• accident;
  • bereavement;
  • depressive state;
  • general health;
  • grief;
  • sudden death;
  • suicide


  1. Top of page
  2. Abstract

Abstract  To evaluate how the suddenness and unnaturalness of death affect general health, depression, and grief, a total of 215 responses to a questionnaire from the bereaved were analyzed. The respondents were divided into five groups: bereaved by suicide, accident, acute illness (<1 day from onset), shorter illness (<1 year from onset), and longer illness. Every sudden-death group indicated averages higher than the clinical threshold on general health scale and depression scale. After statistically controlling for respondents' age, the deceased's age at death, and the months spent with the deceased, differences among groups appeared on all but the subscales of somatic symptoms and of anxiety and insomnia. The difference seemed more apparent on emotional reaction than on physical distress. On pairwise tests suicide was found to be the most distinctive bereavement.


  1. Top of page
  2. Abstract

It is a very common phenomenon that family members suffer grief in bereavement. Grief is considered to be a healthy emotional experience or an active psychological response by which an individual adapts to changes caused by bereavement. It is regarded as pathological, however, when its intensity, onset, duration or manner is inappropriate. Grief is also thought to be a risk factor in poor health because intense and prolonged grief may lead to mental or physical deterioration. Some regard it as identical with a health hazard. It is almost impossible to discriminate between normal and pathological grief symptomatically because both of them are seemingly the same,1 and the criteria for the discrimination differ among cultures.2 Besides grief, many researchers have noted that bereavement itself may result in poor mental or physical health.1,3–5 Although it is still not clear whether grief is an emotional experience or a symptom,1,2,5 the characteristic reaction to bereavement is referred to as ‘bereavement reaction’.

The cause of death has been frequently examined as a factor affecting bereavement reaction. While there is much evidence that unanticipated or unnatural death leads to a severer reaction, it is still unclear how far the impact on survivors differs in a comparison between suicidal loss, accidental loss, loss from endogenous sudden death, loss from chronic illness, and loss from more prolonged illness. The present study, using the action research approach, attained a relatively large sample size that allowed us to examine the effect of death causes on bereavement reaction. The aim of the present study was to explore the differences among death causes by classification in terms of both suddenness and unnaturalness, considering general health, depressive mood, and grief reaction as outcome variables; the former two variables concern physical and mental health, while the latter concerns psychological experiences after bereavement regardless of whether normal or pathological. The detailed aspects of general health and grief reaction are also considered by comparing the scores of each subscale.


  1. Top of page
  2. Abstract

Subjects and setting

Participants in self-help groups and seminars for the bereaved were asked to complete a questionnaire. The responses were returned in the groups or during seminars, except for some by mail.

All respondents were informed of the researcher's identity, the purpose of the study, the voluntariness of their partiapation in the study, and our pledge to preserve their privacy.

Background factors

The questionnaire queried the respondent's sex, age, employment status, months since bereavement, kinship (parent, spouse, child, etc.), age of the deceased at death, months spent with the deceased in the same house, and whether the respondents had been taking medications or not.

Stressful life events have also been considered influential in bereavement reaction.1,4,6 The respondents were asked to address the following six items: “You and your spouse, or your parents are separated”, “Another relative also died or suffered severe illness”, “Your living environment changed”, “You lost or retired from your job”, “You fell ill or got hurt”, and “You celebrated an auspicious event”. Although it was desirable to quantify these events by using some more standardized scale, we decided against doing so to spare the feelings of the respondents.

Cause of death

The respondents were asked to choose the cause of their loss, whether illness, accident, suicide, or others. If the cause was illness, the questionnaire also asked days under medical treatment.

Outcome measures

The General Health Questionnaire (GHQ), which was developed for discrimination in non-organic and non-psychotic disorders,7 is a 2-point self-reporting scale and contains 28 items representing poorer health by higher scores. Scores range from 0 to 28. The clinical threshold is considered to be a score of 5. The scale has the following four subscales of seven items: Somatic Symptoms; Anxiety and Insomnia; Social Dysfunction; and Severe Depression. The present study used the GHQ version translated into Japanese.8

The Self-Rating Questionnaire for Depression (SRQ-D), used to assist in the diagnosis of mild or masked depression,9,10 is a 3-point self-reporting scale and contains 18 items. It has a range of 0–36 and higher scores represent higher severity of depression. The clinical threshold is considered to be a score of 11–15.

The Miyabayashi Grief Measurement (MGM) is a standardized 5-point self-reporting scale for measuring the degree of grief and contains 26 items that are categorized into four subscales: 12 items of Cherished Reminiscence with a score range of 0–48, five items of Alienated Feelings with a range of 0–20, five items of Mood Stability with a range of 0–20, and four reversed items of Adaptive Effort with a range of 0–16.11 The sum (range 0–104) of the four subscales represents the total level of grief reaction.

The questionnaire also asked whether the respondent was taking any medications at that time.

Sample selection and group definition

At first, the data included 371 first-degree relatives. We excluded responses missing age or cause of loss. Suicidal or accidental loss with terminal period >1 day was also omitted from the data. Therefore the sample size was reduced to 307.

It is known in Japan that loss of a parent is less influential in bereavement reaction than other types of loss.12–14 After attempting preliminary analyses, we decided to exclude 92 responses from bereaved children, whose outcomes were significantly lower than those of bereaved spouses or bereaved parents; then the effect of kinship was effectively controlled. Finally, the study sample was composed of 215 responses from bereaved spouses and bereaved parents.

According the cause of death, the responses were divided into five groups: suicide; accident; acute illness (<24 h from onset), shorter illness (>1 day and <1 year from onset), and longer illness (≥1 year from onset). The threshold of “1 year” is the conventional one used in previous Japanese studies.15,16


To compare the five groups either the χ2 test or the General Linear Model (GLM) analysis (anova or ancova) was conducted. Multiple comparison (Bonferroni t) test was performed when the GLM analysis indicated significance. The analyses were undertaken mainly with SPSS version 10 (SPSS, Chicago, IL, USA).


  1. Top of page
  2. Abstract

The study sample consisted of 174 women (80.9%) and 41 men (19.1%). The mean age was 60.5 ± 12.0 years (range, 29–91 years). Among 215 respondents 87 (40.5%) were employed, 116 (54.0%) were unemployed, and 12 (5.6%) did not respond. Months after bereavement ranged from 0 to 175 with a mean of 5.65 ± 4.12 months, and a median of 4.97 years. Regarding kinship, 163 (75.8%) had lost a spouse and 52 (24.2%) had lost a child. The mean age of the deceased at death was 49.0 ± 21.6 years (range, 0–89 years).

Among all 215 respondents, 21 (9.8%) were bereaved due to suicide, 23 (10.7%) due to accident, nine (4.2%) due to acute illness, 74 (34.4%) due to shorter illness and 88 (40.9%) due to longer illness.

After initially attempting to include kinship as a factor in the general linear models, we ultimately opted not to differentiate between conjugal loss and parental loss because this factor had no effect in every analysis.

Background factors

The five groups did not differ significantly in sex, employment status, previously taking medicine (χ2[4] = 5.415, 4.312, 6.708; n.s.), or months since bereavement (F[4,210] = 1.412, n.s.).

The mean age of the respondents differed among the groups (F[4,210] = 4.314, P < 0.01). In the shorter illness group the deceased was older than in the suicide or accident group (P < 0.05). The mean age of the deceased at death differed among the groups (F[4,210] = 14.890, P < 0.001). In the suicide or accident group the deceased was younger than in every illness group (P < 0.05).

The groups differed in mean months spent with the deceased in the same house (F[4,210] = 6.313, P < 0.001). The accident group had spent fewer months with the deceased than those in every illness group (P < 0.01).

The occurrence of each stressful life event did not differ among the groups (χ2[4] = 5.243, 1.193, 3.125, 6.942, 0.999, 5.554; n.s.).

Outcome measures

For each outcome measure, ancova was performed for cause of death with three covariates: respondents' age, deceased's age at death, and months spent with the deceased. The sample sizes were not equivalent due to missing values. Table 1 shows the estimated marginal means and SE on each scale.

Table 1.  Scale scores
Scales:SuicideAccidentsAcute illness (<1 day)Shorter illness (<1 year)Longer illness (≥1 year)
  1. Means were estimated with the overall means of three covariates (i.e. respondents' age, deceased's age at death, and months spent with the deceased).

  2. GHQ, General Health Questionnaire; MGM, Miyabayashi Grief Measurement; SRQ-D, Self-Rating Questionnaire for Depression.

GHQ Total218.981.31237.941.2397.911.88695.720.69824.660.3
 Somatic Symptoms211.810.37231.490.3492.110.53691.500.19841.040.17
 Anxiety & Insomnia212.580.43232.570.4092.440.62692.090.23841.770.20
 Social Dysfunction212.710.39232.200.3691.440.56691.440.20841.050.18
 Severe Depression211.910.37231.730.3591.890.53690.680.20840.680.18
MGM Total2059.115.792253.985.43861.798.607037.702.958242.132.70
 Cherished Reminiscences2131.512.972328.542.79833.714.537121.491.558423.111.41
 Alienate Feeling208.911.27228.931.19811.571.89705.360.65826.110.59
 Mood Stability2011.101.15229.561.0889.331.71706.470.59826.890.54
 Adaptive Effort (rev.)208.360.97227.180.9187.231.44704.520.49825.890.45

As Fig. 1 shows, the mean scores of total GHQ differed significantly among the groups (F[4,196] = 3.032, P < 0.05), and those of all the groups except the longer illness group were higher than the clinical threshold. Multiple comparison tests indicated that the general health of the suicide group was poorer than that of the longer illness group (P < 0.05).


Figure 1. General Health Questionnaire total score. Acute illness, death <1 day from onset; shorter illness, death <1 year from onset; longer illness, death ≥1 year from onset. *P < 0.05; Bonferroni t-test.

Download figure to PowerPoint

Figure 2 shows the mean score of each GHQ subscale. No group difference was found in either Somatic Symptoms (F[4,198] = 1.908, n.s.) or Anxiety and Insomnia (F[4,198] = 1.302, n.s.). Modes of death were significantly different for Social Dysfunction (F[4,198] = 4.412, P < 0.01) with the Suicide group being higher than the Longer Illness group (P < 0.01), and for Severe Depression (F[4,198] = 4.319, P < 0.01), with the Suicide group again scoring higher than each chronic illness group (P < 0.05).


Figure 2. Subscale scores of the General Health Questionnaire. Acute illness, death <1 day from onset; shorter illness, death <1 year from onset; longer illness, death ≥1 year from onset. *P < 0.05; **P < 0.01; ANCOVA and Bonferroni t-test.

Download figure to PowerPoint

As Fig. 3 shows, mean SRQ-D scores differed among the groups (F[4,181] = 6.659, P < 0.001). While the shorter illness group was lower than the clinical threshold, every sudden death group was higher than this, and the longer illness group was as high as the threshold, in terms of the average. Multiple comparison tests showed that the suicide group was more depressed than any chronic illness group (P < 0.001).


Figure 3. Self-Rating Questionnaire for Depression. Acute illness, death <1 day from onset; shorter illness, death <1 year from onset; longer illness, death ≥1 year from onset. ***P < 0.001; Bonferroni t-test.

Download figure to PowerPoint

As Fig. 4 shows, the mean scores of total MGM differed among the groups (F[4,194] = 4.395, P < 0.01). The mean score of the suicide group was higher than that of the shorter illness group (P < 0.05).


Figure 4. Miyabayashi Grief Measurement total score. Acute illness, death <1 day from onset; shorter illness, death <1 year from onset; longer illness, death ≥1 year from onset. *P < 0.05; Bonferroni t-test.

Download figure to PowerPoint

Figure 5 shows the converted mean score of each MGM subscale into a 0.0–1.0 scale. The effect of the cause of death was found on all the subscales. Cherished Reminiscence (F[4,199] = 3.750, P < 0.01) was more apparent in the suicide group than in the shorter illness group (P < 0.05). Alienated Feelings (F[4,194] = 4.253, P < 0.01) were more apparent in the acute illness group than in the shorter illness group (P < 0.05). Mood Stability (F[4,194] = 4.200, P < 0.01) was more obvious in the suicide group than in the shorter illness group (P <0.01) or longer illness group (P < 0.05). Adaptive Effort (F[4,194] = 3.835, P < 0.01) was greater in the shorter illness group than in the suicide group (P < 0.01).


Figure 5. Subscale scores of the Miyabayashi Grief Measurement scale. Acute illness, death <1 day from onset; shorter illness, death <1 year from onset; longer illness, death ≥1 year from onset. The scores were converted into 0.0–1.0 scores. *P < 0.05; **P < 0.01; ***P < 0.001; ANCOVA and Bonferroni t-test.

Download figure to PowerPoint

Modes of death made no difference in taking any medications at that time (χ2[4] = 7.467, n.s.).


  1. Top of page
  2. Abstract

The aim of the present study was to test the effect of cause of death. The study offers certain refinements not present in many bereavement studies. First, the study compared the five categories of loss. Cleiren classified modes of death, which involve conceptual overlap in suddenness and unnaturalness, into three categories: sudden unnatural death, sudden natural death, and anticipated natural death.17 The present study covered all three categories. Second, the outcome was measured multidimensionally. Both health and grief are important aspects of bereavement experience. Nevertheless, there have been relatively few studies evaluating grief using reliable, empirically derived measures.18 Third, the groups did not differ in sex, employment status, previous medication, length of time since the death, and experience of stressful life events, all of which might have complicated the findings. Fourth, the study excluded the confounding effect of kinship. The survivors' age, the deceased's age at death, and the months spent with the deceased were statistically controlled. These variables are inevitably confounded with the cause of death factor because the major cause of death varies among generations. Such a method for controlling extraneous variables is not a perfect one but it may provide more precise examination of the cause-of-death factor.

Many investigators have found unanticipated or unnatural loss to be associated with more severe bereavement.11,14–17,19–27 The present data generally supported the results of previous studies reporting effect of cause of death on bereavement reaction. However, in contradiction to some reports that found that the effect was significant on perceived health,21,26 we found that the effect was not significant on the two GHQ subscales of Somatic Symptoms and Anxiety and Insomnia, nor was the effect significant on medication at that time, which is a measurement of health-related behavior. This suggested that the impact of the cause-of-death factor was more apparent in mental rather than physical manifestations in the present sample, in which the mean duration after death was relatively long. Emotional reactions might be more persistent than physical reactions,28 although it is also suspected that the scales may have varied in sensitivity.

The suicide group significantly differed from any non-suicide group(s) on eight of the 11 indices (Figs 1–5). This partly supported general consideration that suicide was the most devastating bereavement.29 Suicide is often preceded by a history of severe mental problems.30 Shepherd and Barraclough reported that the outcome of suicidal bereavement was equally divided between “better off” and “worse off”.31 Therefore, it is also presumable that suicidal loss is somewhat anticipated and thus has less impact on the bereaved. The present result, however, did not support this presumption. The suicide group was more distinctive than the other sudden death groups in terms of statistical significance. Although it is impossible to determine whether such a distinctiveness of suicidal loss resulted from its suddenness or unnaturalness, at least we may conclude that the anticipation of suicide had no substantial effect on reducing the reaction in the present Japanese sample, even if the suicidal death were anticipated. Intensified reaction may have been consequent on other factors such as the unacceptable character of suicide in society.1,6,17,32

The outcome scores of the chronic illness groups were significantly or seemingly lower than the other groups, suggesting the effect of anticipation. However, no difference was found between the acute illness group and the chronic illness groups, except on Alienated Feeling (Fig. 5). The operationalization of the present study was not so precise as to discern the concept of anticipation. In the context of a bereavement study the suddenness of death depends on subjective rather than objective factors.6,33,34

Regarding descriptive statistics, the mean GHQ and SRQ-D scores showed that many of the bereaved remained in poor health, especially in terms of mental health, despite the fact that most of them had been bereaved for >5 years. This result is consistent with the evidence that the impact of bereavement persists for several years.3,25,29,35–37 The mean scores on these scales exceeded the clinical thresholds, except for those bereaved due to chronic illness. It seems reasonable therefore to regard bereavement as a risk factor for health.

The present sample included a much greater proportion of female subjects, with bereavement generally involving many cases of conjugal loss, which inevitably involves many widows. Such a gender imbalance, often seen in bereavement surveys,38 must be addressed, but it did not prevent us from comparing the groups because no difference was found in respect of sex. Another problem is the sample size of each group. It is difficult in a small sample test to reach a significant level even when substantial differences exist. This prompts the need for future investigations with larger samples to refine the findings.


  1. Top of page
  2. Abstract
  • 1
    Burnell GM, Burnell AL. Clinical Management of Bereavement: A Handbook for Healthcare Professionals. Human Science Press, New York, 1989.
  • 2
    Averill JR, Nunley EP. Grief as emotion and disease. In : StroebeMS, StroebeW, HanssonRO (eds). Handbook of Bereavement: Theory, Research, and Intervention. Cambridge University Press, Cambridge, 1993; 7790.
  • 3
    Parkes CM, Weiss RS. Recovery from Bereavement. Basic Books, New York, 1983.
  • 4
    Parkes CM. Bereavement: Studies of Grief in Adult Life, 3rd edn. Routledge, New York, 1996.
  • 5
    Raphael B, Dobson M. Bereavement. In : HarveyJH, MillerED (eds). Loss and Trauma: General and Close Relationship Perspectives. Taylor & Francis, Philadelphia, PA, 2000; 4561.
  • 6
    Wright B. Sudden Death: A Research Base for Practice, 2nd edn. Churchill Livingstone, New York, 1996.
  • 7
    Goldberg DP, Hillier VF. A scaled version of the General Health Questionnaire. Psychol. Med. 1979; 9: 139145.
  • 8
    Nakagawa Y, Obo I. Nihonban GHQ Seishin Kenkoh Chohsahyoh Tebiki. Nihon Bunka Kagakusha, Tokyo, 1985 (in Japanese).
  • 9
    Kawano T, Suematsu H, Shinzato R. Shinri Test. Asakura Shoten, Tokyo, 1999 (in Japanese).
  • 10
    Narita T, Sato T, Hirano S, Kusunoki K. Reliability and validity of Kasahara's Depression Inventory. Jpn. J. Clin. Psychiatry 1999; 28: 555562 (in Japanese).
  • 11
    Miyabayashi S. Substructure of Japanese grief reaction and grief care. Ochanomizu Med. J. 2003; 51: 2745 (in Japanese).
  • 12
    Kawaai C. Izokuno sohshitsu, hitanno shien. In: OkadoT, UenoH, ShigaN (eds). Byohkito Itamino Shinri.Shibundo, Tokyo, 2000; 243253 (in Japanese).
  • 13
    Sakaguchi Y. Secondary stressors and mental health after bereavement: Comparison between spouses and children. Jpn. J. Fam. Psychol. 2001; 15: 1324 (in Japanese).
  • 14
    Ando K, Matsui Y, Fukuoka K. Psychological responses to bereavement: A preliminary examination. Bull. Sociol. Toyo Univ. 2004; 42: 6384 (in Japanese).
  • 15
    Kawaai C. Haiguushatono shibetsu: Sono shinrito taioh. In: DeekenA, ShigekaneY (eds). Hanryoni Sakidatareta Toki. Shunjuusha, Tokyo, 1988; 457 (in Japanese).
  • 16
    Terasaki A, Nakamura K. Factors influencing grief due to bereavement among elderly widows and widowers. Jpn. J. Public Health 1998; 45: 512525 (in Japanese).
  • 17
    Cleiren MPHD. Bereavement and Adaptation: A Comparative Study of the Aftermath of Death. Hemisphere, Washington, DC, 1993.
  • 18
    Vargas LA, Loya F, Hodde-Vargas JH. Exploring the multidimensional aspects of grief reactions. Am. J. Psychiatry 1989; 146: 14841488.
  • 19
    Gerber I, Rusalem R, Hannon N, Battin D, Arkin A. Anticipatory grief and widows and widowers. J. Gerontol. 1975; 30: 225229.
  • 20
    Ball JF. Widow's grief: The impact of age and mode of death. Omega 1977; 7: 307333.
  • 21
    Demi A. Adjustment to widowhood after a sudden death: Suicide and nonsuicide survivors compared. Commun. Nurs. Res. 1978; 11: 9199.
  • 22
    Vachon MLS, Rogers J, Lyall WAL, Lancee WJ, Sheldon AR, Freeman SJJ. Predictors and correlates of adaptation to conjugal bereavement. Am. J. Psychiatry 1982; 139: 9981002.
  • 23
    Rando TA. An investigation of grief and adaptation in parents whose children have died from cancer. J. Pediatr. Psychol. 1983; 8: 320.
  • 24
    Lundin T. Morbidity following sudden and unexpected bereavement. Br. J. Psychiatry 1984; 144: 4554.
  • 25
    Lundin T. Long-term outcome of bereavement. Br. J. Psychiatry 1984; 145: 424428.
  • 26
    Shanfield SB, Swain BJ, Benjamin GAH. Parents' responses to the death of adult children from accidents and cancer: A comparison. Omega 1987; 17: 289297.
  • 27
    Middleton W, Raphael B, Burnett P, Martinek N. A longitudinal study comparing bereavement phenomena in recently bereaved spouses, adult children and parents. Aust. N.Z. J. Psychiatry 1998; 32: 235241.
  • 28
    Lund DA, Caserta MS, Dimond MF. Gender differences through two years of bereavement among the elderly. Gerontologist 1986; 23: 314320.
  • 29
    Henley S. Bereavement by suicide: The effect of a suicidal death upon surviving relatives. Bereavement Care 1983; 2: 67.
  • 30
    Robins E, Murphy GE, Wilkinson RH, Gassner S, Kayes J. Some clinical considerations in the prevention of suicide based on a study of 134 successful suicides. Am. J. Public Health Nation's Health 1959; 49: 888899.
  • 31
    Shepherd D, Barraclough BM. The aftermath of suicide. Br. Med. J. 1974; 2: 600603.
  • 32
    Sanders CM. Risk factors in bereavement outcome. In: StroebeMS, StroebeW, HanssonRO (eds). Handbook of Bereavement: Theory, Research, and Intervention.Cambridge University Press, Cambridge, 1993; 255267.
  • 33
    Volkan V. Typical findings in pathological grief. Psychiatr. Q. 1970; 44: 231250.
  • 34
    Fulton R, Gottesman DJ. Anticipatory grief: A psychosocial concept reconsidered. Br. J. Psychiatry 1980; 137: 4554.
  • 35
    Barrett CJ, Schneweis KM. An empirical search for stages of widowhood. Omega 1980; 11: 97104.
  • 36
    Lehman DR, Wortman CB, Williams AF. Long-term effects of losing a spouse or child in a motor vehicle crash. J. Pers. Soc. Psychol. 1987; 52: 218231.
  • 37
    Wortman CB, Silver RC. The myths of coping with loss. J. Consult. Clin. Psychol. 1989; 57: 349357.
  • 38
    Stroebe MS, Stroebe W. Who suffers more? Sex differences in health risks of the widowed. Psychol. Bull. 1983; 93: 279301.