Variability

What follows is a review and summary of our analyses concerning the short- and long-term outcomes for the sexually abused females in our research. What will be clear quickly is that there are many differences between the members of the abuse group and the comparison group some of which manifest early on, others of which emerge over time. It is important to make clear at the outset of this review that not only are there many group differences apparent but also there are considerable individual differences or variability in adverse impact. This was clear to us early on. Thus, for example, Putnam, Helmers, and Trickett (Reference Putnam, Helmers and Trickett1993) found not only a mean difference in dissociation levels at T1 between the sexually abused and nonabused females, but also greater scatter (variability) in the dissociation scores for the abuse group compared with the comparison group. Thus, for dissociation scores a number of members of the abuse group were indistinguishable from the comparison group, whereas others had much higher scores than any member of the comparison group. This has been a common finding in this study, as exemplified by the findings reported in a recent publication (Noll, Trickett, Harris, & Putnam, Reference Noll, Trickett, Harris and Putnam2009) where for 13 outcome variables (e.g., depression, substance abuse), the standard deviation was larger for the abuse group than the comparison group in every instance.

Although this variability among victims of sexual abuse may well be in part due to different levels of vulnerability and resilience of the individuals, we also thought it important to consider another possible source. That is, even though our stringent inclusion criteria (recently disclosed, contact sexual abuse of females by a family member) resulted in an unusually homogenous sample, we speculated that differences in the nature and severity of the sexual abuse experiences could result in different degrees of trauma with resultant differences in impact as posited in Trickett and Putnam (Reference Trickett and Putnam1993) and illustrated in Figure 1.

Some research has been conducted investigating the relationships between certain characteristics of sexual abuse and the impact on development. In an earlier work, Trickett, Reiffman, Horowitz, and Putnam (Reference Trickett, Reiffman, Horowitz, Putnam, Cicchetti and Toth1997) reviewed this research that examined (a) severity of the abusive act (i.e., penetration), (b) duration or frequency, (c) presence of force or violence, (d) relationship to perpetrator, and (e) age at onset of abuse. This review indicated the following: first of all, although in each of these areas there were studies showing the predicted results, in all cases there were studies not showing that association. The characteristics most consistently associated with more adverse impact were found to be longer duration of the abuse, force or violence accompanying the abuse, and father or father figure as perpetrator. But even for these characteristics there is research that did not support the association with more adverse outcomes. All of these studies examined only one characteristic of abuse (i.e., identity of perpetrator or penetration) and none examined whether there was a correlation between any of these characteristics (e.g., whether abuse by a father was associated with earlier onset). If such associations exist, it would be difficult to discern which of the characteristics is having the effect.

As a result of this review, we examined the interrelationships of abuse characteristics and relationships with demographic characteristics in our sample (Trickett et al., Reference Trickett, Reiffman, Horowitz, Putnam, Cicchetti and Toth1997). We found that, for this sample, early onset of abuse, abuse severity, and duration of abuse were all positively and significantly intercorrelated. Abuse perpetrated by a biological father was associated with earlier onset and longer duration. Of importance, being abused by an “other father figure” (stepfather or mother's live-in boyfriend) was significantly associated with later onset of abuse and shorter duration. Being abused by multiple perpetrators was associated with physical violence. In these analyses we also found relationships between abuse characteristics and ethnic minority status (minority status was associated with later onset of abuse and shorter duration) and SES (higher SES was associated with earlier onset, abuse severity, longer duration, and perpetrators other than fathers or father figures). Figure 3 illustrates these relationships and the degree of variability in the nature of the abuse experiences by our sample.

Figure 3.

The age of onset and duration of abuse for sample. Each horizontal line indicates one individual in the sample. The left end of the line indicates the age of onset of the abuse, and the length of the line indicates the duration of the abuse. The black lines indicate abuse with penetration and the grey lines abuse without penetration. The pair of vertical lines indicate the average age of onset of abuse in the sample (line on the left) and the average duration (line on the right indicating 2-year duration).

These complex findings of intercorrelations among abuse characteristics suggested to us the possible usefulness of determining whether within our sample of females abused by a family member, meaningful subgroups of participants could be formed and whether these subgroups would show differences in the nature or severity of developmental outcomes. To accomplish this, a hierarchical cluster analysis was performed in order to determine how the sample might be partitioned into meaningful profile groups based on abuse characteristic measures (age of onset, duration of abuse, perpetrator identity, severity of abuse, multiple perpetrator, and physical violence). For details of this analysis see Trickett, Noll, Reiffman, and Putnam (Reference Trickett, Noll, Reiffman and Putnam2001). A three-cluster solution was supported, which we referred to as subgroups. The first cluster (multiple perpetrator [MP] subgroup) comprised females who had been abused by MPs, none of whom were their biological fathers; the abuse was over a relatively short period of time but was likely to have been accompanied by pronounced physical violence. Approximately one-third of the MP perpetrators were nonbiological father figures (stepfathers or mothers' live-in boyfriends) and the remainder were other relatives. The females of the second cluster (single perpetrator [SP] subgroup) were characterized by abuse by an SP who was not the biological father. For this subgroup, the perpetrator was a nonbiological father figure in about half of the cases and another relative in the other half. Duration of the abuse for this profile was relatively short, and violence was not frequent. The third cluster (BF subgroup) was characterized by abuse by the primary father (in all but three cases the biological father) over a long period, beginning at a relatively young age with little violence. Note that these clusters illustrate the difficulty of conceptualizing a single dimension of severity of abuse. That is, the BF subgroup experienced abuse by a BF over a long period with little violence. In contrast, the MP subgroup experience abuse by more than one perpetrator, but not a BF, for a short period but with a high likelihood of violence.

In multivariate analyses we compared the three abuse subgroups and the comparison group on a number of measures of behavior problems and psychopathology symptoms at T1 and T4 (Trickett et al., Reference Trickett, Noll, Reiffman and Putnam2001). At T1, on nine measures derived from mother report and self-report, we found that the BF subgroup showed the most extreme pattern of behavior problems and maladjustment: this group differed from the comparison group on all nine of the outcome variables and also had scores that were elevated relative to the MP and SP subgroups on eight of the nine variables. That is, the scores obtained by the MP and SP subgroup members were intermediate between the BF subgroup and the comparison group and often significantly different from both. At T4, about 7 years on average after the disclosure of the abuse, there were fewer differences on the nine outcome variables than at T1. However, the BF subgroup continued to be the most different with scores elevated relative to the comparison group on five of the nine outcome variables. At this time point the MP subgroup did not differ from the comparison group on any of the outcome variables. In contrast, the SP subgroup showed elevated scores, relative to the comparison group on four of the nine variables and was indistinguishable from the BF subgroup in levels of depression and dissociation, thus exhibiting elevated symptomatology in comparison with the T1 levels for this subgroup.

In other analyses, we have also found that these three subgroups are differentially predictive in late adolescence on reported health problems (gastrointestinal and gynecological) (Sickel, Noll, Moore, Putnam, & Trickett, Reference Sickel, Noll, Moore, Putnam and Trickett2002); sleep problems (Noll, Trickett, Susman, & Putnam, Reference Noll, Trickett, Susman and Putnam2006); and sexual attitudes (Noll, Trickett, & Putnam, Reference Noll, Trickett and Putnam2003).

Psychopathology

The Diagnostic Interview for Children and Adolescents was given at T1, T2, and T3 (Reich, Reference Reich2000). Abused females met criteria for significantly more Diagnostic and Statistical Manual of Mental Disorders (DSM) diagnoses than comparison females. They exhibited greater comorbidity with 53% of the abused females meeting DSM criteria for two or more diagnoses (mean = 3.5 diagnoses) compared with 35% of the comparison females (Jareb, Reference Jareb1995). Batteries of standard self- and parent-report measures of state and trait anxiety, dissociation, posttraumatic stress disorder (PTSD) symptoms, depression, and behavioral problems were also administered across most of the time points. At many points, abused females scored higher on depression, trait anxiety, dissociation, PTSD, and somatic symptoms, as well as behavioral problems such as aggression, delinquent behaviors, and school problems (Bonanno, Noll, Putnam, O'Neill, & Trickett, Reference Bonanno, Noll, Putnam, O'Neill and Trickett2003; Trickett et al., Reference Trickett, Noll, Reiffman and Putnam2001). By adulthood, the major group differences were higher depression and alcohol and drug abuse (Noll et al., Reference Noll, Trickett, Harris and Putnam2009).

Dissociation was assessed at multiple time points using a series of related measures spanning childhood (Child Dissociative Checklist [CDC]; Putnam et al., Reference Putnam, Helmers and Trickett1993), adolescence (Adolescent Dissociative Experiences Scale; Armstrong, Putnam, Carlson, Libero, & Smith, Reference Armstrong, Putnam, Carlson, Libero and Smith1997), and adulthood (Dissociative Experiences Scale; Carlson & Putnam, Reference Carlson and Putnam1993). The sexual abuse group was significantly more dissociative during childhood (Putnam, Helmers, Horowitz, & Trickett, Reference Putnam, Helmers, Horowitz and Trickett1995) and adolescence (Bonanno et al., Reference Bonanno, Noll, Putnam, O'Neill and Trickett2003) than the comparison females. The group difference was no longer significant in adulthood, however (Noll et al., Reference Noll, Trickett, Harris and Putnam2009).

At T1, dissociation was significantly associated with experiences of sexual abuse that were characterized by earlier onsets and multiple perpetrators. At T4 (mean age = 18 years), dissociation was significantly associated with both physical and sexual revictimization, domestic violence, and self-harm, especially self-mutilation (Horowitz, Reference Horowitz1998). Dissociation also predicted sexual ambivalence on the Sexual Activities and Attitudes Questionnaire (SAAQ; Noll, Trickett, et al., Reference Noll, Trickett and Putnam2003). Pathological dissociation, defined by clinical cutoff score on the Adolescent Dissociative Experiences Scale was associated with significantly more PTSD symptoms at T4 (Bonanno et al., Reference Bonanno, Noll, Putnam, O'Neill and Trickett2003). Positive parenting behaviors in G1 mothers were negatively associated with their dissociation scores on the Dissociative Experiences Scale. Harsh discipline and punitive parenting was positively associated with dissociation scores in G1 mothers (Kim, Trickett, & Putnam, Reference Kim, Trickett and Putnam2010). Thus, dissociation, more than depression or anxiety in our sample, is broadly associated with trauma. It is both a sequelae of past trauma and it is associated with the future perpetuation of trauma such as revictimization, self-harm, and harsh parental discipline.

“Unusual” behaviors

In prior research (P.K.T. with physically abused children and F.W.P. researching dissociation in maltreated children) both investigators were impressed by the unusual and sometimes socially inappropriate or precocious behaviors exhibited by the maltreatment participants. One of the secondary aims of this study was to systematically examine verbal and nonverbal behaviors in sexually abused females that might contribute to their problems and symptoms as well as serve as targets for intervention. A number of behavioral scenarios were embedded in T1 to examine the child's responses to structured situations and interpersonal interactions. One of these was the Stranger Task (see Trickett, Reference Trickett1983), which examined how participants responded to an informal interaction with an unfamiliar male interviewer (who was blind to abuse status) during an 8-min warmup procedure prior to the start of psychological testing (Negriff, Noll, Shenk, Putnam, & Trickett, Reference Negriff, Noll, Shenk, Putnam and Trickett2010). The child was seated at the testing table and told she could draw a picture, color in a book, or read a magazine until the interviewer was ready to start. The male interviewer was seated at a desk across the room and busied himself writing and shuffling papers while engaging in light conversation on topics such as school, weather, or sports. After the warmup the child was administered a number of standard tests. Coders blinded to abuse status rated a videotape of the child's behavior during the warmup period for 54 discrete nonverbal behaviors such as “leans toward other,” “legs spread apart,” “crotch touch,” “side head tilt,” and “mouth open.” These behaviors were extracted from anthropological literature as common affiliative and social distancing behaviors (Mausert-Mooney, Reference Mausert-Mooney1992). A factor analysis produced a three-factor solution, wary, affiliative, and coy (internal reliabilities = 0.75, 0.72, and 0.69 respectively).

The abused females were significantly higher on the coy factor, which included simultaneous but contradictory approach and avoidance signals such as full smiling while shrugging their shoulders, or showing their tongues while crossing their legs. Females scoring high on the coy factor at T1 had significantly earlier ages of first consensual intercourse at T4, which was associated with increased sexually transmitted disease (STD) risky behaviors after controlling for sexual abuse status at T4 or T5. In contrast, females scoring high on the affiliative factor manifest healthy patterns of sexual and social interactions.

In a series of studies, Bonanno and collaborators carefully examined the interplay of facial expression with reported affects, level of distress, heart rate, and current and subsequent symptoms and social adjustment (Bonanno et al., Reference Bonanno, Keltner, Noll, Putnam, Trickett and LeJeune2002, Reference Bonanno, Noll, Putnam, O'Neill and Trickett2003, Reference Bonanno, Colak, Keltner, Shiota, Papa and Noll2007; Negrao, Bonanno, Noll, Putnam, & Trickett, Reference Negrao, Bonanno, Noll, Putnam and Trickett2005). Facial expressions were coded using a version (Emotion Facial Action Coding System) of the Facial Action Coding System developed by Ekman (Ekman, Friesen, & Hager, Reference Ekman, Friesen and Hager2002). Expressions of shame, disgust, anger, and humiliation were scored by blinded raters viewing videotapes of an open-ended narrative in which participants are asked to describe a self-selected “most distressing event or series of events” in their life. Interviewers used neutral prompts to keep the participant speaking for at least 6 min. Participants were encouraged to take a moment to think about what they wanted to say and if they drew a blank or ran out of things to say, to take a moment to relax and think about something related to the event.

The nonabused comparison females described the death of a close friend or relative, a divorce or significant family conflict, or a conflict with a close friend in that order. About two-thirds of the sexually abused females described an experience of sexual abuse and were grouped as “disclosers” for purposes of analyses. The final third, the “nondisclosers,” described events that paralleled in nature and frequency the types of traumatic events that the nonabused females described. Child sexual abuse disclosers and nondisclosers differed on facial expressions of disgust, shame, and humiliation. The “disclosers” had more shame, higher levels of PTSD and dissociative symptoms, more polite (non-Duchene) smiles, and earlier onsets of trauma. The “nondisclosers” had more expressions of disgust and scored higher on an index of repressive coping. The comparison participants expressed the highest levels of positive emotion.

Even within the disclosure groups, however, there was significant variability. In the nondiscloser group, females who displayed a greater temporal coherence between their verbal expressions of humiliation and their facial expressions of shame had significantly higher levels of PTSD symptoms. There was also a strong context effect in that a facial expression of positive emotion (Duchene smile) was associated with better social adjustment only if it was appropriate to the topic of the narrative at that moment. Females who displayed Duchene smiles in an inappropriate context (e.g., while describing frightening or repulsive details) had poorer current and subsequent social adjustment. A similar pattern was obtained between a measure of “genuine laughter” and current and subsequent social adjustment.

Sexual distortions and risky sexual behaviors

Through the course of the longitudinal study as participants aged from one developmental stage to another, opportunities to more thoroughly examine aspects of the original conceptual model continually arose. For example, there were questions regarding sexuality that we simply could not address in childhood when participants were younger. There were also questions about sexual activities and attitudes that we did not wish to broach when participants might be in the acute phases of recovery. One advantage of a longitudinal study is that the sample will eventually age into stages where in-depth approaches become appropriate and when sophisticated, more precise assessments can be exercised. Hence, when participants became adolescents, we sought to perform a comprehensive assessment of the development of sexuality. At T4 (mean age = 18), we designed a tool that would assess a wide array of sexual behaviors. Moreover, we understood that many participants would not yet have engaged in sexual activity, but would likely have thoughts and attitudes about sex that we wanted to more fully understand. We learned much from several researchers and theorists about where to start. For example, in their seminal work, Richard Udry, William Friedrich, and colleagues had compiled some questions about sexual attitudes based on work with younger children (Friedrick, Beilke, & Urquiza, Reference Friedrick, Beilke and Urquiza1987; Udry, Reference Udry1988). In 1992, Wolfe and Lehmann produced the Children's Impact of Traumatic Events Scale (Wolfe, Gentile, Michienzi, & Sas, Reference Wolfe, Gentile, Michienzi and Sas1991) that included a subscale specific to attitudes about sex for use with children who had been abused. In the mid-1980s Diana Russell wrote a definitive text portraying the unique emotional and behavioral struggles of incest survivors (Russell, Reference Russell1986).

Our resultant tool, the SAAQ (Noll, Trickett, et al., Reference Noll, Trickett and Putnam2003), is a multimedia instrument assessing a host of sexual activities including age at first intercourse, birth control efficacy, intercourse partners, HIV risk behaviors, STDs, pregnancies and sexual behaviors of peers. Unique to the SAAQ, a large array of sexual attitudes are also assessed including preoccupation, 15 items (α = 0.91) including masturbation, pornography consumption, thinking about sex, and being turned on by sexual themes and fantasies; permissiveness, 12 items (α = 0.96) assessing permissive attitudes toward a relatively normative set of desires and behaviors, including intimate affection, light and heavy petting, and voluntary intercourse; pressure, 6 items (α = 0.70) including gaining maturity and respect from being sexually active and that sex is expected and equated with being loved and wanted; aversion, 10 items (α = 0.85) including sex is dirty and embarrassing, avoidance, loss of respect for self and from friends and worry about STDs and pregnancy; and ambivalence is measured by preoccupation + aversion. The SAAQ factor structure was shown to be invariant across multiple assessments spanning 2 years (Noll, Trickett, et al., Reference Noll, Trickett and Putnam2003) and has been translated into three languages (e.g., Beaudoin, Carbonneau, Godbout, Bouchard, & Sabourin, Reference Beaudoin, Carbonneau, Godbout, Bouchard and Sabourin2007).

Several of these attitudes can be thought of as constituting “sexual distortions” or disturbances in how sex and sexual feelings are approached and regulated. Our data have illuminated some unique results regarding the developmental precursors to sexual distortions. For example, we have reported that childhood anxiety (Noll, Trickett, et al., Reference Noll, Trickett and Putnam2003) and low quality relationships with males throughout childhood (Noll, Trickett, & Putnam, Reference Noll, Trickett and Putnam2000) were predictive of subsequent sexual preoccupation in adolescence. Childhood sexual behavior problems were predictive of subsequent sexual aversion, and persistent, pathological dissociation predicted sexual ambivalence later in development. As can be seen in Figure 4, within-group analyses showed that those experiencing abuse at young ages by a BF (BF cluster group described above) reported the highest levels of aversion and ambivalence, indicating that abuse perpetrated by a BF may be difficult to overcome especially as regards the development of sexuality (Noll, Trickett, et al., Reference Noll, Trickett and Putnam2003).

Figure 4.

The differences in sexual distortion variables across subgroups. BF, biological father subgroup; SP, single perpetrator subgroup; MP, multiple perpetrator subgroup; Comp, comparison group; Preooc, sexual preoccupation; Aversion, sexual aversion; Ambiv, sexual ambivalence. Subjects were different from the comparison group at *p < .05 and different from the SP group at **p < .01.

Furthermore, in another paper (Noll et al., Reference Noll, Trickett and Putnam2000) factors that moderate the relationship between childhood sexual abuse and subsequent sexual activities were identified. For both abused and comparison females a large number of male peers in childhood networks predicted subsequent younger age at first voluntary intercourse, greater numbers of sex partners, and a lack of birth control usage in adolescence. Early sexual relations with boyfriends predicted younger age at first voluntary intercourse. For abused females only, a high quality of relationships with male peers and nonpeers in childhood predicted greater birth control efficacy in adolescence. These results suggest that the emotional depth of a high-quality relationship may be a key component in the cognitive restructuring of relationships with males in terms other than sexual and in the ability to glean emotional, rather than sexual, rewards from romantic attachments. These high-quality relationships may, thus, function to facilitate the reparation of basic trust and the “degeneralization” of all men as abusers.

Data gathered at the T5 assessment indicated that sexually abused females reported significantly higher rates of teen pregnancy (abused = 39%; comparison = 15%; p < .05) and teen motherhood (abused = 23.8%; comparison = 8%; p < .05; Noll, Horowitz, Bonanno, Trickett, & Putnam, Reference Noll, Horowitz, Bonanno, Trickett and Putnam2003; Noll, Newland, & Hulsmann, Reference Noll, Newland and Hulsmann2006, March). Taken together, these results have been instrumental in augmenting a growing body of research focused on how childhood sexual abuse can deleteriously affect optimal sexual development. Sexually abused adolescents report engaging in risky sexual behaviors that are consistent with the contraction of HIV and becoming a teen mother, which are arguably two of the highest priority public heath initiatives facing our youth today. Thus, prevention-focused research that will aid in our understanding of the pathways from childhood abuse to these risk behaviors is vitally important and will likely have a global impact on HIV and teen pregnancy rates.

Revictimization/domestic violence

The prospective, longitudinal nature of our study provided a unique opportunity to contribute to a growing body of literature focused on the continued cycle of violence and the inordinate rates of (re)victimization experienced by women with histories of prior victimization and sexual abuse (Briere & Runtz, Reference Briere and Runtz1987; Messman & Long, Reference Messman and Long1996). Thus, we sought to improve on available assessment tools (e.g., Krinsley et al., Reference Krinsley, Weathers, Vielhauer, Newman, Walker and Kaloupek1994) and devised a measurement instrument that would reliably elicit factual information concerning a comprehensive set of traumatic life events, deduce participants' appraisals of these events, and place these events in developmental context. The resultant Comprehensive Trauma Interview (Barnes, Noll, Putnam, & Trickett, Reference Barnes, Noll, Putnam and Trickett2009) is a semistructured interview that queries specific traumatic experiences across several domains including separations and losses, natural disasters, witnessing violence, physical abuse and/or assault, physical and medical neglect, emotional abuse, self-inflicted harm including suicide attempts, and sexual abuse and/or assault. There are detailed follow-up questions for each traumatic experience including ages and identification of perpetrators, ages at occurrence of events, frequency of occurrence, and extent of victimization. Participants also provide distress ratings for each individual traumatic experience that are anchored to the “worst or most upsetting” event identified at the outset of the interview. Intraclass kappa coefficients showed very high agreement between protective service records and participants' self-reported abuse details 15 years later with respect to severity (i.e., penetration vs. genital contact), age at abuse onset (±1 year), and age when the abuse stopped (±1 year). Moreover, we were able to show remarkable test–retest reliability across two time points occurring 2 years apart (Barnes et al., Reference Barnes, Noll, Putnam and Trickett2009). Not only is the Comprehensive Trauma Interview a highly useful tool for categorizing and contextualizing traumatic life histories, but these reliability and validity results have also been instrumental in reconciling controversy regarding victims' relative ability to recall details of traumatic events (e.g., Fergusson, Horwood, & Woodward, Reference Fergusson, Horwood and Woodward2000; Widom & Morris, Reference Widom and Morris1997; Williams, Reference Williams1994).

Results from T5 and T6 analyses (when participants were mean age 20 and 25, respectively) indicated that sexually abused females were almost twice as likely to have experienced sexual revictimization (odds = 1.99 ± 2.79, p < .05), and physical revictimization (odds = 1.96 ± 2.58, p < .05) compared to victimization rates reported by comparison females. Sexually abused females' revictimizations were also more likely to have been perpetrated by older nonpeers and characterized by physical injury than were victimizations reported by comparison females (Barnes et al., Reference Barnes, Noll, Putnam and Trickett2009). Abused females reported almost four times as many incidences of self-inflicted harm and suicidality (p = .002), and 20% more subsequent, significant lifetime traumas (p = .04) than did comparison females. Sexual revictimization was significantly and positively correlated with PTSD symptoms, dissociation, and sexual preoccupation. Physical revictimization was significantly and positively correlated with PTSD symptoms, dissociation, and sexually permissive attitudes. Self-harm was significantly and positively correlated with dissociation (Noll, Trickett, et al., Reference Noll, Trickett and Putnam2003).

At T6, we also assessed domestic violence using the Domestic Conflict Inventory (Margolin, Burman, John, & O'Brien, Reference Margolin, Burman, John and O'Brien1990), which assesses a host of violent acts (e.g., physically threatened, hit, or beaten) perpetrated by an intimate partner. Domestic violence was defined as having experienced three or more of these acts at the hands of an intimate partner. Results indicate that over 53% of sexually abused females report at least one domestic violence experience compared to 24% of comparison females, F (1, 162) = 9.45, p = .003 (Noll, Barnes, & Trickett, Reference Noll, Barnes and Trickett2010). The Domestic Conflict Inventory also allows categorization of domestic disputes into “mild” (e.g., purposely locked partner out of house; damaged partner's property out of anger) and “severe” (e.g., slapped or physically shaken a partner) incidences and allows quantification of whether the offense was perpetrated on the participant by a partner or whether the participant perpetrated the offense on a partner. Results reported for the first time in this manuscript indicate that, after controlling for age, minority status, SES, and cohabitation status, abused females were more likely to have experienced severe domestic violence perpetrated on them by a partner F (4, 143) = 4.18, p = .04. Further, a significant group interaction was found indicating that abused females who perpetrated mild offenses on a partner where more likely to have a partner perpetrate severe domestic violence, F (7, 143) = 4.79, p = .03. These results indicate that the process of domestic violence may not be a simple one for females with victimization histories. Sexually abused females who have a propensity to enact subtle or mild forms of aggression toward a domestic partner may be the most likely victims of more severe domestic violence.

Taken together, these results suggest that victims of sexual abuse are about twice as likely as comparison females to be revictimized (either sexually or physically) at subsequent times during later adolescence and young adulthood. They also have a propensity to engage in self-harm and suicidal behaviors at higher rates than do their nonabused peers. Severe domestic violence is also a common occurrence for abuse survivors, especially if they engage in subtle forms of perpetration that might provoke extreme responses from domestic partners. Hence, there are likely discernable avenues by which victimization reoccurs and, therefore, the possibility that disruptive processes can be interrupted. To more fully understand these processes and intervene accordingly would likely result in a substantial impact on overall rape and domestic violence rates.

Hypothalamic–pituitary–adrenal (HPA) axis/stress responsivity

At the time the study began, there was emerging evidence from rat and nonhuman primate literature showing how maternal separation and/or deprivation could be detrimental to developing physiology with regard to stress hormones (e.g., Gunnar, Gonzalez, Goodlin, & Levine, Reference Gunnar, Gonzalez, Goodlin and Levine1981). Relatively little was known, however, about HPA axis functioning in humans, especially during childhood and adolescence. We had good reason to suspect that an observational study of children exposed to sexual abuse would constitute the means by which to examine the effects of chronic stress on the developing human in a naturalistic and ethical manner.

A subsample of participants from the parent study were given ovine corticotropin-releasing hormone (oCRH) stimulation as part of a separate protocol in order to ascertain HPA axis regulatory systems several years after the disclosure of abuse (De Bellis et al., Reference De Bellis, Chrousos, Dorn, Burke, Helmers and Kling1994). CRH, a 41-amino acid peptide, selectively stimulates and regulates pituitary ACTH secretion. Results indicated that sexually abused and comparison females showed similar basal and overall stimulated plasma cortisol levels, but that the abused females showed significantly reduced total ACTH responses to oCRH stimulation. Attenuated plasma ACTH with corresponding robust plasma cortisol responses to oCRH stimulation suggests a breakdown in the regulatory HPA system. Further, cortisol responses were similar for both groups until about 30 min poststimulation, at which time cortisol levels in the abused group began to attenuate showing signs of decreased responses while the comparison group continued to show marked increases until 60 min poststimulation. This paper provided some of the first evidence for a breakdown in the regulatory HPA system for sexually abused children.

Given the specific function of cortisol in response to stress, we thought that it would be particularly important to understand the developmental course of cortisol within individuals exposed to severe or chronic stress. Accordingly, we sampled resting cortisol at the same time each morning at each of the six time points as an indication of basal or circulating stress hormone. These data provided the first developmental curves for cortisol spanning age 6 through age 32 (Trickett et al., Reference Trickett, Noll, Susman, Shenk and Putnam2010). Our novel findings provided the first longitudinal evidence to support the hypothesis that the “normative” developmental course for nonstress cortisol levels is, on average, a steady increase from middle childhood into early adulthood after which time there is a leveling off. An additional important and unique finding from this analysis was the interaction between childhood sexual abuse and developmental trajectory indicating a developmental transition from higher levels of cortisol (hypercortisol) in childhood to lower levels of cortisol (hypocortisol) by early adulthood. These results may help to reconcile vast discrepancies in the literature stemming from the reporting of disparate findings across developmental stage; for example, studies of abused children and adolescents showing higher cortisol levels (Carrion et al., Reference Carrion, Weems, Ray, Glaser, Hessl and Reiss2002), and studies of adults retrospectively reporting childhood abuse exhibiting a suppression of cortisol production (Bremner, Vermetten, & Kelley, Reference Bremner, Vermetten and Kelley2007). Moreover, these results lend convincing support for the attenuation hypothesis (Susman, Reference Susman2006), suggesting that early and severe stress leads to an initial heightened stress response, which is in turn, suppressed over time. This suppression may be indicative of an adaptive response given the known consequences of chronic exposure to glucocorticoids including deleterious effects on brain structures (Carrion et al., Reference Carrion, Weems, Ray, Glaser, Hessl and Reiss2002). However, such adaptation may come at a cost as low levels of circulating cortisol have been associated with psychosocial problems including PTSD (Miller, Chen, & Zhou, Reference Miller, Chen and Zhou2007) and antisocial behaviors (Bergman & Brismar, Reference Bergman and Brismar1994), as well as physiological health consequences such as immune and cardiovascular functioning, rheumatoid arthritis, chronic fatigue syndrome, and fibromyalgia (Heim, Ehlert, & Hellhammer, Reference Heim, Ehlert and Hellhammer2000; Raison & Miller, Reference Raison and Miller2003; Sternberg & Gold, Reference Sternberg and Gold2002).

A model proposed by Bauer, Quas, and Boyce (Reference Bauer, Quas and Boyce2002) posited an interactive synergy between multiple stress response systems that should function in consort to produce optimal responses to novel or threatening conditions. This model also asserted the possibility that activation can occur in one system and be understimulated or blunted in another. Such asymmetry in stress response patterns could represent an additional indication of global impairment thereby limiting resources to cope effectively with the demands of a stress or threat. In response to this emerging theory and in recognition that the regulation of arousal and stress involves multiple systems, we expanded our assessment of stress responsivity beyond the HPA to include arousal in the autonomic nervous system via indicators such as heart rate and vagal influences. In a recent paper (Shenk, Noll, Putnam, & Trickett, Reference Shenk, Noll, Putnam and Trickett2010) we report longitudinal findings indicating that childhood sexual abuse predicted an asymmetrical stress response, marked by vagal suppression and blunted cortisol reactivity, 7 years after entry into the study at the T4 assessment. This asymmetrical stress response in turn predicted higher levels of depressive symptoms and antisocial behaviors at T6, which was 6 years after stress reactions were assessed. These results remained significant after controlling for prior levels of psychopathology and baseline levels of physiological indicators. These results underscore the importance of assessing multiple biological systems and suggest that interventions with sexually abused females should consider focusing on the management of mild to moderate stress in order to protect against later psychopathology.

Development of obesity

Research in the past 15 years has provided increasing evidence for a potential association between adverse childhood experiences and the subsequent development of obesity. However, the majority of the extant research is cross-sectional using retrospective reports of past abuse (e.g., Williamson, Thompson, Anda, Dietz, & Felitti, Reference Williamson, Thompson, Anda, Dietz and Felitti2002) precluding strong inference about the connection between childhood abuse and obesity. As a broad indicator of overall health, we objectively obtained height and weight measurements at all six time points. Analyses of these data provided some of the first prospective evidence that sexually abused females are at inordinate risk for developing obesity (Noll, Zeller, et al., Reference Noll, Schulkin, Trickett, Susman, Breech and Putnam2007). Although obesity rates via body mass index calculations were not different across groups in childhood or adolescence, by young adulthood (ages 20–27) abused females were significantly more likely to be obese (42.25%) than were comparison females (28.40%). Growth-trajectory analyses indicated that abused female participants, on average, acquired body mass at a significantly steeper rate from childhood through young adulthood than did comparison female participants after controlling for minority status and parity. We compared these growth trajectories to CDC population trends and showed that the average linear trend for comparison females mirrored that of the population falling almost exactly on the 50th percentile. However, the linear trend for sexually abused females was persistently steeper than the CDC population trend across development and exceeded the 75th percentile by young adulthood.

These results provided strong evidence that victims of sexual abuse might readily adopt lifestyles that are consistent with the development of obesity perhaps due to various abuse sequelae such as depression, body image disturbances, poor peer relations, low self-esteem, and/or the development of binge-eating disorders. However, we assert in this paper the further possibility that sexual abuse victims might be predisposed to obesity due to the high concentrations of cortisol in the formative years of adipose tissue development that is largely responsible for abdominal fat in females (Pasquali et al., Reference Pasquali, Cantobelli, Casimirri, Capelli, Bortoluzzi and Flamia1993; Rosmond, Reference Rosmond2003). We suggest that treating early HPA disruption may have an impact on the overall obesity rate.

Puberty

In our 1993 article (Trickett & Putnam, Reference Trickett and Putnam1993), we were one of the first to point out that historically the peak age of onset of sexual abuse for females is prepubertal (7 or 8 years of age) and the average duration tends to be about 2 years. We surmised that that there may be a directly traceable mechanistic link between the impact of sexual abuse on specific biological processes of pubertal development. Consistent with psychosocial acceleration theory (Belsky, Steinberg, & Draper, Reference Belsky, Steinberg and Draper1991) and its extended paternal investment theory (Ellis, McFadyen-Ketchum, Dodge, Pettit, & Bates, Reference Ellis, McFadyen-Ketchum, Dodge, Pettit and Bates1999), we sought to more fully understand how the experience of childhood sexual abuse might regulate pubertal development. To this end, we assessed secondary sexual characteristics via Tanner breast and pubic hair staging at T1 through T5. Given the ordered categorical nature of the Tanner stage data (i.e., ordinal rating scale ranging from 1 to 5), analytic technology has only recently emerged to facilitate this type of complex growth modeling. Longitudinal cumulative logit analysis via generalized linear modeling was utilized to analyze the rapidity of pubertal development for the sample. Probabilistic interpretations of parameter estimates demonstrate that, on average, abused females reached Tanner breast Stage 2 at 7.5 months earlier (p = .009) and Tanner pubic hair Stage 2 at 6 months earlier (p = .01) than comparisons (Noll & Trickett, Reference Noll and Trickett2009, June). These results suggest that the experience of sexual abuse may trigger biological mechanisms, which in turn accelerate pubertal development. Early pubertal maturation in females has been associated with several negative health conditions and psychosocial outcomes including increased body mass index, reproductive system cancers, adolescent pregnancy, mood disorders, and substance abuse (see review by Mendle, Turkheimer, & Emery, Reference Mendle, Turkheimer and Emery2007; our results underscore how early pubertal maturation might exacerbate an already-turbulent development for victims of sexual abuse).

Summary

In addition to the health outcomes discussed above, we have shown that females in the sexually abused group report greater healthcare utilization and gynecological problems (Sickel et al., Reference Sickel, Noll, Moore, Putnam and Trickett2002), more persistent problems with sleep (Noll et al., Reference Noll, Newland and Hulsmann2006), and higher rates of preterm delivery (Noll, Schulkin, et al., Reference Noll, Schulkin, Trickett, Susman, Breech and Putnam2007) than do females in the comparison group. Our research has been instrumental in illuminating the toll that early abuse has on physiological outcomes. With our longitudinal design, well-matched control group and state of the art analytic models, we have provided some of the most definitive evidence by going well beyond cross-sectional studies or adult retrospective studies. Hence, the literature is better poised to support causal inferences about the impact of sexual abuse on health. HPA dysregulation, obesity, cognitive challenges, HIV risk, teen pregnancy, preterm delivery, and early puberty are among our findings in this vein, many of which arguably constitute the major public health concerns of our time.

The G1/G2 dyad

At T1, we obtained information about G1 mothers' childhood using the Mothers' Developmental History Questionnaire (Trickett & Everett, Reference Trickett and Everett1988), a structured interview protocol developed for this study. From the protocol the variables were derived about the G1s' own childhood abuse histories, their childhood separations from parents, and their reports of their own parents' parenting styles. Other measures obtained at T1 and T2 provided information about G1 mothers' reports of their own parenting styles, family of origin and current family environments, and psychological functioning (depression, anxiety, and dissociation).

Consonant with the findings of others (e.g., Kaufman & Zigler, Reference Kim, Chung and Trickett1987; McCloskey & Bailey, Reference McCloskey and Bailey2000) of an intergenerational association between mothers' childhood sexual abuse and their daughters', we found that almost one-half of the G1 mothers (45%) in the abused sample reported having been sexually abused during their own childhood, whereas 16% of the mothers in the comparison sample did so. Most of this abuse (85%) was reported to be intrafamilial and with mean age of onset of 8.5 years (Kim, Noll, Putnam, & Trickett, Reference Kim, Noll, Putnam and Trickett2007). In order to better understand possible intergenerational mechanisms, analyses were conducted comparing three groups of G1 mothers: (a) mothers of abused daughters who themselves reported childhood sexual abuse, (b) mothers of abused daughters who were not themselves abused, and (c) nonabused mothers of nonabused (comparison) daughters (Kim, Chung, & Trickett, 2007). We found that the mothers of sexually abused daughters who themselves were sexually abused (compared with the other two groups) report the most physical and emotional abuse by their own mothers and fathers; the most separations during childhood from their own mothers; the most residential moves as a child; the lowest current emotional support from their families of origin; the highest (current) depression; and the lowest provision of positive structure to and satisfaction with their daughters. The mothers of sexually abused daughters who were not themselves abused reported the highest use of punitive discipline with their daughters. The mothers of sexually abused daughters (regardless of their own childhood experiences) reported more state and trait anxiety, lower current family cohesion, and higher current family stress (especially stress about financial issues, family violence, and alcohol problems) compared with nonabused comparison group mothers. Thus, these univariate analyses indicate that, as a group, the mothers of the sexually abused females report a lot of current distress (in the 2 years following the disclosure of their daughters' abuse) and problematic parenting, but that those mothers who were themselves sexually abused (i.e., for whom there was an intergenerational association), report the most unstable and harsh childhoods, the most current psychological distress, the least emotional support from family, and the least supportive parenting.

In order to understand further how these many factors might impact on the provision of support and other aspects of parenting, Kim et al. (Reference Kim, Trickett and Putnam2010) examined different explanatory models of the association between G1 mothers' childhood sexual abuse experiences and their later parenting practices (provision of positive structure and use of punitive discipline) using structural equation modeling. In these analyses, the direct effect of mother's childhood sexual abuse on parenting practices was examined while simultaneously controlling for other G1 childhood experiences (harsh upbringing), current resources and problems (satisfaction with social support and level of dissociation), and daughter's sexual abuse status. For the provision of positive structure variable, in these multivariate analyses, there was no direct effect of G1 mother's sexual abuse on parenting. Rather provision of positive structure was predicted by the daughter's sexual abuse status and by the mother's level of dissociation, which was in turn predicted by the mother's harsh upbringing. That is to say, mothers of sexually abused daughters and those with high levels of dissociation reported lower levels of positive structure.

For the parenting variable, punitive discipline, there was a direct, negative effect of G1 mothers' childhood sexual abuse on punitive discipline. There was also a direct, negative effect of satisfaction with social support, and as before a direct effect of level of G1 mothers' dissociation and daughters' abuse group status. Thus, for this parenting variable, G1 mothers reporting childhood sexual abuse, higher levels of satisfaction with social support, and lower levels of dissociation reported lower levels of punitive discipline. In the same multivariate analyses, daughters' sexual abuse predicted higher levels of mothers' punitive discipline. In short these findings show that parenting practices are multiply determined: daughters' child sexual abuse, mothers' childhood experiences (of sexual abuse and of harsh, punitive upbringing), current dissociative symptoms, and current social support were significant predictors of their parenting practices.

Kim (Reference Kim, Noll, Putnam and Trickett2007) also examined the effects of the G1 mothers' social and psychological resources, as reported at T1, on the G2 daughters' behavioral adjustment at T4, at mean age 18 approximately 7 years after the disclosure of childhood sexual abuse. At that time, G2 females reported on their internalizing (i.e., withdrawal/depressed, immature/bizarre symptoms) and externalizing (i.e., delinquent, aggressive behaviors) behavior problems. G1 mothers' reports of support from their family of origin were associated with lower withdrawal/depressed symptoms among comparison females but higher immature/bizarre behaviors among the sexually abused females. The mothers' reported enjoyment of their parental role predicted lower immature/bizarre and aggressive behaviors in the abuse group, whereas such effects were negligible among comparison females. For the sexual abuse group only, the higher the level of the mothers' authoritarian control, the higher the G2 daughters' withdrawal/depressed symptoms.

In sum, these analyses indicate that characteristics of the upbringing of the G1 mothers of the sexually abused and comparison females (including but not limited to their own sexual abuse experiences) are associated with their own mental health and parenting close to the time of their daughters' disclosure of the abuse (T1). In addition, in turn, the G1 mothers' parenting at T1 was related to G2 daughters' mental health symptoms in midadolescence about 7 years later (T4).

The G2/G3 dyad

The prospective nature of our data presented the unique opportunity to demonstrate the inordinate prevalence of various forms of adversity and risk for maldevelopment operating in the lives of offspring born to mothers who experienced childhood sexual abuse. They also could provide a snapshot of the cumulative risk to these offspring, the potential for continued victimization and adversity, and a powerful illustration of the amount of burden that children born into adversity are required to bear. Because of demographic similarities across groups and because these women differed with respect to an objectively determined variable (i.e., substantiated maternal childhood sexual abuse), these data offer inferences about the extent to which offspring burden is at least in partially attributable to maternal childhood abuse. One paper (Noll et al., Reference Noll, Trickett, Harris and Putnam2009) compared the magnitude of the cumulative burden across 91% of all known offspring born to the sample at the T6 assessment. One hundred twenty-three G3 offspring were assessed: 67 born to sexually abused mothers (OA) and 56 born to comparison mothers (OC). Variables examined included various sequelae of childhood sexual abuse that had been upheld in the extant literature as constituting substantial risk to the wellbeing of offspring. Results indicated that the abused G2 mothers were more likely than comparison G2 mothers to have experienced at least one physical victimization, reached clinical cutoff for adult depression, be diagnosed with at least one psychiatric disorder, report a substance dependence, report an alcohol dependence, be a high-school dropout, be the victim of domestic violence, and be obese. G3 offspring born to sexually abused mothers were more likely than offspring born to comparison mothers to have been born to a teen mother, have been born premature, and have been involved in child protective services. The average number of cumulative risks was significantly different across the two groups of offspring; with the OA group averaging 6.88 and the OC group averaging 3.88 F (1, 129) = 5.89, p < .05.

Our intergenerational sample also allowed the unique opportunity to examine the attachment styles of the G2/G3 dyad (Kwako, Noll, Putnam, & Trickett, Reference Kwako, Noll, Putnam and Trickett2010). Utilizing a subset of 35 G2 mothers and their 54 G3 children ranging in age from 11 months to 11.75 years half of whom were female, we performed standard attachment paradigms. We used the Strange Situation procedure classified in two ways: one for infants from 11 to 23 months (Ainsworth, Blehar, Waters, & Wall, Reference Ainsworth, Blehar, Waters and Wall1978), and one for toddlers and preschool-aged children from 2 to 5 years (PAA; Crittenden, Reference Crittenden1992). For children between 6 and 11 years of age, we used the School-Age Assessment of Attachment (Crittenden, Reference Crittenden1997–2005), which is a semiprojective test consisting of picture cards of minor to major threats and coded via a developmentally attuned version of the Adult Attachment Interview discourse analysis (Farnfield, Hautamäki, Nørbech, & Sahhar, Reference Farnfield, Hautamäki, Nørbech and Sahhar2010). Results indicated that children in the OA group were more likely to have extreme strategies of attachment than the children in the OC group typified by anxious attachment.

At the T6 assessment, we also obtained information about the G3 samples' involvement in CPS. Where available, we ascertained this information via local jurisdiction caseworker reports, but for the majority of G3s, we relied on G2 reports. Of the 123 G3 offspring assessed, OA were significantly more likely to have been involved in CPS than were comparison offspring (OA = 17.91%, OC = 1.78%; Noll et al., Reference Noll, Newland and Hulsmann2006, Reference Noll, Trickett, Harris and Putnam2009). The rate of CPS involvement for the OA group was more than twice the national average reported in proximal years. Forty percent of these cases resulted in permanent removal from the G2 home. The majority of the OA cases were neglect cases mostly due to the substance and/or alcohol dependence of the G2. There were two cases of physical abuse perpetrated by the G2 and one case of sexual abuse perpetrated by the maternal grandfather who was the original perpetrator of the G2. To date, we know of three infant deaths, all occurring in the OA group; one died due to complications of prematurity, one died due to being left alone in a bathtub by the G2, and one died shortly after being born to a heroin addicted G2. Overwhelmingly, the strongest predictor of G3 CPS involvement was the teen pregnancy status of the G2 mother with over 57% of CPS cases concerning OA offspring of teen mothers (Noll et al., Reference Noll, Newland and Hulsmann2006). These results highlight the compounded deleterious effects of being born to a teen mother who experienced sexual abuse in her own childhood.

Summary

These intergenerational findings underscore the complex network of risk factors that may be operating in the lives of children born to victims of childhood sexual abuse, as many of these children are at risk for (a) being abused or neglected either at the hands of their own caregivers or by other violent or exploitive individuals who are allowed access to vulnerable children and (b) various deleterious consequences of having a caregiver who suffers from the emotional, psychiatric, and physical sequelae of her own childhood abuse, many of which have health-related implications for offspring. These results also suggest that mothers who were sexually abused do not necessarily become abusers themselves. Most did not abuse or harm their children in any direct manner. Those that did have children involved in CPS were either neglectful (mostly attributable to substance use issues) or in other ways recreated environmental conditions in which abuse was allowed to persist across generations. Our intergenerational results clearly suggest that primary prevention/intervention efforts extending throughout development and focusing on the cumulative risk to offspring (including, but not limited to CPS involvement) will likely improve victim outcomes and curtail intergenerational transmission of violence and adversity.