Mother-Child Interactions at Six Months Postpartum are Not Predicted by Maternal Histories of Abuse and Neglect or Maltreatment Type

Abstract

Objective

A history of childhood maltreatment (CM) is associated with increased rates of maternal psychiatric symptoms and other adverse outcomes in adulthood among postpartum women. However, to date only few studies examined associations between CM and mother-child interactions among a non-clinical sample of postpartum women and the specific potential influence of the type of abuse or neglect is poorly understood. This study aimed to examine the relationships between CM types and observed parenting in a non-clinical group of recently postpartum mothers with maltreatment histories.

Method

Participants were 173 postpartum, non-clinically referred mothers oversampled in the community for CM histories (n=123, 72%) and their infants who underwent during a 6 months postpartum visit a high- and low-stress interactive task, which was video-taped for later independent coding on hostile, controlling and positive parenting. Mothers also provided information on demographics and type of CM (emotional, sexual, physical abuse and neglect). Differences of maternal parenting by history of CM and specific type were analyzed via two-way Univariate General Linear Models.

Results

Inconsistent with a priori hypotheses, no significant differences emerged between overall CM severity or exposure to any particular CM type and hostile, controlling, or positive parenting in the high or low-stress tasks.

Conclusions

Findings suggest that non-clinical postpartum women with CM histories show resilience in regards to postpartum parenting quality and do not differentiate from non-CM maltreated postpartum mothers. We discuss the potential influence of resilience, moderating factors, clinical implications, and recommendations base on our findings.

Introduction

Significant empirical research with mothers indicates a history of abuse and neglect during childhood is associated with adversity in adulthood including amplified risks for postpartum posttraumatic stress disorder (PTSD), depression, suicidal ideation, or use of substances (e.g., Marcenko, Kemp, & Larson, 2000; Muzik, Brier, Menke, Davis, & Sexton, 2016; Oh, Muzik, McGinnis, Hamilton, Menke, & Rosenblum, 2016; Sexton, Hamilton, McGinnis, Rosenblum, & Muzik, 2015). Histories of childhood maltreatment (CM) may place mothers at greater risk for parenting difficulties that influence attachment; such findings are inconsistently reported, generally limited to self-report assessments, and often restricted to survivors of childhood sexual abuse (CSA; e.g., Bailey, DeOliveira, Wolfe, Evans, & Hartwick, 2012; Dilillo & Damashek, 2003; Lyons-Ruth & Block, 1996). Very little research has examined the associations between specific types of CM and later-observed parenting behavior. Previous research in this area is often confounded by recruitment from clinical environments in which maternal or child psychopathology or behavioral difficulties have been identified or from other at-risk groups (Bailey et al.; Harmer, Sanderson, & Mertin, 1999; Lesser & Koniak-Griffin, 2000; Marcenko et al.; Nuttall, Valentino, & Borkowski, 2012). Further research is warranted to expand our understanding of the potential influence of types of distal maltreatment on mother-child interactions.

To our knowledge, only one published study, Bailey et al. (2012), has examined associations between types of maltreatment and self-reported and observed parenting behaviors. Participants included 93 mothers of 4- to 6-year-old children recruited from newsletters, agencies serving high risk families, and clinical environments providing mental health services to mothers and children who endorsed at least one of the following maternal risk factors: single parent, teenaged at time of child’s birth, or below the poverty line. The researchers found neglect and emotional maltreatment (though not physical or sexual abuse) were associated with observed hostility. Moreover, observed maternal sensitivity and intrusiveness were not associated with any type of abuse or neglect. Maternal histories of CSA were associated with self-reported concerns with parental competence, however, as reported with other studies, self-report and observational findings of parenting behavior were not correlated.

Bailey et al.’s (2012) study suggests specific CM types may be associated with particular parenting outcomes in mothers experiencing current adversity. While the recruitment strategy was designed to increase the potential to recruit mothers with histories of CM, the recruitment methodology results in the inextricable confounding of distal and proximal stressors. As noted, CM does increase risk for psychopathology and other stressors in adult mothers. However, this trajectory is not modal; the overwhelming majority of mothers with histories of CM do not evidence significant symptoms in adulthood (i.e., Sexton et al., 2015). Thus, currently available research makes it difficult to ascertain whether CM alone is associated with parental behaviors, whether other confounding factors (i.e., maternal pathology, poverty, single parenthood) are determinant, or whether the presence of these factors in adulthood moderate CM exposure. Assessment of CM exposure type within a non-clinical or otherwise at-risk sample would facilitate greater understanding of the most commonly anticipated maternal behavior trajectories following childhood abuse or neglect.

Additional research focused only on CSA and physical abuse found women with childhood sexual abuse histories self-report difficulty establishing boundaries, more permissiveness, and using harsher physical discipline, and that women with physical abuse histories demonstrate more hostile-intrusive parenting behaviors (DiLillo, Tremblay, & Peterson, 2000; DiLillo & Damashek, 2003; Lyons-Ruth, & Block, 1996). With the exception of the Bailey et al. (2012) and Lyons-Ruth & Block (1996) studies, the data are primarily self-report suggesting potential self-rater biases, and within Bailey et al, (2012) the self-report measures were not related to observational measures suggesting low validity between the measures. However, the literature in this domain is not entirely consistent. Other studies, particularly those using observational methods, have not confirmed direct associations between CM and maternal behaviors (e.g., Fitzgerald, Shipman, Jackson, McMahon, Hanley, 2005; Martinez-Torteya, Dayton, Beeghly, Seng, McGinnis, Broderick, et al., 2014). Martinez-Torteya et al. (2014) explicitly found models specifying direct relationships between CM severity and maternal positive or hostile behaviors did not produce a good fit. However, when CM history was modified by psychiatric symptoms, single parenthood, limited family income, and current stress, indirect pathways between CM history and parenting outcomes were observed. In addition, though previous research has focused on increases in negative parental interactions, relationships between CM and alterations in positive parent-child interactions are particularly scarce and may be associated with the relatively small sample sizes characteristic of several observational studies. The current study will address noted gaps in the literature by examining associations between particular types of childhood abuse and neglect and observational examinations of the positive and negative parenting behaviors during high and low stress tasks.

Given the paucity of research examining the relationships between maltreatment types and parenting behaviors, we aimed to examine the relationships between types of maltreatment and parenting in high- and low-stress parenting tasks (teaching task versus free play activity) among a non-clinical group of recently postpartum mothers with maltreatment histories. We hypothesized that significant and adverse associations would emerge between the severity of CM and hostile, controlling, and positive parenting and that these associations would be more apparent during high stress parenting tasks as task-related stress load increased. We hypothesized that certain types of abuse and neglect would be differentially associated with parenting behaviors but did not speculate a priori as to the specific subtypes of CM for which such differences may emerge.

Materials and Methods

Procedures

Participating mother-infant dyads (N = 173) were engaged in a longitudinal investigation of maternal outcomes following childhood histories of abuse and neglect to evaluate potential differences in maternal and child behaviors and wellbeing. Participants were recruited from an associated investigation of prenatal mental health and via flyers. Enrollment criteria included being 18 years of age or older, English-speaking, and mother of a singleton birth. Participants were recruited via a stratification protocol to engage a high percentage of women endorsing histories of childhood abuse or neglect (based on positive screen on the Childhood Trauma Questionnaire (CTQ; Bernstein et al., 2003). Exclusion criteria included mothers of infants born premature or with severe developmental or health difficulties and mothers with diagnoses of schizophrenia, bipolar disorder, or substance use problems within the three months prior to enrollment. Data for the longitudinal investigation were collected through 18 months postpartum. The results presented in this investigation are limited to data obtained at four months (maternal history of CM, demographic data) and six months (parenting task) postpartum. Participants completed the informed consent process incorporating verbal and written consent and were paid a maximum of $140 for study participation. The [blinded for review] Human Subjects Committee approved the investigative procedures.

Measures

Demographics

Baseline maternal demographic information was collected by phone at four months postpartum and then updated at subsequent data collection time-points. The demographic questionnaire includes questions designed to capture annual household income, race and ethnicity, age, education, and marital status.

Maternal Trauma and Psychopathology

Maternal history of CM was assessed at four months postpartum using the Childhood Trauma Questionnaire (CTQ; Bernstein et al., 2003). The CTQ is a 28-item self-report measure designed to assess the presence and nature of childhood maltreatment experience. Respondents are asked about maltreatment across 5 domains –emotional abuse, physical abuse, sexual abuse, and emotional and physical neglect. Response options across items range from 1=‘never’ to 5=‘very often true’. For the purposes of this study, we utilized CTQ scoring to indicate the presense of any degree of CM for each of the five domains. For the present study, we created dichotomous groups based on using the recommended cut-off scores for each abuse and maltreatment type (Bernstein et al.). The CTQ has demonstrated excellent psychometric properties in past studies (e.g. Bernstein et al., 2003). In the current study, the CTQ evidenced good internal consistency (α=.92).

Observed Maternal Behaviors

Maternal parenting behaviors were observed and coded at six months postpartum during a series of unstructured and structured interaction tasks in mothers’ homes. Specifically, mothers were video-taped engaging in a 10 minute ‘normal’ play with their infants (i.e., unstructured), and also during two 3- minute structured ‘teaching’ tasks (i.e., teaching baby to stack blocks and put balls in a bucket, which is beyond a 6 month olds developmental capacity). The video segments were coded utilizing the MACY Infant-Parent Coding System (Earls, Beeghly, & Muzik, 2009); this coding system reliably (intraclass correlations above 0.8) captures meaningful variations in maternal parenting behaviors associated with maternal and child outcomes (see Muzik et. al., 2013; Martinez et al., 2014 for more detailed descriptions of the activities, coding system, and ICCs).The interaction segments were viewed and scored by coders trained to evaluate behavioral and affective dimensions reflecting quality of maternal parenting behavior.

Coders were blinded to maternal CM history and other study variables and asked to score randomly selected videotapes independently. The present analyses concern three parenting codes derived from the interaction-coding: (1) hostile/rejecting parenting (evidenced by hostile, bitter vocal expressions, unfriendly facial expressions, etc.), (2) overcontrolling/intrusive parenting (evidenced by mothers controlling choice and duration of activities, manipulating infants limbs to accomplish what she wants, etc.), and a (3) composite variable labeled positive parenting behaviors. Positive parenting was defined based on the results of conceptually driven principal components factor analysis (Torteya-Martinez, et al., 2014), and included the following maternal behaviors: Behavioral Sensitivity/Supportive Presence, Engagement, Warmth, Affective Sensitivity, and Flexibility.

Data Analyses

Data were analyzed with SPSS 20.0. Descriptive statistics were conducted with percentages and averages, as warranted. Investigations of potential differences of maternal behaviors in the context of any history of childhood maltreatment and specific abuse and neglect experiences were conducted via two-way Univariate General Linear Models. High- and low-stress parenting behaviors were separately analyzed for positive, over controlling, and hostile parenting behaviors. The a priori significance threshold for analyses was established at p <.05. G*Power 3.0.10 was used to evaluate power to detect medium and large effect sizes differences between groups. For two groups with alpha established at .05, power (1- β) at .8, minimum sample sizes recommendations for large (f = .4) and medium (f = .25) effects were 52 and 128, respectively.

Results

Participant and Trauma Characteristics

Six months postpartum, 173 women completed the dyadic tasks and other assessment instruments. Mothers were generally young adults (mean age = 28.8 years, SD = 6 years). Sixty-three percent identified as Caucasian. Two-thirds were married. Eighty-four percent completed at least some college. Economically, 34% described an annual household income of $25,000 or fewer. Specific to childhood maltreatment, 73% of mothers self-reported a history of CM. Mothers with (n = 126) and without (n = 47) histories of any CM did not differ significantly on demographic characteristics. Of the five types of abuse and neglect assessed, emotional abuse (55%) and neglect (48%) were the most commonly endorsed. Sexual abuse (44%), physical abuse (34%), and physical neglect (29%) were also frequently reported.

Associations between Childhood Maltreatment Type and Maternal Behaviors

Results did not evidence any significant relationships between any particular subtype of CM and the maternal behaviors investigated or between those with and without a history of any CM. These results were identical for both low-stress and high-stress tasks (see Tables in Supplement). We have reported our findings of dichotomous contrasts of mothers with CM of any severity and those without CM. See Table 1 for results. A comparison of those with and without CM exposure in the severe range yielded the same findings (data not reported in this Brief Report).

Table 1.

Relationships between Childhood Maltreatment Types and Maternal Behaviors for Low- and High-Stress Parenting Tasks

CM		Positive Parenting						Over Controlling Parenting						Hostile Parenting
		Low			High			Low			High			Low			High
		M	F	p	M	F	p	M	F	p	M	F	p	M	F	p	M	F	p
EA	Yes	6.4	0.01	.91	6.9	0.30	.59	5.1	0.28	.60	4.1	0.85	.36	2.8	0.06	.81	2.5	0.14	.71
	No	6.4			6.7			4.9			4.4			2.8			2.5
PA	Yes	6.4	0.01	.94	7.0	0.36	.55	5.3	1.93	.17	4.4	0.39	.53	2.9	0.40	.53	2.5	0.19	.67
	No	6.4			6.8			4.8			4.2			2.8			2.5
SA	Yes	6.4	0.09	.77	6.8	0.01	.91	5.0	0.08	.78	4.2	0.03	.86	2.8	0.15	.70	2.4	0.32	.57
	No	6.5			6.8			4.9			4.3			2.8			2.5
EN	Yes	6.4	0.04	.84	6.8	0.21	.64	5.0	0.08	.78	4.2	0.02	.90	2.8	0.08	.78	2.5	0.01	.94
	No	6.5			6.9			5.0			4.3			2.8			2.5
PN	Yes	6.5	0.09	.76	7.0	0.37	.55	4.9	0.37	.63	4.4	0.90	.34	2.6	1.91	.17	2.4	0.53	.47
	No	6.4			6.8			5.0			4.2			2.9			2.5
Any	Yes	6.4	1.62	.20	6.8	0.01	.96	5.1	1.02	.31	4.3	0.25	.62	2.9	1.22	.27	2.5	0.09	.76
	No	6.7			6.8			4.8			4.1			2.7			2.4

Notes. Low = low-stress task; High = high-stress task; M = mean; CM = childhood maltreatment type; EA = emotional abuse; PA = physical abuse; SA = sexual abuse; EN = emotional neglect; PN = physical neglect; Any = any history of childhood abuse or neglect.

Conclusions

Contrary to our hypotheses, the current analyses did not identify differences between types of maltreatment and hostile, controlling, or positive parenting behaviors. Further, contrasting those with any type of CM and those without CM did not result in rejection of any of our null hypotheses. This was the case in both low and high-stress tasks, again contrary to our a priori hypotheses.

These findings underscore other reports in demonstrating that the most likely trajectories experienced by adults with CM histories include recovery and resilience. In addition to research that demonstrates many mothers with childhood trauma histories do not go on to experience significant symptoms of pathology; nor do they appear, as a group, to exhibit parenting deficits in the early postpartum. However, these findings may be disparate from some observations within self-report studies (see Dilillo & Damashek, 2003, for a review), seeming more consistent with previous observational research by Fitzgerald et al. (2005) and Martinez-Torteya and colleagues (2014). In keeping with Bailey et al. (2012), the only other study to specifically assess differential effects of particular types of CM and postpartum parenting, we also did not detect significant relationships between childhood abuse and maternal hostility and elements of positive parenting. However, their finding concerning the relationship between neglect and increased hostility was not replicated, even when parental stress load was taken into account.

There are several potential explanations for disparities between the present findings and other studies of parenting behaviors in the context of maternal CM histories (DiLillo, Tremblay, & Peterson, 2000; DiLillo & Damashek, 2003; Lyons-Ruth, & Block, 1996). Most apparent is the difference in risks factors between our participants and mothers included in other studies. While our study is generally characterized as a non-clinical group of mothers with CM exposure as the primary prominent risk, most of the other research in this area involved mothers with active pathology (i.e., depression, PTSD, substance use), cases involving children with identified behavioral problems, or environments with increased proximal stressors (e.g., limited income, single and/or young mothers). Thus, it may be most parsimonious to conclude either that 1) current psychiatric functioning or environmental factors are more determinant of parenting behaviors than a history of CM, or 2) that proximal factors moderate CM exposure to result in adverse parenting outcomes. In fact, there is some evidence to suggest that the latter possibility has merit. While Martinez-Torteya did not detect a good fit between CM exposure and parenting outcomes at six months postpartum, CM exposure was associated with greater psychiatric symptoms and life stressors that were subsequently correlated with adverse parenting.

Similarly, women in the current study may be presenting with resilience-promoting demographic factors that alter the associations between maltreatment type and parenting. For example, non-minority identification, having stable housing, and a supportive partner all promote wellness among women with CM histories (DuMont, Widom, & Czaja, 2007). Another potential explanation for discrepancies between this study and those that do identify amplified parenting risks may be the age of the children during the assessment. Our study did not identify risks for mother-child dyads when the infant was six months of age. However, this may be altered at other periods of development (e.g., when children develop verbal skills or increased autonomy, periods characterized by power/control dynamics, sexual maturation). Finally, this may represent a “file drawer” bias, such that studies with null findings are less likely to be published. In addition, our study was limited to specific types of mother-child interactions (i.e., positive, over-controlling, and hostile). It may be the case that more direct CM-parenting relationships may be observed in other types of maternal behaviors that may have particular relevance for intergenerational transmission of risk. Finally, another explanation may be that our behavioral tasks (free-play and teaching) were relatively safe and non-threatening thus not eliciting enough stress to challenge mothers with trauma histories and thus their trauma history may not have interfered with “good enough” parenting.

Some limitations and recommendations are worth noting. Within our participants endorsing histories of CM, exposure to multiple types of abuse and neglect were the norm, which is typical in a maltreatment sample (Finkelhor, Omrod, & Turner, 2007). As such, we did not statistically control for other types of abuse and neglect, as any results would likely be non-generalizable to most populations with CM exposure. However, that may have obfuscated some of the significance of types of CM exposure when CM was experienced in isolation. In addition, further research to replicate whether other types of variables moderate CM exposure and parenting associations are warranted to ascertain if CM history yields unique clinical detriments and under which circumstances these occur. Empirical investigations to replicate these findings involving children at varying age groups are also merited. Finally, we acknowledge the rich literature linking maltreatment and disorganized attachments, and believe examining these associations is imperative to understand the development of insecure attachments; however, this is beyond the scope of this study and an aim for future analyses.

Despite these caveats, this study is unique in investigating the relationships between CM types and total exposure with positive and negative parenting behaviors. In addition, the observational design augments the methodological rigor of the investigation. Our findings highlight that resilient outcomes are the most likely outcomes after childhood adversity and underscore the need to further investigate potential moderators.