Sleep problems over a year in sexually abused preschoolers

Abstract

Objectives:

The aim of the present study was to explore sleep problems in sexually abused preschoolers over a year.

Methods:

The sample consisted of 224 abused children and 83 nonabused children aged 3 to 6 years old at enrolment into the study (T1), and 85 abused children and 73 nonabused children at the second evaluation, approximately 1 year later (T2). Sleep problems were assessed using parental reports on the Child Behavior Checklist – Preschool Version.

Results:

Multivariate analysis of covariance revealed that sexually abused preschoolers presented higher scores of sleep problems than nonabused children on all items related to sleep problems at T1. Results from a regression analysis revealed that sexual abuse remained associated with higher levels of sleep problems after controlling for sociodemographical variables. At T2, abused preschoolers still presented more sleep problems than nonabused children on all but one of the sleep items. Results from a repeated measure one-way analysis of covariance revealed that abused preschoolers presented higher total sleep scores at both measurement times. Time was not associated with levels of sleep problems and sleep problems did not evolve differently for abused and nonabused children.

Conclusions:

Sexually abused preschoolers are at risk of presenting with sleep problems. These results are worrisome given the negative impacts of sleep problems on child development. Research and clinical implications are discussed.

Child Sexual Abuse

A number of scholarly reports have identified child sexual abuse (CSA) as a major social problem associated with deleterious consequences. A meta-analysis estimated the prevalence rate of CSA at 18% among girls and 7.6% among boys worldwide (1). Reviews conclude that CSA is a risk factor for a variety of psychological disorders, physical health and medical problems (2,3).

A number of authors have pointed out major gaps in the scientific literature on CSA. The bulk of published studies had focused on identifying long-term outcomes, studying adult populations. Thus, we know little about the mechanisms explaining the links between CSA and early precursors of dysfunctional developmental trajectories. Available studies indicate that sexually abused preschoolers present high levels of internalized (e.g., depression, anxiety, isolation) and externalized (e.g., aggressive behaviours, attention deficits) behaviour problems, as well as higher rates of problematic sexual behaviours and dissociation (e.g., ‘disruption of and/or discontinuity in the normal integration of consciousness, memory, identity emotion, perception, body representation, motor control and behaviour’) (4) than nonabused preschoolers (5–7). Sleep problems might be a relevant variable to consider in trying to identify mechanisms leading to maladaptation.

Sleep Problems in Preschool Children

Sleep disturbances can have detrimental effects on children’s functioning. A meta-analysis evaluated the associations between sleep duration and efficiency, and cognitive abilities and behavioural problems among healthy school-aged children (8). Results revealed that sleep duration was associated with executive and cognitive functioning, as well as school performance. Studies with preschool samples indicate that sleep duration (9), sleep disturbances (9) and night-time fears (10) are associated with inattention and parental reports of child internalized and externalized behaviour problems.

Sleep Problems and CSA

A literature review showed that sleep disturbances were common among CSA survivors (11). Female survivors of CSA surveyed 10 years after the disclosure reported more sleep disturbances than their nonabused counterparts, even after controlling for the effects of traumatic and depressive symptomatologies (12). Furthermore, sleep disturbances were associated with revictimization. Caldwell and Redeker (13) stated that adolescent rape victims presented recurrent sleep disturbances (e.g., frequent wakings, nightmares) impairing daily functioning.

However, little is known on the links between CSA and sleep disturbances in childhood. The available studies were, for the majority, conducted in the 1990. A longitudinal study of children (1 to 16 years old) found that at the initial assessment (T1), 20% of victims presented with important sleep problems (14). Rates increased over time to 34% 9 months later, and 33% 2 years later. In the absence of a comparison group, it remains difficult to qualify these results. Another study, including a psychiatric inpatient sample (2 to 13 years old) compared sleep difficulties in sexually abused, physically abused and nonabused children (15). No differences were found between groups regarding difficulties falling asleep and night awakenings but a greater incidence of parasomnias (e.g., abnormal experiences while asleep like sleepwalking, nightmares and sleep paralysis) was found within the CSA group. Despite the relevance of this study, results may be specific to the clientele surveyed.

Overall, studies indicate that sexually abused children tend to present more sleep disturbances than nonabused children. Nevertheless, no known study has investigated sleep problems in the short and medium term among sexually abused preschoolers. Studies reviewed included a wide age range, despite the age-specific characteristics of sleep and its impacts (8,16). In this context, the aim of the present longitudinal cohort study was to explore sleep problems over a year in sexually abused preschoolers.

METHOD

Participants and procedure

At T1, the sample consisted of 224 children (191 girls, 33 boys), aged 3 to 6 years old, who were referred to the Child Protection Clinic of Ste-Justine Hospital, a tertiary-care paediatric university hospital in Montreal, for evaluation following alleged CSA. As per Quebec laws, cases were reported to Youth Protective Agencies. The comparison group comprised 83 children (49 girls, 34 boys) recruited in kindergarten or daycare centres selected from neighbourhoods with socioeconomic levels comparable to families in the CSA group. At the follow-up assessment (T2), approximately 1 year later, the parents of 158 children (85 in the CSA group), completed the measures used for this study. Exclusion criteria were intellectual disabilities and not speaking French or English.

Parents completed the questionnaires at the intervention centre or at home with a research assistant if necessary. The study was approved by the internal review board of Ste-Justine Hospital and of the Université du Québec à Montréal.

Measures

Abuse-related variables were coded at T1 from the medical files using an adaptation of the History of Victimization Questionnaire (17). Severity refers to the nature of the acts involved, chronicity refers to the duration of the abuse and the number of episode to the number of times the child was sexually abused. The relationship with the abuser, allowing to distinguish between abuse involving an intrafamilial or extrafamilial aggressor, was also coded.

Children’s behavioural problems were assessed using parent reports. The Child Behavior Checklist (CBC), preschool version (18), is a widely used questionnaire designed to assess children’s behavioural difficulties. The CBC has shown good validity and reliability (18). This measure, completed at T1 and T2, comprises 100 items scored on a three-point scale based on the frequency of behaviours in the past 2 months (0=Not true, 1=Sometimes True, 2=Often True). Higher scores reflect greater behavioural difficulties and clinical cut-off scores are provided (T score of ≥70 on any subscale). The CBC T-score is an age-standardized score with a mean of 50 and a standard deviation of 10. The present analysis used results from the 7-item sleep problems subscale (α=0.82). While this questionnaire is not a specialized measure of sleep functioning and cannot be used to thoroughly assess sleep problems and disorders, the convergent, discriminant and external validity of the sleep subscale have been demonstrated through a comparison with validated sleep measures and sleep disorders diagnosis, as well as through its association with external correlates of sleep functioning (19,20).

RESULTS

At T1 and T2 respectively, mean age of participating children was 4.7 (SD=0.8) and 5.5 years (SD=0.6). Chi-square analysis indicated that the family structure was different between the two groups at T1 (χ² [1]=45.23, P<0.001). A higher proportion of sexually abused children lived in single-parent families (62.2%), compared with nonabused children (18.1%). Moreover, living with birth family was more likely in nonabused children (78.3%) than in abused children (20.7%), (χ2 [1]=84.09, P<0.001). The maternal level of education was different between the two groups (χ2 [4]=76.27, P<0.001). A high proportion of abused children’s mother had high school education (49.3%), while a great proportion of nonabused children’s mother had college education (43.4%). Moreover, abused children’s mothers (M=31.4, SD=6.2) were younger than nonabused children’s mothers (M=36.1, SD=5.0; t[184]=6.82, P<0.001).

Associations between sleep problems and sociodemographic information were tested using the whole sample. Negative associations were found between sleep problems and mothers’ age (r=−0.24, n=257, P<0.001), living with birth family (t[303]=10.96, P<0.001) and the maternal levels of education (Mann–Whitney U=3 554,000, P<0.001). There was a positive association between sleep problems and single-parent family status (t[202]=−8.23, P<0.001). Single-parent family status and maternal level of education were included as covariables in subsequent analyses.

Children who participated at T1 and T2 were compared to those who dropped out of the study on total sleep score, gender, family structure and maternal level of education. No differences were found in the comparison group. In the CSA group, children who dropped out had mothers with lower levels of education than children who participated at both assessment (χ2 [4]=11.06, P=0.03).

A majority of abused children sustained more than one episode of CSA (64.1%) and abuse perpetrated by a close family member (59.7%). Close to half of cases involved penetration/attempted penetration (46.6%) or unclothed touching (46%).

Sleep Problems at T1

A one-way multivariate analysis of covariance (MANCOVA) was conducted with the sleep problems items. Results indicated differences between preschoolers with and without experiences of CSA (Wilks’ λ=0.903; F[7, 245]=3.76; P<0.001; η²=0.097). The univariate ANOVAs showed that sexually abused preschoolers had higher impairment scores on all seven items (Table 1). In the CSA group, the percentage of children reaching clinical levels of sleep problems was 25.3%, while for the comparison group it was 1.2%.

Table 1.

Preschoolers’ sleep problems scores with the child behaviour checklist at T1 and T2

Item	T1			T2
	CSA group Mean (SD)	Comparison group Mean (SD)	P	CSA group Mean (SD)	Comparison group Mean (SD)	P
Does not want to sleep alone	0.96 (0.8)	0.61 (0.7)	0.041	0.65 (0.8)	0.47 (0.7)	0.700
Difficulty to fall asleep	0.88 (0.8)	0.53 (0.7)	0.003	0.70 (0.8)	0.32 (0.6)	0.008
Nightmare	0.90 (0.7)	0.36 (0.5)	<0.001	0.57 (0.6)	0.22 (0.4)	0.001
Does not want to sleep at night	0.93 (0.8)	0.55 (0.6)	0.015	0.80 (0.7)	0.45 (0.6)	0.010
Sleeps less than most children in day or night	0.45 (0.7)	0.18 (0.4)	0.018	0.37 (0.7)	0.11 (0.4)	0.025
Talks or screams in his sleep	0.63 (0.7)	0.20 (0.4)	0.001	0.42 (0.6)	0.16 (0.4)	0.002
Wakes up often at night	0.68 (0.8)	0.24 (0.5)	<0.001	0.61 (0.8)	0.10 (0.3)	<0.001
T-scores for the subscale	64.0 (14.1)	53.8 (4.8)	<0.001	58.7 (11.6)	52.1 (3.6)	<0.001

Items are rated on a Likert-scale ranging from 0–2 (see the Method section for a detailed description).

CSA Child sexual abuse; T1 Baseline assessment; T2 Approximately 1-year follow-up; P Statistical significance coefficient

One-way ANOVAs were conducted in order to determine whether abuse characteristics were associated with sleep problems in the CSA group. Results revealed no significant associations.

Table 2 displays the results of the multiple regression analysis performed on the sleep problem T-score at T1. The mean sleep problem T-score in the CSA group was 7.44 (95% CI: 4.00, 10.89) higher than in the comparison group after controlling for child gender, family structure and maternal education. The amount of variance explained in the sleep problem T-score by gender, family structure and maternal education was 7% (R²=0.07, P<0.001), which increased to 14% (R²=0.14, P<0.001) when abuse status was added. None of the characteristics other than abuse status was independently associated with sleep problem T-score.

Table 2.

Association between child characteristics and mean sleep problem T-score at T1

Characteristic	Mean difference in sleep problem T-score (95% CI)
Female child (relative to male)	−1.45 (−4.53, 1.62)
Single-parent household (relative to two-parent)	1.88 (−0.97, 4.72)
Maternal education (per level)	0.09 (−1.39, 1.57)
CSA group (relative to the nonabused comparison group)	7.44 (4.00, 10.89)

Each mean difference that is shown is adjusted for the other characteristics shown in the table.

CSA Child sexual abuse; T1 Baseline assessment.

Sleep Problems at T2

A one-way MANCOVA was also conducted with sleep items at T2. Results still indicated differences between abused and nonabused preschoolers: Wilks’ λ=0.818 (F[7, 147]=4.67, P<0.001, η²=0.182). Subsequent ANOVAs indicated that abused preschoolers had higher scores on all but one of the seven items at T2 (Table 1). At T2, no children in the comparison group reached clinical levels of sleep problems while 14.1% of children in the CSA group did. Of the 19 abused children presenting clinical levels of sleep problems at T1, 8 still did at T2, while 4 abused children presenting nonclinical levels of sleep problems at T1 presented with clinical levels at T2.

To assess the evolution of sleep problems for both groups over the year, a repeated measure one-way analysis of covariance (ANCOVA) was conducted with the T-score of sleep problems. Results showed only a group difference, with abused children presenting more sleep problems at both measurement times (F[1, 152]=22.68, P<0.001) (M_{csa group}: T1=63.9, T2=58.7; M_comparison: T1=53.8, T2=52.1). Time was not associated with levels of sleep problems and sleep problems did not evolve differently for abused and non-abused children.

DISCUSSION

The aim of the present study was to explore sleep problems in sexually abused preschoolers over a year. Results indicated that abused children presented more sleep problems than their nonabused counterparts on every item of the CBC subscale at T1, and on all but one of the items at T2, even after controlling for potentially confounding variables. The abuse status of children also predicted the total score of sleep problems at T1, above and beyond the predictive value of child gender and family sociodemographics. In fact, at the last step of the regression analyses, CSA status was the only variable significantly associated with total sleep problems. At T2, group differences were still observable on the total score. These results are coherent with those of previous studies of adult survivors of CSA and older sexually abused minors (12–15). Our results thus indicate that preschool victims can also suffer from significant sleep disturbances. It is worth noting that an important proportion of abused children presenting clinical levels of sleep difficulties at T1, are still presenting clinical levels at T2. In addition, 7% of abused children without initial clinical levels of sleep problems had developed significant difficulties over the course of the year. These results suggest that these symptoms do not necessarily disappear over time. This is worrisome given the numerous negative impacts of sleep problems on children’s cognitive, behavioural and socioaffective development (8–10).

Sleep disturbances are common among people experiencing stressful events such as a trauma (13), and stress affects sleep quality (21). Indeed, a stress–sleep cycle has been identified: stress is associated with poorer quality of sleep, which is in turn associated with increased stress levels (21). Furthermore, people with high levels of stress report being more negatively affected when they lack sleep (21). Learning theories explain the development and maintenance of sleep difficulties by the fact that during onsets of insomnia, the sleep environment becomes associated with frustration and arousal (11). A similar explanation is proposed regarding sleep disturbances in CSA victims. Indeed, CSA often takes place in the sleep environment creating associations between the later and a sense of threat. This sense of threat acts as a sustaining factor of sleep disturbances in CSA victims (11). This mechanism could be observable with young children. Fortunately, a number of cognitive-behavioural therapeutic strategies can be used (e.g., behavioural exposure), even with young children, in order to unravel this conditioned response (22).

Post-traumatic stress disorder (PTSD) can also explain the presence of sleep problems. Childhood adversity appears to alter the stress physiology and consequently increase the risk of developing a psychopathology over the lifecourse, including PTSD (23). In children younger than 7 years old, the PTSD symptoms of recurrent distressing dreams and hypervigilance could affect sleep quality. Furtermore, one criterion for PTSD directly refers to sleep disturbances such as difficulty falling or staying asleep and restless sleep (4). Thus, children in the CSA group of our sample could have presented PTSD symptoms that resulted in sleep problems.

Abuse characteristics in our sample were not associated with the severity of sleep problems. However, these variables are not consistently associated with CSA outcomes in the literature (24). In the current study, given the young age of the children, it is possible that clinical records reflected partial information, especially in terms of duration and chronicity. Preschoolers have not developed a reliable sense of time yet (25).

Our study entails several limits. First, lack of a prospective design prevents us from drawing conclusions about causality. The small number of boys in the CSA sample limits the interpretation of the findings regarding potential gender effects. The high attrition rate in the CSA group also affected statistical power. The only significant difference between participants at T1 and T2 was the maternal level of education in the CSA group and this variable was statistically controlled for in the analyses. Other relevant variables, such as quality of the family environment and post-traumatic symptomatology, were not examined and should be in future investigations. Even though we statistically controlled for it, group differences in sociodemographic characteristics were evident. Future studies should include a more comprehensive assessment of sleep difficulties (including laboratory measures).

CONCLUSION

In conclusion, documenting the frequency and possible correlates of sleep problems in sexually abused preschoolers is a necessary first step for the design of adequate intervention tackling the needs of young children. Our data yields valuable information toward this end and suggests that future studies should thoroughly investigate the associations between CSA and sleep problems. While future investigations are warranted to corroborate our results, some clinical implications can be suggested. For health care professionals working in child protective services, it is essential to routinely assess sleep problems at intake assessments. Behaviourally based childhood insomnia generally resolves with behavioural interventions such as bedtime routines, extinction, bedtime fading and positive reinforcement (26–28). Paediatricians noticing the presence of sleep problems in young abused children could refer them. As for future investigations, given that there is a certain commonality among sleep problems and some CSA correlates (e.g., behaviour problems, cognitive difficulties), examining how these variables are interrelated using mediation and moderation analyses could be particularly informative. It could also be of relevance for clinicians/paediatricians working with abused children, giving them essential cues to improve case formulation and inform treatment orientation. In the end, pursuing this line of research could provide health professionals with necessary tools to foster resilience in sexually abused preschoolers.