IntroductionRationale

Resilience refers to positive adaptation after exposure to adversity.1 2 It is often conceptualised in terms of health and well-being.1 3 4 A deeper understanding of resilience is highly relevant for public health: what are the factors and circumstances that allow individuals to overcome the increased risk imposed on them by adversity? For exploring questions like these, childhood adversity (CA) is a viable starting point because it is linked to poor health and other negative outcomes far into adulthood.5–8 Resilience can then be understood as the context-dependent, fluctuating process of avoiding these negative outcomes and is shaped by the resources an individual can access, as well as by additional risks.2 Empirically, these resource and risk factors can be operationalised as moderators and mediators of CA. A ‘positive’ moderator is a resource, such as social support, that is even more beneficial in the context of CA than it is in more advantageous circumstances. Understanding these kinds of moderators can guide policy-makers in creating environments that foster resilience. Conversely, a ‘negative’ moderator would exacerbate the impact of CA. A mediator is a factor that lies on the pathway between CA and a negative consequence, for example, lower educational achievement as ‘step’ between CA and adult unemployment. If removing such mediators, one would absorb part of the negative impact of CA. This scoping review, in line with others,9 uses the term ‘resilience factors’ to refer to these mediators and moderators of adversity. Both types of resilience factors are of interest when it comes to painting a full picture of resilience processes and how to foster them. After decades of research on resilience, it is time to draw an interim balance of the current evidence base, identifying weaknesses and untapped potential for further knowledge gains.

There are already several reviews investigating forms of CA, such as childhood neglect,10 in relation to different resilience factors and mental health outcomes in adulthood.10 11 However, there are to the best of our knowledge no reviews combining different types of CA, different types of adulthood outcomes—including, but not limited to mental health—and different types of resilience factors. Limiting reviews to individual types of early adverse exposures and/or individual types of adult health outcomes makes the amount of literature to synthesise more manageable and findings more easily interpretable, but resilience as an overarching concept merits an attempt to go beyond narrow definitions and opening up to an interdisciplinary engagement with it. We, therefore, chose to conduct a scoping review that is open to different resilience definitions and resilience factors, with our only restriction being that resilience was considered the outcome of interest and was not operationalised as an individual trait. The latter would contradict our preferred understanding of resilience as a process shaped by circumstance but achievable for all.

Against this backdrop, we also scope out what is known about gender patterns in resilience processes. Resilience has rarely been approached from a gender perspective, despite the fact that gender, the socially constructed norms and roles linked to being a man or a woman,1 12 is relevant for a number of reasons. First, a gender-sensitive approach can enrich conceptual models of resilience.1 12 13 For example, Ungar and Theron1 view resilience as a multisystemic process and propose that gender should be considered as a possible overarching source of difference in regard to resilience factors. Second, mental ill health and other outcomes considered in the context of CA are often distributed differently between men and women, even in the absence of CA.14 Third, there are empirical studies that point to gender differences in resilience factors, for example, due to gendered norms around acceptable coping strategies.15 All in all, there are many arguments for a more gender-sensitive approach to resilience research,1 12 13 incorporating gender in the formulation of research questions, in the selection of exposures, outcomes and resilience factors, and when interpreting findings.

This scoping review maps resilience as a multicomponent process (a) in terms of resilience factors (b) through which exposure to CA (c) relates to an outcome in adulthood (d) (figure 1). This provides the background against which we also explore gender patterns across studies. Since we wanted to include studies on resilience that are heterogeneous in terms of study designs, effect measures and exposure as well as outcome definitions, we preferred a scoping over a systematic review. Scoping reviews are best suited for an exploration of the research surrounding a multifaceted concept such as resilience, identifying research gaps and possible directions for the future. A systematic review would have required a more stringent definition of exposure, outcome and resilience factors from the outset, limiting our ability to refine search and evidence synthesis in an iterative process.16 Nevertheless, we conducted the literature search and data extraction in a systematic and stringent manner.

Figure 1

Conceptual map of resilience processes in the context of this scoping review.

Objectives

The present study offers an up-to-date review of evidence on resilience factors in the context of CA and adulthood outcomes, with a particular interest in gender patterns. It considers a variety of CA measures and outcomes in adulthood, including outcomes related to mental health, physical health and socioeconomic indicators such as educational attainment. We covered three research questions:

What is known about resilience factors (mediators and moderators) in the association between CA and adulthood outcomes?

What is known about gender differences in the association between CA and adulthood outcomes?

What is known about gender patterns in resilience factors in the context of CA and adulthood outcomes?

Resulting from these questions, we suggest areas of improvement for future research.

Methods

We followed the steps proposed by Arksey et al17 and expanded on by Levac et al,18 with the exception of the optional step of stakeholder consultation. We report the review process and results in line with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR).16 There was no patient or public involvement in this review.

Protocol and registration

The protocol for this scoping review is available on osf.io (DOI 10.17605/OSF.IO/KYHTA).

Eligibility criteria

We included articles where the study population was comprised of adults aged 18 or older who were sampled randomly or as convenience samples from the general population or specific settings such as primary healthcare. We required the population to entail both exposed and unexposed individuals, as well as individuals with and without the outcome in question. We excluded populations where all individuals had received a diagnosis and the study focused on, for example, differences in symptom severity. For the exposure, CA definitions were restricted to studies capturing at least one of the experiences covered in the seminal study by Felitti et al19 on adverse childhood experiences. Adulthood outcomes in included studies could comprise different measures of positive adaptation or resilience, framed either in terms of the absence of a negative outcome (eg, depressive symptoms) or the presence of a positive outcome (eg, well-being). During the screening process, eligibility criteria were increasingly refined and expanded. Most importantly, criteria in relation to the appropriateness of the statistical analyses were added, mainly the existence of an appropriate comparison group and the use of moderation analysis for identifying moderators of CA, and mediation analysis for identifying mediators of CA. We allowed for cross-sectional, case–control and cohort studies but excluded qualitative studies, intervention studies, case studies and case series, and animal studies. Furthermore, studies on biological or genetic resilience factors were excluded, in order to keep the amount of information synthesised in the results manageable. We considered only peer-reviewed articles. An overview of inclusion and exclusion criteria is available in online supplemental material S1.

Information sources and search

After several test searches, a final search was conducted on PubMed, PsycINFO and Scopus on 25 June 2021, including all peer-reviewed studies with no limitation regarding publishing date. A renewed search for additional publications with the same search strings was conducted on 20 July 2023. All search strings can be found in online supplemental material S2. Results were imported to an online screening tool provided by Rayyan, and duplicates were detected automatically. The deletion of duplicates was done manually. Since few studies place an explicit focus on gender patterns in resilience processes, we did not include terms related to gender (such as gender, sex, men, women, males and females) in the search string, to avoid missing out on studies that performed relevant secondary analyses involving sex or gender. No manual search strategies such as snowball search were applied, due to the already large amount of material collected.

Selection of sources of evidence

Titles and abstracts were screened independently by two authors. Articles where both agreed on exclusion were excluded. Articles, where at least one of the authors was unsure about exclusion, were retained for full-text screening, as were articles where both approved inclusion. Each article set aside for full-text screening was screened by at least two authors independently. Conflicting decisions were discussed between all three authors until a decision could be reached. Where needed, adjustments to the inclusion/exclusion criteria were made.

Data charting

Due to the high degree of heterogeneity of included studies, a quantitative pooling of effect measures or measures of association was not appropriate. To retrieve the information of interest for the proposed research questions, as well as characteristics relevant for the quality assessment and description of the studies, an Excel sheet was created and iteratively adapted during the early stages of the screening process. Data charting was done by one author, and the results relevant to answering the three research questions extracted from each study were cross-checked by one of the other authors.

Data items

Data retrieved from the articles are author(s) and year of publication; country; sample characteristics; sample size; study design; exposure (definition, assessment and operationalisation), outcome (definition, assessment and operationalisation); main statistical analysis; proposed mediator(s)/moderator(s) of the exposure–outcome relationship; evidence for mediation/moderation; and any analyses relating to gender.

Critical appraisal of individual sources of evidence

We conducted a quality assessment of the included studies, in order to help the reader gauge generalisability and risk of bias of the evidence synthesised in this review. This information also helps garner an understanding of the methodological shortcomings in this research field. The criteria were based on a quality assessment tool used by Norman et al20 in their systematic review of observational studies on the long-term impact of CA. This tool considers among others the representativeness of the population, assessment of exposure and outcome, selection of unexposed comparison groups, follow-up and response rate, and the appropriateness of statistical analyses. Criteria were modified for the purpose of this scoping review, after discussions among all three authors, and can be found in online supplemental material S3a and b.

Synthesis of results

The results were descriptively and visually summarised in several results tables and figures. Results in relation to the main study characteristics as well as each of the research questions were additionally summarised narratively. To facilitate summarising the results despite the heterogeneous collection of resilience factors, the three authors categorised resilience factors across overarching levels, guided by the Dahlgren and Whitehead model of health determinants.21 This is in line with similar reviews, even though the exact terminology may vary11, and consistent with a public health take on resilience. Specifically, we decided on a higher-order categorisation representing the individual level, relationships with other persons, and structural-level factors, and then agreed on further subcategories, such as coping-related or attachment-related factors.

Results

The initial search resulted in 3300 hits, of which 884 were duplicates. After the title and abstract screening, 354 articles remained for full-text screening. The second search resulted in 51 additional articles for full-text screening. A PRISMA flow chart is accessible in online supplemental material S4. This review included a total of 102 studies investigating CA, resilience factors and adulthood outcomes. Most studies (74%, n=75) considered at least one mental health outcome, and roughly one-third (32%, n=24) of these explicitly focused on or included outcomes related to anxiety, depression or both. The majority of studies (71%, n=72) operationalised resilient outcomes as the absence of a negative indicator, or lower values on a score rating a negative outcome, rather than considering positive indicators alone or in combination with negative indicators. 27% of studies had samples comprising only women (n=26) or only men (n=2) and were therefore not able to compare resilience processes between the two. Among the studies with all female samples, more than one-third (n=9) focused exclusively on sexual abuse as the exposure, compared with two studies with this exposure definition among those with mixed samples. Of the 74 studies with mixed samples, 31% (n=23) investigated gender patterns in relation to the differential impact of CA by gender (gender×CA) and/or gender patterns in terms of mediation or moderation of CA. There were no studies that met our inclusion criteria that included transgender or gender-non-conforming adults, which is why this review was only able to discuss gender patterns as captured by differences between men and women. Most studies were conducted in North America (66%, n=67). Additional study characteristics for all included studies are visualised in figure 2 and presented in more detail in online supplemental material S5.

Figure 2

Study characteristics.

Studies were of varied quality, but mostly poor to moderate, using mainly retrospectively assessed exposure data (91%, n=93) and/or convenience samples (55%, n=56). A strength of many studies was the consistent application of validated scales and clinical assessments of different psychological symptoms or diagnoses (84%, n=86). The quality scores are summarised in figure 3 and presented in more detail in supplementary material (S6).

Figure 3

Quality assessment—overview.

In the following, we present a narrative summary of the results by research question.

RQ 1: what is known about mediators and moderators of the association between CA and adulthood outcomes?

The putative resilience factors covered across studies were highly heterogeneous. We describe and discuss them according to three overarching levels (individual, relational and structural), and subcategories within each of these, such as attachment and coping. An overview of all resilience factors by category is available in online supplemental material S7, and references for all included studies are also listed in online supplemental material S8. In the following, we will provide a narrative summary of some of the most commonly investigated mediators and moderators of CA. A visual overview is presented in figure 4.

Figure 4

Overview of resilience factor counts by category and significance level.

Individual level

At the individual level, coping-related factors were most often investigated (n=41), predominantly as mediators. Emotion-oriented coping emerged as the most robust mediator of CA, with five out of five studies finding evidence for a significant indirect effect in terms of both physical and psychiatric health outcomes.22–26 Similarly, 60% of the studies including factors related to emotion regulation returned significant indirect effects.27–34 Self-esteem was the most robust individual-level buffer of CA, with two out of three studies finding self-esteem to be a moderator in relation to lower psychological distress in a women-only sample35 and decreased risk of internalising disorders in a mixed sample.36 No evidence for moderation was found using an index of different psychopathological outcomes and self-esteem in adolescence.37 Meaning-making factors, including, for example, spirituality, were not part of any mediational models, but were exclusively investigated as moderators. Evidence was mixed, with half of the studies returning non-significant and half returning significant findings.

Relational level

At the relational level, social support measures and different qualities of social relationships were most frequently investigated (n=38 on availability of social support, n=20 on quality of social relationships), mainly as moderators (n=26 and n=18, respectively). Findings were mixed: 46% of social support indicators (n=12) and 33% of indicators capturing the quality of social relationships (n=6) resulted in significant interactions with CA. Among those who did find evidence for interaction, several studies identified support by friends in both childhood38 and adulthood to be a buffer.39 40 Family social support did not act in the same way and was no significant buffer in any of the studies investigating it as a separate social support category.38–40 However, all studies that investigated the buffering effect of a supporting adult during childhood generally identified significant interactions in relation to both physical and mental health outcomes.41–43 Furthermore, while studies found evidence for the moderating effect of emotional support in terms of both aspects of physical and mental health,35 44 this was not the case for instrumental support.35 45

Structural factors

Seven studies covered factors at this level, making it the smallest category by far. With one exception,46 all putative resilience factors in this group were investigated as moderators. In adulthood, all factors captured different neighbourhood characteristics or sense of community, and half of all investigated factors emerged as significant moderators of CA. Especially sense of community seems to be an important buffer of CA in relation to mental health outcomes.47 48 In childhood, two studies covered different aspects of the school environment, and all four factors investigated in this context emerged as significant buffers of CA in terms of mental health outcomes in adulthood.38 49

RQ 2: what is known about gender differences in the association between CA and adulthood outcomes?

The CA×gender interaction was tested in 15 studies and was statistically significant in five of these in relation to at least one of the included outcomes (see figure 5). Ritchie et al50 found that losing a parent during childhood negatively impacted one measure of cognitive functioning in women in later adulthood, but not in men. Similarly, Mian et al8 found women to be more impacted by CA in terms of frailty in late adulthood than men. A study by Neff51 found men to be more resilient to anxiety, but not depression, after a parental history of mental ill health. Similarly, Powers et al40 found that depression symptoms in women with a history of childhood maltreatment were higher than in men, while in the unexposed group, there was no such difference. Horan and Widom52 found an interaction between gender and childhood cumulative risk in relation to educational attainment, but not to mental health symptoms.

Figure 5

Selected results from studies on CA×gender interaction (RQ 2) and gender patterns in moderators and mediators of CA (RQ 3). CA, childhood adversity.

RQ 3: what is known about gender patterns in the mediation and moderation of the association between CA and adulthood outcomes?

In terms of gender patterns in moderators of CA, there were two studies that conducted three-way interaction (CA×gender×moderator) and five studies performing two-way interactions (CA×moderator) stratified by gender. Of the former, only one found statistically significant three-way interactions. Reinert et al53 investigated positive and negative religious coping as moderators. They showed, among men, that the combination of CA and positive religious coping was associated with worse physical health, whereas the combination of CA and negative religious coping had a beneficial influence on physical health. The reverse patterns were demonstrated among women. Four studies that conducted two-way interactions stratified by gender found evidence for potential differences. One found spirituality to be a buffer only in women, in regard to different mental health outcomes.54 Similarly, social support by friends40 and having a child55 were buffering the effect of CA in regard to depression in women, but not in men. Another study reported that abstaining from drinking may be protective against depression, anxiety and somatisation in men with a parental history of alcoholism, but not in women.51

Evidence on gender patterns in mediators of CA was also mixed. Two studies employed multiple-group path analyses, and four conducted gender-stratified mediation analysis. Among the former, one found a significant gender difference, namely that sense of belonging was a stronger mediator in men than in women in relation to mental health.56 All of the stratified analyses resulted in some evidence for potential differences between men and women. Giovanelli et al57 found family support and motivational advantage to be mediators in regard to high school graduation only in boys, while school support emerged as mediator both overall and in girls, but not in boys. Lee et al58 found sleep quality to be a mediator between CA and metabolic syndrome in both men and women, but stress-induced eating was significant only in women. Song et al15 also found more significant mediators for women than men with regard to depression symptoms, with problem avoidance and rationalisation only mediating in women, and problem-solving, self-blame and help-seeking mediating in both men and women. Wang et al59 found loneliness to fully mediate the association of CA and depression in women, but only partially in men. Moreover, coping skills were identified as a mediator in men but not in women. Figure 5 summarises a selection of the findings for RQ3.

Discussion

CA is a major public health issue, with consequences in terms of ill health and socioeconomic difficulty stretching over the entire life course, well into adulthood.5 6 8 Understanding resilience, the process of positive adaptation after adversity, implies mapping factors that influence the association between the adversity and the outcome of interest and is one of the steps we can take to work towards mitigating CA’s impact on individuals and their families. This review contributed by investigating the role of resilience factors in the relationship between CA and a range of adulthood outcomes, paying particular attention to gender patterns.

Despite included studies being heterogeneous in terms of exposures, outcomes and resilience factors, there were some striking similarities when considering more overarching features. Notably, the majority of studies focused on mental health outcomes in combination with individual-level resilience factors, and most studies were set in Western contexts. This is no surprise, given that resilience research in industrialised contexts has a strong tradition in the field of psychology,60 and given that most research continues to be West centric.61 However, these features point to underexploited potential in resilience research, as well as to its limited generalisability for transferring findings across contexts. Furthermore, most studies included in this review did not explore gender patterns in any of the components of resilience processes, which is in line with what another comparable review found.62 A more detailed comparison with previous reviews is difficult since most reviews we are aware of merely comment on the degree to which men and women are represented in the respectively included studies9 or do not mention gender at all in their synthesis, beyond gender in itself being identified as a moderator.10 63 Taken together, it seems that a gender-sensitive approach to resilience continues to be relatively neglected—despite gender being relevant to these processes. For example, there are gender patterns in stressor exposure,53 57 in the distribution of outcomes, such as health problems,38 64 as well as in the resources with the potential to mitigate the impact of CA, such as attachment patterns.65 In the following, we will briefly summarise and discuss the insights generated by this scoping review, in terms of, first, the landscape of empirical studies on resilience factors, and, second, the available evidence on gender patterns in resilience processes. We will provide suggestions for future research throughout.

Resilience factors in relation to CA and adult outcomes: time to expand the focus?

The majority of studies explored individual-level factors, which correspond to the strong representation of psychological literature in this review. This is also in line with what similar reviews found. Specifically, factors related to coping and attachment were among those most frequently investigated, and generating the most robust evidence.9 63 66 Also in our study, coping mechanisms were among the most frequently discussed individual-level resilience factors, mainly in mediational models. Exposure to adversity may shape coping styles in ways that are adaptive at first, but maladaptive in more positive contexts.67 For example, avoiding confrontation in situations where conflict is not safe is adaptive. However, in situations where it is safe, avoidant coping or problem avoidance is no longer the lesser evil but can increase the risk of diagnoses such as depression.67 Accordingly, providing an individual with the resources to adapt their coping behaviour to their circumstances later in life may be a valid way of eliminating part of the indirect effects of CA, and many studies in this review suggest this. It is also noteworthy that some putative resilience resources can be a double-edged sword. Factors related to religiosity or spirituality may, for example, both be harmful or beneficial.53 68 69 These are examples proving that context matters for both how we view adversity and its consequences, and for how useful a given resilience factor is. Generally speaking, these examples may be suggestive of individual-level interventions being the key to enabling resilience in adults with CA experience. Even if we believe this to be true, we can make resilience a public health rather than a purely psychological concern by more frequently considering individual-level factors that society can provide more easily than awareness around coping behaviours. For example, it could be explored whether purpose in life and self-esteem can be bolstered by ensuring access to professional training or education, or community engagement, across the life course, to allow individuals to (re)discover meaning-making activities.

Prior literature, with its focus on individual-level factors, mirrors narratives of individual resistance against and recovery after adversity. As such, it has produced—and will continue to produce—valuable and applicable insights. However, this also suggests that resilience research, despite growing acknowledgement of resilience being a multisystemic process,1 70 71 is still largely focused on these individual-level factors. Accordingly, the corresponding intervention options are often framed on the individual level. This limits our understanding of how to create and sustain wider environments that foster resilience, and it risks placing a disproportionate amount of responsibility on the individual.60 72–74 For example, social support emerged as an important resilience factor in our own and other reviews.10 62 75 We then need to ask how we can ensure that the sources of such support—social networks, families and societies—can successfully function as nurturers of individual resilience. A multisystemic approach to resilience implies imagining interlocking systems that each need to provide enough of the right resources for the other systems they are feeding into.74 For example, what do local communities look like that empower families to navigate their way to resilience? One part of this may well be psychological support, the availability of which is a public health matter in itself. However, we can go much further than this: First, by thinking about resilience in a more participatory way, and second, by transparently discussing more of its upstream determinants. There is now increasing awareness around the need to actively engage disadvantaged populations in resilience research, for example, in work to conceptualise and operationalise resilience.71 74 Furthermore, there have been calls to combine efforts to ‘change the odds’, which include more traditional public health and social policy measures to address social determinants of health, with efforts to ‘beat the odds’ by enabling resilience processes.74 While this is challenging, Ungar et al71 have proposed a sample methodology for conducting resilience research that is not only contextually sensitive and participatory but also addresses the different systems that contribute to how resilience plays out. Furthermore, cross-country comparisons of education or welfare systems could yield evidence on how overarching societal structures can provide support in the face of individual-level adversity and help frame how resilience on an individual level conversely acts as cornerstone for societal and economic resilience at large. Especially when considering the relatively robust evidence for school-level factors38 49 and access to trusted adults such as mentors41 43 being buffers to CA in childhood, it seems clear that a strong case could be made for investing in resilience-promoting schools or similar places of support outside the family context. Additionally, more cross-country or cross-cultural comparison studies, including non-Western settings, could help expand our knowledge of resilience factors across contexts, and open our eyes to different ways of ‘producing’ resilience—outside a Western narrative of self-sufficiency.73

Gender in resilience processes: a blind spot in the empirical literature?

Gender influences all components of resilience processes. Nevertheless, gender is nothing but a confounder in most studies included in this review, limiting our understanding of possible gender patterns in resilience processes. Only a fragment of the included studies investigated whether CA may impact men and women differently. The CA×gender interaction was then mostly tested in studies with mental health outcomes, possibly due to the overall higher rates of mental health problems like anxiety and depression in women76, which may be construed as hint towards the differential impact of CA or other stressors by gender, even though there is no strong evidence supporting this hypothesis.77 Also, in our review, the majority of studies did not find evidence for such an interaction. This may mean that there is simply no such differential impact, or that CA levels the playing field, cancelling out possible advantages men may otherwise have over women in relation to these particular outcomes.

In contrast to studies investigating a possible difference in the impact of CA by gender, more than half of the studies interested in gender differences in mediators and moderators of CA found statistically significant results. Both in terms of moderators as well as in terms of mediators, more resilience factors could be identified for women than for men. There is some evidence that factors related to social support are stronger buffers in women than in men, and conversely, that loneliness is a more potent mediator. One reason for this could be that women tend to have stronger social networks than men.78 79 Similarly, there was evidence for emotional support being more meaningful than instrumental support. One of the enduring norms men are confronted with is the need to appear strong by not disclosing feelings that could be perceived as weakness. This could be another reason why social support does not function as an equally potent buffer in men as in women: men may not only have less access to social support, but they may also be less inclined to use it. Alternatively, it is possible that, due to gender roles and expectations, different resources are prioritised by men, for example, those that are related to economic rather than social status.55 Fitzgerald and Berthiaume54 suggest that in men, sociodemographic and personality factors explain more of the variance in mental health outcomes than is the case for women, which would then explain why interpersonal resources do not seem to be as helpful to men—less of their sense of self-worth is determined by this type of factors. In that sense, one could argue that men face a double vulnerability when exposed to CA because the resources that would be most beneficial to them may be harder to access than the resources women apply. Another study proposed to differentiate between psychological and physiological outcomes, with women being physically more resilient and men psychologically58. This may in turn translate into different resource needs. In terms of mediators, several studies reported women to be more likely to resort to maladaptive coping styles22 25 as well as to exhibit predominantly anxious attachment styles.80 These studies did not themselves test for gender differences in mediation, but Song et al15 found more maladaptive coping behaviours to mediate the effect of CA in women than in men. A gendered use of coping styles could be partly the result of corresponding narratives around self-worth and self-reliance, as well as of actual lack of resources—women are often in more vulnerable economic positions, due to, among other things, continued labour market segregation, which may provide them with less options for problem solving than men. Here, we can see a double vulnerability of women exposed to CA.

Gender, by itself and in intersection with health determinants such as socioeconomic status, shapes individuals’ options for navigating their way to resilience.1 60 71 Resilience processes (or the lack thereof) may be regarded as a way of ‘doing gender’,81 with strategies such as coping, help-seeking, aggression and substance use confirming or challenging normative expectations of masculinity or femininity. While hard to implement in practice, more research should strive to do justice to this complexity. To begin filling this gap, researchers can routinely start by discussing their resilience outcome, as well as any observed gender patterns in it, in relation to possible gender bias. We also suggest to incorporate more outcomes that are, on average, more common in men, such as aggression-related problems or alcohol misuse. Where this is not possible or not of interest, we suggest to dedicate some space in article discussions to acknowledge that equating the absence of, for example, anxiety with resilience may hide men’s suffering and make women appear disproportionately fragile. Some studies do all or part of these points already. For example, Feldman et al38 discuss that women had a higher risk of affective disorders in their sample, but men had a higher risk of any psychiatric disorder as well as of antisocial behaviour and developing dependence. Similarly, a conscious engagement with gender patterns in exposures could make resilience research more gender-sensitive—articles focusing on individual adversity indicators could discuss how far girls or boys may be more often exposed to a specific type of adversity, and why. Lastly, we would suggest that all empirical studies conduct their analyses both in the overall sample, as well as stratified by gender.

Similar to what we discussed in relation to other structural level factors, also in regard to gender patterns in resilience processes, cross-country comparisons could be a helpful exercise. They may shed light on how more egalitarian societies will provide better resilience opportunities for both men and women, helping each reduce their respective ‘double’ vulnerabilities, and at the same time preventing some of the adversity from occurring in the first place. With a smaller power gradient in gender, there might be less violence against women and girls, and with less traditional ideas around masculinity and strength, there might be less exposure to violence for men. Another aspect future studies on gender patterns in resilience may want to incorporate is the idea of key turning points during the life course, which may vary by gender.82 83 For women, this may, for example, be around childbirth. More studies on resilience factors around and after the time of birth could garner insights on how women with previous experience of adversity can be supported. For men, the transition to retirement or involuntary job loss could be experiences that have a particularly isolating impact. Incorporating notions of the gendered life course into resilience research could, therefore, help a more nuanced and longitudinal understanding of adulthood outcomes after CA. Lastly, we would like to mention the untapped potential of intersectionality in resilience research. Gender is a dimension of an individual’s identity, but it is only one of many that shape resilience processes.60 Ideally, future empirical research will not only investigate gender patterns in resilience but also resilience at the intersection of different dimensions of disadvantage.

Limitations

We collated evidence on gender patterns in resilience processes in the context of CA and adulthood outcomes, against the backdrop of an up-to-date comprehensive review of the literature on resilience factors. In contrast to many other reviews on resilience factors, we followed stringent criteria in regard to the statistical methods applied in order to identify said resilience factors, similar to Fritz et al in their systematic review from 2018.9 This is an important strategy to ensure that included factors are investigated in light of their relevance to adversity rather than their general protective potential. We also adhered to the PRISMA-ScR guidelines as closely as possible, making the choices in this review transparent, including their disadvantages. One of the strengths of this review, but simultaneously one of its weaknesses, is the wide range of exposures and outcomes included. While this decision was made in order to be open to different research fields, the resulting heterogeneity of studies made even a narrative summary challenging. At the same time, the introduction of a number of exclusion criteria implies that this scoping review is a snapshot of a certain way of engaging with resilience in research, rather than a comprehensive picture of what is possible. For example, gender should ideally be understood on a spectrum, and as one dimension of many that may shape an individual’s opportunity for resilience. Furthermore, there are strong arguments against understanding gender and sex as clearly delimited concepts, but rather as two aspects of a construct that bidirectionally influence each other.84 This means that studies focusing on biological differences between men and women are relevant to discussing gender patterns but were so different that we decided to exclude them to make it easier to integrate our findings. By choosing to include only quantitative studies, we also inadvertently excluded literature on transgender and gender-non-conforming adults. This resulted in our review discussing gender in terms of the overly simplistic, but still most commonly used binary understanding in empirical studies. We strongly encourage future empirical studies to explore resilience with a more nuanced, ideally intersectional approach to gender. While this is often a difficult undertaking, among other things due to issues with statistical power, there are promising examples of how this may be possible in practice, given enough resources.71 74 Furthermore, we would like to highlight that reviews of qualitative studies on resilience in populations chronically under-represented in quantitative research would be a valuable addition to the literature. More generally, a qualitative engagement with what individuals perceive resilience to be would be an additional angle through which to explore relevant resilience resources and a way of framing the individual as agent rather than as victim. In this context, we would also like to acknowledge that quantitative studies on resilience, and by extension this review, tend to make an implicit value judgement: it is assumed that someone with, for example, a physical or psychological diagnosis is not exhibiting resilience. One could easily argue that the adaptation to said diagnosis is, more than anything, an indicator of resilience in itself. We did not allow for this perspective by excluding, for example, clinical populations. Furthermore, our eligibility criteria come with an inherent gender bias—in the early stages, we decided to exclude studies with outcomes related to criminal behaviour, and many all-male samples were excluded due to their focus on re-exposure to severe stress later in life, which was not the focus of this review. This is a shortcoming that may make the literature appear narrower than it is. A similar limitation is posed by our decision to exclude grey literature and only include peer-reviewed articles.