Background
Methods
Study inclusion
Study selection
Data extraction
Quality appraisal
Analysis
Results
Included studies
Study characteristics
Author, year, country | Study design Sample source | Sample demographics | Mode of bereavement Time since bereavement | Social support measures | Measured outcomes |
---|---|---|---|---|---|
Bailey, 2013, Canada [29] | Cross-sectional Community organisation | n=48 mean age=51.5 100% female | Child lost to gun violence 0.5-12 years | Multidimensional Scale of Perceived Social Support (MSPSS) [30] | Resilience |
Bottomley, 2017, U.S.A. [31] | Longitudinal (6 month follow-up) Support organisation | n=47 mean age=49.7 89.4% female | Family member lost to homicide Mean length= 1.66 years at T1, 2.16 at T2 | Arizona Social Support Interview Schedule (ASSIS) [32] | PTSD, complicated grief, depression, anxiety |
Burke, 2010, U.S.A. [33] | Cross-sectional Support organisation | n=54 mean age=48.6 88.9% female | Family member lost to homicide Mean length= 1.75 years | ASSIS, Inventory of Social Support (ISS) [34] & MSPSS | PTSD, complicated grief, depression |
Cowan, 1985, U.S.A.a [27] | Cross-sectional Death certificates/ court records | n=119 (50 control) mean age=unclear 70% female | Friends and family lost in natural disaster Mean length= .92 years | Coppel Index of Social Support (CISS) [35] | Depression |
Murphy, 1988, U.S.A.a [28] | Cross-sectional Death certificates/ court records Official population records | n= 49 (bereaved) /36 (control) mean age= 30/37 74%/65% female | Friends and family lost in natural disaster Mean length= 3 years (estimate) | CISS | Mental distress, recovery |
Fullerton, 1999, U.S.A. [36] | Cross-sectional Air force squadron | n=71 mean age=33 4.0% female | Squadron members of personnel lost in plane crash Mean length= 0.17 years | Perceived Social Support Scales (Family and Friends) [37] | Depression, initial impact of event |
Heeke, 2017, Colombia [38] | Cross-sectional Humanitarian organisation | n=308 mean age=48.5 61.7% female | Significant other lost in armed conflict Mean length= 12.4 years | DUKE-UNC Functional Social Support Questionnaire [39] | PTSD, Prolonged grief, emotional distress |
Kristensen, 2010, Norway [40] | Cross-sectional Official population records (police deceased list and population register) | n=130 mean age=45.7 51.5% female | Family member lost in natural disaster Mean length= 2.2 years | Crisis Support Scale (CSS) [41] | Complicated grief |
Levi-Belz, 2015, Israel [42] | Cross-sectional Support organisation/online support forum | n=135 mean age=40.3 77.0% female | Family member lost to suicide Mean length = 3.5 years | MSPSS | Stress-related growth |
Levi-Belz, 2019, Israel [43] | Cross-sectional Support organisation/online support forum/online advertising | n=156 mean age=40.7 81.4% female | Family member or friend lost to suicide Mean length = 10 years | MSPSS | Complicated grief |
Li, 2015, China [44] | Cross-sectional Official population records | n=803 mean age=46.7 63% female | Family lost to natural disaster Mean length = 1.0 years | MSPSS | Complicated grief |
Oexle, 2019, U.S.A [45] | Cross-sectional Support organisation/online advertising | n=195 mean age=50 92% female | Immediate family lost to suicide Mean length = 8.9 years | Perceived Social Support Questionnaire (PSSQ) [46] | Depression, personal growth, grief difficulties, suicidal ideation |
Rheingold, 2015, U.S.A. [47] | Cross-sectional Official population records | n=47 mean age=78.7 78.7% female | Immediate family lost to homicide Mean length = 2.1 years | ISS | PTSD, complicated grief, depression |
Spino, 2016, U.S.A. [48] | Cross-sectional Social support group/online advertising | n=44 mean age=44 75% female | Adults bereaved by suicide Length of loss unclear | Norbeck Social Support Questionnaire (NSSQ) [49] | Depression, loneliness |
Sprang, 1998, U.S.A. [50] | Cross-sectional Support organisation | n=171 mean age=34 54.4% female | Immediate family killed by drunk driver Mean length = 2.3 years | Provisions of Social Relations Scale (PSRS) [51] | PTSD, grief, mourning |
Xu, 2017, China [52] | Cross-sectional Official population records | n=176 mean age=54.7 52.3% female | Child lost to natural disaster 6.0-6.3 years | Social Support Rating Scale (SSRS) [53] | PTSD |
Measure | Type of social support assessed by measure | Type of measurement tool | Use of measure in included study | |
---|---|---|---|---|
Arizona Social Support Interview Schedule (ASSIS) [32] | Size and availability of and satisfaction with support network. | Structured interview. | Bottomley 2017 [31] | 12 variables derived. Perceived need for and satisfaction with each of 6 categories: intimate interaction, material aid, advice and information, positive feedback, physical assistance, social participation |
Burke 2010 [33] | 5 variables derived: available support network for family and non-family, actual support network, anticipated and actual negative relationships | |||
Coppel Index of Social Support (CISS) [35] | Structural and functional support | Self-report questionnaire. 15 items on a 5 point Likert scale | Cowan 1985 [27] | Items across domains summed for total score of perceived social support. |
Murphy 1988 [28] | Items across domains summed for total score of perceived social support. | |||
Received social support | Self-report questionnaire. 7 items on a 7 point Likert scale | Kristensen 2010 [40] | Scandinavian version. Factors summed separately to measure positive social support and negative social response. | |
DUKE-UNC Functional Social Support Questionnaire [39] | Functional social support | Self-report questionnaire. 11 items on a 5 point Likert scale | Heeke 2017 [38] | Translated version. Items summed for total score of perceived social support. |
Inventory of Social Support (ISS) [34]a | Perceived social support for grief | Self-report questionnaire. 5 items on a 5 point Likert scale | Burke 2010 [33] | Items summed for total score of available grief support. |
Rheingold 2015 [47] | Items summed for total score of perceived social support. | |||
Multidimensional Scale of Perceived Social Support (MSPSS) [57] | Perceived presence and level of support across three domains: family, friends and significant other. | Self-report questionnaire. 12 items on 7 point Likert scale | Bailey 2013 [29] | Items across domains summed for total score of perceived social support. |
Burke 2010 [33] | Items across domains summed for total score of available general support. | |||
Levi-Belz 2015 [42] | Items across domains summed for total score of available perceived support. | |||
Levi-Belz 2019 [43] | Items across domains summed for total score of perceived support. | |||
Li 2015 [44] | Translated version. Items across domains summed for total score of general social support. | |||
Perceived social support: functional support | Self-report questionnaire. Amount of support from supportive network members listed. | Spino 2016 [48] | Network score, relationship score and both combined for total score. | |
Perceived Social Support Scales, friends and family (PSS-Fr, PSS-Fa) [37] | Perceived social support from friends and family | Self-report questionnaires. 20 items on a 3 point Likert scale | Fullerton 1999 [36] | Items summed for each scale for total score of support from friends and support from family. |
Provisions of Social Relations Scale (PSRS) [51] | Perceived social support | Self-report questionnaire. 18 items on a 5 point Likert scale | Sprang 1998 [50] | Family support and friend support subscales combined for a total score of cognitive appraisal of support. |
Perceived Social Support Questionnaire (PSSQ) [46] | Perceived social support | Self-report questionnaire. 6 items on a 5 point Likert scale | Oexle 2019 [45] | Items summed for a total score of perceived support. |
Social Support Rating Scale (SSRS) [53] | Subjective support, objective support and support availability | Self-report questionnaire developed for Chinese populations. 10 items | Xu 2017 [52] | Three domains of social support combined for a total score and categorised into low, medium and high support. |
Outcome | Study | Exploratory or specific hypothesis | Analysis method | Covariates included in models | Sample size (n) | Findings |
---|---|---|---|---|---|---|
Psychiatric outcomes | ||||||
PTSD | Bottomley 2017 [31] | Exploratory | Regression model with social support as a predictor. | T1 PTSD (at a mean of 1.66 years since loss) | 47 | Of 12 social support variables, need for advice, need for physical assistance and satisfaction with physical assistance were included in the model. Satisfaction with physical assistance was the only significant predictor, negatively predicting PTSD severity at T2 (6 month follow-up) (p<.03, b=-.18). |
Burke 2010 [33] | Exploratory | Correlations | n/a | 54 | Of 6 variables measured, percentage of actual negative relationships significantly correlated with PTSD severity (.28, p<.05). | |
Heeke 2017 [38] | Specific hypothesis | Latent class analysis | Gender, years of education, number of assaultive/accidental traumatic events, relationship to person lost, how loss happened and time since loss. | 308 | Social support was the only factor associated with PTSD symptoms compared to the resilient class (OR= .95, p=.005). | |
Rheingold 2015 [47] | Exploratory | Generalised estimating equations | Variables found to significantly differ by diagnostic status: employment status, deceased contributing to household income. | 47 | Lack of social support was independently associated with increased risk of meeting criteria for PTSD (beta =.19, Wald x2 = 4.64, p<.05). | |
Sprang 1998 [50] | Exploratory | Regression model with social support as a predictor. | Gender, age, race, subjective health status, income, marital status, past experience with death, time since death and religious beliefs. | 171 | Greater social support was associated with lower rates of PTSD symptoms (beta=.415, p<.005; 43.2% of variance). | |
Xu 2017 [52] | Exploratory | Regression model with social support as a predictor. | Ethnicity, residence location, gender, age, monthly income, education level, age of child and gender of child. | 176 | Low social support was a significant risk factor for meeting criteria for PTSD (OR= .244, beta=-1.41, p=.002, 95% CI). | |
Depression | Bottomley 2017 [31] | Exploratory | Regression model with social support as a predictor. | T1 depression (at a mean of 1.66 years since loss) | 47 | Of twelve social support variables, need for advice, need for physical assistance and satisfaction with physical assistance were included in the model but none were significant predictors. |
Burke 2010 [33] | Exploratory | Correlations | n/a | 54 | Of six social support measures, two were significantly correlated with depression severity: grief support (-.27, p<.05) and percentage of anticipated negative relationships (.28, p<.05). | |
Cowan 1985 [27] | Exploratory | Regression model with social support as a predictor. | Stress, age, gender, importance of deceased and perceived preventability of death. | 69 | Perceived social support was associated with greater depression severity (p<.05, b=-.14), accounting for 38% of variance in the model. | |
Fullerton 1999 [36] | Exploratory | Regression model with social support as a predictor. | Age, marital status, social network index, disaster specific social support, family distress, maximum closeness to deceased crew, transience, hardiness, social desirability and initial impact of event (IES). | 71 | Support from friends and support from family were entered as separate predictors in each model. In models controlling for total IES and IES intrusion scores, neither perceived social support variable was associated with depression severity. Controlling for IES avoidance (10%), perceived social support from friends was negatively associated with depression severity (5% of variance; beta=-.03, p=.027). | |
Oexle 2019 [45] | Specific hypothesis | Regression model with social support as a predictor. | Age, gender, pre-loss mental illness, time since loss, relationship to deceased and perceived closeness to deceased. | 195 | Greater perceived social support was significantly associated with a lower level of depressive symptoms (beta=-.53, p<.001). | |
Rheingold 2015 [47] | Exploratory | Generalised estimating equations with social support as a predictor. | Variables found to significantly differ by diagnostic status: age, employment status, deceased contributing to household income. | 47 | Lack of social support was independently associated with increased risk of meeting criteria for MDD (beta =.40, Wald x2 = 14.37, p<.005). | |
Spino 2016 [48] | Specific hypothesis | Regression model with social support as a predictor. | Physical health encumbrance. | 44 | Three social support variables were used as predictors. In a linear regression model, higher network score was associated with a significant decrease depression severity (beta= -0.53, p=.011). In a linear regression model, higher relationship score was associated with a significant decrease depression severity (beta= -0.18, p=.011). In the multiple regression model, higher total support score (beta= -0.02, p=.001) was associated with a significant decrease in depression severity. | |
Complicated grief | Bottomley 2017 [31] | Exploratory | Regression model with social support as a predictor. | T1 complicated grief (at a mean of 1.66 years since loss) | 47 | Of twelve social support variables, satisfaction with physical assistance was the only significant predictor out of the three social support variables included in the model, positively predicting complicated grief severity at T2 (6 month follow-up) (beta=.20, p<.05). |
Burke 2010 [33] | Exploratory | Correlations | n/a | 54 | Of six social support measures, two were significantly correlated with complicated grief severity: percentage of actual negative relationships (.28, p<.05) and available support system (-.28, p<.05). | |
Kristensen 2010 [40] | Exploratory | Regression model with social support as a predictor. | Gender, pre-disaster employment, relationship to deceased, previous experience of loss, time elapsed before death confirmed. | 130 | Two social support variables were included in analysis: low positive social support (OR=.24, p=.012) and high negative social support (OR=3.81, p=.012) were significantly associated with meeting criteria for complicated grief. | |
Levi-Belz 2019 [43] | Specific hypothesis | Regression model with social support as a predictor. | Time since loss, attachment style, self-disclosure and interaction between secure attachment, social support and self-disclosure. | 156 | Greater perceived social support was significantly associated with lower severity of complicated grief (beta=-.30, p<.01). | |
Li 2015 [44] | Exploratory | Regression model with social support as a predictor. | n/a | 803 | Social support was not significantly associated with meeting criteria for complicated grief. | |
Rheingold 2015 [47] | Exploratory | Generalised estimating equations with social support as a predictor. | Variables found to significantly differ by diagnostic status: age, deceased contributing to household income. | 47 | Lack of social support was not significantly associated with increased risk of meeting criteria for complicated grief. | |
Anxiety | Bottomley 2017 [31] | Exploratory | Regression model with social support as a predictor. | T1 anxiety (at a mean of 1.66 years since loss) | 47 | Need for advice, need for physical assistance and satisfaction with physical assistance were included in the model. Satisfaction with physical assistance was the only significant predictor, negatively predicting anxiety severity at T2 (6 month follow-up) (p<.001, b=-.30). |
Prolonged grief | Heeke 2017 [38] | Specific hypothesis | Latent class analysis with social support as a predictor. | Gender, years of education, number of assaultive/accidental traumatic events, relationship to person lost, how loss happened and time since loss. | 308 | The amount of perceived social support did not predict membership of the PGD class. |
Suicidal ideation | Oexle 2019 [45] | Specific hypothesis | Regression model with social support as a predictor. | Age, gender, pre-loss mental illness, time since loss, relationship to deceased and perceived closeness to deceased. | 195 | Greater perceived social support was significantly associated with lower severity of suicidal ideation (beta=-2.87, p<.001). |
Psychological wellbeing outcomes | ||||||
Emotional distress | Heeke 2017 [38] | Specific hypothesis | Latent class analysis with social support as a predictor. | Gender, years of education, number of assaultive/accidental traumatic events, relationship to person lost, how loss happened and time since loss. | 308 | Less social support was a predictor of the emotional distress class (OR= .92, p<.001). |
Grief | Sprang 1998 [50] | Exploratory | Regression model with social support as a predictor. | Gender, age, race, subjective health status, income, marital status, past experience with death, time since death and religious beliefs. | 171 | Greater social support predicted lower extent of grief (beta=-.479, p<.005). |
Grief difficulties | Oexle 2019 [45] | Specific hypothesis | Regression model with social support as a predictor. | Age, gender, pre-loss mental illness, time since loss, relationship to deceased and perceived closeness to deceased. | 195 | Greater perceived social support was significantly associated with decreased grief difficulties (beta=-.47, p<.001). |
Initial impact of event | Fullerton 1999 [36] | Exploratory | Regression model with social support as a predictor. | Age, marital status, social network index, disaster specific social support, family distress, maximum closeness to deceased crew, transience, hardiness and social desirability. | 71 | Neither perceived social support measure (support from friends/ support from family) was a good predictor of total or avoidance IES. Low perceived social support from friends predicted a higher intrusive initial IES score (beta=-.44, p=.044). |
Loneliness | Spino 2016 [48] | n/a | n/a | n/a | Statistical analyses not reported. | |
Mental distress | Murphy 1988 [28] | Exploratory | Regression model with social support as a predictor. | T1 mental distress, age, sex, education, stress, self-efficacy and social support | 49 | Social support did not significantly predict severity of mental distress |
Mourning | Sprang 1998 [50] | Exploratory | Regression model with social support as a predictor. | Gender, age, race, subjective health status, income, marital status, past experience with death, time since death and religious beliefs. | 171 | Greater social support significantly predicted lower extent of mourning (beta=.350, p<.005). |
Personal growth | Oexle 2019 [45] | Regression model with social support as a predictor. | Age, gender, pre-loss mental illness, time since loss, relationship to deceased and perceived closeness to deceased. | 195 | Greater perceived social support was significantly associated with increased personal growth (beta=-44, p<.05). | |
Recovery | Murphy 1988 [28] | Exploratory | n/a | n/a | n/a | Social support was not included in the regression model predicting recovery. |
Resiliencea | Bailey 2013 [29] | Exploratory | Regression model with social support as a predictor. | n/a | 48 | Unadjusted model where traumatic stress predicted greater levels of resilience was significant (b = -.241, p<.049). The adjusted model with social support as a mediator was also significant (b=.297, p=.032). |
Stress-related growth | Levi-Belz 2015 [42] | Specific hypothesis | Regression model with social support as a predictor. | Time since loss, adaptive coping, maladaptive coping, self-disclosure, interaction between time and interpersonal variables. | 135 | Combined with self-disclosure as a predictive interpersonal variable, social support predicted levels stress-related growth (beta=.11, p=.027). |
Quality assessments
Adapted for cross-sectional studies | |||||||||
Selection | Comparability | Outcome | |||||||
Study | Representativeness of sample | Sample size | Non-respondents | Ascertainment of exposure | Confounding factors controlled | Assessment of the outcome | Statistical test | Quality | |
Bailey 2013 [29] | 1 | 0 | 1 | 2 | 0 | 1 | 1 | Poor | |
Burke 2010 [33] | 0 | 0 | 0 | 2 | 0 | 1 | 1 | Poor | |
Cowan 1985 [27]a | 1 | 0 | 0 | 2 | 1 | 1 | 1 | Fair | |
Murphy 1988 [28] | 1 | 0 | 0 | 2 | 1 | 1 | 1 | Fair | |
Fullerton 1999 [36] | 1 | 0 | 0 | 2 | 1 | 1 | 1 | Fair | |
Heeke 2017 [38] | 0 | 0 | 0 | 2 | 2 | 1 | 1 | Fair | |
Kristensen 2010 [40] | 1 | 0 | 1 | 2 | 1 | 1 | 1 | Good | |
Levi-Belz 2015 [42] | 1 | 0 | 0 | 2 | 0 | 1 | 1 | Poor | |
Levi-Belz 2019 [43] | 1 | 0 | 0 | 2 | 2 | 1 | 1 | Good | |
Li 2015 [44] | 1 | 0 | 0 | 2 | 0 | 1 | 1 | Poor | |
Oexle 2019 [45] | 0 | 0 | 1 | 2 | 2 | 1 | 1 | Good | |
Rheingold 2015 [47] | 1 | 0 | 0 | 2 | 1 | 1 | 1 | Good | |
Spino 2016 [48] | 0 | 0 | 0 | 2 | 0 | 1 | 1 | Poor | |
Sprang 1998 [50] | 1 | 0 | 0 | 2 | 2 | 1 | 1 | Good | |
Xu 2017 [52] | 1 | 0 | 0 | 2 | 1 | 1 | 1 | Good | |
NOS for cohort studies | |||||||||
Selection | Comparability | Outcome | |||||||
Representativeness | Selection of non-exposed cohort | Ascertainment of exposure | Outcome of interest not present at start of study | Comparability of cohorts | Assessment of outcome | Follow-up long enough for outcome to occur | Adequacy of follow-up | Quality | |
Bottomley 2017 [31] | 0 | n/a | 1 | 0 | 1 | 1 | 1 | 1 | Fair |
Summary of findings
Number of studies indicating an association between social support and outcome | ||||
---|---|---|---|---|
Positive associationa | Partial positive associationb | No association | Negative association | |
Outcome | ||||
Psychiatric | ||||
Depression (N = 7) | 1 [31] | – | ||
PTSD (N = 6) | – | – | ||
Complicated grief (N = 6) | 1 [33] | 1 [31] | ||
Prolonged grief (N = 1) | – | – | 1 [38] | – |
Anxiety (N = 1) | – | 1 [31] | – | – |
Suicidal ideation (N = 1) | 1 [45] | – | – | – |
Psychological | ||||
Emotional distress (N = 1) | 1 [38] | – | – | – |
Grief (N = 1) | 1 [50] | – | – | – |
Grief difficulties (N = 1) | 1 [45] | – | – | – |
Initial impact of event (N = 1) | – | 1 [36] | – | – |
Mental distress (N = 1) | – | – | 1 [28] | – |
Mourning (N = 1) | 1 [50] | – | – | – |
Personal growth (N = 1) | 1 [45] | – | – | – |
Resilience (N = 1) | 1 [29] | – | – | – |
Stress-related growth (N = 1) | 1 [42] | – | – | – |