Our study shows that outcome oriented and psychometric research approaches on resilience converged to some extent. Yet, each was deficient in its own way. By using different research vistas and diligent control for confounding, we were able to avoid bias and to identify more clearly predictors and correlates of positive mental health despite trauma. Our findings corroborate some previous results but also expand these findings with regard to the elderly who had traumatic experiences during their childhood and adolescence.
With regard to the outcome oriented approach, comparing persons with mild-to-moderate trauma or PTSD with non-PTSD (‘resilient’) persons, a higher number of life-time traumata and current depressive symptoms were found to be associated with current symptoms of posttraumatic stress, corroborating previous results [24, 54]. While the number of previous traumata may be reliably regarded as a variable risk factor, analyses with our matched case–control sample suggest that depression is most likely no true risk factor but rather a concomitant of PTSD and its symptoms. Evidence substantiating this conclusion in the elderly was recently reported by Chaudieu et al. . Symptoms of depression may thus indicate current posttraumatic stress rather than pose a risk factor of PTSD in the elderly. This needs consideration in clinical treatment.
A medium level of education, compared to a low level, also appeared beneficiary for non-PTSD in our study. Previous studies reported conflicting evidence on this issue [11, 24]. Judging from our data, a high level of education does not impede adaptation to trauma . It rather seemed to have no specific beneficial effect compared to a lower level of education. The effects of education need to be investigated more specifically in future research.
Sex did not emerge as a risk factor in our study, corroborating findings by Spitzer et al.  in a community sample of elderly Germans. However, full and sub-threshold PTSD was more frequent among married persons in our study, compared to persons with some symptoms of PTSD but no probable diagnosis. Matching with regard to marital status and a number of other sociodemographic characteristics, persons with full or sub-threshold PTSD were also more likely to have children. These findings are somewhat at odds with previous results on higher levels of perceived social support [18, 24, 27] and greater social engagement  in resilient persons. Yet, our findings may reflect an aspect of help-seeking behaviour in persons with PTSD. Recent research suggests that spouses’ emotion-focused coping strategies may have a beneficial impact on victims’ PTSD symptoms . Stronger familial ties and an increased likelihood to rear children may be a consequence of this kind of help-seeking behaviour. Marital status and a higher likelihood to rear children may in this respect be regarded as consequences of PTSD in the elderly. More research is, however, needed on this topic.
Voluntary work in old age was also associated with a lower probability of full or sub-threshold PTSD. In absence of longitudinal data, we suggest that this may be inversely interpreted as a consequence of the debilitating symptoms of PTSD: persons with PTSD may be less able — because of their symptomatology — to involve themselves in such activities . Volunteerism may thus be understood as an indicator (i.e., a concomitant or consequence) of non-PTSD but not as a protective factor against PTSD.
In contrast to other reports on positive posttraumatic outcome [30, 35], PTSD and non-PTSD were not characterized by differences in social acknowledgement or of having had the opportunity to talk openly about war-time experiences with someone. These conflicting results may be due to sampling differences: Forstmeier et al.  investigated former WWII child soldiers, whereas Maercker and Müller  studied survivors of political imprisonment in former Eastern Germany and recently traumatized crime victims. These samples may have been representative of persons who had some ‘special’ or uncommon traumatic experience. Most civilian WWII child-survivors are not recognized as having a special or in some way outstanding history to tell. Consequently, social acknowledgement and the seeking of such may be generally lower in the cohorts of civilian WWII child-survivors. While social acknowledgement could have been beneficial to them, they might not have had the chance to acquire it.
With regard to the psychometric approach, the 10-item CD-RISC was found to discriminate reliably between PTSD and non-PTSD persons in our study. In the matched case–control sample, differences were accordingly also greatest in items that were related to PTSD symptoms of clusters B and D (Items 6, 7, and 10). Thus, resilience as measured with the CD-RISC evidently mirrored to a large extent only PTSD symptom severity, calling the utility of the CD-RISC somewhat into question . Moreover, being able to adapt to changes (Item 1), considered an essential indicator of psychometrically defined resilience , did not discriminate between matched PTSD and non-PTSD persons. Yet, among those whose psychological health was above average, seeing the humorous side of problems (Item 3) and maintaining the impression that coping with stress can be strengthening (Item 4) were found to be indicative of having dealt successfully with an environmental risk factor in the past. This study thus corroborated that humour is an important component of resilience and coping [34, 58]. Yet, our study also shows that it is complemented by a challenge-oriented attitude towards life. We suggest that these two inter-related cognitive-behavioural characteristics should be regarded as protective factors. Fostering these two factors could thus be important for prevention programmes that seek to boost resistance against posttraumatic stress and PTSD. Experimental and longitudinal research is needed here.
With regard to the initially posed question (‘Is it resilience?’), our study provides no definite answer. We obtained evidence of a risk and a protective factor (number of life-time traumata, medium education), and of a number of likely correlates and consequences of PTSD and non-PTSD in the elderly (symptoms of depression, voluntary work, marital status, likelihood to rear children). From this perspective, only fewer traumata and a medium level of education appeared to promote better mental health and resilience (i.e., showing less likely symptoms of PTSD), replicating previous results . Humour and a challenge-oriented attitude towards life were found to be important aspects in coping successfully with an environmental risk factor in the past. However, these characteristics did not discriminate PTSD from non-PTSD. Thus, our study’s main contribution may lie in pointing out that the question ‘What is resilience?’ needs reformulation. Studies need to examine in more detail which specific factors contribute to good mental health in which specific way. Likewise, studies need to differentiate more systematically between different levels of outcome (e.g., PTSD and non-PTSD), types of correlates (i.e., risk factors, concomitants and consequences), and cognitive-behavioural characteristics of psychometric definitions and operationalizations of resilience that may promote a better outcome. In conclusion, our results underline that the psychometric assessment of resilience needs improvement and should be based on a stringent definition of resilience that avoids too large an overlap with the symptomatology of PTSD. Such an instrument should incorporate — in a ‘multiple pathways’ approach — different and various components that are thought to bring resilience about or for which ample evidence already exists (i.e., humour).
Limitations of our study pertain to its cross-sectional character, which precludes direct inference on causality, problems of reporting bias given the old age of the participants and the large time spans covered, and only limited control over confounding variables that may have introduced further bias, like sampling. In the absence of normative data with regard to the base population of war-exposed Austrians, it is unclear whether our sample was truly representative. The use of ad-hoc scales and single items with regard to social support and acknowledgement on WRTs may have biased results. Associations with the Big Five personality traits and specific coping styles [32, 59], which may help in mapping out the terrain of resilience as a personality trait , were also not considered in this study.