Abstract
Background. Feelings of guilt are common after bereavement. We investigated whether feelings of guilt after the loss of a husband to cancer are associated with the health care provided at the time close to and at the moment of death. Materials and methods. The study population consisted of 506 widows of men who died of prostate cancer in 1995 or of urinary bladder cancer in 1995 or 1996 at the ages 45–74 years. We collected information on the received health care at the time of the husband's death from the widows, through a postal questionnaire. Results. Widows who perceived that their husbands did not get enough pain relief had an increased relative risk of 1.7 (95% CI 1.1–2.8), for guilt feelings, compared to widows who felt that their husbands had adequate pain relief. If a widow considered her husband being exposed to less satisfactory care or treatment, she had an almost two-fold increased relative risk, 1.9 (95% CI 1.2–3.1), for guilt feelings after the husband's death, compared to a widow who thought that satisfactory care or treatment was provided. Discussion. Feelings of guilt after bereavement may occur in response to the perception of inadequate health care during the last months and at the actual moment of death of the significant other.
To lose a loved one is a traumatic life event, regardless of the underlying cause of death, and entails increased risks for both psychological and physical morbidity, and mortality Citation[1–4]. Bereavement is normally followed by grieving, an emotional response including guilt as one of the expressions Citation[5]. Guilt, defined as a remorseful awareness of having done something wrong, however, is a complex and largely unexplored emotion in relation to bereavement. So called ‘survivors guilt’ often occurs after having experienced traumatic events where others die, in war or natural disasters or similar Citation[6]. Parents who have lost a child through stillbirth or neonatal death frequently experience guilt, regardless of the circumstances of their loss Citation[7], Citation[8]. The most persistent and intense form of guilt after bereavement would probably be associated with a perception of somehow having contributed to the death or suffering of the loved one. This perception may result from a feeling of having caused the disease through, for example improper feeding, or lack of affection or support. The perception could also origin from a feeling of not having prevented the disease through for instance letting the person smoke, keep an unhealthy lifestyle, or not pushing physicians hard enough to detect or treat the disease.
Certain individuals may be more prone to react with guilt in situations of bereavement Citation[8], which could lead to a less favourable long-term adjustment after the loss in these individuals Citation[9]. The long-term effect of loss may further be modified by health care related factors, such as the experience of the significant other's unrelieved symptoms Citation[10]. Indeed, in a previous study we found an increased risk of guilt in parents during the first year after having lost a child to cancer if the parents had a perception of inadequate health care Citation[11]. Guilt was particularly evident among parents who were not confident that their child would receive immediate help at the hospital if needed, and among parents who perceived that the health care providers were not competent enough Citation[11].
We undertook the current study to test whether the association between inadequate health care and subsequent feelings of guilt also could be demonstrated in another survivor population, namely among 379 widows who lost their husbands to cancer in the years 1995 to 1996.
Materials and methods
Study population
This population-based nationwide Swedish study utilized The Swedish National Register of Cause of Death to identify 928 men, who died of prostate cancer in 1996 or urinary bladder cancer in 1995 or 1996 at the ages of 45 to 74 years, and who were diagnosed with cancer at least 90 days prior to their death. The female partners/wives (here called widows) of the men were identified from the Swedish Population Register as having the same address as the deceased at the time of his death. Of 601 identified widows, 506 met the inclusion criteria of our study; alive and under 80 years old as of April 1, 1999, and having a listed telephone number. The 506 widows included in our study population were initially approached with an introductory letter explaining the study objective followed by a telephone call to collect informed consent for participation in our study. Those who agreed to participate subsequently received a postal questionnaire. The study was approved by the Regional Ethics Committee at the Karolinska Institute, Stockholm, Sweden.
Questionnaire
The questionnaire was developed through in-depth interviews followed by face-to-face validation, and included 275 questions divided into three sections; the first referring to the period preceding death and the actual moment of death, the second referring to the first six month after the husband's death, and the third included questions regarding the widow's current life situation and health. The validation procedure for the questionnaire has previously been described in detail Citation[4]. Our main outcome – feeling of guilt – was assessed through a single question “Did you have any feeling of guilt after the death of your husband?” with a dichotomous answer “Yes/No” and referred to the six month period immediately after the husband's death. The term guilt was left open to be interpreted by the respondent. The independent variables included various questions about the patient's distress, the health care provided, and the information and support given both to the deceased and his wife during the last months before the husband's death and at the actual moment of his death. Since reports of guilt in widows are likely to be confounded by depression in the widow after the husband's death Citation[12], self-reported information on depression was collected through a well-validated Citation[10], Citation[13–15], seven digit visual-digital scale in response to the question “Have you felt depressed during the past week?”. The widow was considered depressed if her answer corresponded to values between 3 and 7 on the digital scale, and non-depressed if she indicated a value between 1 and 2.
Statistical analysis
For the outcome – presence or absence of guilt feelings – the percentages of subjects in each category of the independent variables were calculated. Subsequently, the relative risk (RR), together with a 95% confidence interval for the RR (using the Mantel-Haenszel formula for the variance) Citation[16], was calculated as the ratio of the percentages of each category of the independent variable among widows with and without the outcome. Multivariable logistic regression models were then used to determine the relative importance of the independent variables for explaining the variation in the outcome. The independent variables were grouped according to different aspects of the health care provided; the widow's access to support and information, the deceased person's access to support, information, and symptoms relief, and the degree to which the deceased person's needs were met by the health care services. Analyses were initially performed separately within each group. We evaluated a possible distorting effect of confounders by forward and backward selection process looking at changes in the chi-square value when introducing or eliminating each new independent variable into the model. Only the variables age and education contributed to a substantial change in χ2 estimate (p < 0.05) and were thus adjusted for in the multivariable models. Apart from these variables we included depression status “Yes/No” in the final multivariable model, since we consider this variable to be a likely confounder of our studied associations. In order to produce adjusted relative risks the procedure Genmod, utilizing a log link and binomial error distribution, were used in SAS version 8.
Results
Among the 506 eligible widows, 379 (75%) returned the questionnaire. Reasons for non-participation included: psychological burden (7%), no given reason (7%), did not return questionnaire despite agreement to participate (5%), illness (3%), not reachable (2%), unknown or other causes (2%), and deceased (<1%). Those who declined to participate did not differ from participants with regards to their age distribution or their area of recidency. The demographic characteristics of the 379 widows who answered the questionnaire are displayed in . Their mean age was 67.6 years and, on average, three years had passed since the death of their husband. The majority of widows had been married for more than thirty years (78%), and had two or more children together (67%) with the deceased. Most widows had completed only elementary school (59%), were employed (47%), and confessed to the Swedish protestant church (86%). Feeling of guilt was reported by about one third (29.4%) of the widows during the six months after the husband's death.
Table I. Characteristics of 379 widows who lost a husband in prostate cancer or bladder cancer during the period 1995–1996.
The younger widows (<65 years old) were more prone to guilt feelings than the older widows (> = 75 years old) (). The increased relative risk of 1.9 (95% CI 1.0–3.6) persisted after adjustment for the widow's education and her self-reported depression. Risk of guilt feeling was not associated with any other demographic characteristic of the widows (). The risk for feelings of guilt in relation to the widow's access to support and information is illustrated in . The widow's perception of not having received honest information from caregivers was associated with an adjusted relative risk of 1.5 (95% CI 1.0–2.4) for feelings of guilt, compared to a perception of having received honest information from caregivers (). Other indicators of the widow's access to support and information during the last three months of the husband's life were not associated with increased risk for guilt ().
Table II. Unadjusted and adjusted relative risks for feeling of guilt among bereaved widows in relation to demographic characteristics at time of the deceased diagnosis.
Table III. Unadjusted and adjusted relative risks for feeling of guilt among bereaved widows in relation to the widow's access to support and information during the husbands last 3 months of life.
Widows who felt that their husbands did not receive enough pain relief during their last three months of life, had a 70% increased adjusted relative risk, 1.7 (95% CI 1.1–2.8), of feeling guilt, compared to those who were entirely sure that enough pain relief was provided (). Also, the perception that the husband did not get enough support or enough information by the caregivers was related to an adjusted relative risk of 1.6 (95% CI 1.0–2.5), and 1.6 (95% CI 1.0–2.6) respectively, for feelings of guilt in widows, as compared to widows considering that their husbands did receive enough support and information ().
Table IV. Unadjusted and adjusted relative risks for feeling of guilt among bereaved widows in relation to the deceased access to support during the last 3 months of life.
A perception of not having good access to doctors during the husband's last three months of life was associated with feelings of guilt in widows, with an adjusted relative risk of 1.6 (95% CI 1.0–2.7), as opposed to being certain that the husband had good access to doctors (). If the widow considered her husband being exposed to less satisfactory care or treatment during his last three months of life, she had an almost two-fold increased adjusted relative risk 1.9 (95% CI 1.2–3.1) of guilt feelings after the husband's death, compared to if satisfactory care was reported ().
Table V. Unadjusted and adjusted relative risks for feeling of guilt among bereaved widows in relation to the health care provided during the husbands last 3 months of life.
Discussion
Our results suggest that inadequate health care during the terminal stage of disease may contribute to the development of guilt feelings in widows. The ramifications of this with regard to the palliative care guidelines have yet to be addressed. Most of the investigated variables showed no or little association with the variations of feelings of guilt. However, we found increased risk for guilt in widows who considered that their husbands did not receive enough pain relief during his last three months compared to widows who thought pain relief was adequate. Furthermore, widows who considered that their husbands were exposed to less satisfactory care and treatment were two times more likely to react with guilt than widows that considered their husband's care being satisfactory. Our results are consistent with some of the findings of a previous investigation of feeling of guilt in relation to the provided health care among parents who lost a child to cancer Citation[11]. That study indicates that feelings that their child had little or no access to pain relief, dietary advice, anxiety relief, or relief of other psychological symptoms increased the risk for subsequent guilt two-fold in the parents. Parents who considered the staff being incompetent and who were unsure whether their child would receive immediate help at hospital were at a particularly high risk for guilt after the child's death.
Both those who are dying from cancer and his or her relatives are generally very dependent on the health care during the last months of life. This fact can in itself be difficult to handle for an individual who has been the main source of support to the deceased for perhaps several decades. It is conceivable that the widows often considered themselves better judges of their husbands’ needs than the health care staff. Yet, clinical experience indicates that some of the widows in our study population belonged to a generation where decisions of doctors and the treatment provided should not be questioned. Nevertheless, unmet needs, such as insufficient symptom control and emotional support are common among terminal cancer patients, and are likely to put them at higher risk of psychological distress Citation[17–19]. This emotional distress could in turn affect the significant other in several ways. Earlier evidence suggests that the patient's unrelieved psychological symptoms (not including pain and other physical symptoms) increased the widow's risk of long-term depression and anxiety Citation[10]. Since depression is also closely related to guilt Citation[12], depression could act as a confounder of our studied associations. We therefore chose to adjust for self-reported depression in our analyses to be able to adequately assess predictors of guilt per se. Interestingly, after adjustment for self-reported depression the increased risk for feelings of guilt was still evident among widows who experienced that their husbands pain relief was unmet.
The health care today can be found demanding; on one hand the decision making and participation in the care by the relatives have increased, on the other hand the treatment options can be plentiful and the decisions to take may have serious consequences. Inevitably, these factors put a lot of pressure on significant others. An important task for the health care providers is thus to be receptive for the relatives’ needs, wishes, and not the least his or her capacity to participate in the care. Particularly, to handle the situation around the moment of death in an adequate way has proven to be important for several reasons. Parents, who sense that their dying child is aware of the imminent death, more often regret not having talked about death with their child than parents who did not sense the child's awareness Citation[20]. Previous results suggest that the most feasible means of reducing bereaved parents long-term psychological burden is to minimize the child's physical pain and improve the care at the moment of death Citation[21]. Furthermore, having the possibility to spend as much time as needed with the stillborn child and collecting a token of remembrance influence later psychological suffering in the bereaved parent Citation[14].
The strengths of our study include our non-selected, nation-wide, population-based cohort of widows, utilizing registries covering virtually 100% of all eligible individuals. Additionally, we collected information anonymously through a well-validated questionnaire, which should ensure honest and adequate answers. The detailed questionnaire included information on several, if not all, potential confounders. Our analyses did not indicate that the relation between an experience of inadequate health care and feelings of guilt could be explained by the distorting effect of one or several confounders. Nevertheless, our results should be interpreted with caution for the following reasons. The retrospective nature of our study, could affect the recall of the offered health care, which would be a possible source of bias. Previous research suggests that the memory among bereaved individuals may be affected by their emotional status Citation[22–24]. Thus, a widow with depression or anxiety could potentially view the past in a negative manner and thus report her husband as having more distress during his last months of life than he actually had. However, in our study, there was no association between the widow's psychological symptoms and the husband's pain or other physical symptoms, which indicates that a depressed widow was as able as a non-depressed widow to report low level of distress in her late husband Citation[10]. Besides, we do not believe that guilt-prone widows would be more likely to over-report inadequate health care compared to other widows. Particularly so, since most of the potential predictors of guilt were not associated with increased risk for guilt. Thus, memory fallacies, if any, would likely be non-differential, and consequently could only lead to an underestimation of our risk estimates.
Another potential limitation with our data is the lack of a confirmed diagnosis of depression. We are dependent on the widow's self-reported feelings of depression. Yet, we are assessing the self-experienced emotion of guilt, thus we feel confident that including self-experienced feeling of depression in the analyses should adequately adjust for any confounding of depression on our outcome. Our follow-up time in this study was a compromise between our objective to investigate long-term morbidity in widows and the widow's expected memory capacity. However, we had only access to information regarding the widows’ feelings of guilt during the first six months following their husbands’ deaths. Regardless of our population-based sample, it is important to stress that our results relate to the health care setting in Sweden in the 1990's. Consequently, the results could not automatically be generalized to other populations where various culture specific factors may influence the relation between an experience of inadequate health care and guilt after bereavement, and where different circumstances for health care exist.
Despite the above mentioned caveats, our finding that provided health care affect later feelings of guilt in the bereaved is intriguing. Stringently, our observational data can not be translated to a prediction of the effect of any intervention. The growing pile of evidence showing that a dying patient's wellbeing during the last weeks in life influence the next-of-kin's long-term psychological status and morbidity gives us arguments to improve the palliative care. Possible increased financial costs in the short-term for improving the palliative care should be balanced by long-term decreased financial costs due to less morbidity for the next-of-kin. Also, investigating effects of a psychological intervention to widows, for instance six months after death, by somehow addressing her feelings of guilt and experience of inadequate health care, may be worthwhile.
In conclusion, our study offers insight into potential mechanisms of guilt feelings after bereavement. Our findings suggest that the health care which is provided to the husband during the last months of life is associated with subsequent guilt in the widow. Particularly, a perception of not receiving enough pain relief and being exposed to less satisfactory care and treatment are important predictors of increased risk for guilt after the loss of a significant other. This knowledge may be of importance for improving the palliative care among individuals with terminal cancer.
Acknowledgements
This work was supported by grants from the Swedish Cancer Society, the Stockholm Cancer Foundation, The Research Council, and the VG-region. We also acknowledge all widows who made this study possible by sharing their experience.
Related Research Data
References
- Stoebe W, Stroebe M. Bereavement and health: The psychological and physical consequences of partner loss. Cambridge University Press, New York 1987
- Li J, Precht D, Mortensen P, Olsen J. Mortality in parents after death of a child in Denmark: A nationwide followup. Lancet 2003; 361: 363–7
- Kreicbergs U, Valdimarsdóttir U, Onelöv E, Henter J-I, Steineck G. Anxiety and depression in parents 4-9 years after the loss of a child owing to malignancy: A population-based follow-up. Psychol Med 2004; 34: 1431–41
- Valdimarsdóttir U, Helgason Á, Fürst C-J, Adolfsson J, Steineck G. Long-term effects of widowhood after terminal cancer: A Swedish nationwide follow-up. Scand J Public Health 2003; 31: 31–6
- Stroebe M, van Son M, Stroebe W, Kleber R, Schut H, van den Bout J. On the classification and diagnosis of pathological grief. Clin Psychol Rev 2000; 20: 57–75
- Rubin SS, Malkinson R, Witztum E. Trauma and bereavement: Conceptual and clinical issues revolving around relationships. Death Studies 2003; 27: 667–90
- Barr P. Guilt- and shame-proneness and the grief of perinatal bereavement. Psychol Psychotherapy: Theory Res Pract 2004; 77: 493–510
- Hasui C, Kitamura T. Aggression and guilt during mourning by parents who lost an infant. Bull Menninger Clin 2004; 68: 245–59
- Weinberg N. Does apologizing help? The role of self-blame and making amends in recovery from bereavement. Health Soc Work 1995; 20: 294–9
- Valdimarsdóttir U, Helgason Á, Fürst C-J, Adolfsson J, Steineck G. The unrecognised cost of a cancer patient's unrelieved symptoms: A nationwide follow-up of the surviving partners. Br J Cancer 2002; 86: 1540–5
- Surkan P, Kreicbergs U, Valdimarsdóttir U, Nyberg U-K, Onelöv E, Dickman P, et al. Perception of inadequate health care and feeling of guilt after the death of a child to a malignancy: A population-based long-term follow-up. J Palliative Med 2006; 2: 317–31
- O'Connor LE, Berry JW, Weiss J, Gilbert P. Guilt, fear, submission, and empathy in depression. J Affect Disord 2002; 71: 19–27
- Onelöv E, Steineck G, Nyberg U-K, Hauksdóttir A, Kreichbergs U, Henningsohn L, et al Measures of anxiety and depression in the oncology setting using visual-digital scales. Acta Oncol 2007 (in press).
- Radestad I, Steineck G, Nordin C. Psychological complications after stillbirth-influence of memories and immediate management: Population based study. BMJ 1996; 312: 1505–8
- Bergmark K, Avall-Lundqvist E, Dickman PW, Henningsohn L, Steineck G. Vaginal changes and sexuality in women with a history of cervical cancer. N Eng J Med 1999; 340: 1383–9
- Rothman KJ, Greenland S. Modern Epidemiology2nd ed. Lippincott-Raven, Philadelphia 1998; 271
- Morasso G, Capelli M, Viterbori P, di Leo A, Alberisio A, Costantini M, et al. Psychological and symptom distress in terminal cancer patients with met and unmet needs. J Pain Symptom Manage 1999; 17: 402–9
- Houts PS, Yasko JM, Harvey HA, Kahn SB, Hartz AJ, Hermann JF, et al. Unmet needs of persons with cancer in Pennsylvania during the period of terminal care. Cancer 1988; 62: 627–34
- Mor V, Masterson-Allenn S, Houts P, Siegel K. The changing needs of patients with cancer at home. Cancer 1992; 69: 829–38
- Kreicbergs U, Valdimarsdóttir U, Onelöv E, Henter J-I, Steineck G. Talking about death with children who have severe malignant disease. New Engl J Med 2004; 351: 1175–86
- Kreicbergs U, Valdimarsdóttir U, Onelöv E, Björk O, Steineck G, Henter J. Care related distress in parents losing a child to cancer: A population-based long-term follow-up. J Clin Oncol 2005; 23: 9162–71
- Broadbent M, Horwood P, Sparks J, de Whalley G. Bereavement Groups. Bereavement Care 1990; 9: 14–16
- Burke M, Matthews A. Autobiographical memory and clinical anxiety. Cogn Emot 1992; 6: 23–35
- Lubchuk M, Kristjanson LJ, Degner L, Blood P, Sloan JA. Perceptions of symptom distress in lung patients: 1 Congruence between patients and primary caregivers. J Pain Symptom Manage 1997; 14: 136–46