Introduction
Global suicide rates are highest among older people (De Leo, Reference De Leo2022; The Institute for Health Metrics and Evaluation, 2019). In 2019, the suicide rate for people aged 70 and over was 24.5 per 100,000 inhabitants and 14.25 for people aged 50–69 (The Institute for Health Metrics and Evaluation, 2019), while it was 9.4 per 100,000 inhabitants, all ages combined. Moreover, suicide rates in older adults increase in each of the 5-year age-bands between the ages of 60–64 years and 90–94 years, for both men and women (Shah et al., Reference Shah, Bhat, Zarate-Escudero, DeLeo and Erlangsen2016). For example, data available from 25 countries showed that the rates of suicide in men increased from 34.7 in the 60–64 age-band to 68.6 in the 90–94 age-band, while in women they increased from 9.8 to 14.8 in the respective age-band (Shah et al., Reference Shah, Bhat, Zarate-Escudero, DeLeo and Erlangsen2016). With the aging of the population, research on suicide in older adults should be the subject of numerous studies, yet this topic is strongly neglected (Okolie et al., Reference Okolie, Dennis, Simon Thomas and John2017). In addition, the suicide of a senior has significant repercussions on relatives, who may come to consider death as a possible solution when they have to face the hardships or challenges associated with their own aging (Michaud-Dumont et al., Reference Michaud-Dumont, Lapierre and Viau-Quesnel2020). Therefore, it is important to examine factors that could help prevent suicide in older adults. Three systematic reviews were published recently on risk factors associated to suicide in old age (Barker et al., Reference Barker, Oakes-Rogers and Leddy2022; Beghi et al., Reference Beghi2021; Fernandez-Rodrigues et al., Reference Fernandez-Rodrigues2022). However, to our knowledge, there are no reviews examining protective factors of suicide among the older population.
For older adults, talking about death may be part of their preparation for the end of their life (Tjernberg and Bökberg, Reference Tjernberg and Bökberg2020). Therefore, in research on suicide, it is necessary to distinguish between normal thoughts about death, and “suicidality” (Jonson et al., Reference Jonson, Sigström, Van Orden, Fässberg, Skoog and Waern2023; Keefner and Stenvig, Reference Keefner and Stenvig2021). Suicidality refers to a broad scope of concepts that extend from suicidal ideation to suicidal behaviors (planning, suicidal attempt, or suicide). Suicidal ideation (SI) is conceptualized as varying along a continuum from passive suicide ideation (PSI), including life-weariness (feeling that life is not worth living) and wish to die (would rather be dead/be better off dead), to active suicide ideation (ASI), which refers to thoughts of and intention to end one’s life (Harmer et al., Reference Harmer, Lee, Duong and Saadabadi2023; Jonson et al., Reference Jonson, Sigström, Van Orden, Fässberg, Skoog and Waern2023). Unfortunately, SI has no consistent operational definition which is often mention as a limitation in meta-analyses associated with suicidality (Berman and Silverman, Reference Berman and Silverman2017; Harmer et al., Reference Harmer, Lee, Duong and Saadabadi2023). Nonetheless, in older adults, PSI usually appears in conjunction with psychological distress and is associated with an increased risk of suicide, even in the absence of depression (Van Orden et al., Reference Van Orden, O'Riley, Simning, Podgorski, Richardson and Conwell2015). Since older adults are less inclined to express ASI and to seek mental health services, and that the majority of them die on their initial attempt, PSI deserves special attention (Harmer et al., Reference Harmer, Lee, Duong and Saadabadi2023).
Research on suicide in older adults report many risk factors (Conejero et al., Reference Conejero, Olié, Courtet and Calati2018), but the complex interactions between factors make it difficult to predict suicidal behaviors in a given person. Depression is clearly the major risk factor for suicidal behavior as well as social disconnection and interpersonal difficulties (Bernier et al., Reference Bernier, Lapierre and Desjardins2020; Fässberg et al., Reference Fässberg2012) or chronic diseases, pain, and functional limitations (Fässberg et al., Reference Fässberg2016). The effect of ageism should not be overlooked; indeed, results show that elderly people who experienced discrimination because of their age were 2.26 times more at risk of having ASI than those who had not experienced it, even after controlling for sociodemographic characteristics, health status, and the presence of depressive symptoms (Kim and Lee, Reference Kim and Lee2020; Ko et al., Reference Ko, Han and Jang2021). Nonetheless, Hawton’s et al. (Reference Hawton, Lascelles, Pitman, Gilbert and Silverman2022), as well as Cramer and Tucker (Reference Cramer and Tucker2021), stated that risk prediction measures are ineffective and that an assessment centered on protective, dynamic (changeable or modifiable), anticipated factors is needed to devise collaborative treatment plans.
Moreover, in the last 10 years, most intervention programs focused on reducing risk factor, especially depression or isolation (Laflamme et al., Reference Laflamme, Vaez, Lundin and Sengoelge2022; Lapierre et al., Reference Lapierre2011; Okolie et al., Reference Okolie, Dennis, Simon Thomas and John2017; Wallace et al., Reference Wallace, Miller, Fields, Xu and Mercado-Sierra2021; Zeppegno et al., Reference Zeppegno, Gattoni, Mastrangelo, Gramaglia and Sarchiapone2018). Only a few researchers have attempted to develop programs focusing on protective factors, such as meaning in life (Heisel et al., Reference Heisel2020; Heisel et al., Reference Heisel, Talbot, King, Tu and Duberstein2015), hope (Hernandez and Overholser, Reference Hernandez and Overholser2021), and the realization of personal projects (Lapierre et al., Reference Lapierre, Desjardins, Dube, Marcoux, Miquelon and Boyer2017; Lapierre et al., Reference Lapierre, Dubé, Bouffard and Alain2007). However, research on protective factors in the older population is still scarce and there is a need for a systematic up-to-date review of the quantitative studies (Cramer and Tucker, Reference Cramer and Tucker2021). The objective of this present study is to synthesize knowledge on protective factors and to examine their predictive associations with a reduced suicide risk in older adults.
Method
Conceptualization of protective factors
In the current study, a protective factor is viewed as a positive psychology construct, and is defined as a resource, an attribute, or an ability that protects the individual from SI and suicidal behaviors. Many studies are using concepts that negatively mirror protective factors (e.g., social disconnectedness, social isolation, and hopelessness). These negative words were not searched because the objective of this review was to find features that can improve the lives of individuals and counteract the negative influence of risk factors. Nonetheless, studies were retained if the authors chose to present the results on protective factors using a negative formulation.
This systematic review is investigating only psychosocial variables. Therefore, studies considering physical health (health-related quality of life) or demographic characteristics, such as religious affiliation, civil and marital status, family structure, and presence of children, grandchildren, or siblings, were not included in order to focus on an individual’s social and psychological functioning. Protective factors related to governmental, organizational, and social policies, programs, or services were not included, although it is clear that the variability of protective factors is closely linked to these macro systems.
One difficulty when conceptualizing the protective factors lies in the unclear demarcation between the factors. Most are inter-related and share a common root. Therefore, even if researchers are using the same name for a protective factor, they often measure different concepts or dimensions. Nonetheless, we identified 15 psychosocial protective factors (details follow below) classified as either intrapersonal or interpersonal. Intrapersonal factors refer to one’s inner perceptions or attitudes. Interpersonal factors stem from interrelations with people and community. Interpersonal characteristics are usually classified into structural and functional aspects. Structural support refers to the extent to which an individual is connected within a social network (often assessed objectively), while functional support corresponds to the role provided by the network and is usually measured by the subjective perception of the help available. Thus, we tried to distinguish between subjective and objective measures of interpersonal factors, although it is sometimes difficult because variables, such as social support, are often defined in terms of both quantity (objective) and quality (subjective) of interrelations.
Search and selection procedures
An a priori protocol was developed and registered on PROSPERO for registration of systematic reviews (CRD42022377759). The search procedure was conducted in two steps. First, selection of protective factors was made by conducting a scoping review of relevant studies to look for pertinent search terms. It was conducted using broad terms such as “protective factors”, based on combinations of keywords (#1 (suicide) + #2 (elderly) + #4 (protective factors) as in Supplementary Table 1) in five databases (PubMed, EMBASE, Cochrane, SocINDEX, and PsycINFO). It provided a total of 15 protective factors, that represented general dimensions that were included for the review: perceived control, well-being and quality of life, life satisfaction, purpose-in-life, resilience, coping, religiosity, hope, self-regulation, sense of belonging, mattering, positive relationship, social support, social connectedness, and social participation.
In the second step, searches were conducted for each factor using specific search terms (#1 + #2 + #3 (specific protective factor) in Supplementary Table 1) in the five previously mentioned databases. Peer-reviewed journal articles met the inclusion criteria if the published study had a population/sample of older adults aged 60+, living in the community or in residential care facilities. No restrictions were made on language, country, or publication year. Case reports, dissertations, psychological autopsies, qualitative-, ecological- and descriptive studies were excluded, as these designs do not provide quantitative estimates of the associations (e.g., the strength of association between protective factors and suicide ideation). Intervention studies were also excluded if they had multimodal components because it would be impossible to identify the influence of a particular protective factor. Also, studies involving specific populations (e.g., suicide attempters, abused or bereaved older adults, patients’ sample, transgender, aboriginal people, or prisoners) were excluded, due to the limitation of generalizability. The selection procedure was carried out independently by two researchers. When there was a disagreement, a third party was involved to resolve it.
Evidence was collected about study population, design, measures, outcomes, and findings. When SI was measured using a single item, a further distinction was made between PSI and ASI. If the item covered statements such as ‘would be better off dead’ (‘rather be dead’, ‘wish to be dead’ or ‘tired of living’), it was classified under PSI, while items representing thoughts about ‘ending one’s life’ (‘killing oneself’ or ‘taking one’s life’), were classified under ASI. Data were extracted by two reviewers who were assigned to each protective factor to increase accuracy. Given the wide range of methodologies and variables in the included studies, we narratively summarized patterns of findings. The reviewers independently assessed the methodological quality of the studies using the NIH quality assessment tool (NIH, n.d.) for cohort and cross-sectional studies (14 criteria), and case-control studies (12 criteria). The quality of each study was assessed with a summary score ranging from 1 to 14 or from 1 to 12, according to their respective design, which was then transformed into a percentage, that led to one of three grades: Poor (<50%), Fair (50–74%), Good (≥75%) (Supplementary Table 4-1 and 4-2).
Results
Study selection
The literature search yielded 9,099 reports from the 15 separate searches of the five databases. After 2,917 duplicates were removed, screening on title and abstract excluded 6,710 papers, leaving 282 studies. On reading the full text, 189 studies were found to be ineligible, the most common reason for being removed was that they were no predictive analyses or available estimates of the association. The PRISMA flowchart (Figure 1) summarizes each step and provides further details on reasons for exclusion. A total of 70 studies were retained, and results were summarized on 13 protective factors (See Supplementary Table 2 for the selection process of an individual factor). Searches on two factors, mattering and self-regulation, did not lead to any findings. Since many studies (n = 18) examined more than two protective factors simultaneously, the analysis was done on a total of 93 different observations/results (see the summary of results in Table 1). Each of the 13 factors is presented individually under one of the main categories: intrapersonal (n = 40 observations) or interpersonal protective factors (n = 53 observations).
Figure 1. PRISMA flow diagram. *The literature search yielded 9,909 studies in total for 15 protective factors from the five databases. †The 93 results were selected from 70 papers.
Table 1. Summary of the included studies in the review on protective factors and suicidality
C-S: Cross-sectional study, C-C: Case-control study, f/u: Follow-up, NS: Not specified, PSI: Passive suicide ideation, ASI: Active suicide ideation, SI: Suicide ideation, SA: Suicide attempt, 2W: 2 weeks, 1Y: 1 year, 1M: 1 month, OR: Odds ratio, IRR: incidence rate ratio, CI: confidence interval.
a The study population is usually a representative sample of the general elderly population, unless a convenience or particular sample is mentioned.
b For the strength of association, if a protective factor is in opposite directions or is reported in multiple categories, the name of variable or category was provided. If the name is not given, it is the same as in the column of measures for protective factor. Additionally, either of two estimates in regression, B (or b, the unstandardized coefficient) and β (standardized coefficient, beta weights) were extracted as presented in the original article. If an article presented both estimates, we prioritized the standardized β.
Study characteristics
Heterogeneity of these quantitative studies was high and the majority (n = 59) were cross-sectional in design and used representative samples (n = 42) (see Supplementary Table 3 for characteristics of included studies). There was an increasing trend over time in the number of studies on protective factors; 36% (n = 25) were carried out recently (in 2020 or after). Most studies were conducted in Asia (n = 38) and North America (n = 14), while some were conducted in Europe (n = 4), Oceania (n = 7), and the Middle East (n = 6), but only one in Africa, and one in South America. A substantial number (n = 20) had small samples (<300 participants), especially those examining intrapersonal factors, such as purpose-in-life and hope. Larger samples (>300) were more often used for interpersonal protective factors, with the exception of sense of belonging. Only a few recruited older adults living in residential care facilities. Quality assessment scores indicated that most studies (n = 61) were rated as fair and only five studies were of good quality, while four were rated as poor. The details of the assessment and the assigned score to each study are provided in the Supplementary Tables 4-1 and 4-2.
Measurement scales were frequently used to assess intrapersonal factors (77%; 31/40). However, scales were not commonly used to measure religiosity, social connectedness, and social participation. Only a small number of studies (n = 6) used deaths by suicide, suicidal attempts or deliberate self-harm as an outcome variable; the majority examined SI. For the latter, many studies (n = 35) used a single item to assess PSI or ASI or both (e.g. Bernier et al., Reference Bernier, Lapierre and Desjardins2020; Liu et al., Reference Liu, Fairweather-Schmidt, Roberts, Burns and Anstey2014). Studies (n = 29) also used validated scales, such as the Geriatric Suicide Ideation Scale (GSIS) or the Beck Suicidal Ideation Inventory, to assess SI. The majority of studies looking at the intrapersonal factors used a scale to measure SI but, for interpersonal factors, the use of a single item was more common.
Overall, associations with SI and suicidal behaviors were consistently observed for all protective factors. The proportion of significant associations was larger for intrapersonal factors (73% = 29/40) compared to interpersonal factors (51% = 27/53). The overall status of association was partly affected by types of measurement (scale vs single item) used to access each protective factor. The use of scales, instead of a single item, to measure protective factors was more efficient to capture the associations. It was clearly the case for studies on life satisfaction, which showed highly consistent associations with suicidality when utilizing the Satisfaction with life scale compared to a single question approach.
The variation of associations by study design, sample size, or the quality level of the studies were not particularly obvious. Studies conducted in Asia (15/17) and in the Middle-East (3/4) consistently showed significant associations between intrapersonal factors and suicidality, while the results were less consistent in Western countries: nine studies showed significant associations, but four studies respectively showed mixed or non-significant associations. Though limited in number, all three studies that included a sample of persons over 70, were significant.
Intrapersonal protective factors
Even if the majority of studies focused on interpersonal factors, the included studies identified 13 different intrapersonal variables that are considered potential protective factors against suicidality: perceived control, self-efficacy, self-esteem, well-being, quality of life, life satisfaction, happiness, purpose-in-life, resilience, coping, religiosity, hope, and self-forgiveness. Surprisingly, there were no studies on self-regulation, even if the literature considers that the ability to control impulsive behaviors and deal with emotional pain is an important factor for suicide prevention (Turton et al., Reference Turton, Berry, Danquah and Pratt2021). The associations of intrapersonal factors with SI were more evident when PSI was used as an outcome (i.e. two out of three studies were significant), compared to ASI (three out of nine studies were significant).
Since suicide prevention has to take into consideration various components simultaneously, some studies looked at the interactions between variables, examining the moderating or mediating effects of protective factors on suicidality or testing conceptual models. For example, purpose-in-life was found to be a moderator between hopelessness and PSI such as this association was significant when meaning in life was low, but not when meaning in life was average or high (Beach et al., Reference Beach, Brown and Cukrowicz2021). Purpose-in-life also came out as a mediator between internal locus of control and risk for suicide, so that a more internal control was associated to higher purpose-in-life, which in turn was related to lower risk for suicide (Aviad and Cohen-Louck, Reference Aviad and Cohen-Louck2021). Meanwhile, all studies showed that people who think that their life is meaningful were less likely to score highly on the SI scale. Therefore, purpose-in-life seems to be an important protective factor for older adults. As for hope, although it is more often studied in a negative way (hopelessness), it seems to be a protective factor against SI since 4 out 5 studies showed significant associations.
Seven studies investigated life satisfaction/happiness, and five examined psychological well-being/quality of life. All results were in the expected directions; however, most studies presented them in a negative way, looking at dissatisfaction, poor quality of life, or unhappiness instead of the positive side. As for psychological well-being (PWB), when depression and number of perceived health problems were entered as covariates in the regression, its predictive strength disappears, while purpose-in-life remained significantly associated with decreased SI (Heisel and Flett, Reference Heisel and Flett2008). It should be mentioned that a global score of PWB was used in the regression even if the multidimensional model of well-being, assessing six dimensions (autonomy, environmental mastery, personal growth, positive relations with others, purpose-in-life, and self-acceptance), is considered better than the single-factor model (Ryff and Keyes, Reference Ryff and Keyes1995). It would have been interesting to see which dimensions were associated with lower SI, but the researchers did not have enough participants to include each of them in the regression.
There were interesting findings related to coping. Even if one of the three studies of this protective factor showed that older adults who were using adaptive coping strategies (religious coping, acceptance, active coping, and positive reframing) exhibited less SI (Ahn and Kim, Reference Ahn and Kim2015), the other two noted that problem-focused coping (managing or altering the circumstance that is causing distress) was not a significant predictor, indicating that it may be less relevant for older adults. As for emotion-focused coping, the associations with SI vary according to the strategies that were assessed for regulating the emotional response to distress. Marty et al. (Reference Marty, Segal and Coolidge2010) found that emotion-focused coping (seeking emotional support, positive reinterpretation, acceptance, humor, and turning to religion) was a protective factor against SI, while Yoon et al. (Reference Yoon, Cummings, Nugent and Forrest-Bank2022) found the opposite results with other emotion-focused coping strategies (self-distraction, behavior disengagement, denial, self-blame and venting). Therefore, as Marty et al. (Reference Marty, Segal and Coolidge2010, p. 1021) stated “simplistic distinction between problem- and emotion-focused coping is not adequate.” The distinction between diverse emotion-focused coping strategies may be especially important for older adults for whom there is often nothing that can be done to actually change the circumstances related to ageing. As for perceived control, a higher level was associated with lower odds of PSI and ASI, and risk of suicide. Interestingly, in a longitudinal study, changes in perceived control were predicting changes in PSI: improving control reduced SI, and decline in control increased it (Stolz et al., Reference Stolz, Fux, Mayerl, Rásky and Freidl2016). It should be noted that among the six studies on resilience, five showed a significant protective association against SI.
Results on the predictive effect of religiosity are inconsistent. Although, Nishi et al. (Reference Nishi, Susukida, Kuroda and Wilcox2017) found an inverse association with SI in participants who give importance to their religious beliefs and attend religious services frequently, Heisel and Flett (Reference Heisel and Flett2008) found a positive association. The authors suggested that older adults experiencing higher levels of SI are more likely to turn to fervent religious observance for help than older adults who are less suicidal.
Interpersonal protective factors
Among the 53 observations compiled on interpersonal factors (see Table 1), the majority (n = 26) reported a significant association, but a substantial number did not (n = 18), or presented mixed results (n = 9). It is interesting to note that both the functional aspects (sense of belonging, positive relationship or social support) and the structural aspects (social connectedness or social participation) showed similar level of associations. When scales were employed to measure SI, significant associations were notable throughout the interpersonal factors (76.5% = 13/17) and these associations were similar, regardless of the outcome measure used (PSI or ASI).
First, it should be noted that there was no study on mattering, the human need to feel significant for others (Flett, Reference Flett2018). This suggests that this is an under-researched area. There were eight eligible studies regarding sense of belonging. All studies used the Interpersonal Needs Questionnaire to measure the association with SI or suicidal behaviors. Five studies reported a significant relationship, but three did not. Among eligible studies, the study with the largest number of participants (n = 669) by McLaren et al. (Reference McLaren, Gomez, Gill and Chesler2015) showed that there was a significant association between sense of belonging and SI (β = −0.007 (0.01), p < .001).
Three studies examined the association between positive relationships and suicidality. There was heterogeneity in terms of type of protective factor and outcome measures; one study looked at the association between having a confidant and PSI (Bernier et al., Reference Bernier, Lapierre and Desjardins2020), another at the relation between interpersonal trust and ASI (Yu et al., Reference Yu2019), and the third, at positive relationships and intentional self-harm (Neufeld et al., Reference Neufeld, Hirdes, Perlman and Rabinowitz2015). Among these studies, the presence of a confidant did not protect against PSI, when social participation, satisfaction with social life, and closeness to others were covariates (Bernier et al., Reference Bernier, Lapierre and Desjardins2020).
There were 19 studies on social support. All studies used a form of scale to measure social support, while outcome measures were mostly confined to a single item of SI with only five studies using a scale (Almeida et al., Reference Almeida2012; Liu et al., Reference Liu, Qin and Jia2018; Shiraly et al., Reference Shiraly, Mahdaviazad, Zohrabi and Griffiths2022; Vanderhorst and McLaren, Reference Vanderhorst and McLaren2005; Won et al., Reference Won, Choi, Ko, Um and Choi2021). Compared to other protective factors, social support showed the least consistent results in terms of significant associations (10/19 studies), probably because relying on support from others can diminish older adults’ sense of competence and increase feelings of being a burden (Thomas, Reference Thomas2010). Studies, which examined men and women separately (Dong et al., Reference Dong, Chen, Wu, Zhang, Mui and Chi2015; Mizuno et al., Reference Mizuno, Hikichi, Noguchi, Kawachi and Takao2019; Sun and Zhou, Reference Sun and Zhou2018; Vasiliadis et al., Reference Vasiliadis, Gagné and Préville2012) did not show gender differences in the direction of association. When instrumental support (availability of help) was explicitly measured, two studies among three showed no association (Awata et al., Reference Awata2005; Saïas et al., Reference Saïas, Beck, Bodard, Guignard and du Roscoät2012). Finally, most studies (except for Mizuno et al., Reference Mizuno, Hikichi, Noguchi, Kawachi and Takao2019) looked at received support, neglecting the role of support given to others, which might be more beneficial for older adults’ well-being (Smith et al., Reference Smith, Cui, Odom, Leys and Fiske2020; Thomas, Reference Thomas2010),
Among 14 eligible studies regarding social connectedness, 11 were conducted in non-Western countries. Associations were inconsistent; six studies exhibited significant protective effects of social connectedness (assessed by size of social network or frequency of social contacts) against SI, but three studies showed no associations and five studies showed mixed results. There were no differences between social network and social contacts in their respective association with lower suicidality, both seem to be equally protective. However, the measures did not evaluate the quality of the interactions or the satisfaction with one’s network. Interestingly, results varied according to the types of relationship: family, children, relatives, or friends. For example, Chang et al. (Reference Chang, Sha, Chan and Yip2018) found that the protective effect of social connectedness against PSI and ASI was observed only within the family but not with friends, while Turvey et al. (Reference Turvey2002) found significant associations for closeness and interactions with friends, but not with children, in predicting lower suicide deaths.
The literature search resulted in nine eligible studies in relation to social participation and the majority showed significant associations with suicidality. However, some presented mixed or no significant results according to types of participation and outcome measures, or in specific sub-groups of population. A Taiwanese study (Yen et al., Reference Yen2005) showed associations between community participation and ASI for the whole sample. However, subsequent analysis showed that, when the sample was separated into subgroups, associations appeared, for example, only in men but not women, and in the low-income group, but not the high-income group. A study from Korea differentiated between types of social participation and showed significant association only for religious involvement, but not for leisure activities, meetings with friends, or instrumental participation for social changes (Ra and Cho, Reference Ra and Cho2013). A study with a large community sample (Jeong, Reference Jeong2020) showed that the association of social participation differed by outcome measures; associations were found for ASI but not for suicide attempt.
Discussion
This systematic review provides confirmation that protective factors are associated with lower SI. All of the included protective factors we examined were associated with a decreased likelihood of PSI and ASI. The most consistent results were among intrapersonal factors. In fact, purpose-in-life and resilience seem to be the most valuable protective factors, showing recurrent positive associations with reduced suicidality. These findings suggest potential value in attending to both purpose-in-life and resiliency when assessing SI and when developing interventions for vulnerable older adults (Heisel and Flett, Reference Heisel and Flett2008). However, except for social connectedness and support, each factor was truly neglected by research. Another problem is that, currently, investigators are mostly focused on deficits or risk factors. They are not examining the resources, attitudes, abilities, and coping strategies that individuals are using to avoid suicide as a possible solution to the suffering associated with the difficulties and losses of old age. Moreover, even if they are looking at protective factors, authors frequently present their results using a negative formulation, reporting that a deficiency in the predicting variable is associated with suicide risk, instead of stating the positive association with reduced suicidality. Though the labelling issue is important because it guides intervention, most protective factors are dynamic and the positive direction of a specific dimension is not sufficient to clearly distinguish it from a risk factor (Heffernan and Ward, Reference Heffernan and Ward2017). Therefore, there is still work that needs to be done on the conceptualization of protective factors.
Some protective factors seem to have been completely overlooked, such as self-regulation (Turton et al., Reference Turton, Berry, Danquah and Pratt2021) and mattering (Flett, Reference Flett2018), while others were examined in a small number of studies (e.g., ≤3 for coping and positive relationships) and it is difficult to draw a generalizable conclusion on their importance for reducing suicidality. Moreover, scientific rigor varied across reviewed articles: while most studies were rated as fair, small sample and cross-sectional design were predominant. Furthermore, the evidence of the protective effects against suicidal attempts or suicide death requires a time span, which suggests that longitudinal designs, integrating quality measures on all outcomes, would be very pertinent. Another possibility is a data linkage between community sample and hospital data or death registry using a probabilistic technique, as shown by Erlangsen et al. (Reference Erlangsen2021). As for the types of measurement, the use of single questions is still a common practice, particularly when assessing religiosity, social connectedness, and social participation. However, number of contacts do not reflect the significance of relationships, nor does frequency of attendance at religious services reveal the importance of faith in the life of older adults (Deuter et al., Reference Deuter, Procter, Evans and Jaworski2016). This limitation may partly be compensated by employing scales or pooling multiple questions into clusters. Since there were more significant associations between intrapersonal factors and PSI than with ASI, it is possible that the former outcome measure is more sensitive to the psychosocial losses experienced by older adults (Jonson et al., Reference Jonson, Sigström, Van Orden, Fässberg, Skoog and Waern2023) and reflect their lower inclination to express ASI (Harmer et al., Reference Harmer, Lee, Duong and Saadabadi2023). However, this explanation has to be verified in future research.
The current review showed that results concerning participants aged 70 and older were more consistent than those obtained with people aged between 60 and 70 years. Previously, age was considered a moderator in the association between independent variables and suicidal outcomes because of a more pronounced impact in the older groups (Innamorati et al., Reference Innamorati2014). The importance of protective factors may be more relevant with older age, because adverse life events occur more frequently and the risk of suicide is higher as older adults advance in age (Shah et al., Reference Shah, Bhat, Zarate-Escudero, DeLeo and Erlangsen2016). Although it was not the objective of this review, it seems that protective factors for older adults are similar to those identified for younger people. For example, Seidler et al. (Reference Seidler2023) explored various protective factors (coping, connecting, self-reported resilience, and selflessness) in men aged 16 years and over (M = 50.3 years) during the COVID-19 pandemic and found significant associations with lower ASI. Future research should explore if some protective factors are unique in old age.
In the current review, the significant associations between intrapersonal protective factors and suicidality were more frequently present in non-Western countries. This suggests that in countries where collective identity is prevalent, intrapersonal factors have an added protective effect against suicidality. Collectivist societies give less importance/value to individual characteristics (Yoon, Reference Yoon2014) and as such, intrapersonal factors might be adding something special and essential to the prevention of suicide for people living within these cultures.
This systematic review was only concerned with psychosocial protective factors. However, each individual is also greatly affected by his/her environment. As such, national suicide prevention programs (Lewitzka et al., Reference Lewitzka, Sauer, Bauer and Felber2019), as well as public policies and social institutions that assist citizens needing help (Stack, Reference Stack2021), have been used to lower suicide rates. Since social environments interact with intrapersonal or interpersonal factors and vice versa, strengthening of protective factors need to be extended to a population-level approach. Quality programs, targeting groups of individuals and communities to enhance protective factors through social connectedness, are highly promising suicide prevention strategies. Good examples of community interventions are proposed (Hou et al., Reference Hou2022), but these have not yet been widely recognized to be embedded in existing national suicide prevention programs, with few exceptions (Substance Abuse and Mental Health Services Administration, 2017).
Since suicidal ideation and behavior develop along a continuum, and often begin with a wish to die, it is important to conceive longitudinal studies that identify protective factors that are most beneficial at each step of the process described by the model of O'Connor and Kirtley (Reference O'Connor and Kirtley2018). Some factors might be more pertinent to prevent the transition between SI and suicide attempts, while others might be more effective to prevent the start of the process entirely.
More importantly, future research should evaluate protective factor models that translate into effective treatment models (Michel, Reference Michel2021). For example, life satisfaction, psychological well-being, and quality of life could be outcome variables of the feeling that one’s life is meaningful or of one’s ability to cope with adversity (resilience). Variables such as character strengths (perseverance, creativity, gratitude, hope, humor, bravery, zest), which have been overlooked by research with older adults, might be considered as protective factors that can contribute to resilience, as it was observed during the COVID-19 pandemic (Lapierre et al., Reference Lapierre, Chauvette, Bolduc, Adams-Lemieux, Boller and Desjardins2023).
Future research could advance knowledge on protective factors by conducting systematic reviews on specific variables, as reported recently by Jeong and Noh (Reference Jeong and Noh2023) for resilience, or by applying an ecological approach to assess the strategies that can momentarily reduce the intensity of suicidal thinking (Stanley et al., Reference Stanley2021). Development of psychological interventions that can improve modifiable intrapersonal factors, such as hope (Hernandez and Overholser, Reference Hernandez and Overholser2021), resilience (Treichler et al., Reference Treichler2020), coping strategies (Gysin-Maillart et al., Reference Gysin-Maillart, Soravia and Schwab2020), and purpose-in-life (Heisel et al., Reference Heisel2020; Lapierre et al., Reference Lapierre, Dubé, Bouffard and Alain2007) could be innovative ways to prevent suicidality in older adults. They might be of particular interest, since they possibly can help shift the narrative away from deficits, toward self-efficacity, adaptation, and growth.
Strengths and limitations
This systematic review was able to synthesize the current knowledge and evidence on the psychosocial factors that protect older adults against suicidality, namely passive and active suicidal ideation, suicidal attempts, and suicide. The foremost strength of this review was that the search was not restricted to the generic term of “protective factors”. Actually, the preliminary scoping review provided a list of 15 psychosocial dimensions (and their related terms), that represented specific protective factors (see Supplementary Table 1). Each was searched separately, producing a large number of studies to examine the predictive associations of each factor with reduced suicidality in older adults. Nevertheless, it is possible that some factors were not identified, as it was the case for self-forgiveness or self-compassion. Self-forgiveness was not an expected variable because it was not among the search terms. Nonetheless, it could be an important protective factor since a systematic review reported significant associations between higher levels of self-forgiveness or self-compassion, and lower levels of SI and self-harm in individuals aged between 11 and 66 years old (Cleare et al., Reference Cleare, Gumley and O'Connor2019).
The main limitation of the review has to do with the conceptualization of each protective factors. Clearly, there is frequent overlap between interpersonal protective factors. For example, sense of belonging share common features with other social constructs, such as social connectedness and social support, as point out by Hatcher and Stubbersfield (Reference Hatcher and Stubbersfield2013) in their own systematic review on the association between sense of belonging and suicide.
Another limitation relates to the assessment of SI. Most selected studies used SI as the outcome variable, probably because it is nearly impossible to examine the role of protective factors in people who attempted or ended their life with suicide. However, as mentioned previously, there is no universally accepted consistent definition of SI, which encompasses everything from life weariness to preoccupation with self-annihilation (Harmer et al., Reference Harmer, Lee, Duong and Saadabadi2023). This leads to ongoing challenges for research since it interferes with the ability to compare findings across studies (Harmer et al., Reference Harmer, Lee, Duong and Saadabadi2023). Although scales to measure SI exist, none is sufficiently reliable to predict suicide (Hawton et al., Reference Hawton, Lascelles, Pitman, Gilbert and Silverman2022). Moreover, suicidal ideation fluctuates over time, and involves varying degrees of intent, motivation, intensity, imagery, and planning (Jonson et al., Reference Jonson, Sigström, Van Orden, Fässberg, Skoog and Waern2023). Therefore, comparative investigations of the predictive value of protective factors would need to sort out how each of the studies defined SI, especially those that use only one item to assess it. It was partly done here, by examining the items that were used in each study and distinguishing between PSI and ASI. Nonetheless, it is difficult to make this distinction, when the item includes both types of SI in one statement (e.g. item 9 of the PHQ-9), or if the assessment of SI includes positive answers to both PSI and ASI items.
Conclusion
Suicide prevention has traditionally concentrated on risk factors and the current study advocated the need to widen the strategies to strengthen protective factors. Research on risk factors was meant to predict suicide attempts and prevent deaths. Research on protective factors can show how to improve well-being and quality of life so that the suicidal process does not even start. In fact, having reasons for living and leading a meaningful life are incompatible with suicide (Lapierre et al., Reference Lapierre, Dubé, Bouffard and Alain2007), while building resilience could reduce the incidence of stress-related disorders (Sher, Reference Sher2019). As Holman and Williams (Reference Holman and Williams2022) suggested, future research should use a network approach to explore the interactions between protective and risk factors to determine which variables are central in order to guide effective targeted interventions.
Conflict of interest
None.
Source of funding
None.
Description of authors’ roles
SL and MK designed the study. MK, BEG, MJH, MKK, MJJ, EJK, and LD conducted literature searches, drafted tables and figures, quality assessments, assisted analysis, and edited supporting files. SL, MK, and BM wrote the manuscript, and updated tables and figures. All authors edited the manuscript and read and approved the final manuscript.
Acknowledgements
None.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S104161022300443X
Open access
