Child maltreatment and associated sociodemographic factors among children affected by HIV/AIDS in Ghana: a multi-informant perspective

ABSTRACT
 Child maltreatment is considered a major public health concern among children because they can cause significant physical and psychological problems. Child maltreatment is widespread but often underestimated. Surprisingly, there is hardly any data on child maltreatment and any associated sociodemographic factors children affected by HIV/AIDS in low-income countries. This study employed cross-sectional, quantitative survey that involved 291 children aged 10–17 years and their caregivers in the Lower Manya Krobo District, Ghana and examined their exposure to and experience of child maltreatment. The results show that at least one form of maltreatment was reported by approximately 90% of the children, and it was significantly higher among orphans and vulnerable children (OVC) as compared with comparison children. Older age, frequent changes in residence, non-schooling and living with many siblings are associated with child maltreatment. The results demonstrate that maltreatment among children affected by HIV/AIDS are not rare, and that the dysfunction family conditions that they find themselves bear systemic risks for maltreatment. It is important that culturally appropriate and evidence-based interventions are implemented to address the maltreatment.


Background
Child maltreatment exist in almost all cultures. Different categories of maltreatment are identified to include physical abuse, sexual abuse, exposure to domestic violence, psychological abuse and neglect. Domestic violence encompassing children witnessing parental quarrels, fights and abuses. Child maltreatment is identified as a factor that places children at risk for developmental problems (Levey et al., 2017). Chen and Gueta (2015) and Levey et al., (2017) suggested that irrespective of the form of maltreatment, it is an increased risk for health problems. Recent literature on children in general found that maltreatment is associated with heightened levels of emotional difficulties such as depression, withdrawal, anxiety and dissociation (England-Mason et al., 2017) and behavioral problems including delinquency, aggression, antisocial behaviors and conduct problems (Katsurada et al., 2017) and the development of high-risk behaviors (Carlson et al., 2015). Arslan (2017) reported higher mental health problems for children who suffer maltreatment from their parents. González et al. (2016) also observed that exposure to domestic violence is a risk factor in childhood psychopathology. Exposure to domestic violence was consistently found to predict internalizing problems in children including post-traumatic stress symptoms (Conners-Burrow et al., 2013). Jackson et al. (2015) found that child maltreatment interferes with the normal development of children, and increase the risks of developing a wide range of mental health difficulties (Guillaume et al., 2017). Increased levels of behavioral problems are mostly reported for children who experienced physical abuse (van der Put et al., 2015) whilst emotional difficulties are mostly identified among children who suffer psychological abuse or neglect (Ban & Oh, 2016). However, Yoon et al. (2017) indicated that externalizing problems are consistently identified among maltreated boys whilst abused girls largely present internalizing difficulties. Beside parental personality types, attitudes, perceptions, practices and values, parental illness and death might also create risks for children to be abused and exploited (Shaw et al., 2015). Cyr and Alink (2017) identified poor and negative family related factors including impaired parent-child relationships and lack of or reduced parental guidance and monitoring as risk factors for child maltreatment. Others include neglectful parenting (Ha et al., 2015), family instability (Breen et al., 2015), poverty (Cluver & Orkin, 2009;Doidge et al., 2017;Thurman & Kidman, 2011), and poor social support network and isolation (Doku et al., 2015). Consistently, it has been demonstrated that orphans and other vulnerable children (OVC) lack adequate care and protection, and frequently live in households characterized by these negative family related factors (Roelen et al., 2017). Yet, to the best of my knowledge and the available literature, no study has examined these observations among Ghanaian orphans and vulnerable children.
Other investigators observed that parents and close family members who are entrusted with providing protection, love and care are the very people that consistently abuse and neglect children (Aplin, 2017;Cluver et al., 2010;Meinck et al., 2015). The evidence, thus, suggests that children affected by HIV/AIDS parental illness and death might be at heightened risk for maltreatment and its subsequent negative consequences. In most African countries, physical punishment of children using sticks and belts is virtually a community norm (Breen et al., 2015;Brown et al., 2009). There have been suggestions that domestic violence (Skeen et al., 2016) and child abuse (Lachman et al., 2014;Thurman & Kidman, 2011) in HIV/AIDS affected households could be high. Hence, a better understanding of the levels and forms/ sources of child maltreatment among AIDS-orphaned and vulnerable children, can help facilitate strategies to support these children. However, no studies to my knowledge has assessed different different types of maltreatment among children affected by HIV/AIDS in comparison with other orphans and non-orphan children in a low HIV/AIDS endemic settings. This article therefore assessed different types of maltreatment among children in low HIV/AIDS endemic communities of Ghana, and compared them with other-orphans and non-orphan children (comparison group). The current study is the first elsewhere to focus on reports from multiple informants (children and their caregivers). This strategy increases the validity and reliability of variables of interest and also captures important information about contextual effects that may have implications for interventions (Lochman, 2004). Because child and caregiver reports could account for unique variances in predicting relevant child outcomes (Ferdinand et al., 2003), the findings in this paper reflect a more accurate picture of child maltreatment across settings that could be generalized.

Data collection
A community based cross-sectional survey design utilizing questionnaires was conducted in the Lower Manya Krobo District of Ghana. There was a pilot preceding the present study to validate (examine the appropriateness and comprehensibility of) the study instruments in the research setting. The results of the pilot study indicated appropriate understanding among the Ghanaian children. Total sampling technique whereby every household in selected communities were approached and asked to participate in the study. This sampling procedure is justified for reasons of feasibility, cost-effectiveness, and accessibility. The total sampling employed also ensured higher participation. Then within each community, a limit of 30 was placed on the number of households recruited. This was done to minimize the effects of over-representation from particular communities. Each participating child and their respective caregiver/parent completed the survey questionnaire that followed the steps described by Thomas (2006). To handle issues of multiple eligible OVC if at least one child in the house was an AIDS orphaned child, then the child was classified as an AIDS-orphaned child and recruited as such. To identify parental cause of death, a verbal autopsy (VA) was used (Lopman et al., 2010). The entire assessment inventory took about 30-45 minutes to complete. The study protocol was approved by the institutional Research Ethics Review Boards of the University of Glasgow and the Ghana Health Service prior to recruitment of children.

Data collection instruments
Items were taken from two previously validated measures (Conflict Tactics Scale and the South African Demographic and Health Survey) to assess child maltreatment. The Conflict Tactics Scale (CTS) was originally developed by Strauss to provide a measure of conflict resolution in events that involve violence (Straus, 1990) by obtaining data on possible dyadic combinations of family members. It has since been used in over 70,000 empirical studies and thoroughly evaluated in over 4000 of them (Veriara, 2014). The scale has strong construct validity. The CTS or any adaptation of it was found to have a reliability ranging from 0.77 to 0.95 and internal consistency between 0.67 and 0.86. The present study utilized adapted versions of both parent-to-child and child-to-parent scales. The Child to-Parent Conflict Tactics Scales (P-CCTS) assessed the child's experience of violence (direct and indirect) abuse, neglect, psychological abuse and corporal punishment (physical abuse) within the household. Parents and caregivers' use of punishment and maltreatment in the home as well as domestic violence were captured by the Parent-to-Child: Punishment, Discipline and Violence (ICAST-P). There were also items from the South African Demographic and Health Survey specifically developed for developing countries (International Household Survey Network, 2013). These items obtained information on both the child and caregiver's exposure to community violence (both victimization and witnessing). A number of sociodemographic factors such as age, gender, family size, number of other minors living at home, number of changes in residence and age at which children were orphaned (where applicable) were also measured. There were also items regarding children's educational level as well as their present education status (presently at school or not).

Statistical analyses
The data was analysed using the IBM SPSS (v21). Differences between the orphanhood groups on socio-demographic factors were established using chi-square tests or analyses of variance (ANOVAs). Then, relationships between the various socio-demographic factors and child maltreatment scores were examined. These were assessed using independent t-tests and chi square (for categorical variables such as gender, religion and being presently in school) and Pearson bivariate correlations, and ANOVA (for continuous socio-demographic such as age, household size and number of children at home). To handle the multi-informant data, paired-sample statistics and Pearson correlations were used to establish whether reports of young people differed from those of their parents and caregivers. Finally, a General Linear Model involving one-way multivariate analysis of covariance (MANCOVA) was performed to investigated the levels of child maltreatment for the various orphanhood types after controlling for relevant socio-demographic factors (adjusted model).

Socio-demographic characteristics of participants
The participants had a mean age of 13.03 years (SD = 2.87), with age range 10-17. There were 51% females. The majority of the children (81.8%) were currently attending school. About 75% had attained primary or junior secondary level education and 12.7% vocational or technical education. The majority of parents and caregivers (62%) worked mainly in farming, driving, trading or as artisans (carpentry, masonry, bead making). In the sampled about 56% of the children indicated that they were Christians, ANOVA shows significant differences [F (287, 3) = 21.131; p < .001] in the age between the four groups. The socio-demographic characteristics of the participants are summarized in Table 1. Interestingly, 89% of OVC reported some form of maltreatment within the household. Eighty percent (80%) of the children reported having been disciplined with a belt, stick or other hard object, 72% had been punched or slapped by adults, 65% have been called funny names and 69% had been threatened with expulsion from their homes.

Association between socio-demographic factors and maltreatment
Concerning socio-demographic cofactors, higher child maltreatment was associated with living in smaller households (r = -.255, p < .001), having more siblings (r = .211, p < .001), frequent changes in place of residence (r = .138, p < .05) and currently not attending school (t = 3.302, p < .001). Increased age was found to be associated with more maltreatment (r = .431, p < .001). There was no difference observed on gender (t = 0.198, p = n. s.).

Cross-informant agreement
The inter-informant correlations for the maltreatment scores in the present sample were low, ranging from .011 to .191 (Table 2). Correlations between children's self-reports and caregivers' accounts on composite maltreatment, neglect, and psychological abuse were significant at alpha levels of p < .001, p < .01 and p < .05 respectively. Correlations on physical abuse and domestic violence of burden between self-report and informants did not reach significant levels. Interestingly, in the present analyses caregivers and parents (informants) reported significantly higher levels of total maltreatment (t = 13.036, p = .001), psychological abuse (t = 27.119, p = .001) and physical abuse (t = 8.914, p = .001) compared with reports from the children (Table 2). However, children reported more domestic violence (t = 11.046, p = .001) and neglect (t = 4.054, p < .001) than their informants.

Differences between OVC groups on domestic violence and maltreatment
One-way multivariate analysis of covariance (MAN-COVA) results revealed statistically significant differences in child maltreatment by the four groups on the composite dependent variableoverall multivariate effect after controlling for relevant sociodemographic variables, [F (4, 284) = 16.824, p < .001; Wilks' λ = .533; partial η2 = .189]. The orphanhood groups accounted for 18.9% of the variance in the combined dependent variable scores of child maltreatment, reflecting a large effect size. On the self-reported levels of overall child maltreatment, children living with HIV/AIDS-infected parents (t = 3.650, p < .001), children orphaned by AIDS (t = 3.304, p < .001) and orphans of other causes (t = 2.750, p < .001) all reported significantly more maltreatment than comparison children. The Univariate tests of between-subject effects were employed to assess the statistical significance of each of the four components of maltreatment under consideration (see Table 3). The Univariate analysis indicates significant group differences between the orphanhood groups on witnessing domestic violence [F (3, 287) = 15.585, p < .001, partial η2 = .14]. Bonferroni adjusted alpha level of (p < .01; i.e., 05/4) was used in all subsequent multiple comparison since there are four components of maltreatment. The results showed that children living with HIV/AIDS-infected parents reported higher adult fights and quarrels (domestic abuse) in the home than both other orphan groups (t = 0.886, p < .01) and comparison children (t = 1.320, p < .001). Children orphaned by AIDS on the other hand also reported significantly more domestic violence than other orphaned children (t = 0.698, p < .01) and comparison children (t = 1.132, p < .001). No further group differences were found. The Psychological abuse subscale of the maltreatment score exhibited the same pattern, where OVC scored higher than comparison children [F (3, 287) = 47.019, p < .001, partial η2 = .33]. However, only children orphaned by AIDS (t = 0.970, p < .001) and children living with HIV/AIDS infected parents (t = 0.718, p < .01) reported significantly higher levels of neglect compared with comparison children. Finally, only AIDS orphaned children reported significantly higher physical abuse compared with comparison children (t = 0.761, p < .01). Cluver et al. (2013) noted that there is no reliable data on exposure to or experiencing of child abuse among families affected by AIDS in developing countries due to poor reporting and recording of abuse incidences. The present analysis provided quantitative examination of child maltreatment disparities among children affected by HIV/AIDS. A key finding from the present analysis is that maltreatment of children was significantly higher among OVC compared with comparison children. Children orphaned by AIDS and children living with HIV/AIDS-infected parents reported more domestic violence than other children. This could be an indicator of the significant physical and emotional  burdens of the HIV/AIDS illness and the strain that the disease has on family relationships. Lachman et al. (2014) noted that HIV/AIDS increases the intensity and frequency of quarrelling and fighting among family members or couples. This is consistent with assertions that children who reside in families affected by HIV/ AIDS are at risk of exposure to abuse (Meinck et al., 2015). Orphans, regardless of the cause of their parental death and children living with HIV/AIDS-infected parents are at high risk of abuse (psychological and physical) and neglect by their caregivers who should be providing protection and guidance. This present evidence contradicts the finding of Nyamukapa et al. (2008) that child abuse was similar among both orphans of AIDS and comparison children but consistent with claims that many OVC suffer cruel and impersonal care-giving from their new caregivers (Sherr et al., 2014). One explanation for the current evidence could be that the traditional family support system and network might have been overwhelmed by the soaring numbers of OVC which increases the likelihood of abuse and neglect (Cyr & Alink, 2017;Guillaume et al., 2017). In the present analysis, the fact that child maltreatment was negatively associated with household size and positively with the number of siblings living in household supports this postulation. That is, interestingly, it is larger numbers of adults in the household that is important to buffer child maltreatment. Another plausible explanation could be the high levels of HIV/ AIDS related stigma, discrimination and social exclusion found among OVC; as increased incidence of child maltreatments is consistently reported among socially isolated families with inadequate community networks and ties (Kim & Maguire-Jack, 2015). These findings are indications that there are differences in the family structures that children affected by HIV/ AIDS and comparison children find themselves in. Approximately 90% of OVC reported some form of maltreatment within the household. Subsequently, 80% of the children reported having been disciplined with a belt, stick or other hard object, 72% had been punched or slapped by adults, 65% have been called funny names and 69% had been threatened with expulsion from their homes. These prevalence rates are two to three times higher than those found among OVC in a community survey in South Africa (Thurman & Kidman, 2011), highlighting the worrying nature of the present finding for Ghanaian children. However, it is important to note that many severely maltreated and neglected OVC are not captured in community surveys because, as part of South Africa's government sponsored policies, they are sent to residential care (Levine, 2000). Ghana has no such residential policy for maltreated OVC and so they are found within the community. This is evident by the fact that higher levels of abuse are reported in South African orphanages and institutional care accommodating OVC than in the community (Williamson, 2005).

Discussion
Contrary to public health assumptions, caregivers and parents in the present study reported higher child maltreatment compared to the children's own report whilst children, however, reported higher levels of domestic violence than their guardians. This may be a reflection of parents' social and cultural acceptance of harsh child rearing practices. It could also be deduced that parents are being selective (social desirability bias) by under-reporting experiences of domestic violence because they might be victims or perpetuators of it (Sandberg, 2016). The number of changes in place of residence (household migration) was significantly associated with more maltreatment. Speculatively, it may be that children who are being abused by their guardians are constantly on the move to try and escape the violence. Consistent with findings in South Africa, in the present study increased age was associated with more maltreatment (Thurman & Kidman, 2011). Finally, in the present study, living with many siblings in the same households was related to higher maltreatment whilst larger household size was associated with less maltreatment. This suggests that it is better for OVC to live in households with many adults than in households with several minors. Certainly, these identified significant associations between child maltreatment and relevant socio-demographic factors may be important in developing developmentally and contextually appropriate interventions that would alleviate child maltreatment among children affected by HIV/AIDS.

Limitations and future research
It is worth noting that the retrospective, self-reporting and cross-sectional nature of this study could not allow for any conclusions to be drawn regarding causal relationships because exposure (contextual factors and HIV/AIDS exposures) and event (child maltreatment outcomes) were measured at the same time. Despite these limitations, the study shows that although child maltreatment is a serious and common problem in Ghana, surveys with rigorous research design, such as stratified random sampling method and multi-informant perspectives used here, are not common. In conclusion, this study found that among OVC, the prevalence of maltreatment was 90%. Furthermore, children who live with many minor siblings, are of older age, changed residence frequently and are currently not attending school were identified as being at significant risk of maltreatment. These significant predictors of child maltreatment could provide important information for planning future preventive measures. More attention should be paid by health professionals to children affected by HIV/AIDS as they represent the at-risk children for maltreatment.