Compassion in three perspectives: Associations with depression and suicidal ideation in a clinical adolescent sample
Article Category: Research Article
Published Online: Dec 07, 2023
Page range: 120 - 127
DOI: https://doi.org/10.2478/sjcapp-2023-0012
Keywords
© 2023 Emmy Jonsson et al., published by Sciendo
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
Mental health among adolescents is a major public health concern (1). In recent decades, research has indicated a steady increase in the prevalence of depressive disorders among adolescents (2), especially in adolescent girls (2–4), and this trend seems to be most significant in Northern Europe, especially in the Scandinavian countries (4).
There are many known complications of adolescent depression. Adolescence is a critical period for both physical and mental development, and during this time adolescents need to develop social and emotional skills such as coping, problem-solving, and interpersonal skills, as well as healthy sleeping, eating, and exercising habits (1). The development of these skills and habits, which are all important for mental well-being, can be disrupted by the burdens of depression. Depression in adolescents may also increase the risk of mental health complications later in life (5), and one of the most serious consequences of depression is suicide.
In Sweden, suicide makes up a third of all deaths in the age-group 15–24 years old (6). There are many known risk factors for suicide, but one of the most important risk factors is mental disorders (7). 90% among people who have commit suicide, have suffered from at least one mental disorder (8). Mental disorders could be for example eating disorders and affective disorders, of which depressive disorders are the most important and have been found to constitute up to 74% of suicide risk (8, 9). Depression is also the major contributor to suicide worldwide (10). However, in addition to the known risk factors for suicide, there are also some known protective factors. These include social connectedness, the quality and density of relationship ties, and learned skills such as assertiveness, empathy, and the ability to manage stress (11). Another factor that may have a protective effect against suicidal ideation is compassion (12, 13).
Compassion has in some instances been defined as an ability to recognise suffering in self and others with a commitment to try and alleviate and prevent it (14). Most previous studies have only focused on one perspective of compassion, namely “self-compassion”, which is the compassion we show and feel towards ourselves. These studies have found self-compassion to be strongly associated with mental well-being in both adolescents and adults (15, 16) and that there is a significant relationship between self-compassion and psychopathology, primarily depression, anxiety, and stress (17–20). However, there have recently been studies exploring other perspectives of compassion as well, including the three perspectives of compassion: self-compassion, compassion for others, and compassion from others (14, 21).
This study explored three perspectives of compassion – self-compassion, compassion for others, and compassion from others (14) – and their relationship to depression and suicidal ideation in a clinical adolescent sample (ages 16–22). The aim was to investigate whether the participants’ levels of compassion significantly correlated with levels of depression and suicidal ideation. The study also included known risk factors for depression – age, gender, and socioeconomic status (SES) – to see whether they would moderate the outcome of the analyses.
This study was conducted as part of the research project “Adolescents’ experience of mental illness – Psychometric properties of new Swedish versions of tests (UPOP).” The study has received ethical approval from the Swedish Regional Ethical Review Boards in Umea (Ref. no. 2018/59-31).
Data were collected through a survey that was available both online and on paper. Participants were recruited from two Child and Adolescent Psychiatric (CAP) clinics in Northern Sweden as well as from primary care. The survey took around 30–60 minutes to complete, and a gift card for around 18 euro (SEK 200) was given to participants as a reward for participating. All participants received both verbal and written information about the research project, and written consent was obtained from all participants. Recruitment was handled by research assistants.
Around 870 adolescents (ages 12–22) were asked to participate in the larger research project, and N = 549 (63.1%) agreed to participate. Out of the 549 respondents, 204 participants (ages 12–15) were removed from the present study due to their version of the survey not containing the compassion questionnaire. Another 68 participants (participants answering the survey on paper) were also removed due to the paper version not containing the suicide questionnaire. Lastly, another 18 participants were removed due to missing data.
The final sample consisted of N = 259 adolescents (219 girls, 40 boys) ages 16–22 (M = 18.08, SD = 2.08). All participants were Swedish speaking, and most were Swedish born (n = 244, 94.2%). The distribution of socioeconomic classification is presented in Table 1.
Gender differences in mean total scores of RADS-2, SIQ-JR, and the three CEAS-Y scales, independent-samples t-test, and effect size.
| Total (N = 259) M (SD) | Girls (n = 219) M (SD) | Boys (n = 40) M (SD) | 95% CI | p | Cohen’s d | |
|---|---|---|---|---|---|---|
| RADS-2 | 78.27(16.22) | 80.13(15.57) | 68.13(16.13) | 6.70; 17.30 | <.001 | .767 |
| SIQ-JR | 23.62(21.07) | 24.78(21.82) | 17.25(15.03) | 1.96; 13.10 | .009 | .360 |
| Self-compassion | 53.85(15.83) | 53.01(15.75) | 58.43(15.69) | -10.75; -.09 | .046 | -.344 |
| Compassion for others | 83.49(13.50 | 84.62(13.01) | 77.30(14.61) | 2.83; 11.81 | .001 | .552 |
| Compassion from others | 59.75(18.87) | 59.87(19.33) | 59.08(16.30) | -5.60; 7.20 | .806 | .042 |
Note. M = mean; SD = standard deviation; CI = confidence interval; RADS-2 = Reynolds Adolescent Depression Scale – Second Edition; SIQ-JR = Suicidal Ideation Questionnaire – Junior Version.
RADS-2 is a self-assessment questionnaire that measures four dimensions of depression: 1. Dysphoria, 2. Anhedonia/negative affect, 3. Negative self-evaluating, and 4. Somatic complaints. The questionnaire contains 30 items on a 4-point scale with a possible raw score ranging from 30 to 120. The items consist of brief self-statements with answers ranging from “Almost never” to “Most of the time”. Higher scores indicate a higher severity of depressive symptoms. Reynold’s recommend a cut-off total raw score of 76 (22); however, an alternative cut-off raw score of 67 has been presented (23). The scale has demonstrated strong psychometric properties and has proven to be a reliable instrument for interpreting the severity of depressive symptomology (23), including in Sweden (24). In the present study, Cronbach’s alpha was 0.93 (95% CI [0.91 – 0.94]). This study only used the total raw score as a measurement of depressive symptoms.
The SIQ-JR is a self-assessment questionnaire used for measuring suicidal ideation in adolescents. The scale contains 15 items on a 7-point scale with a possible raw score ranging from 0 to 90. The grading indicates how often a certain thought related to suicidality has occurred in the individual answering the questionnaire, ranging from “Never” to “Almost every day”. Higher scores indicate a higher severity of suicidal ideation. A clinically significant level of suicidal ideation is reflected by a cut-off raw score of 31, although a more liberal alternative cut-off score of 23 is also presented in the SIQ professional manual (25). Recently, a cut-off score of 20 has also been suggested (26). Previous studies have shown it to be reliable (27), and to be significantly associated with a history of suicide attempts (28). Cronbach’s alpha was 0.96 (95% CI [0.95–0.97]) in the present study. This study only used the raw total score as a measurement of suicidal ideation.
CEAS-Y is a self-assessment questionnaire that measures compassion in three scales: 1. Self-compassion, 2. Compassion for others, and 3. Compassion from others. Each of the scales contains two subscales that measure compassion in two different orientations: 1) engagement with distress/suffering (items E1, E2, E5, E6, and E8), and 2) actions aimed to prevent and alleviate distress/suffering (items A1, A2, A4, and A5). Each scale contains 13 items on a 10-point scale. Four items, however, do not contribute to the final score, three of them (items E3, E7, and A3) for being inverted, and one (item E4) for not achieving sufficient validity in the Swedish version of the questionnaire (21). The items consist of brief self-statements with answers ranging from “Never” to “Always”, with possible raw scores ranging from 9 to 90. Higher scores indicate higher levels of compassion. CEAS-Y has previously been validated and has demonstrated good psychometric properties in Sweden (21). In the present study, Cronbach’s alpha was 0.89 (95% CI [0.86–0.91]) for Self-compassion, 0.91 (95% CI [0.89 –0.92]) for Compassion for others, and 0.92 (95% CI [0.91– 0.94]) for Compassion from others.
The survey included two questions where the participants were asked to state their age and legal gender (female/male). SES was measured using the following question “What does your parent/legal guardian do for a living”, and the answers were then categorized into socioeconomic groups according to the Socioeconomic Classification Index, which is a socioeconomic classification system constructed by Statistics Sweden that consists of the following seven larger groups: 1) non-economically active, 2) workers with less than 2 years of education such as cleaners and service workers, 3) workers with more than 2 years of education but less than 3, such as painters, 4) assistant non-manual employees such as administrators, 5) employed (e.g. teachers and nursers) and self-employed professionals, 6) higher civil servants (e.g. jurists and physicians) and executives, and 7) the self-employed (other than professionals) (29). For the sake of simplicity, these answers were converted into three categories: 1) low socioeconomic status (groups 1–3), 2) middle socioeconomic status (groups 4, 5 and 7), and 3) high socioeconomic status (group 6).
Self-reported descriptive information regarding the participants’ contact cause was obtained using the following question: “If you have been in contact with CAP or primary care, describe with a few words your contact cause (i.e., depression, ADHD)”.
A subset of the total sample also had accessible medical journals from which information regarding psychiatric diagnoses was obtained. Research assistants collected the information around six months after the web survey. Physicians or psychologists had assessed and decided the diagnoses in the medical journals. The diagnoses were set 0-154 days before or after the web survey, M = 19.57, SD = 25.39). Both primary and secondary diagnoses were obtained.
Data checks and screening were first utilized. Descriptive statistics were calculated.
Mean differences and effect sizes between girls and boys on RADS-2, SIQ-JR and CEAS-Y total scores, were calculated with independent-sample t-tests. Effect sizes were interpreted according to Cohen’s
We used a MANOVA to calculate the possible SES differences in depression and suicide ideation.
Tests of assumptions were calculated. Continuous variables were checked for normality using histograms and scores for skewness and kurtosis. We investigated the assumptions of linearity and homoscedasticity with residual p-plots and scatterplots.
Multiple linear regression analyses were conducted to test if the independent variables significantly predicted participants’ ratings of depression symptoms (RADS-2) and suicidal ideation (SIQ-JR). SES was not included in the regression analysis due to missing data. Statistical analyses were conducted using SPSS version 28.0.
The primary contact causes were as follows (response rate: 83.4%, n = 216): 139 participants searched for affective concerns including depression, 40 searched for ADHD or ADD, 15 searched for eating disorders or problems, seven searched for autism concerns, 6 searched for trans-related concerns, four searched for compulsive disorder concerns, and five searched for other concerns. 136 participants (52,5%) out of 216 answers, searched for multiple concerns (comorbidity).
The primary diagnoses for a subset of the total sample (n = 93, 35.9%) were as follows: 55 had received an affective diagnosis including depression, 17 had ADHD or ADD, five had autism syndrome, six had eating diagnoses, five had compulsive diagnoses, four had trans-related diagnoses, and one had no diagnosis. 49 participants (52.7%) out of 93 patients, had received a second diagnosis, including depression. A summary of both primary and secondary diagnoses were that 52.7% of the participants (n = 49) had received a depressive disorder, while 71% (n = 66) had received a depressive or an anxiety disorder.
Girls, on average, reported a significantly higher level of depressive symptoms than boys, see Table 1.
Girls, reported higher levels of suicidal ideation than boys, see Table 1.
Girls, reported lower levels of self-compassion and higher levels of compassion for others than boys, see Table 1.
SES was classified as low, middle and high SES. 42 participants were classified as belonging to low SES, 89 were classified as middle SES and 93 were classified as high SES. 35 (13.5% of the total sample) participants had unknown SES.
A MANOVA was performed to investigate possible differences in depression and suicidal ideation depending on SES. Two dependent variables were used – total raw scores for RADS-2 and total raw scores for SIQ-JR. The independent variable was SES divided into low, middle and high socioeconomic status. There was no statistically significant difference between SES on the combined dependent variables (F (4, 442) = .707, p = .588, Pillai´s Trace = .013, partial eta squared = .006). There was also no significant difference for either variable when results for the dependant variables were considered separately.
Tests of assumptions were calculated before conducting the regression analyses. Scores for RADS-2, SIQ-JR, Self-Compassion, and Compassion for others were skewed, but the sample size was large enough that violations of this assumption were tolerated (31). Residual p-plots and scatterplots indicated that the assumptions of linearity and homoscedasticity were satisfied. Independent variables showed no significant multicollinearity, with no Variance Inflation Factor above 2.
The results of the regression indicated the predictors explained 35% of the variance (Adj.R2 = 0.35; F (5, 253) = 28.97; p < 0.001). Gender, Self-compassion, and Compassion for others significantly predicted depressive symptoms, see Table 2. Because gender predicted higher RADS-2 scores, separate analyses were performed for girls and boys. Age made no statistically significant contribution to the model once other variables were added. Compassion for others showed no statistically significant correlation with depressive symptoms (see Table 2).
Results from multiple regression for RADS-2 and SIQ-JR for the total sample.
| Depression (RADS-2) | Suicidal ideation (SIQ-JR) | |||
|---|---|---|---|---|
| Adj.R2 | β | Adj.R2 | β | |
| Model | .352*** | .170*** | ||
| Age | -.077 | -.083 | ||
| Gender1 | -.216*** | -.081 | ||
| Self-compassion | -.360*** | -.284*** | ||
| Compassion for others | .096 | .119* | ||
The results of the regression indicated the predictors explained 17% of the variance (Adj.R2 = 0.17; F (5, 253) = 11.56; p < 0.001). Self-compassion, Compassion for others, and Compassion from others significantly predicted suicide ideation. Age and gender were not significant, see Table 2.
The results of the two regression analyses showed that the model for boys could explain more variance in depressive symptoms (Adj.R2 = 0.41; F (3, 36) = 10.06; p < 0.001) than the model for girls (Adj.R2 = 0.28; F (3, 215) = 28.84; p < 0.001). See Table 3 for results.
Results from multiple regression for RADS-2 for girls and boys separately.
| Girls (n= 219) | Boys (n= 40) | |||
|---|---|---|---|---|
| Adj.R2 | β | Adj.R2 | β | |
| Model | .277*** | .411*** | ||
| Self-compassion | -.354*** | -.490** | ||
| Compassion for others | .083 | .171 | ||
| Compassion from others | -.296*** | -.310* | ||
Self-compassion predicted the outcome in all analyses, see Tables 2 and 3. The explained variance was larger for depressive symptoms than for suicidal ideation and was larger for boys compared to girls. Both Self-compassion and Compassion from others had a negative correlation with the outcome in all analyses, while Compassion for others showed a positive correlation. This positive correlation was only statistically significant for the outcome for suicidal ideation and not for depressive symptoms.
The results from this study support previous research where self-compassion have been linked to mental well-being (15–17), and in particular where lower levels of self-compassion have been associated with depression (14, 18–20). The findings from this study also support what is already known about gender differences in depressive symptoms, suicidal ideation, and compassion. The prevalence of depressive disorders is known to be higher in females than males (10), and while suicide deaths may be more common in men, suicidal ideation and suicide attempts are more common in women (32, 33). In many previous studies, girls have also generally reported lower levels of self-compassion compared to boys (19–21, 34), which is all reflected by the findings of this study.
Interestingly though, while girls reported both higher levels of depressive symptoms and lower levels of self-compassion, the correlation between depressive symptoms and low self-compassion was stronger in boys. This may suggest that self-compassion has a stronger protective effect against depressive symptoms for boys than for girls, and there are some previous studies that may support this theory. One previous study found the protective effect of self-compassion on anxiety symptoms to be greater in boys compared to girls (34), and another found the relationship between self-compassion and resilience, with resilience being defined as “the positive adaptation to adverse life experiences”, to be stronger in males than in females (35).
In addition to self-compassion, Compassion from others also had a similar relationship to both depressive symptoms and suicidal ideation. Being able to receive social support and compassion from others is commonly regarded as an essential resource for maintaining mental well-being, and there is a known association between poor social support and the onset and relapse of depression (36). Research has shown that the ability to receive compassion from others buffers symptoms of depression, anxiety, and stress (37, 38) and that compassionate social support has a significant impact on resilience to stress and on the outcomes of numerous medical illnesses (39). As for the association with suicidal ideation, research shows that social support has a great impact in reducing suicidal behaviour (40), and the Swedish National Board of Health and Welfare also recognize the importance of social support and compassionate treatment of suicidal patients (41). All of this supports the findings of this study where higher levels of Compassion from others correlated with lower levels of depression and suicidal ideation.
The only perspective of compassion that did not negatively correlate with the dependent variables of this study was Compassion for others. This is consistent with the findings of Gilbert (14), where Compassion for others displayed low or non-significant correlations with depression, anxiety, and stress. Interestingly though, in this study there was a significant positive correlation between Compassion for others and suicidal ideation. While the findings of this study are insufficient to draw any conclusion regarding this correlation, it is still possible to speculate why this happened.
There is an important distinction between “pathological” compassion and “healthy” compassion. Pathological compassion, or concern, can be defined as favouring the needs of others over the needs of yourself, which may be associated with feelings of unfairness, fragile self-image, depression, and anxiety (42). The combination of low Self-compassion and high levels of Compassion for others, as seen in this sample, may be an indicator for “pathological” compassion, which based on the findings of this study may be associated with suicidal ideation. In order to draw any solid conclusions regarding these findings, further research is needed.
In recent years, several types of psychotherapies focusing on compassion have been introduced (43–47). The effectiveness of psychotherapies for the treatment of mental disorders has in many ways been well-established, and the Swedish National guidelines for treatment of depressive disorders in children and adolescents from the Swedish National Board of Health and Welfare recommend psychotherapy as a treatment with the same level of recommendation as antidepressant medication (48). They also identify psychotherapies, with examples such as Dialectical Behavioural Therapy and Acceptance and Commitment Therapy, as treatments of choice for suicidal ideation (41). There are, however, areas where the current treatments for depression and suicidal ideation are known to be lacking. There is still a significant number of patients who show no improvement with current treatment strategies for depression and anxiety (49), and there are studies that highlight the need for improvement (50, 51). One of the main issues is that a large number of patients receive no treatment at all (52). For those who do receive treatment, research suggests that different types of psychotherapies may be effective treatments (40, 53–55); however, as of now, there are no Swedish national guidelines for specific treatments of suicidality.
As for compassion-focused psychotherapies, there are studies indicating that these could be effective in the treatment of depressive disorders (45–47, 56–58), while the research on their effectiveness on suicidal ideation is lacking. However, with further research proving significant associations between compassion and suicidal ideation, this may be a field worth exploring.
One strength of this study includes that we explored three different perspectives of compassion, instead of just self-compassion. A second strength is that we used reliable and valid self-report scales as variables. We also included other known risk factors in the analyses to examine how much of the variance could be attributed to them. Another strength is that we used a clinical sample. Finally, we investigated the correlation between compassion and suicidal ideation, which has been quite unexplored. However, we also acknowledge some limitations within the present study.
The cross-sectional design of this study limits the ability to draw valid conclusion about the causality between the dependent and independent variables. There are also limitations regarding the independent variables of age, gender, and SES. The recommended number of cases for reliable equations when performing regression analyses is about 15 cases per predictor (59), which would equal a minimum of n = 45 cases for the separate gender analyses. Gender made a statistically significant contribution to the regression analysis for depressive symptoms, but any conclusions based on this should be made with caution considering the limited sample of boys.
SES did not achieve any statistical significance, for which there could be various reasons. The stratification into social economic classes was based solely on the occupation of the participants’ parents or legal guardians. There were no income disclosures or disclosures of overall wealth available. This limits the accuracy of which any valid conclusion can be drawn regarding the economy of the participants’ households, as well as actual SES. Because the sample was not geographically stratified, it is also possible that there was simply not enough underlying variance of SES for it to impact the outcome of the analyses.