Life Course Models of Child Maltreatment: Effects on General Psychopathology Outcomes in a Longitudinal Sample

Abstract

The study compared life course models (LCM; accumulation, recency, and sensitive period) of child maltreatment and general psychopathology in a large, national longitudinal data set of 1354 youth ages birth-16 years (657 boys, 53.2% Black, 59.7% <$40K caregiver income). Previous research has supported the accumulation and recency models, albeit with shorter or fewer time periods of outcome measurement. We extend this work by modeling the impact of combined abuse and neglect allegations on a general psychopathology factor (dysregulation profile). Cross-sectional structural equation models were constructed using LCMs and tested across two-year periods from 4–16 years old and compared using Akaike Information Criterion weights. The recency variable generally explained the greatest proportion of variance in psychopathology. Notably, maltreatment more proximal to the time of outcome measurement had the strongest effect, suggesting that more recent maltreatment may have stronger effects on general psychopathology. These results lend support to a recency effect of maltreatment on psychopathology outcomes, although substantive overlaps with the accumulation model are noted.

Keywords

child maltreatment psychopathology structural equation modeling

Child maltreatment (CM) is a widespread problem, with nearly 300 million children aged 2–4 years old having been physically or emotionally abused (WHO, 2020). The National Incidence Study of Child Abuse and Neglect tracks incidents of abuse and neglect via reports by mandatory reporters; its fourth iteration (NIS- 4) recorded a rate of 3.95% of children in the United States being subject to any kind of CM. CM is associated with severe deleterious consequences across time. Aside from the physical injuries incurred by physically and/or sexually abused children, many of these children go on to suffer social and behavioral problems, cognitive and school problems, poor mental health outcomes, and poor overall medical health (Widom, 2014). The mainstream ecological-transactional model of CM suggests that CM brings about deviations from the average expectable environment of the child, leading to failures to resolve salient developmental issues at each developmental stage with the consequent poor outcomes (Cicchetti & Valentino, 2015).

Incorporated at the microsystem level of the ecological-transactional model are the dimensions of CM. Occurrences of CM have a multifaceted temporal dimension, encompassing when CM occurs (developmental timing/sensitive periods), how proximal the CM occurrences are to the time of measurement (recency), and accumulation. Accumulation is a broad construct that can refer to any of the following: how many distinct CM exposures occur (total frequency), the number of developmental periods the CM occurrences span (periodic frequency), and the duration between the time of first CM event and last CM event (duration; Jonson-Reid et al., 2012). It can also refer to the degree to which CM occurrences are closely clustered in time (conglomeration). These temporal dimensions, also termed life course models (LCM; Dunn et al., 2018), are key features of the etiological models of CM, since a wide array of evidence suggests that these subdivisions of the temporal dimension are influential over CM outcomes.

While the temporal dimension seemingly affects outcomes, it remains unclear which facets are the most explanatory. Sensitive periods posit that maltreatment occurring in certain periods are more explanatory of outcomes than maltreatment occurring in other periods. A review of CM studies found that while sensitive periods had effects on a variety of child outcomes (such as mental disorders, correlated neuroanatomical changes, and other psychological functions), no consistent sensitive periods could be identified (Schaefer et al., 2022). Effects indicating a dose-dependent relationship between CM and mental health outcomes/mental functioning have been found in a variety of studies (Harpur et al., 2015; Jaffee & Maikovich-Fong, 2011; Jonson-Reid et al., 2012; Russotti et al., 2021; Young-Southward et al., 2020), supporting the role of accumulation. Several studies have also found recency effects. For instance, Shanahan et al. (2011) found that CM occurring in the same period as outcome measurement was predictive of child-onset and adolescent onset depression, while antecedent CM was not. Other studies examining diverse outcomes like delinquency, substance use, and mental health have found that CM in adolescence was more explanatory of negative late adolescent outcomes than CM in childhood (Ireland et al., 2002; Thornberry et al., 2001; Tracy et al., 2019). These studies indicate the need to examine CM effects on broader outcomes, as well as the need to test multiple facets of these temporal dimensions in the same study.

Theoretical shifts in developmental psychopathology suggest traditional disorder-based models may be less explanatory than factors accounting for general risk for psychopathology. Just as a general intelligence, or g, appears to account for associations between test scores across cognitive tests for the same individual, Caspi et al. (2014) suggested that p might serve a similar function, reflecting various dimensions of psychopathology. p is thought to be clinically useful as an index of overall impairment, which can support treatment planning and goal setting (Martel et al., 2017; Patalay et al., 2015; Smith et al., 2020). The dysregulation profile (DP) is a similar construct found exclusively in children, typically comprising one general DP factor superordinate to three specific factors of anxiety/depression, aggressive behavior, and attention problems (Deutz et al., 2018). Deutz et al. (2020) noted that while p and DP have different operationalizations and factors (such as attentional problems and thought-problems for g), they concluded that qualitative differences in their operationalizations negligibly impact the association with antecedent characteristics and outcomes, suggesting that DP may similarly be useful as a general factor of psychopathology. Further, Deutz et al. (2016) found that a bifactor DP model was the best fitting model for modeling CBCL factors in adolescence, suggesting the validity of the measure in both adolescence and childhood. Given our focus on the development of psychopathology across childhood into adolescence and the broad risks for psychopathology conferred from multiple forms of CM, we focus on the general DP factor in this investigation.

Current Study

To our knowledge, two direct comparisons have been made in the literature between different temporal models of childhood adversity. Dunn et al. (2018) investigated relations between LCMs of childhood adversity exposure and later mental health outcomes in a prospective longitudinal sample of children from birth to eight years old. They compared the extent to which each of the three LCMs — recency, accumulation, and sensitive period — were predictive of psychopathology symptoms at eight years old. Childhood adversities were defined as caregiver emotional or physical abuse, sexual or physical abuse (by anyone), financial stress, and parental legal problems. The accumulation model was defined as the sum of the number of 12-month periods in which exposure to maltreatment occurred. The recency model was defined using the accumulation model but weighted the sum, such that periods closer to the final measurement point had higher weights. Finally, the sensitive period model was defined as an indicator variable (exposed/not exposed) for each period. They analyzed the results using a least angle regression procedure for the models among each type of adversity. Overall, they found that across adversities, child psychopathology symptoms were better explained by accumulation and recency of exposure than sensitive periods. With both caregiver abuse and abuse by anyone, more recent exposures, followed by accumulation, had the most influence. Only two sensitive periods emerged: during very early childhood for financial stress in girls and very early childhood for parental legal problems (Dunn et al., 2018).

Russotti et al. (2021) operationalized the temporal dimension of CM in terms of chronicity (operationalized similarly to accumulation as defined by Dunn and colleagues), developmental timing (recent and/or early-onset), and as a binary variable (any maltreatment vs. no maltreatment), examining these effects on the internalizing and externalizing factors of the CBCL from 10–12 years old and subsequent psychopathology presentations around 20 years old. Maltreatment types used included physical, sexual, emotional abuse and neglect. They then tested the three dimensional models separately: the binary variable, chronicity/accumulation (as a number of developmental periods), and the four categories of early-onset recent maltreated, early-onset nonrecent maltreated, late-onset maltreated, and non-maltreated. After conducting confirmatory factor analysis on the factor structure of the latent CBCL factors being examined, they estimated three structural equation models using the three dimensional models separately as regression predictors. Results found that the chronicity/accumulation model better captured the developmental pathways across time as compared to the binary variable. The late onset-only group predicted externalizing symptoms, while the early-onset recent group had greater risk of internalizing and externalizing symptoms. These results lend support to the accumulation model, the school-age sensitive period model, and provide evidence against an early sensitive period.

Integrating and expanding upon previous efforts, we explored the degree to which LCMs (recency, accumulation, and sensitive period; Dunn et al., 2018) of combined maltreatment (physical, sexual, emotional abuse, and neglect; Russotti et al., 2021), explained general psychopathology outcomes using the DP framework. As psychopathological symptoms commonly manifest in childhood and express more fully as disorders in adolescence and emerging adulthood (Zeman et al., 2020), more nuanced examinations of timing of CM across childhood, using developmentally appropriate conceptual frameworks, like DP, are needed. The use of a large, national prospective longitudinal sample of children (LONGSCAN; Runyan et al., 2020) afforded the opportunity to extend the time periods of assessment and increase the number of observation periods from both Dunn’s and Russotti’s work, as well as provide a more comprehensive view of psychopathology.

Methods

Data Source

LONGSCAN is a multi-site consortium of research studies on CM with coordinated data collection procedures, assessment measures, and schedules (Runyan et al., 2020). The comprehensive and longitudinal nature of LONGSCAN makes it ideal for secondary analysis of this type, with measures collected at ages 0, 2, 4, 6, 8, 10, 12, 14, and 16 years and a total sample size of 1354 children. The attrition rate across sites from baseline to age 16 was 33.9% (Runyan et al., 2020). The overall sample was geographically representative, with the studies conducted in the East (EA), Midwest (MW), Northwest (NW), Southwest (SW), and South (SO) regions of the United States. The EA cohort (n = 282) consisted of children from low-income families served at three pediatric clinics, with groups within this sample having either had inadequate growth in the first two years of life, parents who had a HIV infection or used drugs, or no other identified risk factors. The MW cohort (n = 245) contained children from families reported to child protective services (CPS), comprising two-thirds of the sample; half of these were provided access to comprehensive services (including mental health services), while the other half only received an intervention from CPS. The remaining third were controls from the neighborhood. The NW cohort (n = 254) comprised children aged 0 to 4 judged to be at moderate risk of CM by CPS after their families were reported to CPS, with 60% of the reports being substantiated. The SW cohort (n = 330) included children who were put in out-of-home placement with a relative or foster family, while the SO cohort (n = 254) was drawn from children identified as high risk of CM by a state-level public health tracker, with a ratio of 2:1 between non-reported children to reported children.

Demographic characteristics of the analytic sample are outlined in supplemental materials (Supplemental Table 1). Overall, the sample contained an approximately even mix of girls and boys, with over half of the sample identifying as African American. The majority of the sample’s caregiver income was below $40K. Proportions of children exposed to any kind of maltreatment (physical, sexual, emotional, neglect) prior to the point of measurement ranged from 55.8% (at Wave 4) to 67.1% (at Wave 10). Proportions stabilized from Wave 10 onwards.

Measures and Procedures

Measures include the temporal dimensions of CM from the Derived Maltreatment Dataset as coded by trained LONGSCAN investigators from CPS records and general psychopathology as captured by the Achenbach Child Behavior Checklist (CBCL) 2–3 and CBCL 4–18 Caregiver Report. Details regarding the measures used are drawn from the LONGSCAN Measures Manuals - Volume I (Hunter et al., 2001), Volume II (Hunter et al., 2002), III (Knight et al., 2010), IV (Knight et al., 2014a), and V (Knight et al., 2014b). Analyzed scales were administered during face-to-face interviews (before age 8), computer-assisted face-to-face interviews (age 8) and through the Audio-Computer Assisted Self Interview delivery system from age 12 onward. Full data collection procedures are outlined in Runyan et al. (2020).

Achenbach CBCL

The Achenbach Child Behavior Checklist 4 to 18 (CBCL; Achenbach & Ruffle, 2001) caregiver report form assesses eight areas of problems in 4 to 18 year olds: Social Withdrawal, Somatic Complaints, Anxiety/Depression, Social Problems, Thought Problems, Attention Problems, Delinquent Behavior, and Aggressive Behavior. In LONGSCAN, the scale was generally reliable, with a Cronbach’s alpha of 0.62–0.92 for boys aged 4–11 and 0.66–0.92 for girls aged 4–11. The CBCL Dysregulation Profile (DP), developed by Biederman et al. (1995), is comprised by the Anxiety/Depression Problems, Attention Problems, and Aggressive Behavior scales. The latent factor CBCL-DP was constructed from a bifactor model and used in analysis at wave 4 (M = 21.58, SD = 14.34), wave 6 (M = 23.13, SD = 15.41), wave 8 (M = 23.19, SD = 16.58), wave 10 (M = 21.65, SD = 17.43), wave 12 (M = 23.39, SD = 17.84), wave 14 (M = 23.09, SD = 18.96), and wave 16 (M = 21.09, SD = 18.90). Means and standard deviations provided here are of the raw summed scores, as opposed to the latent factor.

Derived Maltreatment Dataset

The LONGSCAN researchers consolidated CM data regarding time of referral, number of CM allegations, and substantiated allegations within time periods, type of CM for each, and whether the type of CM experienced was singular or multiple. CM data was drawn from multiple sources. Maltreatment data was extracted and coded by trained coders from Child Protective Services Agency narratives using the Modified Maltreatment Classification System (English & The LONGSCAN Investigators, 1997). Additionally, self-report measures for physical and sexual abuse and emotional maltreatment were administered at age 12 (Measure Manual of Functioning and Outcomes in LONGSCAN: Middle Adolescence [Age 16]; LONGSCAN Coordinating Center, 2014). These sources were combined by LONGSCAN investigators into a single combined maltreatment variable for analysis. Types of maltreatment included in the current study include physical and sexual abuse, emotional maltreatment, and neglect. Definitions of maltreatment types were defined in the Modified Maltreatment Classification System (English and The LONGSCAN Investigators, 1997). Here, physical abuse is defined as the intentional infliction of an injury upon a child by a caregiver or responsible adult. Sexual abuse is any attempted or actual sexual contact between a caregiver of other responsible adult and a child for the purposes of the caregiver’s or responsible adult’s sexual gratification or financial benefit. Emotional maltreatment is defined as a parent’s relentless or extreme thwarting of a child’s basic emotional needs and/or acts by parents that are insensitive to a child’s developmental level, causing harm. Basic emotional needs include psychological safety and security, acceptance and self-esteem, and age-appropriate autonomy. Neglect is defined as physical neglect that either involves a failure to provide a minimum degree of care to meet the child’s physical needs or a lack of supervision involving inadequate precautions for a child’s safety relative to their developmental level. CM is recorded at intervals of two years, from birth to 16 years of age. No clear age cut-off dates were apparent for the intervals since the LONGSCAN researchers collected data from each person at a different date; however, because the data is recorded as counts of CM occurring within a particular age interval, there should be no overlaps in the intervals.

Data Analytic Plan

Preliminary data cleaning and analysis were conducted in STATA (16.1, StataCorp LLC, College Station, TX) and R version 4.0.3 (2020), with advanced modeling in MPlus (Version 8, Mac; Muthén & Muthén, 2017). Due to substantial variance in the attrition and retention of participants in the sample, the sample for each age group is comprised of overlapping but different sets of individuals – hence, we represent these age groups as ‘waves’ rather than ages. The LCMs were constructed similarly to the variables utilized by Dunn et al. (2018; Table 1). To summarize, the accumulation variables were the sum of all two-year period indicators – 1 or 0 – of each of physical, sexual, emotional maltreatment and/or neglect allegations. In other words, if each type occurred, then the two-year period would contribute 4 to the accumulation variable. Recency variables involved the same sum, but weighted two-year period indicators more highly if they were closer to the time of measurement. Finally, the sensitive period variables were defined as indicators – 1 or 0 – depending on whether any of physical, sexual, emotional maltreatment and/or neglect allegations had occurred. Accumulation and recency variables defined at each wave were time-varying – for example, the accumulation variable at wave 8 was the sum of two-year period indicators of combined maltreatment allegations leading up to age 8 for each individual. Key differences from Dunn et al. (2018) include: 1. leveraging additional assessment periods available in this set (eight 2-year periods: birth-16); 2. ability to model LCMs at each age due to more frequent CBCL assessments; 3. focus on physical and sexual abuse, emotional maltreatment, and neglect; 4. specified early childhood (ECSP; 0–4), middle childhood (MCSP; 4–8), pre-teen (PTSP; 8–12), teenage (TSP; 12–16) sensitive periods. The literature lacks consensus on how to divide age ranges into potential sensitive periods to be tested, since sensitive periods for maltreatment differ across different outcomes (Schaefer et al., 2022). As such, we divided our age range equally into these four age groups. Correlations between the LCMs can be found in the supplemental materials (Supplemental Tables 2–8). Correlations indicated that the accumulation and recency variables were very high across all seven waves (.90+), while sensitive periods were moderately correlated (range: .37–.63) with the accumulation and recency variables and were poorly correlated with other sensitive periods (range: .17–.34).

Table 1.

Description of Life Course Models Operationalized as Manifest Variables in the SEM.

Type	Description	Definition
Accumulation	The sum of indicator variables of each of physical, sexual, emotional maltreatment and neglect within two-year time periods between zero to the age of measurement	CombInd_0_to_2 = IndPhys_0_to_2 + IndSex_0_to_2 + IndEmo_0_to_2 + IndNeg_0_to_2 Accumulation_variable = CombInd_0_to_2^a + CombInd_2_to_4 + … + CombInd_ $(n - 2$ )_to_ $n$ ^b
Recency	The weighted sum of indicator variables of physical, sexual, emotional maltreatment and neglect within two-year time periods between zero to the age of measurement	Recency_variable = CombInd_0_to_2 +2 $\times$ CombInd_2_to_4 + … + $(\frac{n}{2}) \times$ CombInd_ $(n - 2$ )_to_ $n$
Early childhood sensitive period (ECSP)	An indicator variable for whether allegations of physical, sexual, emotional maltreatment and neglect appeared in the period between 0 to 4 years old	Ind_0_to_4 = 1 if CombInd_0_to_2 ≥ 1 OR CombInd2_to_4 ≥ 1 ECSP_variable = Ind_0_to_4
Middle childhood sensitive period (MCSP)	An indicator variable for whether allegations of physical, sexual, emotional maltreatment and neglect appeared in the period between 4 to 8 years old	MCSP_variable = Ind_4_to_8
Pre-teen sensitive period (PTSP)	An indicator variable for whether allegations of physical, sexual, emotional maltreatment and neglect appeared in the period between 8 to 12 years old	PSP_variable = Ind_8_to_12
Teenage sensitive period (TSP)	An indicator variable for whether allegations of physical, sexual, emotional maltreatment and neglect appeared in the period between 12 to 16 years old	TSP_variable = Ind_12_to_16

Note. This table is based on a table by Dunn et al. (2018) that defines the life course theoretical models accordingly.

^aThe indicator variable (“Ind”) equals 1 when any maltreatment allegation of any type is present during the specified period, and 0 when no maltreatment allegation is present during the specified period.

^b“n” represents the age of measurement in years.

Confirmatory factor analysis (CFA) using maximum likelihood parameter estimation with robust standard errors was applied as the data was distributed normally. The root mean square of approximation (RMSEA) and the standardized root mean square residual (SRMR) indexed fit. RMSEA <0.05 indicated close fit, RMSEA <0.08 reasonable fit (Browne & Cudeck, 1993), and SRMR <0.08 good fit (Hu & Bentler, 1999). Comparison of structural models at each wave was determined using Akaike Information Criteria (AIC), with the lowest AIC being selected. AIC weights were calculated from the AICs to demonstrate the probabilistic difference between the models. AIC weights are calculated using the difference between the model’s AIC and the smallest AIC among the models, then calculated as a proportion of the sum of each of these differences (Wagenmakers & Farrell, 2004). They can be understood as the “probability that [the model] is the best model, given the data and the set of candidate models” (Wagenmakers & Farrell, 2004, p. 194). AIC and related information criteria provide estimates of out-of-sample prediction error; that is, they anticipate how well a model can predict future data. They accomplish this by including penalty terms for the number of free parameters, which corrects for the ability of more complex models to more easily overfit the current data. A model with more parameters (and thus more flexibility) must not only outperform simpler models, but do by enough to also overcome this overfitting penalty. However, AIC only estimates out-of-sample prediction error up to a constant for a given sample of outcome values. As such, the absolute scale of AIC scores is meaningless; only the relative values matter among a collection of models that have been applied to the same sample, with the lowest AIC (i.e., lowest estimated error) being favored (for details, see McElreath, 2018).

To examine potential covariates, relations between demographics and the manifest DP variable (summed CBCL Aggressive Behavior, Attention Problems, Anxious/Depression subscales) indicated that gender had a minor effect on the manifest DP variable, while both race and site had a major effect at each wave. As site closely captured the racial patterns, χ2 (12, Ns = 868–1221) = 310.48–431.14, ps < .001), only site was subsequently included as a covariate in the cross-sectional models. Given the variability in items assessing treatment exposure across sites and substantive differences in access, models including both covariates failed to converge. As such, we use site as a proxy for treatment as well. The latent DP factor was regressed upon the LCMs and covariates in structural models at each wave. If the R² estimate was statistically significant using a Bonferroni-corrected critical p-value (p < .05/𝑛 comparisons), the model result estimates, standard errors, standardized model result estimates, and model result p-values for that LCM are reported.

Results

Measurement Models

A measurement model for DP was built for the 0–4 assessment wave and then applied to each subsequent wave to provide a stable operationalization of DP across time (supplemental Figure 1). This was modeled as a bifactor model as per previous operationalizations of the CBCP-DP (Deutz et al., 2016). Table 2 provides fit information for the CFA at each time point, before and after modification. RMSEA and SRMR indicated a close to good fit to the data for all measurement models. We allowed individual items on the CBCL to covary only if the item content supported theory to strengthen model fit at all assessment waves: CBCL 21 (destroys others’ things) with CBCL 20 (destroys own things); CBCL 10 (restless) with CBCL 8 (can’t concentrate); CBCL 14 (cries a lot) with CBCL 68 (screams a lot); CBCL 45 (high-strung) with CBCL 46 (nervous movements); CBCL 23 (disobedient at school) with CBCL 22 (disobedient at home); CBCL 95 (temper tantrums) with CBCL 68 (screams a lot); CBCL 74 (showing off) with CBCL 7 (boasting); CBCL 87 (sudden changes in mood) with CBCL 86 (irritable); CBCL 61 (poor school work) with CBCL 23; CBCL 80 (can’t concentrate) with CBCL 8 (can’t concentrate); CBCL 80 (can’t concentrate) with CBCL 13 (seems to be in a fog). Modification indices supported these changes.

Table 2.

Model Fit Parameters of Confirmatory Factor Analysis for DP Across Base and Modified Models.

Wave	Type	N	Chi Square	df	RMSEA	90% CI	SRMR
4	Base	1220	2476.18	857	0.039	0.038 0.041	0.046
4	Modified	1220	2045.051	845	0.034	0.032 0.036	0.043
6	Base	1221	2713.358	857	0.042	0.040 0.044	0.045
6	Modified	1221	2071.66	845	0.034	0.033 0.036	0.042
8	Base	1124	2670.266	857	0.043	0.042 0.045	0.047
8	Modified	1124	2035.161	845	0.035	0.033 0.037	0.044
10	Base	1017	2511.029	857	0.044	0.042 0.046	0.050
10	Modified	1017	2027.725	845	0.037	0.035 0.039	0.047
12	Base	953	2681.722	857	0.047	0.045 0.049	0.052
12	Modified	953	2125.838	845	0.040	0.038 0.042	0.049
14	Base	936	2655.309	857	0.047	0.045 0.049	0.051
14	Modified	936	2121.121	845	0.040	0.038 0.042	0.048
16	Base	867	2516.442	857	0.047	0.045 0.049	0.055
16	Modified	867	2067.854	845	0.041	0.039 0.043	0.052

Structural Models

Across waves, absolute fit indices suggested an excellent fit to the data using the criteria described above (Table 3). AIC weights were the highest for the recency model for the most number of waves (Age 6, w_i(AIC) = 0.647; Age 8, w_i(AIC) = 0.742; Age 10, w_i(AIC) = 0.677; Age 14 w_i(AIC) = 0.776), indicating it was the best fitting model overall. Other models selected for were the ECSP LCM at Age 4 (w_i(AIC) = 0.394), PTSP LCM at wave 12 (w_i(AIC) = 0.999) and the TSP at wave 16 (w_i(AIC) = 0.999). R² values across LCMs indicated that a small but statistically significant proportion of the variance was explained. Between 4.5%–11.2% of the variance in CM was explained by the LCMs, with the highest proportion of the variance explained by the accumulation and recency models and the teenage sensitive period model in Wave 16 (Figure 1 top). The MCSP, PTSP and TSP LCMs each outperformed the accumulation and recency LCMs at age 8, 12 and 16 respectively, but the MCSP and PTSP LCMs dropped off in performance.

Table 3.

Prediction of DP by Each LCM Sorted by Wave.

Wave 4	R ²	p-Value	B	SE	β	p-Value	χ ²	χ² df	RMSEA	90%CI	SRMR	AIC	w_i(AIC)
ACCUMULATION	0.030	0.004	—	—	—	—	2546.342	1060	0.034	0.032 0.036	0.043	68,643.605	0.376346672
RECENCY	0.029	0.005	—	—	—	—	2544.975	1060	0.034	0.032 0.036	0.043	68,644.590	0.229984225
ECSP (0–4)	0.03	0.004	—	—	—	—	2574.614	1060	0.034	0.033 0.036	0.043	68,643.515	0.393669104

Wave 6	R ²	p-Value	B	SE	β	p-Value	χ ²	χ² df	RMSEA	90%CI	SRMR	AIC	w_i(AIC)
ACCUMULATION	0.053	0.000^†	0.038	0.010	0.142	0.000*	2629.533	1060	0.035	0.033 0.036	0.042	72,048.242	0.350228694
RECENCY	0.054	0.000^†	0.020	0.005	0.144	0.000*	2637.761	1060	0.035	0.033 0.037	0.042	72,047.016	0.646508504
ECSP (0–4)	0.045	0.001^†	0.229	0.080	0.112	0.004*	2615.891	1060	0.035	0.033 0.036	0.041	72,057.594	0.003262802

Wave 8	R ²	p-Value	B	SE	β	p-Value	χ ²	χ² df	RMSEA	90%CI	SRMR	AIC	w_i(AIC)
ACCUMULATION	0.068	0.000^†	0.029	0.010	0.119	0.003*	2500.144	1060	0.035	0.033 0.037	0.042	64,774.647	0.240502657
RECENCY	0.070	0.000^†	0.013	0.004	0.126	0.002*	2497.377	1060	0.035	0.033 0.036	0.042	64,772.393	0.742283419
ECSP (0–4)	0.060	0.000^†	0.144	0.081	0.070	0.074	2484.795	1060	0.035	0.033 0.036	0.042	64,783.444	0.002957165
MCSP (4–8)	0.072	0.000^†	0.314	0.085	0.131	0.000*	2501.110	1060	0.035	0.033 0.037	0.042	64,780.298	0.014256759

Wave 10	R ²	p-Value	B	SE	β	p-Value	χ ²	χ² df	RMSEA	90%CI	SRMR	AIC	w_i(AIC)
ACCUMULATION	0.085	0.000^†	0.028	0.009	0.131	0.002*	2489.222	1060	0.036	0.035 0.038	0.045	57,608.319	0.289922198
RECENCY	0.086	0.000^†	0.010	0.003	0.134	0.001*	2490.758	1060	0.036	0.035 0.038	0.045	57,606.623	0.676960282
ECSP (0–4)	0.073	0.000^†	0.107	0.082	0.052	0.190	2484.810	1060	0.036	0.035 0.038	0.045	57,620.318	0.000719005
MCSP (4–8)	0.080	0.000^†	0.240	0.083	0.101	0.004*	2498.284	1060	0.037	0.035 0.038	0.045	57,612.702	0.032398515

Wave 12	R ²	p-Value	B	SE	β	p-Value	χ ²	χ² df	RMSEA	90%CI	SRMR	AIC	w_i(AIC)
ACCUMULATION	0.100	0.000^†	0.035	0.009	0.166	0.000*	2539.571	1060	0.038	0.037 0.040	0.046	53,856.674	0.000107738
RECENCY	0.104	0.000^†	0.011	0.003	0.172	0.000*	2540.483	1060	0.038	0.036 0.040	0.046	53,852.948	0.000694162
ECSP (0–4)	0.083	0.000^†	0.175	0.089	0.084	0.049*	2538.375	1060	0.038	0.036 0.040	0.046	53,872.845	3.31804 × 10⁻⁸
MCSP (4–8)	0.093	0.000^†	0.307	0.091	0.126	0.001*	2549.283	1060	0.038	0.036 0.040	0.046	53,863.465	3.61179 × 10⁻⁶
PTSP (8–12)	0.119	0.000^†	0.543	0.101	0.211	0.000*	2562.782	1060	0.039	0.037 0.040	0.046	53,838.404	0.999194455

Wave 14	R ²	p-Value	B	SE	β	p-Value	χ ²	χ² df	RMSEA	90%CI	SRMR	AIC	w_i(AIC)
ACCUMULATION	0.074	0.000^†	0.039	0.008	0.194	0.000*	2603.646	1060	0.039	0.038 0.041	0.047	51,386.369	0.199247855
RECENCY	0.077	0.000^†	0.010	0.002	0.193	0.000*	2601.808	1060	0.039	0.038 0.041	0.047	51,383.651	0.775532238
ECSP (0–4)	0.053	0.000^†	0.232	0.088	0.112	0.008*	2605.074	1060	0.039	0.038 0.041	0.047	51,405.782	1.21315 × 10⁻⁵
MCSP (4–8)	0.050	0.001^†	0.174	0.091	0.072	0.055	2588.507	1060	0.039	0.037 0.041	0.046	51,408.867	2.59427 × 10⁻⁶
PTSP (8–12)	0.07	0.000^†	0.422	0.101	0.164	0.000*	2588.330	1060	0.039	0.037 0.041	0.046	51,390.504	0.025205181

Wave 16	R ²	p-Value	B	SE	β	p-Value	χ ²	χ² df	RMSEA	90%CI	SRMR	AIC	w_i(AIC)
ACCUMULATION	0.078	0.000^†	0.027	0.009	0.145	0.003*	2480.041	1060	0.039	0.037 0.041	0.049	44,840.853	2.63687 × 10⁻⁷
RECENCY	0.083	0.000^†	0.008	0.002	0.158	0.001*	2480.317	1060	0.039	0.037 0.041	0.049	44,836.512	2.3106 × 10⁻⁶
ECSP (0–4)	0.066	0.000^†	0.173	0.094	0.083	0.065	2481.801	1060	0.039	0.037 0.041	0.049	44,851.343	1.39064 × 10⁻⁹
MCSP (4–8)	0.067	0.000^†	0.178	0.095	0.073	0.061	2468.333	1060	0.039	0.037 0.041	0.049	44,850.842	1.78651 × 10⁻⁹
PTSP (8–12)	0.072	0.000^†	0.275	0.104	0.109	0.007*	2488.476	1060	0.039	0.037 0.041	0.049	44,845.721	2.31215 × 10⁻⁸
TSP (12–16)	0.112	0.000^†	0.676	0.124	0.234	0.000*	2462.579	1060	0.039	0.037 0.041	0.049	44,810.556	0.999997399

Note. Model results are not provided when the R² p-values are non-significant. For R² p-values, significance is indicated with ^†p ≤ .001. For parameter estimate p-values, significance is indicated with *p < .05. SE, standard error; df, degrees of freedom; RMSEA, root mean square error of approximation; CI, confidence interval; SRMR, standardized root mean square; AIC, Akaike information criterion. w_i(AIC), Akaike weight; w_i(AIC)s that are bolded indicate the highest weighted model in that wave.

Figure 1.

Model R² (top; p < .001) and standardized parameter estimates for statistically significant LCMs (bottom; p < .05).

There was a general upward trend in effect size for the accumulation and recency variables as age increased, but a dip at wave 16 (Table 4; Figure 1 bottom). Standardized parameter estimates ranged from 0.05–0.23, with the highest effect sizes for the accumulation and recency variables at wave 14. The sensitive period variables were most effective in prediction at the wave they were introduced, with downward trends as time passed. The exception occurred for ECSP, which had the lowest standardized estimates of all the variables at each time period. A temporal pattern for TSP could not be determined since it only had one standardized estimate at Wave 16 (Figure 1 bottom) but evidenced the highest standardized estimate across LCMs.

Table 4.

Standardized Model Estimates for Each LCM Across Each Wave.

	4	6	8	10	12	14	16
Accumulation	—	0.142*	0.119*	0.131*	0.166*	0.194*	0.145*
Recency	—	0.144*	0.126*	0.134*	0.172*	0.193*	0.158*
Early childhood sensitive period (0–4)	—	0.112*	0.070	0.052	0.084*	0.112*	0.083
Middle childhood sensitive period (4–8)			0.131*	0.101*	0.126*	0.072	0.073
Pre-teen sensitive period (8–12)					0.211*	0.164*	0.109*
Teenage sensitive period (12–16)							0.234*

Note. ECSP, early childhood sensitive period; MCSP, middle childhood sensitive period; PTSP, pre-teen sensitive period; TSP, teenage sensitive period. Model estimates are not provided when not applicable. *p < .05.

Discussion

We examined relations between LCMs and general psychopathology, using the DP framework. Three LCMs with strong empirical and theoretical support were tested: the accumulation, recency, and sensitive period LCMs. We estimated the degree to which each CM LCMs explained general psychopathology outcomes from birth to 16 years of age. All three models related to psychopathology and explained larger proportions of the variance than found previously. Overall, the recency LCM best explained general psychopathology with slightly larger effect sizes than the accumulation LCM across age. Furthermore, the model with the recency variable had the best model fit across the most waves. The sensitive periods most proximal to the DP assessment were the most robust, consistent with Russotti et al. (2021) where recent CM was most predictive of future psychopathology. Unique to this study, the ECSP and MCSP LCMs influenced psychopathology in adolescence as well as in earlier childhood, suggesting a delayed effect. Our findings are consistent with previous findings that CM has small direct causal effects on mental health, whether assessed cross-sectionally or longitudinally, for different mental health outcomes, and across sex and age (Baldwin et al., 2023).

These results provide support for both the recency and accumulation theories. The recency model was generally the best performing, and proximal sensitive periods served as the most explanatory LCM for those periods. Taken in combination, this appears to provide support for the recency theory - namely, that CM proximal is more influential in explaining psychopathology contemporaneously than CM occurring earlier. The mechanism underlying this pattern is unclear. However, resilience theory suggests that people exposed to traumatic events may initially experience symptomatology, then after some time evidence a trajectory of gradual recovery to reach normal functioning (Bonanno et al., 2011).

It is unclear whether the recency model’s effects were a product of recency itself or its overlap with the accumulation model. Recency and accumulation models were very highly correlated, with values greater than 0.90 across all waves, limiting our ability to interpret these models separately. The recency model effect sizes were equal or slightly better than the accumulation model effects, but these differences were not substantial. As these models generally outperformed the sensitive period models, this lends support to the cumulative (summative frequency) model of CM (Appleyard et al., 2005). This aligns with previous research finding stronger associations between the dose-response/periodic frequency models of CM and psychopathology than sensitive period models (Jaffee & Maikovich-Fong, 2011; Jonson-Reid et al., 2012; Mackenzie et al., 2011; Russotti et al., 2021). Recent evidence suggests emotional regulation may mediate relations between childhood adversity and psychopathology (Miu et al., 2022). Emotional dysregulation plays a central role in DP (Masi et al., 2015), forming part of the persistent vulnerability profile for psychopathology. As emotion regulation changes across the lifespan (Ruba & Pollak, 2020), this may explain why the accumulation model generally outperformed other LCMs; patterns of pervasive CM may lead to consistent disruptions in the healthy development of emotional regulation, leading to worse psychopathology outcomes over time.

These results provide only limited support the sensitive period model. This model predicts that CM experienced in a specific period are more influential in terms of future functioning than CM experienced in any other period; our results do not align with this prediction. All four sensitive period LCMs had their strongest effects when first introduced, with reduced influence on psychopathology as time passed (excluding TSP assessed at a single timepoint). This finding is consistent with previous findings showing a lack of consistent sensitive periods across a variety of child outcomes (Schaefer et al., 2022). Since there was no consistent sensitive period found to be influential at multiple timepoints, but rather only when the sensitive period was proximal to the outcome measurement timepoint, this finding was interpreted as supporting the recency model rather than the sensitive period model. One explanation for this inconsistency could be that different child outcomes may have different sensitive periods. For instance, a systematic review of the link between CM and depression found that exposure to CM in middle childhood led to the highest risk of developing depression at any age (Li et al., 2023), while exposure to violence between birth and three years old was associated with greater externalizing symptoms than exposure to violence at older ages (Dunn et al., 2020). Sensitive periods may also be less detectable in models incorporating composite measures of maltreatment rather than individual types (Schaefer et al., 2022). In any case, our study examined which temporal pattern of CM best explained general psychopathology, and the sensitive period model provided the worst account for these changes.

An important theoretical point is that none of these temporal models of CM are mutually exclusive. Temporal patterns of psychopathology may be the result of a combination of the effects of accumulation, recency, and sensitive periods. Our study indicates that both accumulation and recency effects may be present, each potentially explained by different mechanisms. Advancing our theory in this regard will require further longitudinal work testing potential mechanisms involved in accumulation and recency, such as disruptions to emotional regulation development. Trajectory approaches are particularly informative when considering such time-varying predictors.

This study demonstrates several strengths. This is the first study to utilize a measure of general psychopathology in testing the effects of the temporal dimension of CM. Given the value of general psychopathology factors as indicators of clinical impairment, this study demonstrates developmental associations between combined maltreatment and impairment broadly construed. The use of a large, national prospective longitudinal sample of children afforded the opportunity to extend the time period of assessment from previous work (Li & Godinet, 2014) and provide a more holistic view of psychopathology. The two-year assessment periods also afford a greater number of observational periods than Russotti et al. (2021) work. As the data was captured prospectively from case files, we avoid the self-presentation bias and memory-related issues associated with retrospective self-report (Greenhoot, 2011; MacMillan et al., 2003) and rule out the possibility that recency effects can be explained solely by recall bias. The LONGSCAN sample was intentionally selected to maximize the degree of CM measured, thus increasing the number of incidents available and the capacity for the LCM model to predict psychopathology outcomes. We specifically accounted for site differences in the models, given the variability between sites in study conditions, sample characteristics, data collection, and access to treatment (Runyan et al., 2020).

This work also has limitations. Since the LONGSCAN sample is comprised of children at high-risk of CM, the generalizability of findings to lower-risk populations is limited. Furthermore, we were unable to assess the effects of mental health treatment on the sample. Given that access to treatment was conflated with study location, and that operationalization of treatment varied greatly across years and locations, model convergence was negatively impacted as a result. Location was the covariate used in analysis to address this limitation but is not optimal. Additionally, we had the choice of allegations, validated reports, and substantiated reports to operationalize CM. Empirical work has shown that the risk experienced by children who are the subject of unsubstantiated reports is often equally high as substantiated reports, reflected in rates of entry into special education (Jonson-Reid et al., 2004), subsequent placement into foster care (Kohl et al., 2009), and a wide range of developmental, cognitive, emotional, and substance use outcomes (Casanueva et al., 2012). Moreover, we largely adhered to the operationalization of the LCMs by Dunn et al. (2018) which had substantial overlap between accumulation and recency models, leading to very high correlations between the models in our sample. Other operationalizations of accumulation and recency should be explored in future work. This issue contributed to collinearity of the LCMs, requiring us to run these models separately. Another issue concerning the operationalization of the LCMs was that the accumulation and recency models had greater variability in their scores compared to the sensitive period operationalization, which may have advantaged the former LCMs over the latter in the model comparisons. A potential future direction would be to examine the LCMs under different operationalizations so that the models can be tested simultaneously, as this would allow stronger claims to be made for favoring one model over another. Finally, the nature of this study aimed to capture the general effects of a broadly defined CM on a general factor of psychopathology, DP. The accumulation and recency variables captured the quantities of different types of maltreatment in their operationalizations, while the sensitivity period was operationalized as an indicator of whether any of the different types of maltreatment occurred. Given that these operationalizations aggregated multiple types of maltreatment and types of outcomes in different ways, a new approach may be necessary to determine causal risk mechanisms of specific types of maltreatment for specific outcomes.

Our findings have important clinical implications. Clinicians should seek to capture timing information in their CM screening to improve prediction of general psychopathology (English et al., 2005). This information may help clinicians tailor their therapeutic strategies in accordance with the expected trajectory of psychopathology symptoms. While prevention efforts should ideally target children of all ages at risk of CM, these results imply that we should prioritize efforts targeting children who are likely to experience abuse across multiple developmental periods (Hindley et al., 2006). Risk factors for a high periodic frequency of CM should be investigated and identified to fine-tune effective prevention efforts. Furthermore, the acute effects of proximal CM on psychopathology suggests that additional clinical attention should be paid to children who have recently experienced CM, regardless of age. Our results underline the need for a stronger theoretical base for the chronic and recency models of CM and reinforce the explanatory value of the temporal dimension of CM for general psychopathology.

Supplemental Material

Supplemental Material - Life Course Models of Child Maltreatment: Effects on General Psychopathology Outcomes in a Longitudinal Sample

Supplemental Material for Life Course Models of Child Maltreatment: Effects on General Psychopathology Outcomes in a Longitudinal Sample by Dylan Wong, and Kristen G. Anderson in Child Maltreatment.

Footnotes

Acknowledgments

The authors would like to thank Gregory Jensen, PhD for his feedback on statistical analyses.

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) received no financial support for the research, authorship, and/or publication of this article.

Ethical Statement

ORCID iD

Kristen G. Anderson

Supplemental Material

Supplemental material for this article is available online.

References

Achenbach

T. M.

Ruffle

T. M.

(2000). The child behavior checklist and related forms for assessing behavioral/emotional problems and competencies. Pediatrics in Review, 21(8), 265–271. https://doi.org/10.1542/pir.21-8-265

Appleyard

Egeland

van Dulmen

M. H. M.

Sroufe

L. A.

(2005). When more is not better: The role of cumulative risk in child behavior outcomes. The Journal of Child Psychology and Psychiatry and Allied Disciplines, 46(3), 235–245. https://doi.org/10.1111/j.1469-7610.2004.00351.x

Baldwin

J. R.

Wang

Karwatowska

Schoeler

Tsaligopoulou

Munafò

M. R.

Pingault

J.-B.

(2023). Childhood maltreatment and mental health problems: A systematic review and meta-analysis of quasi-experimental studies. American Journal of Psychiatry, 180(2), 117–126. https://doi.org/10.1176/appi.ajp.20220174

Biederman

Wozniak

Kiely

Ablon

Faraone

Mick

Mundy

Kraus

(1995). CBCL clinical scales discriminate prepubertal children with structured interview-derived diagnosis of mania from those with ADHD. Journal of the American Academy of Child & Adolescent Psychiatry, 34(4), 464–471.

Bonanno

G. A.

Westphal

Mancini

A. D.

(2011). Resilience to loss and potential trauma. Annual Review of Clinical Psychology, 7(1), 511–535. https://doi.org/10.1146/annurev-clinpsy-032210-104526

Browne

M. W.

Cudeck

(1993). Alternative ways of assessing model fit. In Bollen

K. A.

Long

J. S.

(Eds.), Testing structural equation models (pp. 136–162). Sage.

Casanueva

Dolan

Smith

Ringeisen

(2012). NSCAW child well-being spotlight: Children with substantiated and unsubstantiated reports of child maltreatment are at similar risk for poor outcomes OPRE Report #2012-31. Office of Planning, Research and Evaluation, Administration for Children and Families, U.S. Department of Health and Human Services.

Caspi

Houts

R. M.

Belsky

D. W.

Goldman-Mellor

S. J.

Harrington

Israel

Meier

M. H.

Ramrakha

Shalev

Poulton

Moffitt

T. E.

(2014). The p factor: One general psychopathology factor in the structure of psychiatric disorders? Clinical Psychological Science: A Journal of the Association for Psychological Science, 2(2), 119–137. https://doi.org/10.1177/2167702613497473

Cicchetti

Valentino

(2015). An Ecological-Transactional Perspective on Child Maltreatment: Failure of the Average Expectable Environment and Its Inf luence on Child Development. In Cicchetti

Cohen

D. J.

(Eds.), Developmental Psychopathology: Volume Three: Risk, Disorder, and Adaptation, Second Edition. Wiley.

10.

Deutz

M. H. F.

Geeraerts

S. B.

Belsky

Deković

van Baar

A. L.

Prinzie

Patalay

(2020). General psychopathology and dysregulation profile in a longitudinal community sample: Stability, antecedents and outcomes. Child Psychiatry and Human Development, 51(1), 114–126. https://doi.org/10.1007/s10578-019-00916-2

11.

Deutz

M. H. F.

Geeraerts

S. B.

van Baar

A. L.

Deković

Prinzie

(2016). The Dysregulation Profile in middle childhood and adolescence across reporters: Factor structure, measurement invariance, and links with self-harm and suicidal ideation. European Child & Adolescent Psychiatry, 25(4), 431–442. https://doi.org/10.1007/s00787-015-0745-x

12.

Deutz

M. H. F.

Shi

Vossen

H. G. M.

Huijding

Prinzie

Deković

van Baar

A. L.

Woltering

(2018). Evaluation of the strengths and difficulties questionnaire-dysregulation profile (SDQ-DP). Psychological Assessment, 30(9), 1174–1185. https://doi.org/10.1037/pas0000564

13.

Dunn

E. C.

Soare

T. W.

Raffeld

M. R.

Busso

D. S.

Crawford

K. M.

Davis

K. A.

Fisher

V. A.

Slopen

Smith

A. D. A. C.

Tiemeier

Susser

E. S.

(2018). What life course theoretical models best explain the relationship between exposure to childhood adversity and psychopathology symptoms: Recency, accumulation, or sensitive periods? Psychological Medicine, 48(15), 2562–2572. https://doi.org/10.1017/S0033291718000181

14.

English

D. J.

The LONGSCAN Investigators . (1997). Modified maltreatment classification system (MMCS). As modified from the maltreatment classification system outlined in Barnett, D., Manly, J.T., & Cicchetti, D. (1993). Defining child maltreatment: The interface between policy and research. In Cichetti

Toth

S. L.

(Eds.), Advances in applied developmental psychology: Child abuse, child development and social policy (pp. 7–74). Ablex Publishing Corp. For more information visit the LONGSCAN website at. https://www.iprc.unc.edu/longscan

15.

English

D. J.

Upadhyaya

M. P.

Litrownik

A. J.

Marshall

J. M.

Runyan

D. K.

Graham

J. C.

Dubowitz

(2005). Maltreatment’s wake: The relationship of maltreatment dimensions to child outcomes. Child Abuse & Neglect, 29(5), 597–619. https://doi.org/10.1016/j.chiabu.2004.12.008

16.

Greenhoot

A. F.

(2011). Retrospective methods in developmental science. In Laursen

Little

T. D.

Card

N. A.

(Eds.), Handbook of developmental research methods (pp. 196–210). Guilford Press.

17.

Harpur

L. J.

Polek

van Harmelen

A.-L.

(2015). The role of timing of maltreatment and child intelligence in pathways to low symptoms of depression and anxiety in adolescence. Child Abuse & Neglect, 47, 24–37. https://doi.org/10.1016/j.chiabu.2015.05.019

18.

Hindley

Ramchandani

P. G.

Jones

D. P. H.

(2006). Risk factors for recurrence of maltreatment: A systematic review. Archives of Disease in Childhood, 91(9), 744–752. https://doi.org/10.1136/adc.2005.085639

19.

Bentler

P. M.

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6(1), 1–55. https://doi.org/10.1080/10705519909540118

20.

Hunter

W. M.

Cox

C. E.

Teagle

Johnson

R. M.

Mathew

Knight

E. D.

Leeb

R. T.

Smith

J. B.

(2001). Measures for assessment of functioning and outcomes in longitudinal research on child abuse Volume 1: Early childhood. In LONGSCAN coordinating center. University of North Carolina at Chapel Hill.

21.

Hunter

W. M.

Cox

C. E.

Teagle

Johnson

R. M.

Mathew

Knight

E. D.

Leeb

R. T.

Smith

J. B.

(2002). Measures for assessment of functioning and outcomes in longitudinal research on child abuse Volume 2: Middle childhood. In LONGSCAN coordinating center. University of North Carolina at Chapel Hill.

22.

Ireland

T. O.

Smith

C. A.

Thornberry

T. P.

(2002). Developmental issues in the impact of child maltreatment on late delinquency and drug use*. Criminology, 40(2), 359–400. https://doi.org/10.1111/j.1745-9125.2002.tb00960.x

23.

Jaffee

S. R.

Maikovich-Fong

A. K.

(2011). Effects of chronic maltreatment and maltreatment timing on children’s behavior and cognitive abilities: Effects of chronic maltreatment. The Journal of Child Psychology and Psychiatry and Allied Disciplines, 52(2), 184–194. https://doi.org/10.1111/j.1469-7610.2010.02304.x

24.

Jonson-Reid

Drake

Kim

Porterfield

Han

(2004). A prospective analysis of the relationship between reported child maltreatment and special education eligibility among poor children. Child Maltreatment, 9(4), 382–394. https://doi.org/10.1177/1077559504269192

25.

Jonson-Reid

Kohl

P. L.

Drake

(2012). Child and adult outcomes of chronic child maltreatment. Pediatrics, 129(5), 839–845. https://doi.org/10.1542/peds.2011-2529

26.

Knight

E. D.

Smith

J. B.

Martin

L. M.

Lewis

The LONGSCAN Investigators . (2010). Measures for assessment of functioning and outcomes in longitudinal research on child abuse Volume 3: Early adolescence. LONGSCAN Coordinating Center, University of North Carolina at Chapel Hill.

27.

Knight

E. D.

Smith

J. B.

Martin

L. M.

Lewis

The LONGSCAN Investigators . (2014a). Measures for assessment of functioning and outcomes in longitudinal research on child abuse and neglect Volume 4: Middle adolescence (age 16). LONGSCAN Coordinating Center, University of North Carolina at Chapel Hill.

28.

Knight

E. D.

Smith

J. B.

Martin

L. M.

Lewis

The LONGSCAN Investigators . (2014b). Measures for assessment of functioning and outcomes in longitudinal research on child abuse and neglect Volume 5: Young adulthood (age 18). LONGSCAN Coordinating Center, University of North Carolina at Chapel Hill.

29.

Kohl

P. L.

Jonson-Reid

Drake

(2009). Time to leave substantiation behind: Findings from a national probability study. Child Maltreatment, 14(1), 17–26. https://doi.org/10.1177/1077559508326030

30.

Godinet

M. T.

(2014). The impact of repeated maltreatment on behavioral trajectories from early childhood to early adolescence. Children and Youth Services Review, 36, 22–29. https://doi.org/10.1016/j.childyouth.2013.10.014

31.

Gao

Zhang

Yang

D’Arcy

Meng

(2023). The timing effect of childhood maltreatment in depression: A systematic review and meta-analysis. Trauma, Violence, & Abuse, 24(4), 2560–2580. https://doi.org/10.1177/15248380221102558

32.

LONGSCAN Coordinating Center . (2014). Measures for assessment of functioning and outcomes in longitudinal research on child abuse and neglect Volume 4: Middle adolescence (age 16). LONGSCAN Coordinating Center, University of North Carolina at Chapel Hill.

33.

MacKenzie

M. J.

Kotch

J. B.

Lee

L.-C.

Augsberger

Hutto

(2011). A cumulative ecological–transactional risk model of child maltreatment and behavioral outcomes: Reconceptualizing early maltreatment report as risk factor. Children and Youth Services Review, 33(11), 2392–2398. https://doi.org/10.1016/j.childyouth.2011.08.030

34.

MacMillan

H. L.

Jamieson

Walsh

C. A.

(2003). Reported contact with child protection services among those reporting child physical and sexual abuse: Results from a community survey. Child Abuse & Neglect, 27(12), 1397–1408. https://doi.org/10.1016/j.chiabu.2003.06.003

35.

Martel

M. M.

Pan

P. M.

Hoffmann

M. S.

Gadelha

do Rosário Mari

J. J.

Manfro

G. G.

Miguel

E. C.

Paus

Bressan

R. A.

Rohde

L. A.

Salum

G. A.

(2017). A general psychopathology factor (P factor) in children: Structural model analysis and external validation through familial risk and child global executive function. Journal of Abnormal Psychology, 126(1), 137–148. https://doi.org/10.1037/abn0000205

36.

Masi

Muratori

Manfredi

Pisano

Milone

(2015). Child behaviour checklist emotional dysregulation profiles in youth with disruptive behaviour disorders: Clinical correlates and treatment implications. Psychiatry Research, 225(1–2), 191–196. https://doi.org/10.1016/j.psychres.2014.11.019

37.

McElreath

(2018). Statistical rethinking: A Bayesian course with examples in R and stan. Chapman and Hall/CRC.

38.

Miu

A. C.

Szentágotai-Tătar

Balázsi

Nechita

Bunea

Pollak

S. D.

(2022). Emotion regulation as mediator between childhood adversity and psychopathology: A meta-analysis. Clinical Psychology Review, 93, 102141. https://doi.org/10.1016/j.cpr.2022.102141

39.

Muthén

L. K.

Muthén

B. O.

(1998-2017). Mplus user’s guide. (8th ed.). Muthén & Muthén.

40.

Patalay

Fonagy

Deighton

Belsky

Vostanis

Wolpert

(2015). A general psychopathology factor in early adolescence. The British Journal of Psychiatry: The Journal of Mental Science, 207(1), 15–22. https://doi.org/10.1192/bjp.bp.114.149591

41.

Ruba

A. L.

Pollak

S. D.

(2020). The development of emotion reasoning in infancy and early childhood. Annual Review of Developmental Psychology, 2(1), 503–531. https://doi.org/10.1146/annurev-devpsych-060320-102556

42.

Runyan

D. K.

Dubowitz

(2020). Longitudinal studies of child abuse and neglect (LONGSCAN) assessments 0–18. LONGSCAN Coordinating Center. University of North Carolina at Chapel Hill.

43.

Russotti

Warmingham

J. M.

Duprey

E. B.

Handley

E. D.

Manly

J. T.

Rogosch

F. A.

Cicchetti

(2021). Child maltreatment and the development of psychopathology: The role of developmental timing and chronicity. Child Abuse & Neglect, 120, 105215. https://doi.org/10.1016/j.chiabu.2021.105215

44.

Schaefer

J. D.

Cheng

T. W.

Dunn

E. C.

(2022). Sensitive periods in development and risk for psychiatric disorders and related endpoints: A systematic review of child maltreatment findings. The Lancet Psychiatry, 9(12), 978–991. https://doi.org/10.1016/S2215-0366(22)00362-5

45.

Shanahan

Copeland

W. E.

Costello

E. J.

Angold

(2011). Child-, adolescent- and young adult-onset depressions: Differential risk factors in development? Psychological Medicine, 41(11), 2265–2274. https://doi.org/10.1017/S0033291711000675

46.

Smith

G. T.

Atkinson

E. A.

Davis

H. A.

Riley

E. N.

Oltmanns

J. R.

(2020). The general factor of psychopathology. Annual Review of Clinical Psychology, 16(1), 75–98. https://doi.org/10.1146/annurev-clinpsy-071119-115848

47.

Thornberry

T. P.

Ireland

T. O.

Smith

C. A.

(2001). The importance of timing: The varying impact of childhood and adolescent maltreatment onmultiple problem outcomes. Development and Psychopathology, 13(4), 957–979. https://doi.org/10.1017/S0954579401004114

48.

Tracy

Salo

Slopen

Udo

Appleton

A. A.

(2019). Trajectories of childhood adversity and the risk of depression in young adulthood: Results from the Avon Longitudinal Study of Parents and Children. Depression and Anxiety, 36(7), 596–606. https://doi.org/10.1002/da.22887

49.

Wagenmakers

E.-J.

Farrell

(2004). AIC model selection using Akaike weights. Psychonomic Bulletin & Review, 11(1), 192–196. https://doi.org/10.3758/BF03206482

50.

WHO . (2022). Child maltreatment. WHO. Retrieved October 3, 2020, from. https://www.who.int/news-room/fact-sheets/detail/child-maltreatment

51.

Widom

C. S.

(2014). Longterm consequences of child maltreatment. In Korbin

J. E.

Krugman

R. D.

(Eds.), Handbook of child maltreatment (Vol. 2). Springer. https://doi.org/10.1007/978-94-007-7208-3_12

52.

Young-Southward

Eaton

O’Connor

Minnis

(2020). Investigating the causal relationship between maltreatment and cognition in children: A systematic review. Child Abuse & Neglect, 107, 104603. https://doi.org/10.1016/j.chiabu.2020.104603

53.

Zeman

Suveg

West

K. B.

(2020). Developmental psychopathology: Basic principles. In Maddux

J. E.

Winstead

B. A.

(Eds.), Psychopathology: Foundations for a contemporary understanding. Routledge.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.17 MB