The Childhood Trauma Questionnaire–Short Form (CTQ-SF) (Bernstein et al., 2003) is a widely used retrospective and self-report questionnaire with 28 items. It assesses the frequency of five forms of CM occurring during childhood–like being hit (PA), verbally mistreated (EA), not having enough food (PN), being forced to assist or do something sexual (SA) and not having a supportive family (EN) (ranging from 1 to 5 respectively - never, rarely, sometimes, often, or, very often true). The CTQ-SF presented good psychometric properties with Cronbach's alpha varying between .61 for PN and .95 for SA in clinical populations (Bernstein et al., 2003), and between .58 for PN and .94 for SA (Scher, Stein, Asmundson, McCreary, & Forde, 2001) in community samples. Results for time stability and convergent and divergent validity were considered good (Bernstein et al., 2003). In the Portuguese population, the original CTQ-SF factorial structure was confirmed in a community sample of 746 subjects and Cronbach's alpha varied between .47 for PN and .79 for EN, with .84 for the overall scale (Dias et al., 2013). In the current sample, Cronbach's alfa were .45 for PN, .71 for SA, .74 for EA, .77 for PN, .79 for EN and .84 for the overall CTQ-SF. We used the Bernstein and Fink (1998) cut-off points to classify CM with clinical significance in each scale: EA ≥13; PA ≥10; SA ≥8; EN ≥15; PN ≥10. For the overall scale we used the cut-off point of one standard deviation above the average.

The Posttraumatic Diagnostic Scale is a 49-item scale to assess PTSD as defined by DSM-IV (PDS) (Foa, Cashman, Jaycox, & Perry, 1997). It comprises of four parts: exposure to TEs; symptoms of re-experiencing, avoidance and hyperarousal, and PTSD severity (17 items, classified in a four-point frequency scale (0 - Not at all or only one time; 1–Once a week or less/once in a while; 2 - 2 to 4 times a week/half the time; 4 - 5 or more times a week/almost always), and the impairment for various life-domains. Psychometric properties of the original version showed good internal consistency (alpha coefficient of .92), good test-retest reliability and a convergent validity of .82 (Foa et al., 1997). In a community sample of 404 Portuguese subjects, the original factor structure was confirmed. Internal consistency was found to be good, varying between .81 for hyperarousal and .92 for severity (Ribeiro et al., 2011). In the analyzed sample, PDS had an internal consistency of .83 for avoidance, .85 for hyperarousal, .89 for re-experiencing and .93 for PTSD severity.

The PTSD diagnosis was assessed applying the criteria of the DSM-IV: exposure to TEs accompanied by at least one symptom of re-experiencing, three or more of avoidance and two or more of increased arousal, with significant impact in daily activities.

In this ex post facto investigation, subjects were considered revictimized when they identified subsequent TEs as assessed by the PDS after being exposed to CM, as measured by the CTQ-SF. The research protocol included informed consent, questions about socio-demographics listed in Table 1 and the instruments mentioned above. Research was conducted under the Declaration of Helsinki. The data collection was approved by the Ethical Commission of the Military Hospital of Coimbra and by the Portuguese Commission for Data Protection (CNPD) in Portugal, and all participants signed an informed consent.

We found that the probability of PTSD diagnosis increased more than three times in CM subjects (OR=3.3, 95% CI [2.1, 5.2]), as described in Table 2. Emotional abuse, PA and SA increased the probability of PTSD to the greatest extent. Emotional abuse was found in 31.6%, a larger percentage than that of subjects exposed to PA or SA.

In addition, we studied the risk for PTSD using logistic regression and including demographic variables (age, education, marital status and type of job), forms of CM and the number of TEs. The analyzed model significantly predicted subjects with and without PTSD diagnosis (chi-square=133.70, p ≤ .001; Nagelkerke's R2=.24) and identified 90.9% of subjects correctly (99.2% without and 11% with diagnosis). From the included set of predictors, marital status, type of job, EA and the number of TEs were significant (Wald p<.05) (Table 4).

Our study adds evidence about the long-term consequences of CM in adults in the community. Self-reported CM was associated with a two-fold risk for exposure to new traumatic events and three-fold for PTSD diagnosis. More than 30% of adults with PTSD reported moderate or severe EA during childhood. Moreover, EA remained significantly associated with revictimization and PTSD after adjusting for demographics. While previous studies underlined the consequences of PA and SA, our findings strengthen the assumption that EA can be even more harmful than other forms of CM. Although it figures among the most frequent CM forms (Stoltenborgh et al., 2015) it is rarely a motive for specific measures for child protection.

Clinicians and researchers need to be aware of the injurious potential of other less evident risk factors that can contribute to traumatic exposure and PTSD across the lifespan. The identification of pre-trauma risk factors for PTSD is valuable in the sense that it can help to tailor intervention protocols targeting more efficiently PTSD and related comorbidities (Britvić et al., 2015; DiGangi et al., 2013). Health professionals need to look “outside the box” and pay more attention to the identification of etiologic and maintenance factors for PTSD and its associated features (Keeley et al., 2016; Timimi, 2014).

In our sample, a 9.1% risk for PTSD was found. This was slightly larger than De Albuquerque et al. (2003) found in the Portuguese population (7.8%). A life-threatening disease, accidents or having someone close suffering them were mentioned as most frequent TEs. However, subjects exposed to sexual assault, war exposure and torture had the highest percentages of risk for PTSD. These findings clearly indicate that war is a relevant traumatic event for the Portuguese society. Nevertheless, civilians were at higher risk for revictimization and PTSD than military personnel. This is remarkable as military duties involve exposure to potential TEs and as the Portuguese military was exposed to many hardships in the Colonial Wars. A possible explanation is that military training exerts a protective effect in terms of the perception of threatening events, providing subjects with higher thresholds for experiencing events as traumatic. This finding deserves further investigation as it can elucidate about potential resilience mechanisms for PTSD.

Exposure to accidents, the second most frequent TE, was found to be significantly associated with all CM forms except SA. This result is particularly relevant because of the high incidence of car accidents in Portugal. It is likely that CM persons may have impaired skills for risk perception (Chu et al., 2016) and for emotional regulation (Messman-Moore et al., 2010) and consequently, a greater vulnerability for exposure to accidents. Another potential explanation refers to the lack of sleep in CM persons suffering from PTSD and related comorbidities (Britvić et al., 2015; Kendall-Tackett, 2002). When these hypotheses are confirmed, specific actions are needed to assess and improve the quality of sleep, the risk perception and emotion regulation in vehicle motorists exposed to CM.

Remarkably, PA and SA did not predict PTSD severity in our sample, as previously found (Breslau et al., 2014). This finding confirms that consequences for CM forms may vary depending for instance, on the cultural background of analyzed samples.

To our knowledge, this is the first study on the relations between CM, traumatic exposure and PTSD among Portuguese adults. We confirmed that CM increases the risk for traumatic exposure and PTSD in adults in the community, and that EA is an important predictor, although previous studies underlined more often the injurious potential for PA and SA. Because we analyzed diverse forms of traumatic exposure in adults, we learned that common life events such as accidents can be linked to CM. As we used a community sample, our findings are less biased than those focused on subjects with substantiated CM or clinical groups. However, uncertainty about the representativeness of the Portuguese sample imply caution in generalizing our findings.

We used robust self-report instruments validated in Portuguese samples, although the PN subscale of the CTQ-SF was found to have low internal consistency. It is well known that recall bias can reduce the reliability and effectiveness of retrospective assessments (Bernstein et al., 2003). Although the use of anonymous self-report questionnaires may facilitate the disclosure of CM experiences, one should be careful with statements about causality. The PTSD diagnosis has been defined according to DSM-IV, and not to DSM-5, but current PTSD prevalence was found not to differ significantly when both editions were compared (O’Donnell et al., 2014).

Our findings reinforce that prevention of CM, particularly EA, must be improved. Agencies for child protection and health care need to join efforts to stop EA and to broaden interventions to prevent CM-associated risk for revictimization and PTSD across the lifespan. The routine screening for diverse CM forms among clients suffering from PTSD and related comorbidities can contribute to the improvement of therapeutic interventions. The prospective investigation about how emotional processing, sleep and risk detection occurs in maltreated versus non-maltreated adults can shed light about pathways linking CM, revictimization and PTSD, and better inform preventive and intervention programs.