|Home | About | Journals | Submit | Contact Us | Français|
Little research has been conducted into the physical health implications of suicide bereavement compared to other causes of death. There is some evidence that suicide bereaved parents have higher morbidity, particularly in terms of chronic illness. This systematic review aims to examine the physical and psychosomatic morbidities of people bereaved by a family member’s suicide and compare them with family members bereaved by other modes of death.
MEDLINE, EMBASE, CINAHL, and PsycINFO were searched from 1985 to February 2016. The search was re-run in March 2017. Peer-reviewed English language articles comparing suicide-bereaved family members to non-suicide bereaved family members on measures of physical or psychosomatic health were eligible for inclusion. Cohort, cross-sectional, case-control and cohort-based register studies were eligible for inclusion. A modified version of the Newcastle Ottawa Scale was used for quality assessment. Results were synthesised using narrative synthesis.
The literature search located 24 studies which met the inclusion criteria. Seven studies found statistically significant associations between physical health and suicide bereavement. Five of the studies found that suicide-bereaved family members were more likely to experience pain, more physical illnesses and poorer general health. They were also at increased risk of cardiovascular disease, hypertension, diabetes and chronic obstructive pulmonary disease. In contrast, another study in Denmark found that those bereaved by suicide had a lower risk of a number of physical health disorders, including cancers, diabetes, cardiovascular and chronic lower respiratory tract disorders compared to those bereaved by other causes of death. Additionally, a further study conducted in the United States found that suicide-bereaved children visited a GP less frequently than non-suicide bereaved children.
Review findings are relevant for clinicians working with people bereaved by suicide as they highlight that such clients are at increased risk of several adverse physical health outcomes. Future research should examine health risk behaviours of suicide-bereaved and non-suicide bereaved family members as they may confound the association between exposure and outcome.
The review protocol has been registered on PROSPERO, registration number CRD42016030007.
The online version of this article (10.1186/s12889-017-4930-3) contains supplementary material, which is available to authorized users.
Bereavement is a significant stressor that can initiate or compound existing mental and physical disorders . Grief is a reaction to bereavement, encompassing thoughts, feelings, behaviours and physiological changes which may fluctuate and change in intensity over time . Over 800,000 people die by suicide worldwide every year, leading to an estimated 45–500 million people experiencing suicide bereavement annually [3, 4]. Suicide has an emotional impact on those bereaved but it also has a societal impact, in terms of economic effects . While a number of studies have considered the economic impact of suicide [6–8], it is important to understand the effects of suicide bereavement in terms of survivors’ physical and psychosomatic health.
There have been mixed results regarding how suicide bereavement differs from other forms of bereavement [9, 10]. A seminal review  posited that suicide bereavement can be differentiated by three over-arching themes. Firstly, the qualitative aspects of grief may be different, with those bereaved by suicide experiencing higher levels of guilt, blame, responsibility and rejection. Secondly, social processes may differ for those bereaved by suicide, where they feel more isolated or stigmatized due to their loved one’s suicide. Finally, a pre-existing dysfunctional family environment may have contributed to the development of suicidal thoughts and behaviour in the deceased. This dysfunction coupled with the suicide may contribute to the occurrence of psychiatric conditions amongst the surviving family members.
Also, people bereaved by suicide are at increased risk of engaging in suicidal behaviour themselves . Researchers have put forward several explanations as to why those bereaved by suicide are at particular risk of suicidal behaviour. Firstly, they have established that suicidal behaviour is familial and may be partly explained by genetics [12, 13]. Research indicates that proband suicide attempt increased the odds of offspring suicide attempt by nearly 5-fold, when controlling for a number of factors including baseline history of suicide attempt . Some research suggests that the intrafamilial transmission of impulsive aggression, childhood maltreatment and mood disorder may be possible mediators . However, the exact mechanism underlying this genetic transmission is still unclear.
In addition, social stigma and blame represents a significant challenge for those bereaved by suicide, which may motivate some families to conceal the cause of death [14, 15]. People bereaved by suicide are at increased risk of suicide, depression, substance abuse, complicated grief and feelings of shame and guilt [10, 16–18]. They are also at increased risk of negative physical outcomes, including cardiovascular disease (CVD), chronic obstructive pulmonary disease (COPD), hypertension, diabetes and pancreatic cancer [19, 20]. While it is important to consider such specific physical health conditions, it may take many years for such conditions to develop. Therefore, it is also critical to consider somatic and psychosomatic symptoms and complaints which may be more likely to be present in the short-term following bereavement [21–24].
To date, a synthesis of research on the effects of suicide bereavement on physical health problems and psychosomatic symptoms has not been conducted. The rationale for this review is to contribute to the evidence around the societal impact of suicide bereavement as borne by the families and health services, as well as informing clinicians who support those bereaved by suicide. The population of interest is bereaved family members and the exposure of interest is suicide bereavement. Therefore, people bereaved by suicide will be compared to people bereaved by other causes of death to examine any differences in physical and psychosomatic health between the two groups. The aim of this paper is to examine the physical and psychosomatic morbidities of people bereaved by a family member’s suicide compared with family members bereaved by other modes of death.
This review was conducted by adhering to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist . The PRISMA checklist has been completed (see Additional file 1). The review protocol has been registered on PROSPERO, registration number CRD42016030007.
MEDLINE, EMBASE, CINAHL, and PsycINFO were searched for articles published between 1st January 1985 and 15th February 2016. Searching was re-run on 27th March 2017 to locate additional articles published in the interim (n=666). This search found one recently published study that met the study inclusion criteria. The following MeSH terms were exploded to define exposure: “suicide”, “bereavement”, “genetic predisposition to disease” and “family characteristics”. Searches for the following keywords were also run to define exposure: “grief”, “familial”, “family history” and “genetic predisposition”. The following MeSH terms were exploded to define the population of interest: “family” and “friends”. The term “friends” was included in order to ensure inclusion of all relevant articles that may have included family members also. The term “survivors” was not exploded as it would have included survivors of terminal illness and long-term HIV survivors. Searches for the following keywords were also run to define the population: “relative*”, “parent*”, “mother*”, “father*”, “sibling*”, “offspring*”, “child*”, “brother*”, “sister*”, “family”, “friend” and “survivor*” (see Additional file 2). Searches were limited to English language articles only and articles published from 1985 to 15th February 2016 for the first search. As previously stated, the search was re-run in March 2017. There were slight modifications to this search strategy when searching other databases, where needed/appropriate. Reference list searching was employed for all included studies. The search strategies used for each of the databases is provided in Additional file 2. Among the full-texts of articles retrieved, sixty were subsequently excluded. The citations of these articles along with the reasons for their exclusion are provided in Additional file 3.
Studies that met the following eligibility criteria were included in the review: (1) the population of interest comprised family members bereaved by suicide, including those related by blood and also including spouses; (2) controls were family members bereaved by a non-suicide death; (3) authors specified at least one physical or psychosomatic health outcome; (4) original cross-sectional, case-control, cohort and registry-based studies.
Studies that exclusively used non-bereaved controls as the comparison group were excluded, as it is impossible to say if any negative health effect observed is attributable to suicide bereavement or to bereavement in general . Case reports, cases studies, reviews, randomised controlled trials and studies with no control groups were excluded. If multiple articles meeting the inclusion criteria were published based on the same study, the article(s) containing the most complete or new information was used.
The first author (AS) conducted the initial searches and screening process. Three authors (AS, CL and KMS) screened the full-text articles to assess for eligibility; disagreements were discussed and resolved with a fourth reviewer (EA). One author who was contacted regarding missing relevant information provided further analyses to meet the inclusion criteria for this review . One author (AS) extracted the following information from relevant articles:
A modified version of the Newcastle Ottawa Scale which was used in a previously published systematic review was chosen to assess risk of bias of individual studies at the study level . Scores range from zero (high risk of bias) to three (low risk of bias). Definitions then follow in order to determine what constitutes low, moderate and high risk of bias. Two authors (AS and FR) independently assessed the quality of each included article. AS resolved any disagreements through discussion with another reviewer (EA). Articles were not excluded based on the quality assessment. Results of included studies were synthesised in narrative form.
Figure 1 highlights the process of identifying relevant articles. A total of 6959 records were identified across the four databases, with four additional records identified from other sources, namely reference list searching. Eighty-six full-text articles were assessed for eligibility. The search was re-run in March 2017 which retrieved 666 articles that were published in the interim. One of these met the criteria for the study. Therefore, 24 papers meeting the inclusion criteria, representing 27 studies were included in the review. Three papers were published using the same study sample. Where this occurred, the most up-to-date or most comprehensive information and results were included. This was done to ensure that information was not duplicated in the review [21, 28, 29]. Of the 24 included studies, five studies examining aspects of physical health [19, 30–33] found that family members bereaved by suicide had statistically poorer health outcomes than the non-suicide bereaved comparison. Two further studies found statistically significant associations in the opposite direction, whereby the suicide-bereaved were at lower risk than the non-suicide bereaved comparison [34, 35]. No studies examining psychosomatic health outcomes found statistically significant results.
Table 1 outlines details of the 24 included studies. These were conducted in the United States (n=9), Sweden (n=4), Canada (n=2), the Netherlands (n=2), Denmark and Sweden (n=1), Slovenia (n=1), Denmark (n=1), China (n=1), Japan (n=1), Norway (n=1), and England (n=1). The majority of the studies (13 studies) were conducted more than 10 years ago (1988–2003) [22, 24, 31, 33, 34, 36–43], with 11 studies published in the last ten years (2006–2017) [19, 20, 23, 26, 30, 32, 35, 44–47]. Included studies were 10 cross-sectional studies [23, 24, 30, 32, 33, 36, 38, 41–43], 11 cohort/registry-based studies [22, 26, 31, 34, 35, 37, 39, 44–47] and three case-control studies [19, 20, 40]. Eleven studies examined aspects of physical health, including general health, [19, 20, 26, 30, 31, 35, 39, 44–47], eight studies examined somatic complaints/reactions [24, 32, 33, 36, 37, 40, 42, 43], three studies examined psychosomatic health outcomes [22, 23, 41], while the remaining two studies examined both physical and psychosomatic health outcomes [34, 38]. Sample size varied considerably between studies ranging from 13 people bereaved by suicide to large-scale registry-based studies with 31,672 people bereaved by suicide. Length of follow-up was also diverse, ranging from nine months to 45 years.
The modified version of the Newcastle Ottawa Scale used has nine questions relating to five domains of evaluation, namely: selection of study participants (selection bias); controlling for confounding (performance bias); statistical methods (detection bias), measuring outcome variables (information bias); and subject follow-up (only for follow-up studies; attrition bias). Following the assessment of included studies using the modified version of the Newcastle Ottawa Scale, a number of study limitations emerged. Firstly, some studies (4/24) recruited study participants by advertising or recruiting from self-help groups, which may have introduced selection bias into the studies. Nearly half (11/24) of all included studies had small sample sizes, with the smallest sample being 13 suicide-bereaved widows. This small sample size may have reduced the likelihood of being able to identify a statistically significant difference between the suicide-bereaved and non-suicide bereaved groups with respect to physical and psychosomatic health. Seven of the studies had suicide-bereaved sample sizes of 30 participants or less. Over a quarter (6/24) of included studies either did not control for any confounding factors (4/24) or only adjusted for limited confounding factors (2/24). A small minority of studies (4/24) controlled for various factors including pre-bereavement functioning, kinship, cause of death, decedent’s gender and age and time since death. Overall, statistical analysis conducted across the papers was good, with the use of appropriate statistical methods. However, it was noted that over a quarter of studies (6/24) carried out multiple testing that was not accounted for, had inconsistent or no reporting of p-values and 95% confidence interval thresholds, and data was not presented for some analyses that were conducted. Half of the studies (12/24) contained heterogeneous control groups, where family members bereaved by violent (accident, homicide) and non-violent deaths (natural anticipated, natural unanticipated) were analysed together. This may have introduced selection bias as research indicates that health consequences differ when the death is violent or nonviolent .
A comparison of characteristics of responders and non-responders was present in a minority of the studies (2/24), with the majority of papers not presenting this information (13/24). Six studies were register-based studies and therefore, the issue of non-response bias is not applicable. One study did not have any information on non-responders beyond gender, age, mode of death and place of residence of deceased, due to confidentiality reasons. A further study compared excluded cases to included cases on a number of variables including victim’s age, race, sex and method of death and concluded there was no evidence of sample bias. Finally, one study compared bereaved offspring that remained in the study to those lost to follow-up. Bereaved offspring lost to follow-up were more likely than those who remained in the study to have a caregiver with a history of alcohol or substance disorder (32.1% vs. 16.7%), to have a caregiver of minority status (28.4% vs. 11.7%), and to have had a proband with a history of an anxiety disorder (28.3% vs. 16.4%). Overall, selection bias emerged as an important methodological consideration in the included papers (Table 2).
Seven studies examined general health. Two cross-sectional studies found significant associations between suicide bereavement and general health [30, 32]. The first study by De Groot and colleagues included a Dutch sample of 223 bereaved family members. Suicide-bereaved family members scored more negatively than family members bereaved by natural death on a number of domains of the RAND scale, including pain, general health and experiencing a change in health following the death when compared to those bereaved by a natural death immediately before the death . When analyses were adjusted for demographics, neuroticism and expectedness of death, pain was the only health-related measure that remained significant (95% CI: −.7, −.003) . Selection bias may be an issue in this study because just 45% of approached suicide-bereaved families took part in the study. The second study showed that the general health of those bereaved by suicide (n=21) was significantly poorer than those bereaved by a long-term illness (n=88) (p <0.05) . However, participants were recruited from self-help groups and seminars for the bereaved and thus represent a biased sample of bereaved individuals.
Two cohort studies found statistically significant associations between general health and suicide bereavement , with one study finding an inverse association . The first study of 60 bereaved family members found that those bereaved by suicide (n=30) reported “more physical illnesses” and greater frequencies of appointments with healthcare professionals than those bereaved by accidental death (n=30) . Length of follow-up was relatively short, with the first and second interview occurring a mean 5.8 (range 4–8) and 9 (range 7–11) months after the death. In addition, nearly one-third of the suicide-bereaved sample (30.6%) refused to take part in the follow-up interview which may have introduced attrition bias. The second cohort study consisted of 26 suicide-bereaved and 332 non-suicide bereaved children and adolescents conducted in the United States by Cerel and Colleagues. In contrast to Séguin’s findings, the suicide-bereaved youth had significantly fewer visits to a doctor at 13 months (mean±SD=0.7±1.1 versus 2.0±3.3; t=2.71, df=24.5, p<.05) and 25 months post-bereavement (mean±SD=1.8±2.7 versus 6.0±8.5; t=3.50, df=18.4, p<.005) . Séguin and colleagues found that suicide-bereaved parents visited health professionals more frequently than accident-bereaved parents . The study by Cerel and colleagues also found that school-reported health problems did not vary between the suicide and non-suicide bereaved groups. In addition, non-suicide-bereaved offspring had missed significantly more days from school than suicide-bereaved offspring (2.8±3.8 versus 0.8±0.8; t=2.78, df=10.1, p<.05). This study had a very small sample of suicide-bereaved (n=26) in comparison to an over-representation of non-suicide bereaved (n=332) participants. A greater number of suicide-bereaved participants would have been preferable in order to have more balanced exposure groups for comparison purposes.
Two further cohort studies conducted in Slovenia  and The Netherlands  failed to find any significant association between suicide bereavement and physical health. The first study had a small sample size of 30 suicide-bereaved, 23 accident bereaved, and 20 spouses bereaved by a long-term illness (no p-value reported) . The second study included a sample of 309 people bereaved by the death of a first-degree relative at four and 14 months after the death, but only controlled for limited confounders including sex, kinship and mode of death . Lastly, a cross-sectional study conducted in the United States by Demi and Miles failed to find a significant difference between 59 suicide and 61 non-suicide bereaved parents with regard to physical health problems [F(4, 95=1.52, p=.20] . However, the suicide-bereaved sample was recruited from self-help groups through various means of contact, indicating that a response rate for the suicide-bereaved group could not be reported. In addition, since only parental age was controlled for in the analysis, other important confounding factors were not taken into account.
Eight studies examined specific physical disorders and suicide bereavement. Four cohort/register-based studies examined the possible association between various forms of cancer following suicide bereavement. One additional registry-based study examined a number of physical health conditions, including cancer, cardiovascular disease, diabetes and suicide bereavement .Two Swedish register-based studies concluded that, being bereaved by the suicide death of an offspring conferred a higher risk (RR: 1.24, 95% CI: 1.01–1.49) of infection-related cancers  and pancreatic cancer in parents  (OR: 1.23, 95% CI: 1.03–1.46) compared to those bereaved by a non-suicide death . Nevertheless, these findings were not statistically significant when compared with non-suicide bereaved parents. Similarly, two other large national studies found no statistically significant association between loss of a parent due to suicide when compared to other unexpected causes of death with a maximum of 15 years and 40 years of follow-up, respectively [26, 45]. These four studies successfully met most of the quality assessment criteria [20, 44, 45] with the exception of one paper having a relatively small number of suicide-bereaved participants .
No significant differences were found in categories of BMI (normal, overweight, obese) by Weinberg and colleagues at the 5-year assessment in offspring bereaved by suicide (n=45), accident (n=27) or sudden natural death (n=51) . However, the sample size was relatively small with some participants being recruited via advertising. Using data provided in the paper by Wilcox and colleagues cohort study, additional calculations did not show any statistically significant differences in the risk of sickness absence due to somatic diagnosis between suicide-bereaved, accident-bereaved and naturally bereaved. . This study met most of the quality assessment criteria with some minor limitations related to selection bias and outcome measurement.
A Canadian case-control study conducted by Bolton and colleagues found that suicide-bereaved parents (n=1415) had a significantly increased risk of a number of specific physical health disorders both before and after their offspring’s death compared to 1132 accident-bereaved parents . These include cardiovascular disease (2 years pre-death ARR: 1.54: 1.16–2.03; 2 years post-death ARR: 1.63: 1.23–2.16), hypertension (ARR 1.37: 1.19–1.59; ARR 1.32: 1.15–1.52), diabetes mellitus (ARR 1.45: 1.20–1.76; ARR 1.66: 1.37–2.00) and chronic obstructive pulmonary disease (ARR 1.68: 1.20–2.37; ARR 2.01: 1.40–2.90) . In addition, suicide bereaved parents had an increased risk of visiting a physician for a physical illness (ARR 1.38: 1.15–1.65; ARR 1.39: 1.18–1.63) and also being hospitalised for a physical illness (ARR 1.49: 1.01–2.20; ARR 1.52: 1.07–2.16) . This paper met most of the quality assessment criteria. In contrast, a Danish register-based study found that spouses bereaved by suicide (n=15,607) had a lower risk of receiving a subsequent diagnosis of a number of physical health disorders compared to spouses bereaved by a non-suicide death (n=788,778) . These included cancers (men: IRR, 0.8; 95% CI, 0.7–0.9; women: IRR, 0.8; 95% CI, 0.7–0.9), diabetes (men: IRR, 0.6; 95% CI, 0.4–0.7; women: IRR, 0.6; 95% CI, 0.5–0.8), cardiovascular (men: IRR, 0.9; 95% CI 0.8–0.9; women: IRR, 0.9; 95% CI, 0.8–1.0), and chronic lower respiratory tract disorders (men: IRR, 0.8; 95% CI, 0.7–1.0; women: IRR, 0.7; 95% CI, 0.6–0.8). Suicide-bereaved were less likely to take sick leave (men: IRR, 0.8; 95% CI, 0.7–0.9; women: IRR, 0.8; 95% CI, 0.7–0.8), while men were less likely to visit a general practitioner than those bereaved by other causes of death (IRR, 0.9; 95% CI, 0.8–1.0). Also, suicide-bereaved women had lower use of somatic hospitals (IRR, 0.9; 95% CI, 0.8–1.0). Similarly, this study scored highly across all of the assessment domains .
Eight studies examined physical symptoms/somatic complaints. Three American cross-sectional studies found no significant difference in somatic complaints for suicide-bereaved and accident-bereaved widows  and next-of-kin  and suicide-bereaved and cancer-bereaved children . Sample sizes for the suicide-bereaved were a particular issue for two of the studies with a sample of 13  and 16 , respectively. A further cross-sectional study conducted in Norway found that parents bereaved by SIDS (Sudden Infant Death Syndrome) experienced significantly fewer problems (p <.05) than parents bereaved by suicide and accidents on the general health questionnaire (GHQ). Therefore, those bereaved by suicide and accident significantly differed from the SIDS sample with respect to their level of complaints on the GHQ (suicide: M=9.8, SD=8.3; accident: M=10.4, SD=7.8; vs. SIDS: M=5.8, SD=7.1, F=4.17, p <.05) . Cleiren and colleagues also found no significant difference in somatic complaints between 91 suicide-bereaved, 93 road traffic accident or 125 long-term-illness-bereaved first-degree relatives in a Dutch 10-month cohort study (data not presented in original paper) . This study controlled for sex, kinship and mode of death only, which may have biased the results. Kinship, in this review refers to the type of familial relationship (parent-child, spousal, sibling, child-parent), including blood and non-blood relationships, between two people.
An American cross-sectional  and an English cohort study  found that somatic reactions did not significantly differ between suicide-bereaved and non-suicide bereaved participants. Specifically, the first study consisted of 14 suicide-bereaved (M=12.86, SD=4.57, p>.05), 15 accident-bereaved (M=12.40, SD=4.01, p>.05), 15 unanticipated naturally-bereaved (M=12.67, SD=3.27, p>.05) and 13 expected naturally bereaved widows/widowers (M=11.08, SD=3.01, p>.05) . The second study included 20 suicide-bereaved and 18 naturally-bereaved children (M=10.7 versus M=9.9) . Sample size was a significant limitation in both studies.
One final cross-sectional study conducted in Japan  did not find any significant group differences in somatic symptoms and complaints when comparing suicide-bereaved and non-suicide bereaved family members. The paper had hugely different sample sizes within the bereavement groups: suicide (n=21), accidents (n=23), acute illness (<1 day) (n=9), shorter illness (<1 year) (n=74) or longer illness (≥ 1 year) (n=88) . This may have had an impact on identifying potential group differences.
Five studies examined somatisation and suicide bereavement. Three were cross-sectional studies, two of which were conducted in the United States [38, 41], and one was conducted in China . The American cross-sectional study conducted by Demi and Miles did not find significant differences between 59 suicide-bereaved and 61 non-suicide bereaved parents on the scale measuring distress, which included a somatisation measure [F (5, 111)=.45, p=.81] . Participants were recruited via self-help groups and only parental age was adjusted for in the analysis. Similarly, the second American cross-sectional study concluded that mean scores on somatisation did not significantly differ between 85 suicide-bereaved (M=9.9, SD=9.9), 56 homicide-bereaved (M=9.7, SD=9.6), 135 accident-bereaved (M=10.2, SD=10.1), 167 sudden natural death bereaved (M=9.7, SD=9.7) and 106 long-term illness-bereaved widows (M=10.8, SD=10.8) . The cross-sectional study conducted in China found no significant difference in somatisation between 92 suicide-bereaved and 64 accident-bereaved immediate family members (p=0.87) . Both of these studies met most of the quality assessment criteria. Cerel and colleagues conducted a cohort study in the United States which found no differences between 26 suicide-bereaved and 322 non-suicide bereaved children and adolescents with respect to somatisation . Interviews were conducted with participants at 1, 6, 13 and 25 months post-parental death . Sample size for the suicide-bereaved was a limitation in the study, together with the limitation that no confounding factors were adjusted for in the analysis. Similarly, the American cohort study conducted by Farberow and colleagues found that suicide-bereaved (n=108) and naturally bereaved spouses (n=199) did not differ significantly on the somatisation subscale (no p-value given) . This study met most of the quality assessment criteria. However, it appears that there was a high rate of loss to follow-up in the study.
The current systematic review found 24 studies that fit the inclusion criteria. Of these, seven studies found statistically significant associations between aspects of physical health and suicide bereavement. Five studies noted that people bereaved by suicide had an increased risk of a number of adverse physical health outcomes. Two further studies found an association in the opposite direction for a number of physical health outcomes  and healthcare utilisation  for those bereaved by suicide.
This review of physical and psychosomatic health outcomes found tentative evidence supporting an association between bereavement by suicide and some physical health outcomes, although there are inconsistencies. Cardiovascular disease, COPD, hypertension, diabetes, increased pain and poorer general health were more frequently reported adverse physical health outcomes among people bereaved by suicide [19, 30–33] compared to those who experienced other types of bereavement. Some studies found that suicide bereavement conferred a lower risk of various physical and psychosomatic health outcomes [34, 35]. However, the majority of studies found no significant differences in physical and psychosomatic health outcomes following suicide bereavement [20, 22–24, 26, 36–47].
The prevalence of physical health issues in those bereaved by suicide [19, 30–33] may lead to more healthcare utilisation. There are varying findings with respect to healthcare utilisation amongst the suicide-bereaved. Suicide-bereaved adults were more likely to experience a health change after the death, have more appointments with healthcare professionals and also to be hospitalised more often for physical illnesses compared to non-suicide bereaved family members. In contrast, men bereaved by the suicide of a spouse were less likely to visit a general practitioner than those bereaved by other causes of death. In addition, both men and women bereaved by a spouse’s suicide were less likely to take sick leave than those bereaved by other causes of death . Not seeking medical attention for physical health problems may be due to being preoccupied by grief [35, 48]. This underlies the importance for health care practitioners to be aware of the unique challenges of suicide bereavement and its associated health issues in their patients. Similarly, children bereaved by suicide missed significantly fewer days from school and also had fewer visits to a doctor compared to non-suicide bereaved children . Suicide-bereaved children may have less familial environmental stressors and higher levels of functioning, including grief responses which have been shown to be important in moderating long-term outcomes for parentally bereaved children [16, 49]. Moreover, suicide-bereaved people experience more perceived stigma than those bereaved by both sudden unnatural and sudden natural death . Shame and stigma have been linked to a number of avoidance behaviours, including poorer help-seeking in the suicide-bereaved [50–52]. Where people experience high levels of shame and stigma, this may impact negatively on their help-seeking behaviour which may in turn impact negatively on health outcomes. Additionally, a recent systematic review highlighted that stigma experienced by people bereaved by suicide was strongly correlated with increased somatic reactions, including headaches and stomach pain . Therefore, it may also be plausible that shame and stigma may moderate the relationship between physical and psychosomatic health outcomes following suicide bereavement.
Following the synthesis of results, a number of issues associated with the included studies became apparent. Firstly, sample size was a significant limitation across a number of the studies, resulting in studies being underpowered, with some of the suicide-bereaved samples being as low as thirteen participants . Some of the studies recruited participants from advertising, self-help and bereavement support groups, which biases the sample recruited as this group may be significantly different to those who do not attend support groups in terms of their own characteristics and grief responses [24, 32, 38, 46]. A number of studies did not adjust for any confounding factors [34, 36, 39, 42], and some adjusted only for a limited number of confounders including basic demographics of the deceased and/or surviving relative [32, 37, 38]. Only two of the included studies examined pre-bereavement physical health, which examined outcomes both before and after offspring death [19, 47]. Therefore, the majority of the studies included in this review only focus on changes to physical health after bereavement, and consequently, are subject to recall bias. Length of follow-up for the cohort/registry-based studies was generally considerable, with the shortest follow-up being nine months after the death. However, two of the studies conducted interviews with bereaved participants two and three months post-death. This short time span may bias results as acute grief reactions may still be present. Half of included studies had heterogeneous control groups, where both violent and nonviolent bereavements were included [20, 26, 32, 33, 36, 37, 39, 41, 44, 46, 47]. Research indicates that sense-making is significantly more challenging for people bereaved by violent deaths versus non-violent deaths . In addition, those bereaved by suicide and drug-related death appear to be more affected by grief and mental health problems compared to those bereaved by accidental and natural deaths . Therefore, the presence of heterogeneous control groups in these studies may underestimate the true impact of suicide bereavement on physical and psychosomatic outcomes.
Overall, the evidence to support an increased risk of adverse physical health outcomes following suicide bereavement is growing but further longitudinal controlled studies are needed. No study examining psychosomatic outcomes and suicide bereavement found a positive association. The use of objective measures of physical health is warranted in future studies, as much of the research conducted in this area have used self-reported measures of health which are subject to recall bias. Furthermore, more studies need to examine pre-bereavement physical health, which examines outcomes both before and after the death. Following on from this, uncertainty remains regarding psychosomatic health and suicide bereavement.
This is the first review to synthesise all relevant papers related to suicide bereavement and physical and psychosomatic health outcomes, using a rigorous, exhaustive and comprehensive search strategy. The PRISMA checklist guided the reporting of this review. This review also has some limitations. Firstly, only English-language studies published from 1985 to March 2017 were included. Only quantitative papers were included; differing results and conclusions may have been found with the additional inclusion of qualitative studies. It is possible that some differences in suicide bereavement may only be revealed through in-depth qualitative interviews as opposed to quantitative methods. The evidence indicates that suicide bereavement is associated with some adverse physical health outcomes, but there are inconsistencies across the studies. In addition, studies relating to psychosomatic health outcomes did not show an association with suicide bereavement. There was also an imbalance of studies reporting on physical health outcomes, with a small minority of papers solely focussing on psychosomatic health outcomes. This needs to be addressed in future research. Some of the papers investigating psychosomatic health outcomes had small sample sizes, selection bias and did not control for confounding factors. We need further research addressing the uncertainty regarding the association between physical and psychosomatic health outcomes and suicide bereavement as well as the specificity of these outcomes. Register-based and cohort studies are the most appropriate means of examining this research question; selecting an appropriate control group, people bereaved by sudden and violent deaths, is essential. Future studies also need to allow for sufficient time to follow-up as some of the outcomes may not be present shortly after bereavement.
This systematic review found that a small number of studies demonstrated associations between suicide bereavement and adverse physical health outcomes, including cardiovascular disease, diabetes, chronic obstructive pulmonary disease, hypertension and poorer general health. However, most studies failed to conclude that people bereaved by suicide were at higher risk for a number of physical health conditions compared to non-suicide bereaved individuals. No studies found a significant association between suicide bereavement and psychosomatic health outcomes. Thus, the findings of this review indicate that, in terms of psychosomatic health issues at least, those bereaved by suicide may closely resemble people bereaved by other causes of death. Inconsistencies in results may be due to methodological shortcomings in the available studies, including inappropriate selection of control groups, small sample size and failure to control for confounding factors. Further longitudinal controlled studies need to be conducted in order to better understand the health implications of suicide bereavement, specifically compared to bereavement after sudden and violent deaths, including accident and homicide deaths.
I would like to thank Dr. Alexandra Pitman for giving me permission to use her search strategy. I would like to acknowledge the help of Mr. Joe Murphy, assistant librarian with the Mercy University Hospital Library, for his excellent input thereby optimising the search strategy utilised. I would also like to thank Professor John Browne who taught an excellent postgraduate module on systematic reviews in University College Cork. This proved to be invaluable in conducting this research.
This work was conducted as part of the SPHeRE Programme under Grant No. SPHeRE/2013/1.
AS contributed to the conception and design of the review. AS conducted all of the literature searches, screened the abstracts and assessed full-text articles for potential relevance and inclusion. CL, KMS, PC and EA contributed to assessing full-text articles for potential relevance and inclusion. AS reviewed the full-texts of the papers for data extraction. AS and FR carried out quality appraisals on included articles independently. AS prepared the first draft of the paper. KMS, CL, PC, FR and EA reviewed the manuscript. All authors read and approved the final manuscript.
The authors declare that they have no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Electronic supplementary material
The online version of this article (10.1186/s12889-017-4930-3) contains supplementary material, which is available to authorized users.
Ailbhe Spillane, Phone: 00353 21 4205544, Email: firstname.lastname@example.org.
Celine Larkin, Email: ude.demssamu@nikraL.enileC.
Paul Corcoran, Email: ei.ccu@narocrocp.
Karen Matvienko-Sikar, Email: email@example.comK.
Fiona Riordan, Email: firstname.lastname@example.orgF.
Ella Arensman, Email: ei.ccu@namsnerae.