Search strategy and study selection

A comprehensive and systematic search strategy was constructed by adapting strategies in previous studies coupled with input from the project team. This was trialled in a sample database, edited based on comments from the team, and implemented by the information specialist [GS]. The search string explored the three main concepts of ‘LGBT+’ [e.g. Lesbian; Gay; Bisexual; Transgender], ‘older’ and ‘resilience’ using synonyms, controlled vocabulary and Boolean operators as appropriate (see S1 Appendix. Example of search terms used in Medline (Ebsco)). CINAHL (Ebsco), Embase (Elsevier), Medline (Ebsco), PsycInfo (Ebsco), Social Science Database (ProQuest), and Web of Science (Clarivate) were all searched from inception until 17 June 2022. The protocol for conducting this review was registered with PROSPERO (registration number: CRD42021249093).

Eligibility criteria

Given the aims of the review, the focus was on peer-reviewed papers in English only. Studies which reported primary research with older LGBT+ individuals who were aged 50 and above were included. Depending on the type of review people at the outset may define their concept or phenomenon of interest very tightly and use that definition as an inclusion criterion. However, as our review was focused on how researchers conceptualised and defined resilience within their research, we did not have a strict definition of resilience, although our reading and knowledge of the concept did inform our thinking on the search terms used. Studies included could be qualitative, quantitative or mixed methods as long as there was a clear mention of resilience within the paper. Resilience could be a primary or secondary focus of the research study or have emerged from participant’s narratives in qualitative studies. Systematic reviews, case studies, randomised controlled trials, intervention studies and descriptive/discussion papers were excluded (see S1 Table. Inclusion/Exclusion criteria).

Screening and selection

Following the database searches, all citations were uploaded into Covidence screening software (https://www.covidence.org) and any duplicates removed. Each title and abstract were independently assessed against the inclusion and exclusion criteria for eligibility by two members of the review team. Where there was difference in assessment, a third member of the team reviewed the title and abstract and made the final decision to include or exclude. Out of the 7101 papers reviewed at title and abstract level, 254 were put forward for full text review. These papers where once again assessed against the inclusion and exclusion criteria for eligibility by two members of the review team. Out of the 254 citations in total 27 papers were eligible for inclusion. Fig 1 represents the PRISMA diagram.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. PRISMA flow diagram of selection process.

https://doi.org/10.1371/journal.pone.0277384.g001

Quality assessment

Quality appraisal in a mixed study systematic review has been described as complex [39] and different approaches have been used [40]. We used the Mixed Method Appraisal Tool or MMAT [41], specifically designed for quality appraisal in mixed study reviews. The MMAT includes two general screening questions and a further five methodological quality criteria for common types of methodologies which permits the critical appraisal of mixed study reviews within a single tool and specific quality criteria for mixed methods studies [42]. An affirmative response to both the initial screening questions (are there clear research questions and do the collected data allow to address the research questions) indicate that the paper meets initial criteria for screening. As per the MMAT user instructions, two of the authors independently assessed the quality of the papers and met to discuss and reach consensus about the quality of the individual papers. Excluding papers that are of low methodological quality is discouraged by the authors of the MMAT.

All papers (n = 27) met the baseline for empirical studies following initial screening. Of the 27 papers included there were 13 qualitative and 14 quantitative papers. Collectively the quality of the papers ranged from average to good using the MMAT. All quantitative papers were appraised as cross-sectional descriptive studies based on their methodology. For the quantitative studies, only one study met all the quality indicators [43] with the others failing to meet at least one of the MMAT criteria. For the most part, studies were not representative of the target population or used non-probability convenient sampling. In addition, it was not clear in eight of the quantitative papers if the risk of non-response bias had been addressed [44–51]. Only four qualitative studies met the full criteria following initial screening [52–55]. All papers were subsequently included in the extraction and analysis phase as advocated by Hong et al. [42].

Data extraction and analysis

Two authors designed and piloted the data extraction template, which included the aim of the research, sample characteristics, and details on the study methodology. Any data on resilience, particularly how resilience was defined and conceptualised, if resilience was the primary or secondary focus (objective) of the study, and whether it was an entry or an outcome of the study was extracted. For the purpose of this review, studies that had resilience as a stated or inferred objective were classified as entry point studies and those where resilience was a topic inductively emerging from the data were classified as outcome studies. Details on the tools to measure resilience and conceptual frameworks used, were also extracted (see Table 1). In the qualitative papers, data findings relating to the review aims were extracted verbatim, whereas in the quantitative papers, data was transformed (‘qualitised’) in line with the convergent synthesis design that was adopted [37]. Data was extracted manually from each paper by two authors who then met to discuss the extraction and finalise the included data. Once the templates were populated with data, meetings were held with all authors who tabulated and summarised the data given the aims of this review, using a descriptive approach. This process included establishing an audit trail which articulated the process from paper selection through to data extraction, tabulation and data visualization which was checked by an author who acted as moderator who critiqued the process. In addition, regular meetings with all team members ensured that the team were clear on issues such as inclusion criteria, data selection, analysis and presentation of findings.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Overview of included papers.

https://doi.org/10.1371/journal.pone.0277384.t001

Overview of studies

The 27 papers included represented 22 studies as four papers came from the ‘Caring and Aging with Pride’ study [47, 48, 68, 69] and three papers further analysed data from the ‘Aging with Pride’ study sample [10, 43, 75], which collected data in 2010 and 2014. The majority of the papers came from the USA (n = 17), three from Canada, two from Portugal, one from the Republic of Ireland and one from Australia. Three papers involved more than one country: USA/Canada (n = 1), USA/UK (n = 1) and USA/UK/Canada (n = 1). The papers were almost equally divided between quantitative (n = 14) and qualitative (n = 13) study designs. Quantitative studies focused on measuring variables that the researchers of the included papers considered represented resilience and the designs generally used descriptive, comparative and cross-sectional designs (n = 6), descriptive and cross-sectional designs (n = 6) and cross-sectional designs (n = 2). The emphasis of qualitative studies was on exploring participants’ perspectives or narratives on a variety of issues using qualitative descriptive (n = 4), grounded theory (n = 3), narrative methodology (n = 2), phenomenological (n = 1) approaches and secondary analysis (n = 3) of previously collected data. Data were collected mostly using one approach, namely individual interviews (n = 10), although in Bower et al’s [65] study participants were also asked to share artefacts they felt enriched their narrative interview. Other methods used were focus groups (n = 1), online survey with open-ended questions (n = 1) and one mixed methods study using interviews, focus groups, and a town hall meeting (n = 1).

Data analysis in four studies were informed by Charmaz’s [76] grounded theory approach [52, 53, 57, 66], guidelines on phenomenological research [56], thematic analysis [58, 59, 61], process coding [60], comparative analysis [54] and Merriam and Tisdfell’s [77] method for data analysis using open and analytical coding [55]. Some studies used theoretical frameworks to support the analysis, such as ambiguous loss theory and the model of minority stress [52], generativity theory with a critical feminist perspective and queer theory [65], the Model of Global and Situational Meaning [57], the Anderson’s Health Care Utilization Model [61] and the Resilience properties in Aging Framework [62]. Quantitative studies applied methods of analysis correspondent to the research designs, such as t-tests, chi-square test, correlations tests, linear and multivariate logistic regression among others.

Most studies reported the participant’s demographics, including ethnicity, age, gender identity, and sexual orientation, while education and income were less commonly reported. In terms of demographics, a variety of terms were used such as gender identity, gender, sex, sexual orientation and sexuality. When reporting gender identities besides male ((cis-)man) or female ((cis-)woman) some papers used terms such as transgender [46, 59, 60, 65, 68, 69], trans [66], transwoman and transman [45], while others additionally reported a variety of identities such as female with trans histories, pangendered, queer femme and trans, and male to female and back [54, 58]. To avoid misrepresenting or misinterpreting the original authors, the terminology used in the data extraction table is what is reported within the papers.

In those studies that reported ethnicity, the majority of participants were white and/or white non-Hispanic ethnicities, and the mean age varied from 52.5 to 67 years. While six of the quantitative studies mention gender identities besides men and women, five studies [43, 44, 51, 71, 75] focused exclusively on gay and bisexual men but none of the studies specifically state if all identified as cisgender. Six of the quantitative studies reported including people who identified as transgender or were reported as ‘other’ than female or male [10, 45, 46, 48, 68, 69]. These constituted between 0.6% [48] to 39.8% [45] of the sample. While Fredriksen-Goldsen et al in one paper [48] reported that 0.61% of their sample identified as other than female or male in earlier papers on the same study sample [68, 69] they report on a sample of 4.1% and 7% of transgender people. One paper by Fredriksen-Goldsen et al. [68] focused specifically on transgender people, while two studies reported included LGBT couples and spouses [52, 62], but only one [52] reported gender identity.

The predominant sexual orientation reported was gay (men), although one qualitative study focused only on lesbian women [56], and some studies focused on people who identified as bisexual [48, 54] or bisexual individuals constituted most of the sexual minority sample [50]. Only two studies reported queer identities in their sample [54, 60].

Definition of resilience

Eleven of the papers offered definitions of resilience as either a trait, a process or an outcome [43, 46, 47, 52, 53, 55, 57, 59, 62, 69, 71]. However, there is no uniformity in the definitions offered. In the remainder of the papers, resilience was not clearly defined although it was alluded to and perceived as a strength or a positive attribute generally [10, 44, 45, 48–51, 54, 56, 58, 60, 61, 65, 66, 68, 75]. Where resilience was defined as a process, it was mainly seen as a process of adaptation where older LGBT+ individuals responded to adversities over time. Adversities were perceived as the context within which LGBT+ resilience was operationalised and in some of the papers being a member of the LGBT+ community was associated with adversity in itself. Handlovsky et al. [53], writing about gay men’s exposure to systemic discrimination, describe resilience as a response to risk exposure where environmental harms are mediated through internal and external protective processes. This allows older LGBT+ individuals to cope with adversity using both internal and external processes. King and Richardson [71] also refer to resilience as a coping mechanism, but refer to its health-promoting capacity as well its capacity to buffer physical and psychological distress. Nelson-Becker and Thomas [62] viewed resilience as ‘the capacity to manage significant difficulty and stress and is both a process and an outcome’ (p. 2), while also describing it as a character trait. As a trait, it is seen as the capacity to manage stress which is facilitated by personality characteristics and environmental supports which mirror Handlovsky et al’s. [53] definition. However, some definitions include outcome and process elements, such as Nelson-Becker and Thomas [62], who define resilience ‘as an ability to integrate life learning and expand coping repertoires, reaching a new understanding that encompasses what came before but also moves beyond it’ (p. 2) and as ‘the ability to access one’s inner wisdom and strength enhanced by time and experience’ (p. 2). Both Dziengel [52] and Nelson-Becker and Thomas [62] also bring in the notion of adversity shifting older LGBT+ individuals off-balance with resilience providing the capacity to help them to restore equilibrium.

Foster et al. [57] also define resilience in the context of it being a trait which is the capacity to respond and successfully negotiate stressful life events in addition to it being a set of protective factors. Foster et al. [57] incorporate the ability of resilience to build capacity to also respond to future challenges which stresses the developmental nature of resilience. Higgins et al. [59] refer to Foster et al’s [57] definition of resilience as well as using Ungar’s [14] trait and process orientated definition which states that resilience is the capacity to navigate resources that build and sustain wellbeing. Fleishman et al. [46] use the American Psychological Associations definition of resilience as a process of adaptation to adversity, while Fredriksen-Goldsen et al. [47, 69] discuss resilience in the context of behavioural patterns, functional competence and cultural competence that individuals and communities use in stressful situations. Emlet et al. [43] refer to resilience as a resource as well as a process, outcome and trait.

Six qualitative papers (classified as being outcome-focused) did not set out to research resilience, therefore a conceptualisation or definition of resilience was not required at the outset of these studies. In one example, the researcher referred to resilience emerging only briefly in participants’ narrations in relation to how historical experiences led the participants developing resilience in later life [66], while other authors included resilience in the discussion section of their papers [54, 56, 58, 60, 66]. In these papers, the reference to resilience is most often related to the experiences of discrimination, oppression and surviving the HIV/AIDS epidemic. These lived experiences led LGBT+ older adults to develop coping skills, which leads to resilience in later life. For example, Jen and Jones [54] relate resilience to the broader cultural and historical events and intersectional identities evident within the participants’ narratives while also recognising resilient communities’ efforts around HIV/AIDS. Green and Wheeler [61] discuss how the participants in their study demonstrated resilience as they aged which acted as a buffer to negative health care experiences related to their minority sexual orientation and HIV status. Others only briefly mention resilience as part of participants’ strengths [56, 60] and as a potential topic for training that would shift practices from pathology to seeking indicators of resiliency [58]. Only Rowan and Butler [56] set out to define resiliency. They define it by using Blundo’s [78] understanding of resiliency as ‘finding a stronger and more meaningful way to deal with life difficulties and stressors’ (p. 191).

How resilience was studied

Adopting a clear definition appeared to be influenced by how the researchers approached the study of resilience. Twenty-one (n = 21) studies were classified as entry point studies as they had resilience as a stated or inferred objective and six studies were outcome studies as resilience inductively emerged from the data. In all the quantitative studies (n = 14), [10, 43–51, 68, 69, 71, 75] resilience was an entry point, while the qualitative studies, were divided between entry (n = 7), [52, 53, 55, 57, 59, 62, 65] and outcome studies (n = 6), [54, 56, 58, 60, 61, 66]. Entry point studies were further classified based on the focus researchers gave to studying resilience. Studies that utilised resilience theory and aimed to explicitly research resilience were classified as primary-focused. These studies specifically explored resilience in terms of how it was developed and the factors that influenced it. Where resilience was a secondary focus or a secondary research objective, resilience per se was researched in the context of other phenomena, such as multiple stigmas in relation to LGBTQ aging and cultural generativity [65], integration of sexual orientation and faith [57], physical and mental health [45, 50], as one of the several mental health and well-being indicators [49], and others.

Two primary focused studies specifically researched spiritual resilience [57, 62], where the focus was on the struggles LGBT people had in reconciling their sexuality with their sense of spirituality or how the persons spiritual beliefs or practices supported them in addressing life’s challenges. Others examined more general protective processes or factors that foster positive health outcomes and decreased psychological distress associated with sexual identities [53]. Resilience was also studied as an adaptive factor [44] that helps older LGBT+ people to mediate the impact of stigma and discrimination and adapt to prejudice and loss [59]. As a primary focus, resilience was also studied alongside other predictors of mental distress in gay men such as stigma, discrimination and internalised homophobia [71] and as a predictor of sexual satisfaction [46]. Resilience, along with the concept of mastery, were studied as two separate psychological resources that relate to each other but may be influenced by different factors in the structural-environmental and individual contexts [43]. In addition, Haldane et al. [55], set out to study how prominent ‘sexual orientation minority elders’ perceived the LGBTQ+ community as developing hope and resiliency in relation to major events that led to the development of rights for LGBTQ+ people.

In papers where resilience was a secondary focus, it was often seen as something that was exercised or demonstrated in the face of adversities. In Bower et al’s [65] study for example, they set out to research multiple stigmas and cultural generativity in relation to LGBTQ and ageing. They also set out a secondary objective to discuss how the experience of stigma and discrimination support a generational legacy of resilience. Similarly, Foster et al. [57] explored the integration of sexual orientation and faith and discussed how spirituality may become a method for meaning-making and resilience. Resilience as a secondary focus was more prominent in quantitative papers and often researched in relation to the presence of different physical and mental health characteristics, which were used as proxy measures of resilience. Cortes et al. [45], for example, set out to find and confirm evidence of resilience in older LGBT adults by measuring levels of anxiety, depression and alcohol use and then comparing them with younger LGBT adults. Similarly, Monin et al. [50] examined whether sexual minority status confers vulnerability or resiliency in older adulthood by using a variety of scales to measure mental health, social support and exposure to trauma. Included papers from the ‘Caring and Aging with Pride’ and ‘Aging with Pride’ studies, in most cases did not report measuring resilience specifically, but used a variety of tools to measure key health indicators, risk and protective factors and health outcomes. In these studies, higher scores in the variables measured suggest the exercise of resilience. Emlet et al. [43, 75] also drew their findings from ‘Aging with Pride’ but differed because they reported directly measuring resilience.

Tools and conceptual frameworks used

In the studies by Batista and Pereira [44] and Pereira and Silva [51], resilience was a primary focus and was measured using the 10-item Connor-Davidson Resilience Scale that utilises a four-point Likert scale (CDRISC-10, [79]). This self-reporting scale takes the perspective that resilience is a personal quality that reflects the ability to cope with stress [80]. Fleishman et al. [46] and King and Richardson [71] used Wagnild and Young’s Resilience Scale [72], although the former used a 25-item scale, the latter used a 14-item scale without a clear rationale for omitting some of the items on the original scale. Wagnild and Young’s [72] self-reporting scale views resilience as a positive personal characteristic that enhances individual adaptation. Emlet et al. [43, 75] used a three-item scale adapted by Fredriksen-Goldsen and Kim [73] from the Brief Resilience Scale (BRS–six item) by Smith et. al [74]. The BRS was designed to assess the person’s ability to bounce back or recover from stress [74] and was used in its original form in the study conducted by Lyons et al. [49].

A number of studies used a conceptual framework to guide the research design or the analytical process. Fredriksen-Goldsen et al. [47] used a Resilience Conceptual Framework that incorporated risk and protective factors. The framework included five components: ‘(1) background characteristic (sexual orientation, gender, race etc); (2) key health indicators (access to health care, health behaviours); (3) risk factors (victimisation, stigma, concealment); (4) protective factors (social support, network size); and (5) health outcomes (general health, disability, depression)’ [47, p. 665]. While this framework was used to inform a number of the ‘Caring and Aging with Pride’ studies [47, 68, 69], there was variation in the title given to the framework and in how the dimensions of the framework were reported within the different papers published from this study.

Fredriksen-Goldsen et al. [10, 48] and Emlet et al. [75] used the Health Equity Promotion Model [70] that situates health more broadly within the life course, asserting that optimal ageing is linked to the availability of opportunities over time to promote health. ‘This model highlights how (a) social positions and (b) structural and environmental contexts intersect across the life course with (c) health-promoting and risk processes (psychological, social, behavioural, and biological) to culminate in the health and well-being of LGBT adults as they age’ [10, p. S73]. The authors argue that the key variables in the model influence resilience, therefore higher scores in health behaviours suggest the exercise of resilience. Finally, Foster et al. [57] adapted the Model of Global and Situational Meaning [81] to develop a theoretical model for Lesbian and Gay Christian Spiritual Resilience [57].

Strengths and limitations of the review

A member of the team with specific expertise in literature searching and retrieval conducted the database searches without any time limitations being applied (GS). This helped to ensure that the review was as comprehensive as possible. In addition, the review process, data extraction and quality assessment were conducted by more than one author, making the process less susceptible to bias. However, many of the included studies had methodological weaknesses which may influence the quality of the findings in this review. Only peer-reviewed papers in the English language were included. Furthermore, papers that included resilience might have been excluded where resilience was not mentioned in the title/abstract/keywords section of the paper or in cases where researchers used terminology other than resilience although our search strategy sought to minimise this. While some might consider the inclusion of multiple papers from the same study a limitation, with the potential to skew the findings, their inclusion shows how even within the same study resilience is differently applied and conceptualised across the papers. Deciding on whether resilience was a primary or secondary focus had an element of subjectivity attached. For example, we classified Haldane et al. [55] study as a primary focused paper, as the study of resiliency was stated as an aim; however, as resiliency was not included in the subsequent research questions or reflected in the interview schedule others may have classified this paper differently.

Future research

In terms of future research this review points to the need for researcher to explore in greater depth from an emic perspective how older LGBT+ people conceptualise resilience, including the socio structural factors that promote or hinder its development. It is also important that future studies are designed in a way that takes account of peoples’ multiple identities (e.g. gender, sexual orientation, ethnicity) might intersect and influence their understandings and experiences of resilience. The review demonstrates that white, gay and middle-class men were more prevalent in the samples of the included studies, therefore the question of whom is being studied needs to be considered in any future research on resilience in LGBT+ later life. In addition, future researcher should be more explicit about how they are using resilience theory, including the discourse or perspective they are taking when employing the concept. As highlighted in the review more studies are also required to make explicit how concepts such as resilience, resiliency, coping and adaptation differ. If the study of resilience is to move towards a strengths-based approach, there is a need for researchers to consider their choice of measurement scales and move beyond scales that measure symptomatology. There is also a need for future resilience studies to move beyond the ‘individual’ adaptation discourse to one that addresses or politicises resilience in the context of socially embedded discrimination experienced by the older LGBT+ community. Finally, while not specifically focused on resilience, to avoid confusion there is a need for researchers to be cognisant of the language they use to report demographic characteristics, especially when reporting on sex, gender and sexual orientation.