School Absenteeism Associated With Child Protection System Involvement,Maltreatment Type,and Time in Out-of-Home Care

Abstract

Greater school absenteeism is associated with numerous negative educational outcomes. We used a retrospective cohort design with linked administrative data on 296,422 children to examine the relationship between school absenteeism and child protection system (CPS) involvement. Children with substantiated maltreatment had 4.1 times more unexplained and problem absences than children with no CPS involvement. In multivariate analyses, children with substantiated maltreatment had significantly greater “chronic” truancy (OR = 3.41) and less “acceptable” levels of absences (OR = 0.74) compared to children with no CPS involvement. Greater absenteeism was seen for children with substantiated neglect and who had their first CPS notification earlier in life. Being in out-of-home care for 3+ years was a protective factor for children who had a CPS notification before age 5. Additional adversities had a strong additive effect with CPS involvement on absenteeism and chronic truancy. This study demonstrates the potential scope for reducing problem absenteeism and helps inform the public debate regarding how the type and timing of CPS involvement might ameliorate or exacerbate harm for children.

Keywords

child maltreatment school functioning child protective services child abuse cohort studies

A considerable body of research has indicated that children with a history of maltreatment have poorer educational outcomes compared to children not known to have experienced maltreatment (Romano et al., 2015). Studies have found that child maltreatment is a risk factor for lower scores on mathematics and verbal tests, higher rates of absenteeism, reduced high school completions, more behavioral problems and suspensions at school, and reduced likelihood of going on to university or college (Coohey et al., 2011; Eckenrode et al., 1993; Fantuzzo et al., 2011; Hagborg et al., 2018; Maclean et al., 2017, 2018; Mills et al., 2019; Stone, 2007; Veltman & Browne, 2001). Children who have been placed in out-of-home care (OOHC) as a result of child maltreatment have often been found to be at particular risk of adverse educational outcomes, leading to increased risk of long-term economic and social disadvantage (Tessier et al., 2018).

School absenteeism and truancy is generally considered to be one of the most important indicators of poor educational outcomes. It is associated with significant long-term deleterious effects such as criminal activity (Rocque et al., 2017) and adult psychiatric problems (Reid, 2003) and is a better predictor of school dropout than poor academic performance (Balfanz & Byrne, 2012). Absenteeism here refers to nonattendance of a scheduled school day (or part thereof) for any reason, whereas truancy normally refers to absenteeism that is unexplained or not accepted as a legitimate reason to be absent from school.

The association between truancy and child maltreatment is complicated. It may be a direct or indirect consequence of the physical and/or psychological damage caused by abuse, with a child unable or unwilling to attend school because of that maltreatment. It may also be a component of parental neglect. Parents with mental illnesses, drug or alcohol problems, or other serious compromising challenges may have both an inability to adequately care for their children and be unable to adequately support active, regular school attendance. With some children, accumulated stresses and/or adverse childhood experiences (ACEs) may lead them to fall behind in learning outcomes, which can increase the likelihood of negative behaviors at school, leading to suspensions, negative school attitudes, and further absenteeism. In addition, and consistent with social learning theories, violent, aggressive, or abusive interpersonal relationships in the home may lead to displays of learned inappropriate behaviors in school, initiating a downward spiral of disciplinary action and poor educational outcomes. A further pathway is via the potential effect of toxic stress on brain development, which may result in challenging behaviors such as hypervigilance, low impulse control, increased internalizing, and poorly developed executive functioning (Child Welfare Information Gateway, 2015; Shonkoff & Garner, 2012).

Attributing causality for poorer educational outcomes of maltreated children, irrespective of the role of OOHC, is also complicated. Often maltreated children have a range of personal, familial, and social challenges which may compromise educational achievements independent of maltreatment or OOHC. In some cases, maltreatment may be considered a consequence of these issues in addition to a risk factor for other adverse child outcomes (Maclean et al., 2016). Studies have shown an attenuation of the association between child maltreatment experience and educational outcomes when controlling for familial and social circumstances (Boden et al., 2007; Maclean et al., 2016).

The nature of the abuse and neglect experienced by a child might also serve to modify the associations between maltreatment and absenteeism. Fantuzzo et al. (2011) found that substantiated neglect, in comparison to physical abuse, was associated with poorer classroom learning behaviors, poor social skills, and low attendance. Eckenrode et al. (1993) found children who were neglected had lower reading and mathematics achievement scores. Other research has linked poorer educational outcomes specifically with childhood sexual abuse (Chandy et al., 1997; Einbender & Friedrich, 1989; Frothingham et al., 2000). Summarizing these mixed results, Stone (2007) concluded that children who experience neglect seem to be especially at risk of general academic deficits, physical abuse seems to be related to behavioral difficulties at school, and all maltreated children appear to be at risk of grade retention. There is some indication then that children suffering serious neglect might be most at risk of greater absenteeism and truancy.

Another possible factor related to school absenteeism for children who have suffered maltreatment is the age at which they engage with the child protection system (CPS). Some studies have found that later CPS involvement is associated with poorer education outcomes (McGlung & Gayle, 2010; Tessier et al., 2018). Other research has failed to confirm these associations (Flynn et al., 2013) or reported mixed findings (O’Higgins et al., 2017).

While a small number of studies have examined and found associations between child maltreatment and school absenteeism (Fantuzzo et al., 2011; Hagborg et al., 2018; Rouse & Fantuzzo, 2009; Rouse et al., 2011), the classification of absenteeism varies and the studies have important limitations. For example, Hagborg et al. (2018) assessed both maltreatment and truancy of Swedish adolescents using child self-report, which may introduce social desirability bias. Slade and Wissow (2007) used “8 or more days absent” from school over an entire year as a measure of poor attendance, but the reason for selecting this threshold was not explained and limited the statistical power for measuring the association with a “maltreatment index” based on participant recollection. Arbitrary distributional measures of attendance have also been used, such as the percentage of attendance one standard deviation below the mean to define “high absenteeism” (Rouse & Fantuzzo, 2009), the highest quartile of days absent for “poor attendance” (Fantuzzo et al., 2011), and absences for any reason at >20% of days enrolled in a single semester of 1 school year to indicate “high levels of absence” (Maclean et al., 2018). Only a few studies have used an objective or educationally relevant categorization of absenteeism. Rouse et al. (2011), for example, adopted a cut point of 25 or more unexcused absences in a single school year to indicate truancy, in line with definitions used by the local school district. Their study found that the odds of children with substantiated maltreatment demonstrating truancy were 1.83 times greater than for children without substantiated maltreatment.

Addressing many of the limitations noted above, our study was designed to empirically examine school absenteeism in a large cohort of children who had varying types of contact with the CPS, in comparison to children with no CPS involvement. While there have been decades of research indicating poorer educational outcomes for child victims of maltreatment, much of the available evidence comes from retrospective self-reports or smaller cross-sectional studies. By using longitudinal whole population linked administrative data sets, we were able to address some of the limitations in previous work.

Based on the understanding that child maltreatment is frequently chronic in nature and not defined by or limited to the instance of first contact with the CPS, nor necessarily with the specific timing of any given event, we aimed to compare rates of absenteeism and truancy across all years of available school data for children without and with different types of CPS involvement. Additionally, we aimed to look at whether associations varied by child maltreatment type and the number of additional cumulative adversities. Finally, given the relative lack of information in this area, we aimed to investigate the interaction between school absenteeism and two potentially important CPS factors, age of first contact with the CPS, and the amount of time children had spent in OOHC.

Method

Study Design and Data Linkage

This study investigated a population cohort using linked administrative data from South Australia (SA), Australia. Data linkage was facilitated by an authorized, independent body (SA-NT DataLink) that acted as an intermediary between researchers and data custodians. SA-NT DataLink is the sole authorized data linkage organization in SA as part of the Australia-wide national data linkage network, the Population Health Research Network. The organization follows best practice guidelines which initially include deduplication processes and deterministic linkage, followed by probabilistic linkage using industry-leading algorithms, and with detailed clerical review. In a recently published paper using data from SA-NT DataLink, it was stated that Australian data linkage systems estimate linkage error rate at approximately 0.1%–0.5% (Procter et al., in press). The data sets made available to the researchers included child protection data from the SA Department for Child Protection (DCP), child birth information from the SA Births Registry and the SA Perinatal Statistics Collection, and education data from the SA Department for Education (DfE). Privacy was maintained using the well-established separation principle, in which approved researchers are only provided access to de-identified data (Kelman et al., 2002).

Study Population

The sample of children was drawn from a larger birth cohort forming the “Impacts of Child Abuse and Neglect” (iCAN) study, which comprises all children born in SA between January 1, 1986, and December 31, 2017. Because absenteeism data from the SA DfE were available for children from 2007 to 2018, we selected all children who were part of the birth cohort and who attended school for at least one term over that 12-year period.

Dependent Variables

Absenteeism and truancy

DfE data included date, reason, and part or whole day absent for all absences for all children attending government schools for at least one term within our observation period. In Australia, most schools have four terms each year, which comprise periods of continuous school attendance lasting about 10 or 11 weeks broken on either side by a school holiday period. Government schools are publicly funded schools, available for attendance by all Australian children and most noncitizen children, and comprise about two thirds of all registered schools operating in the country. For the calendar years 2007–2014 as well as 2018, absenteeism data were available for the first two (of four total) school terms only, while for 2015–2017 absenteeism data were available for the first three school terms. Dependent variable measures of absenteeism were adjusted for observation length, as described below.

Mean number of days with recorded absences for each term (whether full or part day) was calculated for both “problem” absences as well as for absences for any reason. For the purpose of this study, a child was considered as absent for problem reasons if the reason was recorded by their school as being unexplained or unsatisfactory, they did not have the permission of the parent/s, they were sent home for disciplinary reasons or were suspended or excluded from school. Absences recorded where a child had an approved exemption from school by a principal, for family/social reasons condoned by the parents, or who was ill with or without a medical certificate, were not included as a problem absence.

Using DfE definitions, we also created two categories of nonattendance: (1) “acceptable” levels of absence, where absences were less than 5 days a term across all school terms and (2) “chronic truancy,” where students were absent for 10+ days in any term during their schooling, where there was no acceptable reason provided for the absences.

Independent Variables

CPS involvement

In SA, involvement in the CPS is generally categorized as no contact with the CPS, a notification not progressed further within the DPS (screened out), a screened-in notification, an investigation, a substantiation, and OOHC (Pilkington et al., 2017). We elected to use this general categorization sequence but to expand screened-out notifications into two distinct categories and to also use categories formed by different combinations of substantiations and OOHC (in order to reflect specific CPS practices and client characteristics). Therefore, and based explicitly on categorizations used by the DCP, children were categorized into eight mutually exclusive and conceptually meaningful groups as (1) having never had any CPS involvement, (2) having only CPS notifications that did not reach a minimum threshold of DCP concern (“notifier only concern” [NOC]), (3) having received at least one notification indicating possible threat or risk to a child but which fell outside of DCP remit (e.g., maltreatment was not by parents, adolescent self-harming), (4) having received at least one notification considered to be a child protection matter (CPM) but which was not subsequently investigated (a CPM notification), (5) having had at least one investigated maltreatment notification but with no substantiation ever recorded, (6) having had at least one substantiated case of maltreatment but no time in OOHC, (7) having had at least one case of substantiated maltreatment and at least one OOHC placement (of any duration), and (8) having had at least one OOHC placement but with no substantiated maltreatment ever recorded. A child with any OOHC experience could only be allocated to Group 7 or 8.

Maltreatment type

Each case of substantiated maltreatment is classified by the SA DCP as primarily “physical abuse,” “sexual abuse,” “emotional abuse,” or “neglect.” Because children could have multiple maltreatment substantiations across their childhood, possible combinations of primary maltreatment type were used to form six categories: (1) physical abuse only, (2) sexual abuse only, (3) emotional abuse only, (4) neglect only, (5) multiple maltreatment types including neglect, or (6) multiple maltreatment types excluding neglect.

Time in OOHC

Time in OOHC was summed across all placements and categorized as follows (1) “no OOHC,” (2) “up to 4 weeks in OOHC,” (3) “from 4 weeks to up to 3 years OOHC,” and (4) “3+ years OOHC.” Categories were created based on the structure of the data and the objective to create groups reflecting some but minimal OOHC, relatively significant time in OOHC, and long-term OOHC.

Child variables

Child variables included gender and Aboriginality. For determining Aboriginal status, we used a multistage median algorithm that draws on all available data sets, as recommended by Christensen et al. (2014). Birthweight has long been associated with abuse and neglect, and we adopted the usual convention of dichotomizing into normal (2,500 g or greater) or low (<2,500 g; Spencer et al., 2006). Birthweight was obtained from the perinatal data set. Physical or intellectual disability was determined directly from DfE school census data and based on child medical conditions as reported by a parent or guardian. Recorded disabilities included autistic/Asperger’s disorders, language and communication disabilities, global development delay, hearing disabilities, intellectual disabilities, physical disabilities, and visual disabilities. A child was categorized as having a disability if any condition was ever recorded.

School level and number of school terms

Research indicates that absenteeism is consistently lower and more stable for children in primary school than in secondary school (Hancock et al., 2013). To control for variations in the observation period, we calculated both the total number of school terms for which absenteeism data were available and the percentage of school terms in which the child was in secondary school (Grade 8 or higher, in SA).

Parent and family variables

Based on research on the associations between young maternal age at child birth and both subsequent maltreatment (Scannapieco & Connell-Carrick, 2016) and education outcomes for maltreated children (Maclean et al., 2016; Tanaka et al., 2015), we categorized maternal age as (1) <21 years (2) 21–30 years, or (3) or >30 years. Each mother’s marital status as recorded in the child’s birth registration was categorized as (1) “married/defacto” or (2) “never married, widowed, divorced, separated, marriage not indicated.” Maternal smoking status at the child’s birth was derived from the SA Midwives Perinatal Data Collection and categorized as (1) “nonsmoker, (2) “currently smoking any,” or (3) “unknown.”

Parents’ occupation and highest level of schooling completed were obtained from the earliest recorded information from a child’s school census records. Occupations were recorded by the DfE as (1) “senior management or qualified professionals”; (2) “other business managers or associate professionals”; (3) “trades, clerical, sales, and service staff” (4), “other laborers, office assistants, and so on”; (5) “not employed in the last 12 months”; and (6) “not stated/unknown.” The highest level of parental education completed at the earliest available school census record was categorized as (1) “less than Year 12,” (2) “Year 12 but no higher degrees,” (3) “certificate or diploma,” (4) “bachelor degree or above,” or (5) “missing.” Because information was also available for a second parent or guardian, we used the highest occupational category of either parent or guardian to determine parent occupation and highest education of either parent or guardian to determine parent education.

Area-level variables

Area-level variables included measures of socioeconomic status (SES), residential remoteness, and parent country of birth. The first recorded residential postcode of the child while at school was used to assign area-based SES as well as an indicator of level of remoteness for each child. For SES, we used the 2011 Index of Relative Social Disadvantage for Australia, which is based on geographic area-level access to material and social resources (Australian Bureau of Statistics [ABS], 2008). Postcode was also used to define geographical remoteness using the Australian Statistical Geography Standard, with five remoteness area categories: (1) “major cities of Australia,” (2) “inner regional, (3) “outer regional,” (4) “remote,” and (5) “very remote.”

Self-reported information was also available from the school census on parents’ place of birth, which may indicate cultural or language diversity and SES. Consistent with reporting categories used by the ABS, we categorized each parent’s place of birth as being “Australia,” “other main English-speaking country,” “other country,” or “unknown.” Because we potentially had information on two parents or guardians, we created the following categories for parents’ country of birth: (1) “only Australia”; (2) “Australia + other main English-speaking”; (3) “Australia + other country”; (4) “only other main English-speaking or one main-English speaking + one other country”; (5) “only other country”; (6) or “not known for either parent.”

Cumulative adversities

Cumulative adversities were calculated by summing the number of selected adverse individual, family, and social characteristics for each child. For the purpose of this study, 10 risks of adverse outcomes were included: being Aboriginal, having a disability, being low birthweight, having a mother aged <21 at birth, having a mother that smoked at birth, parents not married/defacto at birth, neither parent having completed Year 12 of high school, neither parent working, being in the lowest SES quintile and living in a remote or very remote location. Based on the distributional aspects of the data, we created five categories of cumulative adversities: 0, 1–2, 3–4, 5–6, and 7+.

Analysis

Descriptive statistics were computed for each of the independent variables (IVs) and dependent variables (DVs) and bivariate associations computed. Associations between categorical IVs and continuous DVs were initially examined using one-way analysis of variance (ANOVA), while bivariate associations between categorical IVs and DVs were examined using cross-tabulations and Pearson χ² tests. Due to the nature and distribution of the data, generalized linear models assuming a γ distribution with a log-link function were used to test associations between CPS involvement and both “any absences” and “problem absences” controlling for other predictor variables, and the estimated marginal mean (EMM) number of absences for each category of CPS involvement. Multivariate logistic regression was used to examine associations between CPS involvement and both chronic truancy and acceptable levels of absence, controlling for the other explanatory variables. Univariate ANOVA was used to examine the interactions between absenteeism and CPS variables, age at first notification, and total time in OOHC, as well as the interactions between absenteeism, number of adversities, and type of maltreatment.

Ethics

Ethics approval for the study was obtained from the University of South Australia Human Research Ethics committee, the South Australian Department for Health and Aging Human Research Ethics Committee, and the Aboriginal Health Research Ethics Committee.

Results

In total, 296,422 children were recorded as having attended a public school for at least one term between 2007 and 2018 and were included in this study. The mean number of school terms with recorded absenteeism data for a child was 11.6 (range: 2–25), and the mean number of absences per term was 4.24 (range: 0.0–50.5). Approximately 9% of children had no recorded absences. In relation to CPS involvement, 70.9% had no involvement, 9.4% had only NOC notifications, 3.8% had one or more non-CPM notifications, 4.8% had one or more CPM notifications that did not proceed to an investigation, 4.7% had a nonsubstantiated investigation, 3.9% a substantiation, 1.9% substantiation with OOHC, and 0.6% OOHC without substantiation.

Univariate descriptive statistics and bivariate associations between the individual and family-based IVs and all absenteeism DVs are provided in Table 1. Greater absenteeism, whether for all or problem reasons, was notably higher for children who were Aboriginal, had a disability, and were lower birthweight, with mothers who were younger, smoked during pregnancy, and who were not married or in a defacto relationship at child birth, and whose parents had lower educational and occupational attainment, or were not employed.

Table 1.

Associations Between School Absences and Individual- and Family-Level Explanatory Variables.

Individual- and Family-Level Explanatory Variables	Total		Absences per Term		Chronic Truancy	“Acceptable” Absences
	Total		Any Reason	Problem	Chronic Truancy	“Acceptable” Absences
	n	%	Mean	Mean	%	%
Gender
Male	152,216	51.4	4.27	1.49	18.4	20.4
Female	144,206	48.6	4.21	1.33	18.0	21.0
Aboriginal status
Aboriginal	15,730	5.3	8.51	4.32	51.9	8.7
Non-Aboriginal	264,121	89.1	4.11	1.28	16.8	19.5
Unknown	16,571	5.6	2.26	0.80	8.7	50.3
Child disability ever indicated
No	266,674	90.0	4.04	1.32	17.0	21.7
Yes	29,748	10.0	6.06	2.25	28.9	11.0
Child birthweight
2,500 g or greater	276,125	93.5	4.20	1.39	17.9	20.8
<2,500 g	19,088	6.5	4.84	1.72	22.2	18.5
Maternal age at child birth
<21 years	27,411	9.3	5.80	2.66	36.5	13.9
21–30 years	161,641	54.9	4.21	1.44	18.6	21.4
>30 years	105,557	35.8	3.78	0.98	12.2	21.6
Mother marital status at child birth
Married/defacto	250,164	84.4	3.94	1.21	15.3	21.8
Never/other/unknown	46,258	15.6	5.87	2.52	33.9	14.6
Mother smoking status at child birth
Nonsmoker	134,027	45.2	3.95	0.91	10.0	14.8
Unknown	120,819	40.8	3.85	1.60	22.4	32.1
Smoking any amount	41,576	14.0	6.32	2.52	32.6	6.3
Highest parental education
Less than Year 12	50,048	16.9	6.04	2.61	35.7	13.1
Year 12 but not higher	29,283	9.9	4.32	1.40	18.9	16.8
Certificate or diploma	115,874	39.1	4.33	1.32	17.3	14.9
Bachelor or higher	55,677	18.8	3.27	0.63	6.7	21.9
Not known	45,540	15.4	3.16	1.31	14.9	44.5
Highest parental occupation
Senior managers and professionals	45,399	15.3	3.33	0.72	8.5	22.0
Business managers and assoc. prof.	58,348	19.7	3.63	0.93	11.8	18.5
Trades, clerical, sales, and service	55,379	18.7	4.19	1.26	17.1	15.1
Other occupations	52,510	17.7	4.92	1.74	24.1	12.6
Not working last 12 months	24,846	8.4	7.09	3.12	41.9	9.7
Not known	59,940	20.2	3.79	1.56	17.9	38.5

Note. All bivariate associations significant at p < .001.

Associations between the area-based explanatory variables and the four measures of absenteeism are shown in Table 2. Absenteeism and chronic truancy were associated with greater socioeconomic disadvantage, living in a very remote region, and with having parents born in Australia. Acceptable levels of absence were least common for children from the most disadvantaged and geographically remote families.

Table 2.

Associations Between School Absences and Area-Level Variables.

Area-Level Explanatory Variables	All Children		Absences per Term		Chronic Truancy	“Acceptable” Absences
	All Children		Any Reason	Problem	Chronic Truancy	“Acceptable” Absences
	n	%	Mean	Mean	%	%
Postcode-based SES
Quintile 1 (most disadvantage)	82,576	27.9	5.30	2.16	28.0	16.7
Quintile 2	66,699	22.6	4.44	1.53	20.0	19.6
Quintile 3	50,053	16.9	3.95	1.18	15.0	20.9
Quintile 4	49,822	16.9	3.47	0.89	11.0	24.0
Quintile 5 (least disadvantage)	46,485	15.7	3.22	0.75	9.3	25.4
Geographical remoteness
Major city	194,676	65.7	4.15	1.36	17.9	21.9
Inner regional	37,031	12.5	4.17	1.34	17.0	19.0
Outer regional	50,567	17.1	4.46	1.55	19.5	18.3
Remote	10,262	3.5	4.33	1.55	18.2	16.9
Very remote	3,761	1.3	6.49	2.81	29.1	13.8
Parental place of birth
Australia only	218,633	73.8	4.44	1.52	19.6	18.4
Australia + other main English-speaking	26,981	9.1	3.98	1.22	16.9	19.3
Australia + other country	13,544	4.6	3.95	1.10	13.9	19.9
Other main English-speaking	9,749	3.3	4.19	1.44	19.8	22.2
Other country only	15,624	5.3	3.65	1.02	11.0	25.7
Neither known	11,891	4.0	2.26	0.80	8.2	58.7

Note. All bivariate associations significant at p < .001. SES = socioeconomic status.

Bivariate Associations Between CPS Involvement and Absenteeism

Two thirds of the children in the study sample had no history of CPS involvement. Children with any level of CPS involvement had more mean absences per term than children with no CPS involvement (Figure 1A). The strongest associations were shown for “problem absences.” In particular, children with any substantiated maltreatment (mean = 3.67) had 4.1 times the problem absences as children with no CPS involvement (mean = 0.90). There were also strong associations between CPS involvement and each category of absenteeism (Figure 1B). While only 10.7% of children with no CPS involvement had any school term with chronic truancy, this was 47.6% for children with at least one substantiated case of maltreatment and no OOHC and 46.3% for children with a substantiation and a period in OOHC. In relation to acceptable levels of absence (never five or more absences in a school term), almost one quarter of children with no CPS involvement had acceptable levels of absences across all terms, compared to only 12.4% of children with substantiated maltreatment and no OOHC.

Figure 1.

Panel A: Mean school absences per term (and 95% confidence intervals) by types of child protection system involvement. Panel B: Percentages of children with chronic truancy (any term) and acceptable absenteeism ever recorded (and 95% confidence intervals) by types of child protection system involvement.

Multivariate Analyses of CP Involvement and Absenteeism

Table 3 presents results from multivariate models of mean number of days absent per term by type of CPS involvement after accounting for all other explanatory variables. A similar pattern of results is evident to that found in the bivariate comparisons, although the associations are attenuated. For example, in the unadjusted model, children with substantiated maltreatment had 4.1 times more problem absences as children with no CPS involvement, whereas this was reduced to 1.85 times greater for the adjusted mean number of problem absences per term.

Table 3.

Summary Statistics for Multivariate Analyses of Type of CPS Involvement on School Absences.

Type of CPS Involvement	Total		Any Absences		Problem Absences
Type of CPS Involvement	n	%	EMM [95% CI]	OR	EMM [95% CI]	OR
No CPS involvement [Ref.]	210,281	70.9	4.76 [4.71, 4.82]	—	1.52 [1.49, 1.55]	—
NOC notification only	27,826	9.4	6.18 [6.10, 6.27]	1.30	2.58 [2.53, 2.64]	1.70
1 + non-CPM notification	11,303	3.8	6.14 [6.03, 6.24]	1.29	2.48 [2.42, 2.55]	1.63
1 + CPM notification	14,115	4.8	6.30 [6.20, 6.41]	1.32	2.68 [2.61, 2.74]	1.76
1 + Investigation	13,987	4.7	6.60 [6.49, 6.70]	1.39	2.80 [2.73, 2.86]	1.84
1 + Substantiation	11,495	3.9	7.06 [6.94, 7.18]	1.48	3.19 [3.11, 3.27]	2.10
Substantiation + OOHC	5,532	1.9	6.29 [6.16, 6.43]	1.32	2.73 [2.65, 2.82]	1.80
OOHC, no sub.	1,883	0.6	6.01 [5.81, 6.23]	1.26	2.38 [2.26, 2.51]	1.56

Note. All analyses use generalized linear modelling controlling for gender, Aboriginal status, and disability of the child; marital status, birthweight, smoking status, and maternal age at child birth; parent’s birthplace, socio economic status of residence, geographic remoteness, highest occupation, and highest education at first school census record; and total number of school terms and percentage of school terms in high school. EMM = estimated marginal mean (linear mean adjusted for all other variables in the model); OR = odds ratio; CI = confidence interval; CPS = child protection system; NOC = notifier only concern; OOHC = out of home care.

Output from the unadjusted and multivariate logistic regression of chronic truancy and acceptable levels of absences on CPS involvement are shown in Table 4. Children who had one or more substantiations had 7.59 times greater odds of having had at least one term of chronic truancy, compared to children with no CP involvement. After controlling for all other explanatory variables, the odds ratio (OR) was attenuated to 3.41. In addition, children with CPS involvement had reduced odds of having acceptable levels of absences. However, in relation to absolute percentages of children having unfavorable and favorable absenteeism outcomes, 20.3% more children with one or more substantiations had at least one term of chronic truancy compared to children with no CP involvement, but only 2.9% less had acceptable levels of absences.

Table 4.

Adjusted Percentage of Children Affected and Odds Ratios for CPS Involvement on Overall Chronic Truancy and “Acceptable” Absenteeism.

Type of CPS Involvement	Chronic Truancy				Acceptable Level of Absence
	Adjusted Statistics				Adjusted Statistics
	Unadj. OR [95% CI]	OR [95% CI]	B	%	Unadj. OR [95% CI]	OR [95% CI]	B	%
No CPS involvement [Ref.]	—	—	—	21.0	—	—	—	23.0
NOC notification only	3.50 [3.40, 3.61]	2.51 [2.42, 2.60]	0.92	32.5	0.41 [0.39, 0.43]	0.70 [0.67, 0.73]	−0.35	19.2
1 + non-CPM notification	4.22 [4.05, 4.40]	2.49 [2.38, 2.61]	0.91	34.5	0.54 [0.51, 0.57]	0.71 [0.67, 0.75]	−0.34	19.3
1 + CPM notification	4.31 [4.15, 4.48]	2.72 [2.61, 2.85]	1.00	34.7	0.35 [0.33, 0.37]	0.69 [0.64, 0.73]	−0.38	19.3
1 + Investigation	5.99 [5.78, 6.21]	2.85 [2.73, 2.98]	1.05	37.6	0.47 [0.45, 0.50]	0.67 [0.63, 0.71]	−0.40	18.7
1 + Substantiation	7.59 [7.30, 7.89]	3.41 [3.25, 3.57]	1.23	41.3	0.44 [0.42, 0.47]	0.74 [0.69, 0.80]	−0.30	20.1
Substantiation + OOHC	7.20 [6.81, 7.60]	2.90 [2.71, 3.09]	1.06	38.0	0.43 [0.40, 0.47]	0.83 [0.75, 0.91]	−0.19	20.6
OOHC, no substantiation	5.22 [4.75, 5.74]	2.01 [1.81, 2.24]	0.70	32.1	0.76 [0.68, 0.86]	0.85 [0.74, 0.97]	−0.16	20.0

Note. All analyses use binary logistic regression controlling for gender, Aboriginal status, and disability of the child; marital status, birthweight, smoking status, and maternal age at child birth; parent’s birthplace, socioeconomic status of residence, highest occupation, and highest education at first school census record; and total number of school terms and percentage of school terms in high school. Adjusted percentages obtained from univariate analysis of variance using the same categories as the logistic regression and with absenteeism categories 0 = no, 1 = yes. OR = odds ratio; CI = confidence interval; CPS = child protection system; NOC = notifier only concern; CPM = child protection matter; OOHC= out of home care.

Absenteeism by Child Maltreatment Type, Age at First CP Notification, and Total Time in OOHC

Children who had suffered neglect or neglect combined with sexual, physical, or emotional abuse had significantly more school absences than maltreated children who had not had neglect recorded as a maltreatment type (Figure 2). A similar pattern was found for chronic truancy, with 51.9% of children who had suffered serious neglect having at least one school term with 10+ unexplained absences compared with 38.6% and 33.0% of those children physically or sexually abused, respectively. Only 9.9% of children with substantiated neglect (only) had acceptable levels of absences recorded.

Figure 2.

Mean school absences per term (and 95% confidence intervals) and chronic and “acceptable” absenteeism by substantiated maltreatment types.

Examining only those children with CPS involvement, both age at first CPS notification and total time in OOHC had significant associations with any absences and “problem absences” (Table 5.) Mean absences per term were significantly higher for those children who received a notification before the age of 1 year than for those who received their first notification after the age of 12, with respective differences of 37.7% and 36.2% higher for any absences and problem absences. In relation to total time in OOHC, children with no time in OOHC and those spending more than 3 years in OOHC had fewer absences than children who had up to 4 weeks in OOHC or who had between 4 weeks and 3 years in OOHC. There was also a noticeable interaction between age at first CPS notification and total time spent in OOHC. Especially for those children who received their first notification before the age of 1, having spent more than 3 years in care was associated with substantial relative reductions in absences per term of 31.9% for any absences and 46.7% for problem absences, in comparison to children with up to 4 weeks in OOHC. For children receiving their first notification between the age of 1 and 4, differences were reduced, while for children who did not receive their first CPS notification until the age of 13 or older, relative differences were 4.4% and 5.3%. Similar associations were found in relation to chronic truancy, where more than 3 years in OOHC was associated with reduced chronic truancy but only for children aged less than 1 or between 1 and 4 (Figure 3).

Table 5.

Mean School Absences Per Term by Age of First CPS Notification and Total Time Spent in OOHC.

Total Time in OOHC	Age of First CPS Notification
	<Age 1 (n = 12,350)		Age 1–4 (n = 25,994)		Age 5–12 (n = 34,907)		Age 13+ (n = 12,517)		Total (n = 85,768)
	Mean	95% CI	Mean	95% CI	Mean	95% CI	Mean	95% CI	Mean	95% CI
Any absences
No OOHC	7.44	[7.34, 7.54]	6.46	[6.40, 6.53]	5.85	[5.80, 5.91]	5.30	[5.21, 5.39	6.14	[6.10, 6.17
Up to 4 weeks	8.40	[7.95, 8.84]	7.40	[7.06, 7.75]	6.43	[6.01, 6.84]	5.69	[4.70, 6.67]	7.28	[7.01, 7.56]
>4 Weeks up to 3 years	7.94	[7.62, 8.27]	7.89	[7.58, 8.20]	6.28	[5.82, 6.74]	6.09	[4.84, 7.35]	7.56	[7.33, 7.79]
>3 Years	5.72	[5.46, 5.98]	6.08	[5.75, 6.42	5.76	[5.17, 6.34]	5.44	[3.34, 7.55]	5.84	[5.66, 6.02]
Total	7.31	[7.21, 7.42]	6.53	[6.47, 6.60]	5.87	[5.82, 5.92]	5.31	[5.23, 5.39]	6.20	[6.16, 6.23]
Problem absences
No OOHC	3.23	[3.17, 3.30]	2.74	[2.70, 2.78]	2.47	[2.44, 2.51]	2.34	[2.28, 2.39]	2.62	[2.60, 2.65]
Up to 4 weeks	4.22	[3.93, 4.51]	3.82	[3.60, 4.05]	3.62	[3.35, 3.90]	3.00	[2.35, 3.65]	3.82	[3.64, 4.00]
>4 Weeks up to 3 years	3.82	[3.61, 4.03]	4.07	[3.86, 4.27]	3.24	[2.94, 3.54]	3.00	[2.18, 3.82]	3.79	[3.63, 3.94]
>3 Years	2.25	[2.09, 2.42]	2.85	[2.63, 3.08]	2.89	[2.51, 3.28]	2.84	[1.46, 4.22]	2.52	[2.41, 2.64]
Total	3.20	[3.13, 3.27]	2.83	[2.79, 2.87]	2.51	[2.47, 2.54]	2.35	[2.29, 2.40]	2.68	[2.66, 2.70]

Note. Only includes children with at least one CPS notification; CI = confidence interval; CPS = child protection system; OOHC = out-of-home care.

Figure 3.

Percentage of children (and 95% confidence intervals) with chronic truancy (any term) and acceptable absences (every term) by age of first notification and total time spent in care.

Cumulative childhood adversities had a profound effect on the associations between CPS involvement and both mean problem absences (Figure 4A) and chronic absenteeism (Figure 4B). Within each level of CPS involvement, there were strong stepwise associations between additional adversity and poorer school attendance. Using children with no CP involvement and no adversities as a “best case” scenario (mean problem absences per term = 0.57, chronic truancy any term = 6.0%), children with substantiated maltreatment and 7+ adversities had 12 times more problem absences (mean = 6.81) and were 13.3 times more likely to have at least one term of chronic truancy (mean = 80.0%). Children with CP involvement also had more adversities than those without CP involvement. For example, 18.4% of children with substantiated maltreatment had five or more adversities and only 6.6% had no adversities. In contrast, only 0.9% of children with no CP involvement had five or more adversities, while 37.8% had no adversities.

Figure 4.

Panel A: Problem absences (and 95% confidence intervals) by type of child protection system involvement and number of adversities. Panel B: Percentages of children with any chronic truancy (and 95% confidence intervals) by type of child protection system involvement and number of adversities.

Discussion

This is the first large-scale population-based study to undertake a detailed examination of the relationships between child maltreatment and school absenteeism. We found a strong association between greater CPS involvement and higher rates of absenteeism and chronic truancy. Associations were only partially attenuated after controlling for child, family, and area-level variables. We also found that absenteeism was greatest for children who had substantiated neglect, for children who if removed had spent less total time in OOHC, for those who had their first CPS notification earlier in their life, and for those experiencing higher levels of other childhood adversities.

While we found associations between CPS involvement and absences for any reason, the strongest associations were found for problem absences. For example, while children with substantiated maltreatment had 7.5 absences on average per term, which was more than twice that for children with no CPS involvement, problem absences were 4.1 times higher than for children with no CPS involvement. After controlling for other explanatory variables, including SES and parental education, maltreated children had 3.4 times the odds of having at least one term of chronic truancy than non-CP children. Although school systems might consider absences from school for any reason to be problematic, findings indicate that unexplained or unacceptable absences, in particular, have stronger associations with academic achievement (Zubrick, 2014). In this study, we were able to explore the extent of school absenteeism by broad reason in contrast to other studies examining the relationship between maltreatment and school absenteeism, which have not distinguished by reason for absences (e.g., Fantuzzo et al., 2011; Slade & Wissow, 2007). This allowed us to describe more fully the extent of school absenteeism experienced by children with CPS involvement.

Similar to previous research (Fantuzzo et al., 2011), our study showed that children with some CPS involvement, including those who never had a substantiated investigation, had greater school absenteeism than did children with no CPS involvement, with the associations between CPS involvement and absenteeism remaining in multivariate models. In contrast to other work (e.g., Kohl et al., 2009), we found differences in absenteeism between children with and without substantiated maltreatment, as well as differences between different types of CPS involvement, which remained after controlling for other variables. While other risk factors for problem absenteeism and chronic truancy are more common in children who experience substantiated maltreatment (Johnson-Reid et al., 2010), as was also found in this study, CPS involvement (as an indicator of likely maltreatment) still has an additive impact.

Similar to the results of Fantuzzo et al. (2011), children who had suffered substantiated neglect had significantly more days absent from school than did children who had been sexually, physically, or emotionally abused. This is consistent with other research on the effects of serious child neglect (e.g., Hildyard & Wolfe, 2002) and shows the importance of considering the circumstances of maltreatment when investigating possible deleterious outcomes. Neglected children, by definition, lack appropriate parental support and involvement, which might include insufficient concern and follow-through regarding school attendance.

There is a large literature on the profound effect of ACEs on a range of deleterious outcomes in childhood and, subsequently, in adulthood. More recently, ACEs have been found to be related to chronic school absenteeism (Stempel et al., 2017). Our results show a similar pattern of disadvantage in relation to the effect of adverse child circumstances and school absenteeism. Within any category of CPS involvement, children with more adverse circumstances had significantly more absenteeism. These results demonstrate that while child maltreatment might result in poorer school attendance overall, the associations are strongly affected by the individual circumstances of the children involved.

While most of the measures of absenteeism used in this study (and in almost all other studies) relate to poorer educational outcomes, we also included a measure of acceptable level of absence, which was defined as no school term with five or more absences for any reason. It is not necessarily the case that predictors of poor outcomes, or associations between risk factors and poor outcomes, are the same as those with positive outcomes. In seeking to reduce educational inequalities for maltreated children, it might be useful to focus not just on the reduction of negative outcomes but also on promoting positive outcomes. We found that 24.1% of children with no CPS involvement had acceptable absenteeism ever recorded, compared with only 12.4% of children with substantiated maltreatment. However, this difference was appreciably attenuated in multivariate analyses. Overall, our results indicate that greater benefits might be achieved by focusing on reducing the high amounts of problem absenteeism in maltreated children, than in trying to bring about consistent low levels of absenteeism.

An important new finding of this study was the interaction between a child’s age of first CPS notification and length of time in OOHC on school absenteeism. We found that children with more than 3 years in OOHC had on average lower absenteeism and truancy than other children with CPS contact. These results were strongest for children who had their first CPS contact at a younger age, although there was still an advantage if the first CPS contact was before age of 5 for chronic truancy or before age of 13 for problem absences. Indeed, for children who received their first CPS notification before the age of 1, unexplained absences for those who had more than 3 years in OOHC was less than half that of children who had been in OOHC but for no more than 4 weeks. Mean problem absences were half that for children with CPS contact before age of 1 but with no OOHC experience. Similarly, chronic truancy in any school term was just 29.5% in children with CPS contact before age one and more than 3 years in care compared with 58.2% for children experiencing no more than 4 weeks in care.

The implication of this finding is that the earlier maltreated children come to the attention of the CPS and are provided long-term care in an alternative home environment (even if not in a single placement), this can be protective in terms of reduced school absenteeism. This finding is consistent with the purpose of placement into OOHC, to prevent further harm to children who have been maltreated in their family homes. And yet, because OOHC is often a last recourse for ensuring child safety, children placed in OOHC will have been exposed to highly distressing home environments and will have suffered considerable trauma. It is significant then that we found longer time in OOHC to be strongly associated with reduced school absenteeism.

Because most children have initial CPS contact early in life and do not commence school until the age of 5, this study was unable to examine the timing of CPS involvement and school absenteeism before and after such involvement. While a simple hypothesis might be that absenteeism would reduce following CPS intervention, based on an assumption that there would be some improvement in the child’s family circumstances, we have no data on the nature of CPS interventions beyond placement of the child in OOHC and have no information regarding the effectiveness of any interventions. Also, because abuse and neglect are often chronic, relating child maltreatment events obtained in administrative data to changes in absenteeism is plagued with problems. Yet it is the aim of child protection agencies to improve conditions for abused and neglected children. Studies specifically directed toward assessing the extent of such changes are sorely needed.

This study has several significant strengths over previous published research, including the large sample of almost 300,000 children, multiple years rather than single year of school absenteeism data, information from both primary and high school, multiple definitions of absenteeism, more granular classifications of CPS involvement based on agency determinations, and high-quality data on many possible confounders.

We acknowledge several limitations. Although we had school absenteeism data from the first two or three school terms each year, there is evidence that absences increase in the latter half of the year. It is possible, therefore, that the effect of child maltreatment on school absenteeism might have been underestimated if the increase in the emotional burden of school attendance across the course of a year increased differentially for children with CPS involvement compared to those with no CPS involvement. Another limitation of our study is that the data do not include the 35% of SA children attending nongovernment (Catholic or private) schools. Given both the known association between SES and child maltreatment (e.g., Pelton, 1981) and also the significant family SES differences of children attending government compared to nongovernment schools in Australia (Beavis, 2004), it is likely that our study represents a more disadvantaged and at-risk population than would be found in the entire population. Finally, although data linkage inevitably involves some linkage error, which may be a potential source of bias in results (Harron et al., 2017), the separation of linkage and analysis processes makes it impossible for us to assess the extent of any possible bias. We note, however, that the authorized data linkage units within Australia follow best practice procedures, use extensive clerical review to resolve linkage errors, and that the results of this study were especially robust, making them less susceptible to problems caused by linkage errors.

In conclusion, study findings indicate significantly more absenteeism for children with CPS involvement, with the highest levels for those children who had substantiated maltreatment and who had suffered confirmed neglect. Given the importance of school attendance on future educational and occupational outcomes, there are significant policy challenges in addressing these inequities for maltreated children, across government, but particularly for child protection and education agencies. There is hope in the finding that children who were identified by the CPS early in their life and placed into long-term OOHC had better absenteeism outcomes than other children with CPS involvement. While this in no way supports the removal of maltreated children from their family home as a primary option, it most assuredly underlines the value of having a stable and supportive home environment for positive educational outcomes and reinforces that early intervention is beneficial to children.

It is imperative that more effort, resources, and services are made available to help families create supportive and nurturing environments to reduce the prevalence of maltreatment, so that children exposed to maltreatment can reach their potential by reducing the educational disparities associated with unwanted misfortune in their early life circumstances.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: National Health and Medical Research Council (grant 1103439).

ORCID iD

Jason M. Armfield

Emmanuel Gnanamanickam

Leonie Segal

References

Australian Bureau of Statistics. (2008). 2033.0.55.001—Census of population and housing: Socio-economic indexes for areas (SEIFA), Australia—data only, 2006. http://www.abs.gov.au/AUSSTATS/abs@.nsf/DetailsPage/2033.0.55.0012006

Balfanz

Byrnes

(2012). Chronic absenteeism: Summarizing what we know from nationally available data. Johns Hopkins University Center for Social Organization of Schools.

Beavis

(2004). Why parents choose private or public schools. Research Developments, 12, 1–4.

Boden

J. M.

Horwood

L. J.

Fergusson

D. M.

(2007). Exposure to childhood sexual and physical abuse and subsequent educational achievement outcomes. Child Abuse & Neglect, 31, 1101–1114.

Chandy

J. M.

Blum

R. W.

Resnick

M. D.

(1997). Sexually abused male adolescents: How vulnerable are they? Journal of Child Sexual Abuse, 6, 1–16.

Child Welfare Information Gateway. (2015). Understanding the effects of maltreatment on brain development. U.S. Department of Health and Human Services, Children’s Bureau. https://www.childwelfare.gov/pubPDFs/brain_development.pdf

Christensen

Davis

Draper

Mitrou

McKeown

Lawrence

(2014). Evidence for the use of an algorithm in resolving inconsistent and missing indigenous status in administrative data collections. The Australian Journal of Social Issues, 49, 423–443.

Coohey

Renner

L. M.

Hua

Zhang

Y. J.

Whitney

S. D.

(2011). Academic achievement despite child maltreatment: A longitudinal study. Child Abuse & Neglect, 35, 688–699.

Eckenrode

Laird

Doris

(1993). School performance and disciplinary problems among abused and neglected children. Developmental Psychology, 29, 53–62.

10.

Einbender

A. J.

Friedrich

W. N.

(1989). Psychological functioning and behavior of sexually abused girls. Journal of Consulting and Clinical Psychology, 57, 155–157.

11.

Fantuzzo

J. W.

Perlman

S. M.

Dobbins

E. K.

(2011). Types and timing of child maltreatment and early school success: A population-based investigation. Children and Youth Services Review, 33, 1404–1411.

12.

Flynn

R. J.

Tessier

N. G.

Coulombe

(2013). Placement, protective and risk factors in the educational success of young people in care: Cross-sectional and longitudinal analyses. European Journal of Social Work, 16, 70–87.

13.

Frothingham

T. E.

Hobbs

C. J.

Wynne

J. M.

Yee

Goyal

Wadsworth

D. J.

(2000). Follow up study eight years after diagnosis of sexual abuse. Archives of Disease in Childhood, 83, 132–134.

14.

Hagborg

J. M.

Berglund

Fahlke

(2018). Evidence for a relationship between child maltreatment and absenteeism among high-school students in Sweden. Child Abuse & Neglect, 75, 41–49.

15.

Hancock

K. J.

Shepherd

C. C. J.

Lawrence

Zubrick

S. R.

(2013). Student attendance and educational outcomes: Every day counts. Report for the Department of Education.

16.

Harron

K. L.

Doidge

J. C.

Knight

H. E.

Gilbert

R. E.

Goldstein

Cromwell

D. A.

Van Der Meulen

J. H.

(2017). A guide to evaluating linkage quality for the analysis of linked data. International Journal of Epidemiology, 46, 1699–1710.

17.

Hildyard

K. L.

Wolfe

D. A.

(2002). Child neglect: Developmental issues and outcomes. Child Abuse & Neglect, 26, 679–695.

18.

Jonson-Reid

Emery

C. R.

Drake

Stahlschmidt

M. J.

(2010). Understanding chronically reported families. Child Maltreatment, 15, 271–281.

19.

Kelman

C. W.

Bass

A. J.

Holman

C. D. J.

(2002). Research use of linked health data—A best practice protocol. Australian and New Zealand Journal of Public Health, 26, 251–255.

20.

Kohl

P. L.

Jonson-Reid

Drake

(2009). Time to leave substantiation behind: Findings from a national probability study. Child Maltreatment, 14, 17–26.

21.

Maclean

M. J.

Taylor

C. L.

O’Donnell

(2016). Pre-existing adversity, level of child protection involvement, and school attendance predict educational outcomes in a longitudinal study. Child Abuse & Neglect, 51, 120–131.

22.

Maclean

M. J.

Taylor

C. L.

O’Donnell

(2017). Relationship between out-of-home care placement history and educational achievement: A population level linked data study. Child Abuse & Neglect, 70, 146–149.

23.

Maclean

M. J.

Taylor

C. L.

O’Donnell

(2018). Out-of-home care and the educational achievement, attendance, and suspensions of maltreated children: A propensity-matched study. The Journal of Pediatrics, 198, 287–293.e2.

24.

McGlung

Gayle

(2010). Exploring the care effects of multiple factors on the educational achievement of children looked after at home and away from home: An investigation of two Scottish local authorities. Child & Family Social Work, 15, 409–431.

25.

Mills

Kisely

Alati

Strathearn

Najman

J. M.

(2019). Cognitive and educational outcomes of maltreated and non-maltreated youth: A birth cohort study. Australian and New Zealand Journal of Psychiatry, 53, 248–255.

26.

O’Higgins

Sebba

Gardner

(2017). What are the factors associated with educational achievement for children in kinship or foster care: A systematic review. Children and Youth Services Review, 79, 198–200.

27.

Pelton

L. H.

(1981). The social context of child abuse and neglect. Human Sciences Press.

28.

Pilkington

Grant

Chittleborough

Gialamas

Montgomerie

Lynch

(2017). Child protection in South Australia. BetterStart Child Health and Development Research Group, School of Public Health, The University of Adelaide. https://health.adelaide.edu.au/betterstart/publications/reports/child-protection-in-south-australia.pdf

29.

Procter

A. M.

Pilkington

R. M.

Lynch

J. W.

Smithers

L. G.

Chittleborough

C. R.

(in press). Potentially preventable hospitalizations in children: A comparison of definitions. Archives of Disease in Childhood. Advance online publication. doi:10.1136/archdischild-2019-316945

30.

Reid

(2003). Tackling truancy in schools: A practical manual for primary and secondary schools. Routledge.

31.

Rocque

Jennings

W. G.

Piquero

A. R.

Ozkan

Farrington

D. P.

(2017). The importance of school attendance: Findings from the Cambridge study in delinquent development on the life-course effects of truancy. Crime & Delinquency, 63, 592–612.

32.

Romano

Babchishin

Marquis

Fréchette

(2015). Childhood maltreatment and educational outcomes. Trauma, Violence, & Abuse, 16, 418–437.

33.

Rouse

H. L.

Fantuzzo

J. W.

(2009). Multiple risks and educational well being: A population-based investigation of threats to early school success. Early Childhood Research Quarterly, 24, 1–14.

34.

Rouse

H. L.

Fantuzzo

J. W.

LeBoeuf

(2011). Comprehensive challenges for the well being of young children: A population-based study of publicly monitored risks in a large urban center. Child & Youth Care Forum, 40, 281–302.

35.

Scannapieco

Connell-Carrick

(2016). Correlates of child maltreatment among adolescent mothers with young children. Journal of Evidence-Based Social Work, 13, 59–75.

36.

Shonkoff

J. P.

Garner

A. S.

(2012). The lifelong effects of early childhood adversity and toxic stress. Pediatrics, 129, 232–246.

37.

Slade

E. P.

Wissow

L. S.

(2007). The influence of childhood maltreatment on adolescents’ academic performance. Economics of Education Review, 26, 604–614.

38.

Spencer

Wallace

Sundrum

Bacchus

Logan

(2006). Child abuse registration, fetal growth, and preterm birth: A population based study. Journal of Epidemiology and Community Health, 60, 337–340.

39.

Stempel

Cox-Martin

Bronsert

Dickinson

L. M.

Allison

M. A.

(2017). Chronic school absenteeism and the role of adverse childhood experiences. Academic Pediatrics, 17, 837–843.

40.

Stone

(2007). Child maltreatment, out-of-home placement and academic vulnerability: A fifteen-year review of evidence and future directions. Children and Youth Services Review, 29, 139–161.

41.

Tanaka

Georgiades

Boyle

M. H.

MacMillan

H. L.

(2015). Child maltreatment and educational attainment in young adulthood: Results from the Ontario child health study. Journal of Interpersonal Violence, 30, 195–214.

42.

Tessier

N. G.

O’Higgins

Flynn

R. J.

(2018). Neglect, educational success, and young people in out-of-home care: Cross-sectional and longitudinal analyses. Child Abuse & Neglect, 75, 115–129.

43.

Veltman

M. W. M.

Browne

K. D.

(2001). Three decades of child maltreatment research. Trauma, Violence & Abuse, 2, 215–239.

44.

Zubrick

(2014). School attendance: Equities and inequities in growth trajectories of academic performance. Australian Council for Education Research. https://research.acer.edu.au/research_conference/RC2014/4august/7