Background
Caregiving for people with health-related care needs is still primarily performed by relatives or friends, i.e. informal caregivers (Statistisches Bundesamt, 2020). Caregiving can increase meaning in life and can even improve cognitive functioning (Yu et al., Reference Yu, Cheng and Wang2018; Zwar et al., Reference Zwar, König and Hajek2018). However, it can also be very demanding as is reflected in the negative impact on health and well-being of caregivers (Bom et al., Reference Bom, Bakx, Schut and Van Doorslaer2019). One factor that has not received much attention, however, is suicide ideation among informal caregivers.
The World Health Organization categorizes suicide mortality as a public health priority (World Health Organization, 2021). Lifetime prevalence of suicidal ideation, i.e. thoughts about suicide or wishing to die, is much higher than that of suicide attempts or committed suicide (Cao et al., Reference Cao2015; Nock et al., Reference Nock2008), but it is one of the main determinants of suicide behavior (Favril et al., Reference Favril, Yu, Uyar, Sharpe and Fazel2022). Despite being highly associated with mental disorders, it can occur independently of depression as well (Campos et al., Reference Campos2016).
Weekly to lifetime prevalence of suicide ideation among dementia caregivers has been reported with a broad range (4.69–77.78%; Solimando et al., Reference Solimando2022). While a few studies have investigated the association between caregiving and suicide ideation and attempted or committed suicide, questions regarding individual or contextual influential factors of the caregiver situation remain unanswered (Molina et al., Reference Molina2019; O’Dwyer et al., Reference O’Dwyer2021). Also, only very few longitudinal studies have been conducted and their findings were mixed (e.g. Joling et al., Reference Joling, Ten Have, De Graaf and O’dwyer2019; Rosato et al., Reference Rosato, Tseliou, Wright, Maguire and O’reilly2019). For example, a study from the UK found a lower risk of suicidal ideation among caregivers compared to non-caregivers (Rosato et al., Reference Rosato, Tseliou, Wright, Maguire and O’reilly2019). Others found a higher risk of suicide among caregivers (Czeisler et al., Reference Czeisler2021; Nakanishi et al., Reference Nakanishi2022). A recent study from Mexico and Colombia found an association between suicidal ideation of care recipient and caregiver (McKee et al., Reference Mckee2021). However, no significant differences between caregivers and non-caregivers were found in the Netherlands (Joling et al., Reference Joling, Ten Have, De Graaf and O’dwyer2019). Many of the studies, including longitudinal research, focused on suicide ideation after bereavement (Molina et al., Reference Molina2019; Rosengard and Folkman, Reference Rosengard and Folkman1997). Thus, it remains unclear if suicidal ideation occurred due to caregiving or due to the loss of the care recipient.
In sum, there is not sufficient evidence on the association between informal caregiving and suicide ideation and the risk or protective factors of the care situation. Moreover, more longitudinal studies are needed. The association between caregiving and suicide ideation can only be analyzed with observational data, and they can easily be biased by unobserved variables. Longitudinal data analyzed with fixed effects (FE) regression analysis can reduce this danger of bias considerably and enable the calculation of consistent estimates (Cameron and Trivedi, Reference Cameron and Trivedi2009). Thus, this study aims to fill the research gaps by analyzing the association between informal caregiving and suicidal ideation with a longitudinal design and FE regression analysis. We will also investigate the significance of factors specific to the care situation for this association. These findings are expected to help identify vulnerable groups and develop preventive and supportive measures for informal caregivers.
Theoretical framework: theory on suicide thoughts and behavior applied to informal caregiving
The integrated motivational-volitional model of suicidal behavior (IMV) is one of the most prominent ideation-to-action-theories (O’Connor et al., Reference O’Connor, Cleare, Eschle, Wetherall and Kirtley2016; Forkmann, Reference Forkmann2021). It is a biopsychosocial model that takes individual and contextual factors into account. The theory proposes that suicidal behavior is the result of a three-step process. In the pre-motivational phase, biological, sociodemographic, and personality factors contribute to the individual’s vulnerability. Those with a higher level of vulnerability are more easily triggered to transition into the motivational phase. Trigger events can be stressful life events, which result in significant changes in life and have a high emotional impact (Howarth et al., Reference Howarth2020). Informal caregiving can be such a stressful life event, if the requirements of caregiving outweigh the caregiver’s resources (Lazarus and Folkman, Reference Lazarus and Folkman1987). In the motivational phase, suicidal ideation develops, which can eventually progress to suicidal behavior (volitional phase). Suicidal ideation can develop if individuals feel defeated and trapped in their situation, if their situation is perceived as hopeless and without any positive thoughts toward the future (O’Connor et al., Reference O’Connor, Cleare, Eschle, Wetherall and Kirtley2016; Forkmann, Reference Forkmann2021). Informal caregivers have reported such thoughts. For example, the desire to escape, needing a reprieve from caregiving, or feeling overburdened, was often reported as reasons for suicide thoughts (O’Dwyer et al., Reference O’Dwyer, Moyle and Van Wyk2013; O’Dwyer et al., Reference O’Dwyer2021). Worries and wishing to prevent ending up in the same situation as their care recipient were also mentioned (Anderson et al., Reference Anderson, Eppes and O’dwyer2019), indicating that the confrontation with the care recipient’s situation generated fear of the future. Thus, suicidal ideation among caregivers can result from and be indicative of being unable to cope with the caregiving situation, i.e. experiencing high levels of distress.
According to care-specific stress theory (Pearlin et al., Reference Pearlin, Mullan, Semple and Skaff1990), whether caregiving is experienced as stressful depends on the caregiver’s sociodemographic background and the care situation. Individual and contextual factors can influence the vulnerability relevant to developing suicide ideation as well (O’Connor et al., Reference O’Connor, Cleare, Eschle, Wetherall and Kirtley2016). Thus, when analyzing the association between informal caregiving and suicidal ideation, it is important to take contextual (e.g. caregiving context) and individual factors (e.g. caregiver’s age and gender) into account.
Previous research supports this. For example, in Europe the suicide risk is higher among women and older adults (Carrasco-Barrios et al., Reference Carrasco-Barrios2020). Moreover, cross-national differences can occur due to different welfare systems. This cross-country variation is found in global comparison as well as in comparison within Europe (World Health Organization, 2021). Thus, apart from gender and age, the country of residence is important for the suicide risk.
Adding to this, these factors are also of relevance in the caregiving context. For example, female caregivers have higher distress and worse mental health issues than male caregivers (Pinquart and Sörensen, Reference Pinquart and Sörensen2006). Findings regarding age as risk factor are mixed (van den Kieboom et al., Reference Van Den Kieboom, Snaphaan, Mark and Bongers2020), and this may be associated with who provides care for whom. The care relationship is often of relevance for the consequences for health (Pinquart and Sorensen, Reference Pinquart and Sorensen2011) or social support (Zwar et al., Reference Zwar, König and Hajek2022): that is, spousal caregivers are usually affected worse than adult-child caregivers. Further factors of the care situation, which may influence the appraisal of the situation and the suicide risk, could be the location of informal care provision, inside or outside the household. Previous findings indicated that inside-household caregivers are impacted worse, while outside-household-caregivers can even be positively affected (Kaschowitz and Brandt, Reference Kaschowitz and Brandt2017).
The welfare system of the country of residence of the caregiver is relevant to caregiving and its consequences as well. The welfare systems in Europe can be clustered into the Scandinavian, Bismarckian, Southern, and Eastern welfare systems (Eikemo et al., Reference Eikemo, Huisman, Bambra and Kunst2008; Ferrera, Reference Ferrera1996). The Scandinavian system is characterized by a strong public responsibility, with universal, generous, and egalitarian welfare state provision and public services (Greve, Reference Greve2022). It is financed by taxes and public sector provisions, and the state plays a vital role. Thus, many support opportunities are available for the population and for caregivers, which may ease their level of stress and the feeling of entrapment. The Bismarckian system provides social security benefits based on wage-dependent contributions and is coordinated by social partners, not the state (Greve, Reference Greve2022). It is a conservative model, which focuses on supporting a single breadwinner family model. Social benefits and alternatives to caregiving are available. For example, in Germany a care lump sum is provided by the statutory nursing care insurance based on the level of care needs; however, additional services need to be financed privately (Federal Ministry of Justice, 1994). The Southern system builds on this conservative model as well. It has undergone several changes during the last decades but many challenges remain, such as an unbalanced distribution of social expenditure and high job insecurity. The health care system (especially the long-term care policy), the labor market, and family policies remain underdeveloped. Care responsibility is therefore primarily placed on the family, mostly women, who are less well-integrated into the labor market and caregivers have fewer perceived alternatives and support than in the other systems (Greve, Reference Greve2022).
Purpose and hypotheses
This study used the biopsychosocial approach of the stress model (Pearlin et al., Reference Pearlin, Mullan, Semple and Skaff1990) and the IMV on suicide behavior (O’Connor et al., Reference O’Connor, Cleare, Eschle, Wetherall and Kirtley2016; Forkmann, Reference Forkmann2021) as basis to analyze the association between informal caregiving and suicide ideation as well as care-specific risk and protective factors.
First, we analyzed whether the association between informal caregiving and suicidal ideation was found in both caregiving locations, inside and outside the household. We expected adults transitioning into care inside to have worse suicidal ideation, whereas we expected no change among those transitioning into care outside. This differentiation was used as basis for the other analyses.
Second, we analyzed age and gender as moderators. Based on aforementioned findings, we expected female and older caregivers to have higher suicidal ideation than male and younger caregivers and expected the odds to be higher among those providing care inside the household than among those providing care outside the household.
Third, we analyzed the caregiver relationship. We expected suicidal ideation to be increased among care relationships with higher levels of closeness and expectations (e.g. partners).
Fourth, we took the broader context of caregiving into account by analyzing the association between care and suicidal ideation in three European welfare systems. We assumed caregivers in Scandinavian countries to have the lowest odds and those in Southern countries to have the highest odds.
Method
Sample
This analytical sample was drawn from the cross-national Survey of Health, Ageing, and Retirement in Europe (SHARE) (Börsch-Supan et al., Reference Börsch-Supan2013). SHARE collects data from adults aged ≥50 years and their partners (who could be <50 years) of the same household with a probability sampling method. In this study, we included participants from waves 1, 2, 4, 5, and 6 (collected from 2004–2015) from Austria, Germany, Sweden, Netherlands, Spain, Italy, France, Denmark, Switzerland, and Belgium. Other waves and countries were excluded, since they did not collect the analyzed variables consistently. SHARE uses ex-ante harmonization, i.e. one questionnaire is developed and translated into the national language (Börsch-Supan et al., Reference Börsch-Supan2013). Participants younger than 40 years were excluded, since the main period of caregiving is in the second half of life (Rothgang and Müller, Reference Rothgang and Müller2018), which left a sample of 171,848 observations (Obs.) pooled over all waves. FE regression analysis was used to analyze the research questions. This method allows us to analyze the association between changes within individuals in the variables over time (Brüderl and Ludwig, Reference Brüderl, Ludwig, Best and Wolf2015). Due to this, only individuals who experienced a transition into the analyzed variables were used for estimating the coefficients. For example, the basis for calculating the estimate of the association between caregiving inside and suicide ideation was 4,165 transitions from no caregiving to caregiving inside. All who were constantly providing care or constantly not providing care during the analyzed time frame were not used for this estimation. However, they are kept in the sample to allow for the calculation of the other coefficients of the model. More details on this can be found in Statistics.
Ethics committees or institutional review boards of the countries, where data are collected, review the SHARE project continually. The Ethics Committee of the University of Mannheim approved the first four waves, and the Ethics Council of the Max Planck Society approved the continuation of the project (detailed information at https://share-eric.eu/data/faqs-support).
Variables
Suicidal ideation was measured with one item from the EURO-D questionnaire (“In the last month, have you felt that you would rather be dead?”; no/yes; Prince et al., Reference Prince1999), which has good psychometrics (Tomás et al., Reference Tomás, Torres, Oliver, Enrique and Fernández2022). This single-item approach is in accordance with other studies’ approaches in this field (O’Dwyer et al., Reference O’Dwyer2021). Caregiving outside the household included caregiving for relatives, friends, or other acquaintances outside the household within the last twelve months in terms of personal care, practical household help, or help with paperwork (“Now I would like to ask you about the help you have given to others. […] In the last twelve months have you personally given any kind of help listed on this card to a family member from outside the household, a friend or neighbor?”; no/yes). Caregiving inside the household referred to regular (daily or almost daily) care for relatives or non-relatives during at least the last three months (“Let us now talk about help inside your household. Is there someone living in this household whom you have helped regularly during the last twelve months with personal care, such as washing, getting out of bed, or dressing?”; no/yes). Only participants living with at least one other person in their household were asked this question. To analyze the caregiver relationship, two variables were created, which categorized caregivers into five categories describing their relationship with their main care recipient (0 not providing care, 1 providing care for partner or ex-partners, 2 providing care for step-/parents/-in-law, 3 providing care for other relatives, and 4 providing care for non-relatives) either inside or outside the household. Based on the country, where the participant lived, they were categorized into a welfare system according to the typology from Ferrera (Reference Ferrera1996). The typology divides European countries into four systems, depending on how social benefits are provided to ensure an acceptable standard of living for all: the Southern system (Spain, Italy), the Bismarckian system (France, Germany, Switzerland, Belgium, Austria, Netherlands), and the Scandinavian system (Denmark, Sweden) as well as the Eastern system. We focused on the first three systems, since we had no data for the included variables and waves for Eastern Europe.
Participants provided information on sociodemographic factors (age, gender, marital, and employment status). Also, self-rated health [1 excellent to 5 poor health, single-item from the SF-36 questionnaire (Ware and Gandek, Reference Ware and Gandek1998)] and number of self-reported chronic diseases [ever diagnosed by a physician, sum score (theoretical range: 0–14)] were measured.
Statistics
Logistic FE regression analysis was performed, since they provide a major advantage when using longitudinal observational data (Brüderl and Ludwig, Reference Brüderl, Ludwig, Best and Wolf2015; Gunasekara et al., Reference Gunasekara, Richardson, Carter and Blakely2014). Longitudinal data allow us to distinguish between time-constant and time-varying errors. However, time-constant confounders often cannot be measured in observational studies (Brüderl and Ludwig, Reference Brüderl, Ludwig, Best and Wolf2015); thus, the assumption of methods, such as random effects regression, that there are no unobserved confounders is rarely fulfilled, which can severely bias the estimates (Gunasekara et al., Reference Gunasekara, Richardson, Carter and Blakely2014). The FE regression analysis relies on the more conservative assumption that there are unobserved time-constant confounders. Therefore, only time-varying variables are used for estimation and all time-constant (observed and unobserved) variables are controlled. Thus, FE regression analysis is a within-estimator and calculates an average treatment effect on the treated by using only the variation within individuals, i.e. analyzing only differences over time within the same individuals (Brüderl and Ludwig, Reference Brüderl, Ludwig, Best and Wolf2015). This conservative approach is a significant advantage, since it significantly reduces the risk of biased estimates, reduces the risk of a type-I error and, if the assumption that all time-varying confounders are controlled is fulfilled, consistent estimates can be calculated (Brüderl and Ludwig, Reference Brüderl, Ludwig, Best and Wolf2015; Wooldridge, Reference Wooldridge2010).
In the main models, we analyzed transitions into caregiving inside and outside the household in association with suicidal ideation, while adjusting for time-varying sociodemographic and health variables, based on previous research (Favril et al., Reference Favril, Yu, Uyar, Sharpe and Fazel2022). The exception were time-constant variables (e.g. gender), which were already controlled for by FE regression. Further analyses included an interaction effect between caregiving and gender, respective, age. Additionally, the main predictor of caregiving was replaced by a variable measuring the transition into a specific relationship. The main models were calculated again after stratifying the sample by welfare system. Also, two moderator analyses were conducted, which included an interaction effect between caregiving and welfare system with the smallest and the largest category as reference. These analyses and the model analyzing the associations in the Scandinavian sample included a dichotomized variable of marital status (relationship yes/no) and a trichotomized variable of employment status (retired, employed, unemployed). The full tables are provided in the Supplementary Material (Tables A1 and A2).
We did not control for depression in the main analyses, because the item that measured suicide ideation was part of the depression instrument. High correlations between these variables and consequently biased results would have been likely. Instead, self-rated health and chronic diseases were controlled and a sensitivity analysis with an adapted measure of depression (without the item measuring suicide ideation) was conducted (see Supplementary Material, Tables A3 and A4).
Missing values were below 5% in all variables (see Table A5, Supplementary Material). Low levels of missing values usually do not bias the results significantly; thus, it is recommended not to use imputation in this case (Allison, Reference Allison2001; Van Buuren, Reference Van Buuren2012). Moreover, FE reduces the danger of bias due to unobserved time-constant variables, which reduces the danger of bias due to missing values further (Brüderl and Ludwig, Reference Brüderl, Ludwig, Best and Wolf2015; Brüderl, Reference Brüderl, Wolf and Best2010; Wooldridge, Reference Wooldridge2010).
The level of significance was set at alpha level of 0.05. All analyses were conducted with Stata 16.1 (Stata Corp., College Station Texas).
Results
Descriptive results
In Table 1, we describe the complete analytical sample (Obs. = 171,848) and the two caregiver samples (inside the household, Obs. = 10,148, or outside the household, Obs. = 44,998), pooled over all five waves. Of the complete sample, 23.15% of the participants lived in a Southern welfare system, 59.03% in a Bismarckian welfare system, and 17.82% in a Scandinavian welfare system. Of the complete sample, 26.18% provided care outside the household. Caregivers outside the household were on average 62.89 years (SD = 8.83) of age, 56.48% were female, and 45.17 % were retired. Of those participants living not alone, 7.74% were caregivers inside their households. The majority of them provided care for partners (64.52%). Caregivers inside the household were on average 67.74 (SD = 10.85) years old, 58.28% were female, and 54.88% were retired.
Description of the complete sample (Obs. = 171,848), the sample of caregivers inside the household (Obs. = 10,148) and outside the household (Obs. = 44,998) pooled over all five waves
Obs. = Observations; ISCED 1997 = International Standard Classification of Education 1997; self-rated health [1 excellent to 5 poor health, single-item from the SF-36 questionnaire (Ware and Gandek, Reference Ware and Gandek1998)]; number of chronic diseases (sum score of diseases, which had ever been diagnosed by a physician (theoretical range: 0–14) including chronic diseases: heart attack, high blood pressure or hypertension, high blood cholesterol, stroke, diabetes or high blood sugar, chronic lung disease, asthma, arthritis, osteoporosis, cancer, stomach or duodenal ulcer, peptic ulcer, Parkinson’s disease, cataracts, hip fracture or femoral fracture), mean, and standard deviation are given for the continuous variables; frequency and percentage are given for the categorical variables.
a Only those who were not living alone were asked if they were providing care inside the house.
b 10.28% of the complete sample had missing information regarding caregiving because they were living alone and therefore not asked about care provision inside their household.
Suicidal ideation was reported by 6.76% of the complete sample, 9.53% of caregivers inside the household, and 5.92% of caregivers outside the household.
In total, 4,165 transitions into caregiving inside the household and 9,667 transitions into caregiving outside the household were observed. The number of transitions into different caregiving relationships and within each welfare system can be found in Table A6 (Supplementary Material).
Results of regression and moderator analyses for care inside the household
Transitioning into caregiving inside the household (Table 2) was significantly associated with increased odds of suicidal ideation by 36% (OR = 1.36. CI [1.18; 1.58], model 1). Age and gender did not significantly moderate this association (models 2 and 3). Transitioning into care for partners (OR = 1.36, CI [1.15; 1.62]) and for parents (OR = 1.67, CI [1.11; 2.50]) was significantly associated with increased odds for suicidal ideation (model 4). Stratification by welfare system indicated that the odds of suicidal ideation were significantly increased among those transitioning into caregiving inside the household in a Southern welfare system (OR = 1.53, CI [1.22; 1.91], model 5) and in a Bismarckian welfare system (OR = 1.24, CI [1.02; 1.51], model 6). No significant association was found in the Scandinavian welfare system (model 7). Moderator models showed no significant differences when using Southern (model 8) and Bismarckian welfare systems (model 9) as references for comparison.
Results of fixed effects regression analyses with informal caregiving inside the household as main predictor
Odds ratio and 95% confidence intervals are provided in the table. All models are adjusted for age, marital status, employment status, self-perceived health, and number of chronic diseases; the moderator analyses including the moderator gender, age or welfare system are also adjusted for these variables. Main effects of gender and welfare systems were omitted due to the lack of transitions. Levels of significance: ***p < 0.001, **p < 0.01, *p < 0.05.
a Models 7, 8, and 9 included a dichotomized variable of marital (married or in a registered relationship vs. not married, widowed, or divorced) and a trichotomized variable of employment status (retired, employed, or unemployed) to account for the small sample and the low number of transitions in the Scandinavian welfare system.
Results of regression and moderator analyses for care outside the household
Transitioning into caregiving outside the household (Table 3) did not significantly change the odds of suicidal ideation (model 1, OR = 1.09, CI [0.99; 1.20]). Gender (OR = 1.23, CI [1.00; 1.50], model 3) and age (OR = 0.99, CI [0.98; 1.00], model 4) moderated the association, indicating higher odds of thoughts of suicide among female caregivers and lower odds of thoughts of suicide among all caregivers with higher age. Significantly increased odds for thoughts of suicide were only found among those transitioning into care outside the household for non-relatives (OR = 1.16, CI [1.02; 1.31], model 5). Stratified analyses by welfare system indicated a significant association with increased odds of suicidal ideation by 32% among those transitioning into caregiving in the Southern welfare system (OR = 1.32, CI [1.07; 1.64], model 5). The association was not significant in the Bismarckian or the Scandinavian system (models 6 and 7). The Southern welfare system differed significantly in this association (model 8) from the Bismarckian (OR = 0.74, CI [0.58; 0.95]) and the Scandinavian welfare systems (OR = 0.70, CI [0.51; 1.00]). This was also found in the second moderator analyses with the Bismarckian welfare as reference category (Southern OR = 1.35, CI [1.06;1.72]), while the Scandinavian and Bismarckian systems did not differ significantly (model 9).
Results of fixed effects regression analyses with informal caregiving outside the household as main predictor
Odds ratio and 95% confidence intervals are provided in the table. All models were adjusted for age, marital status, employment status, self-perceived health, and number of chronic diseases; the moderator analyses including the moderator gender, age or welfare system were also adjusted for these variables. Main effects of gender and welfare systems were omitted due to the lack of transitions. Levels of significance: ***p < 0.001, **p < 0.01, *p < 0.05, +p < 0.10.
a Models 7, 8, and 9 included a dichotomized variable of marital (married or in a registered relationship vs. not married, widowed, or divorced) and a trichotomized variable of employment status (retired, employed, or unemployed) to account for the small sample and the low number of transitions in the Scandinavian welfare system.
Discussion
This longitudinal study provides new evidence to previous research, which had rarely used longitudinal designs to study the association between informal caregiving and suicidal ideation and had not been able to provide information on biopsychosocial risk or protective factors specific to the caregiving situation (O’Dwyer et al., Reference O’Dwyer2021). Our findings fill these research gaps by showing that care-specific aspects including the location of caregiving, the caregivers’ age and gender, their relationship to the care recipient, and their welfare system are significant for suicidal ideation among informal caregivers.
Adults who transitioned into caregiving inside the household had higher odds of suicidal ideation, which was not found among those transitioning into caregiving outside the household. This confirms our first hypothesis regarding the caregiving location and strengthens former findings of coresidential caregiving being more problematic (Kaschowitz and Brandt, Reference Kaschowitz and Brandt2017). In line with the IMV theory (O’Connor et al., Reference O’Connor, Cleare, Eschle, Wetherall and Kirtley2016), caregivers inside the household may feel more trapped than those who leave the house to provide care. The lack of physical distance may provide fewer opportunities for respite. Caregivers outside the household may be involved in less personal care tasks, and therefore, suicide ideation worsened only among caregivers inside the household. In our questionnaire, caregiving inside focused only on personal care, which is the most stressful care task (Lopez-Anuarbe and Kohli, Reference Lopez-Anuarbe and Kohli2019).
We could not confirm our second set of hypotheses on the significance of the caregiver’s characteristics for care inside the household. Although these caregivers were mostly characterized by risk factors of suicidal ideation in Europe (Carrasco-Barrios et al., Reference Carrasco-Barrios2020), i.e. on average older and most were female, the association was not significantly influenced by age or gender. However, the care relationship was relevant. Only care for partners and for parents was associated with higher odds of suicidal ideation, which confirms our hypothesis. Interestingly, the odds among those looking after parents were twice as high as among those looking after partners inside their household. These findings are not in line with findings of worse mental health among spousal compared to adult-child caregivers (Pinquart and Sorensen, Reference Pinquart and Sorensen2011). While the majority of inside caregivers support partners, there are still 14% providing care for their parents in their household. This emphasizes the need to analyze depressive symptoms separately from suicidal ideation and take the location of caregiving into account. Sharing a household with care-dependent parents may present additional challenges that are particularly relevant to suicidal ideation. In contrast to our hypothesis, age and gender were of relevance for the association between caregiving outside the household and suicide ideation. Caregivers outside the household had lower odds of suicidal ideation with increasing age. This can be seen as positive for older caregivers, since suicide risk is usually higher among older adults (Carrasco-Barrios et al., Reference Carrasco-Barrios2020). Caregiving outside the household can counteract loneliness, in particular among older adults (Hajek and König, Reference Hajek and König2020), by providing meaning to life and social interaction (Dahlberg et al., Reference Dahlberg, Mckee, Frank and Naseer2022), and thereby prevent suicidal ideation. However, it may also reflect that older adults are less likely to report suicide thoughts than those of younger age (Lee, Reference Lee2023; Nock et al., Reference Nock2008).
Women had higher odds of suicidal ideation when transitioning into caregiving outside the household. Women usually provide more personal care (Pinquart and Sörensen, Reference Pinquart and Sörensen2006) and may do this inside as well as outside the household. Thus, female caregivers outside the household may feel a similar entrapment as caregivers inside.
Another relevant factor for caregivers outside was the care relationship. In contrast to our hypothesis, only caregivers for non-relatives outside the household had higher odds of suicidal ideation. Caregivers of non-relatives may have a very close relationship with the care recipients and provide care due to a lack of available family caregivers, thus, resulting in a similar situation without alternatives as has been found among family caregivers inside the household. Caregiving to non-relatives and the consequences for them has not received as much investigation, despite the increase of this group in recent years (Klaus and Ehrlich, Reference Klaus and Ehrlich2021). Further investigation focused on them is recommended.
Last, we investigated caregiving in three European welfare systems. As hypothesized, odds for suicidal ideation were significantly higher among adults transitioning into caregiving inside and outside the household in the Southern welfare system. The Southern welfare system is still developing, in particular, the long-term care system currently relies mostly on the family (Greve, Reference Greve2022). Caregivers in this system may be aware that there are no (affordable) alternatives to informal caregiving, which can foster the feeling of being trapped in their caregiver role and not perceiving any improvement in their situation in the (near) future (O’Connor et al., Reference O’Connor, Cleare, Eschle, Wetherall and Kirtley2016). We also found higher odds of suicidal ideation among caregivers inside, but not outside the household, in the Bismarckian system. This is also a conservative system, relying mostly on family as caregivers; however, it has better social security options, which can improve the general situation if they are used (Greve, Reference Greve2022). Thus, the care systems in both Bismarckian and Southern welfare systems are apparently not providing enough support to prevent suicidal ideation among caregivers inside the household. The Southern system in particular needs to improve their support options to prevent suicidal ideation among caregivers inside and outside the household.
We compared our findings with the odds ratio calculated by Chen et al. (Reference Chen, Cohen and Chen2010) as a benchmark similar to Cohen’s d to interpret the size of our effects. We use the lifetime prevalence of suicide ideation among a cross-national European adult sample as disease rate of the nonexposed population (i.e. non-caregivers) (5.5%, Castillejos et al., Reference Castillejos, Huertas, Martin and Moreno Kuestner2021). Accordingly, odds ratios of 1.52, 2.74, and 4.72 indicate small, medium, and large effects. Our findings therefore indicate small effects. Still, a broad range of previous research has indicated the vulnerability of informal caregivers (Bom et al., Reference Bom, Bakx, Schut and Van Doorslaer2019) and suicide ideation is a major risk factor for suicide behavior (O’Connor et al., Reference O’Connor, Cleare, Eschle, Wetherall and Kirtley2016; Forkmann, Reference Forkmann2021). Thus, even small effects are of relevance in the context of informal caregiving and add to aforementioned literature indicating the risks of informal caregiving for health and well-being of caregivers.
There are a few limitations to the study. Sensitivity analyses indicated that depressive symptoms seem to account for age and gender differences formerly attributed to caregiving. However, the other effects remained significant even if their size was smaller. Thus, caregiving contributes to changes in suicide ideation in addition to mental health.
We used a dichotomized single-item to measure suicidal ideation, although a multi-item measure would be preferred due to its psychometric advantages (Sarstedt and Wilczynski, Reference Sarstedt and Wilczynski2009). However, using single-item measures is an established method to analyze suicidal ideation (O’Dwyer et al., Reference O’Dwyer2021). Also, we measured passive suicide ideation (i.e. wishing one was dead), which is to be differentiated from active suicide ideation (e.g. thinking about ending one’s life, Wastler et al., Reference Wastler2023). Thus, results might not generalize to more severe forms of suicide ideation. Last, selection bias could be problematic but is quite small in SHARE data (Börsch-Supan et al., Reference Börsch-Supan2013), and the FE regression method reduces the bias due to time-constant variables further.
Conclusion and implications
Our findings contribute innovative and important new insights about the informal caregiving situation and its association with suicidal ideation. The findings indicate that especially caregivers inside the household are vulnerable to suicide ideation. Our findings are of particular relevance for older groups and individuals with lower socioeconomic status, who are more likely to be involved in caregiving (Quashie et al., Reference Quashie, Wagner, Verbakel and Deindl2022; Rothgang and Müller, Reference Rothgang and Müller2018) despite higher risk and underreporting suicide thoughts (Kim et al., Reference Kim, Ha, Yu, Park and Ryu2014; Pitkälä et al., Reference Pitkälä, Isometsä, Henriksson and Lönnqvist2000). Our biopsychosocial approach showed that individual factors, such as age and gender, and social aspects, such as the caregiver relationship, are of relevance for the association. Moreover, contextual factors, such as the location of caregiving and the welfare system, need to be taken into account, too. This adds to previous research, which had not analyzed the context of caregivers in relation to suicide ideation.
Suicidal ideation was more likely among caregivers inside the household and caregivers in Southern (and Bismarckian) welfare systems. Additionally, primarily caregivers of parents inside their own household showed higher suicidal ideation, as well as younger and female caregivers outside the household. Thus, more care options that allow to provide care for parents outside, instead of inside the household, would be helpful. As our study shows, inside-household caregivers are mainly spousal caregivers and a small but (in regard to suicidal ideation) highly vulnerable group of parental caregivers. Spousal caregivers are usually less inclined to look for and be provided with support (Pinquart and Sorensen, Reference Pinquart and Sorensen2011; Zwar et al., Reference Zwar, König and Hajek2022). More outreach and low-threshold services could therefore help to support these hidden groups in their seemingly hopeless situation and prevent the development of suicidal ideation.
Also, aspects of poverty prevention and a broader range of affordable ambulatory care options could help with this. The most vulnerable caregivers were indicated to be part of the Southern welfare systems, which are not well positioned with regard to these social security measures. However, developments in recent years give reason to hope that further improvements will be made in this direction (Greve, Reference Greve2022), which are urgently needed to prevent the increased suicide ideation risks found among these caregivers.
Moreover, support for non-relative caregivers, in particular younger and female caregivers, seems to be needed. Often caregiving support options refer only to family caregivers; however, there is a growing group of non-related caregivers outside the household (i.e. friends and neighbors) (Klaus and Ehrlich, Reference Klaus and Ehrlich2021). Likely, the increasing group of singles and widowed women with care needs draws upon support from them. Our findings further emphasize the need to direct more attention in terms of research and support to these caregivers.
Conflict of interest
None.
Description of authors’ roles
LZ contributed to conception, design, and analysis of the data and drafted the manuscript. HHK and AH contributed to review and editing. All authors read and approved the final manuscript.
Acknowledgments
This study uses data from the Survey of Health, Ageing, and Retirement in Europe (SHARE), release 8.0.0 (10th Feb. 2022) of waves 1 (DOI: 10.6103/SHARE.w1.800), 2 (DOI: 10.6103/SHARE.w2.800), 4 (DOI: 10.6103/SHARE.w4.800), 5 (DOI:10.6103/SHARE.w5.800), and 6 (DOI: 10.6103/SHARE.w6.800). Data are available free of charge for registered users at http://www.share-project.org.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S1041610223000601
Open access
