months and 62 percent by 7.5 years after initial referral [50]. Thus, repeated admission noticed from hospital admission database should raise the alarm about possible child abuse.

Furthermore, based on the evidence, children who present with certain characteristics of the children and the injuries should be assessed for risk of abuse. The distribution of abuse is skewed toward the younger age group, most probably due to their vulnerability and inability to flee [18, 37]. Also, presentation of more blunt injuries might be attributed to accessibility of blunt objects at home. Several studies commented that superficial injuries and bruising were found in abused children [51, 52, 53]. Injury from abuse is often located at the head region due to child’s small stature which makes the head the closest part to an adult’s hand, followed by the skin, which is perceived as a lesser vital organ to harm.

Finally, the current findings suggest that integration between health and welfare system data is crucial in early detection of child abuse and neglect. While neither system captures the cases completely, concordance of databases provides insight into the potentially undetected cases. Many supportive studies showed good linkage of hospital records with CPS data. An Australian study found nearly all children with hospital assigned abuse codes (93.1%) had registered under CPS (16). Another study in Missouri also demonstrated 88 percent concordance between child abuse hospitalization data with CPS data [54]. The number of non-concordant cases, however small, necessitates efforts in improving the linkage to achieve 100 percent concordance.

The present review has several limitations. First, as child injury databases were reviewed, the focus was only on detection of physical abuse instead of other forms of abuse such as neglect, sexual and emotional abuse, which can be equally harmful to the child. Second, the articles reviewed were mostly from developed countries which had more advanced health systems and thorough child injury databases. Thus, the findings might not be representative of the global population of child injury cases. Third, since articles reviewed came from different countries, different diagnostic criteria of injury were applied. Although International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes were commonly used in diagnosis of child abuse from injury, the accuracy might vary. A previous study has shown variability in diagnosis of abuse when the same child abuse case was reviewed by physicians [55]. Fourth, the exclusion of grey literature might omit potentially important unpublished information and recent data which restricted the literature thus not being able to provide a more complete view of available evidence.

Conclusion

Existing studies showed that there are still limitations in databases for identifying children at risk of abuse. A centralized system which is able to identify and flag potentially abused children based on risk factors, characteristics of abused children, frequency of repeated admission and presentation to the ED is recommended. The set-up of a standardized and integrated database not only among hospitals, but also between hospitals and other relevant agencies, would enable swift collaboration between healthcare professionals and welfare agencies from the first point of contact. The resulting efficiency will translate to better detection and subsequently, more lives of children can save. This is in accordance to Malaysia’s commitment to the Millennium Developmental Goals (MDGs) and the Child’s Rights Convention (CRC) to provide better, safer and more sustainable environment for our children to grow.