Mortality

Notwithstanding methodological shortcomings in both retrospective and prospective controlled studies, it is clear from the epidemiologic evidence that some people are at increased risk for mortality following bereavement. The most important evidence is from studies that demonstrate an increase in overall mortality among the recently bereaved (see Table 1 for a summary of the studies discussed in this section).

TABLE 1

Summary of Epidemiologic Evidence for Mortality Following Bereavement.

Kraus and Lilienfeld ³⁹ reported one of the earliest systematic studies on mortality in the bereaved. They retrospectively calculated mortality rates for widows and widowers, matching data from the National Office of Vital Statistics for 1956 with data from the 1950 census. Death rates of widowed subjects were compared to the rates for married men and women matched for age, sex, and race. The mortality ratios of widowed to married were strikingly higher at younger ages; as age increased the differences in mortality between the widowed and married decreased for men and women and for all races. Mortality rates for males who were widowed were consistently higher than those of female widowed.

Specifically, younger widows and widowers (ages 20-24) had the highest ratio of mortality for eight causes of death: vascular lesions of the central nervous system, arteriosclerotic heart disease, non rheumatic chronic endocarditis and other myocardial degeneration, hypertension with heart disease, general arteriosclerosis, tuberculosis, and influenza and pneumonia. When all these disease groups were combined, the mortality rate was at least seven times greater among the young widowed group (under age 45) than for the matched young married control group. The mortality rate for death from cardiovascular disease was 10 times higher for young widowers than for married men of the same age. Kraus and Lilienfeld concluded that as a group, the recently bereaved were at greater risk for mortality. Although provocative, these data did not take into account either the duration of widowhood or the fact that widows and widowers who do not remarry may have been in poorer health.

Cox and Ford ¹⁷ reanalyzed government records of 60,000 widows receiving pensions for the first time in 1927, and then identified from vital statistics those who died over the next five years. They compared the actual and expected numbers of deaths for the first five years of widowhood. Only during the second year following bereavement was there some excess in mortality, and it appeared most pronounced in women between the ages of 60 and 68.

Several studies of relatively small cohorts of widows and widowers ^9,10,25,75 found no significant increase in mortality in the first or subsequent years following bereavement. However, Young et al. ⁷⁷ found that among recent widowers over the age of 55, mortality rates were significantly higher for the first six months of bereavement than in married controls of the same age. No differences were found after six months.

Rees and Lutkins, ⁶³ in a prospective study, followed a cohort of 903 relatives of 371 residents of a village in Wales who had died during the previous year. These 903 persons were compared with a cohort of 878 nonbereaved individuals matched for age, sex, and marital status. These cohorts were followed for six years. Mortality rates were slightly higher for all types of bereaved relatives, but significant differences were only found among spouses during the first year. The mortality rate for the bereaved spouses was significantly greater than for the control spouses—12 percent of bereaved spouses died compared with 1 percent of nonbereaved spouses.

The most definitive study on mortality after bereavement was conducted by Helsing and his colleagues. ^30-32 In a 1963 health census conducted in Washington County, Maryland, data were obtained on 91,909 persons—98 percent of the noninstitutionalized population. The population was followed prospectively to 1975. The widowed population in 1963 was matched to a married population of the same race, sex, year of birth, and geographic category of residence. Any member of the married population who became widowed after entry into the study was withdrawn from that category and enrolled in the widowed population as of the date of death of the spouse.

Widowed men aged 55-74 exhibited a highly significant increase in mortality. Those younger, 19-54, also showed a difference in the relative risk of death, but because there were so few deaths in this group the differences did not achieve statistical significance. Interestingly, widowers over 75 did not have a significantly increased mortality. This finding is consistent with Clayton's ¹¹ observation that older widowed men who survive their spouses may be in better health than married men of the same age. There was no evidence that men's mortality was higher in the first year than in subsequent years of the study; for women, however, the data suggest that the risk of mortality is greatest in the second year following bereavement.

Although there were some differences in the mortality rates by age among females, and for widows as compared with married women, when education, social class, cigarette smoking, and other potential risk factors were controlled the differences disappeared. Control of these factors did not affect the significance of the mortality rates in men. ³²

At least half the men who were widowed before the age of 55 remarried during the course of the study. As would be expected, given the demographic composition of the population, ¹⁶ remarriage among women was far less common, with the remarriage rate in any age group being similar to that among males 20 years older. The differences in mortality rates between the widowed males who remarried and those who did not remarry were substantial. In fact, age-specific mortality rates among widowed males who remarried were lower than the rates among married males. The ratio of remarried/not remarried mortality rates for males ranged from about 1:8 to 1:2; the small numbers of widowed females who remarried plus the already low mortality rate among the widowed females made their remarried/not remarried ratios meaningless. Clearly, remarriage must be taken into account in any mortality study of bereavement. It should be noted, however, that it is not known whether marriage itself protects against ill health or whether good health is what permits remarriage.

Helsing and his colleagues ³² also found that there was a significant mortality difference for both sexes by change of address after widowhood. This high mortality rate among subjects who moved was due largely to those who moved into nursing homes, retirement homes, and chronic care facilities (presumably indicators of poor health). Finally, living alone was also associated with a higher mortality than living with others.

It has often been suggested that vulnerability to illness and death is increased following the suicide of someone close. Shepherd and Barraclough ⁶⁶ reported that spouses of suicides have no increased mortality in the first year of bereavement. Over a longer period, 58 months, there were 10 deaths in 44 spouses (23 percent), a trend that indicates survivors of suicidal deaths are at greater risk of mortality than survivors of other deaths. Compared with mortality rates of the married rather than widowed, this excess mortality is even more significant. Half the spouses who subsequently died were seriously ill at the time of the suicide. The authors felt that in addition to contributing to their own deaths the consequences of these illnesses might have precipitated the suicide. This is an example of the complicated interactions that arise in studying mortality following bereavement, and is consistent with similar findings about poor health and mortality in the Ward ^74,75 data.

The mortality studies reviewed in this section provide examples of somewhat different results due to the methodological and population variations mentioned earlier. Unless differences are very great they will not reach statistical significance in small samples. Comparisons of the retrospective study from vital statistics by Kraus and Lilienfeld ³⁹ and the prospective study by Helsing and Szklo ³¹ reveal somewhat different findings with regard to mortality rates and specific causes of death. These may be attributable to differences in sample characteristics— Kraus and Lilienfeld studied only widowers who had not remarried, whereas Helsing and Szklo included widowers who had remarried and examined the interaction between remarriage and mortality. The different findings may also be traced to changes in general mortality characteristics (such as reduction in cardiovascular death rates) during the 20 years between these two studies, or to differences stemming from retrospective and prospective study designs.

These differences notwithstanding, the weight of the evidence indicates that, up to age 75, widowed men are about one-and-one-half times more likely to die prematurely than their married counterparts. Although especially pronounced during the first year, the mortality rate for men who do not remarry continues to be elevated for many years. For widows, there is no increase in risk the first year, but several reliable studies find excess mortality in the second year.

It should be noted that there are very few studies dealing with mortality following any bereavement other than conjugal loss. The effect of the death of parents, children, or siblings has been virtually unstudied. With regard to parents, there is one highly controversial study. Levav ⁴¹ reanalyzed the data from Rees and Lutkins, ⁶³ focusing specifically on bereaved parents in the original sample. The 35 bereaved parents were compared with 29 control parents. When the accumulated deaths were compared over the five-year study period it was found that 34.3 percent of bereaved parents had died compared with 6.9 percent of nonbereaved parents. Although loss of a child is generally considered hazardous, ¹¹ these mortality figures seem very high, perhaps because of some unspecified characteristics of the particular population that was studied. The effects of this type of loss are discussed in detail in Chapter 4.

Death by Suicide. National statistics for the United States in the years 1949-1951 established that suicide rates were higher among the widowed than the married. ⁵¹ Since that time numerous studies from vital statistics and from survey data (e.g., Kraus and Lilienfeld, ³⁹ Bock and Webber, ² Stroebe et al., ⁶⁹ and Helsing et al. ³⁰) have confirmed this increase, especially among elderly men. Among women the suicide rate is not as high for widows as for the divorced or separated.

Several hypotheses have been tested in recent studies, including that bereavement itself and the circumstances of widowhood predispose people to suicide, that those who remain widowed (that is, do not remarry) following the death of their spouse have preexisting characteristics (such as alcoholism and depression) that predispose them to suicide, and that the nature of the death of the spouse (especially suicide) predisposes the surviving spouse to suicide.

MacMahon and Pugh ⁴³ found that the suicide rate among a widowed population was 2.5 times higher in the first six months after bereavement and 1.5 times higher in the first, second, and third years after bereavement than in the fourth or subsequent years, thus suggesting that bereavement itself is a powerful etiologic factor in death by suicide. These figures were based on a study of 320 widowers and widows who had committed suicide (excluding homicide-suicide combinations) in Massachusetts between 1948 and 1952. The suicide sample was compared with a control group of widows and widowers matched for age, sex, and race who died from causes other than suicide. The age-standardized suicide rate for widowed men was 3.5 times higher than among married men and for women the rate was twice as high. Generally, men were found to have a higher suicide rate than women.

In a striking study, Bunch et al. ⁶ reported that half of 75 people who committed suicide in West Sussex, Great Britain, had experienced maternal bereavement within the last three years. This high maternal bereavement rate in suicides was compared to a 20 percent rate among controls who were matched for age, sex, marital status, and geographical location. Moreover, 22 percent of the suicides, compared with 9 percent of the controls, had experienced loss of their fathers within the previous five years. In addition, although married suicides and controls were not significantly different with respect to loss of a mother, the single male suicides showed greater recent maternal bereavement (60 percent) than the single male controls (6 percent). The authors hypothesize that single men may be a high-risk group for suicide following their mother's death because of a higher proclivity of males to act out via self-destructive impulses, whereas women more easily seek out medical and psychiatric help.

In a small study, Murphy and Robins ⁴⁹ found that among a group of alcoholics who committed suicide, 17 percent (5 out of 31) had experienced the death of someone close in the previous year. An additional 41 percent of the alcoholics who committed suicide had experienced another type of loss such as separation or divorce. Thirty-two percent of these other types of loss occurred within six weeks of the suicide. In a later study ⁶⁴ it was found that in a group of people who were diagnosed after their suicides as having had an affective disorder (chiefly depression), 5 percent (3 out of 60) had experienced bereavement in the year before they killed themselves. An additional 12 percent with affective disorders had experienced other types of losses. The authors' main conclusion was that alcoholics were a high-risk group for suicide after the loss of an affectional relationship.

Morbidity and Health Care Utilization

There is considerable controversy over the nature and extent of morbidity associated with bereavement and the concomitant burdens on the health care system. General health status, specific medical and psychiatric disorders, health-related behavior, and health care utilization have all been studied in an attempt to determine the impact of bereavement on health and the use of services. The findings from these studies, however, are frequently inconsistent and inconclusive, in part because of very small sample sizes in many studies.

Psychiatric Morbidity. An increase in psychiatric morbidity in the first year of bereavement could be signaled by higher rates of emotional and mental symptoms sufficient for diagnosable mental disorder; consumption of pills or alcohol; and use of psychiatric services, both inpatient and outpatient. As described earlier in this chapter, all studies have documented that distress and depressive symptoms dominate the emotional life of the bereaved during the first year. But how often do the bereaved meet criteria for true psychiatric illness?

Stein and Susser ⁶⁸ studied widowhood and mental illness among outpatients in Salford, England. They looked at the transition into widowhood and first entry into psychiatric care and compared the widows with a general population control group. Significantly more widows entered psychiatric care in the first year after the death of a spouse than in subsequent years. Thus, widows who enter psychiatric care are more likely to do so early in widowhood than later. This is the only study with controls, though not age-matched, showing a relationship between widowhood and psychiatric care.

Parkes and Brown, ⁵⁶ in a prospective study, followed for four years a group of Boston widows and widowers under the age of 45. They found higher rates of depressive symptoms among the widowed and more use of counseling by the young widowed than by controls. Other studies also mention this use of counseling, ⁶² but the appropriate data are sparse.

Data on psychiatric hospitalization are also very limited. Clayton's prospective studies of the widowed ^9,10 found that psychiatric hospital ization occurred in three (2 percent) of the bereaved and one of the controls. Two were alcoholics and the third was previously diagnosed as depressed.

Of all the psychiatric conditions, clinical depression would appear to be the one most likely to occur with greater frequency among the bereaved. As noted earlier in this chapter, Clayton and others have found that a fairly substantial proportion of bereaved individuals—estimated at approximately 17 percent—still have a constellation of depressive symptoms such that they could be diagnosed as depressed at one year. Conceptually this makes sense. Loss through death would be expected to be an important life stressor precipitating clinical depressive disorders, particularly in those predisposed by virtue of family history, personality, or previous life experience.

In view of this, it is surprising how few studies have systematically attempted to test this commonly believed clinical hypothesis. Of the three studies reported, only Paykel et al. ⁵⁸ found an increased risk of clinical depression following bereavement. The authors studied life events in 185 depressed outpatients and inpatients and a group of matched community controls. Sixteen of the depressives and four of the control group reported the death of an immediate family member in the six months prior to the onset of the illness or the interview. There were five patients who had experienced the death of a child, which is surprisingly high but is consistent with other research indicating that this type of bereavement is extremely traumatic.

Hudgens et al. ³⁴ studied 40 hospitalized patients with affective disorders and 40 matched nonpsychiatric hospital controls for precipitating events and found no deaths of spouses in either group. In the year preceding, 13 percent of the psychiatric patients and 3 percent of the controls had lost a first-degree relative, a difference that was not significant. When the sample was expanded to 100 psychiatric patients and 100 controls, 7 percent of the former and 3 percent of the latter experienced the death of first-degree relatives, though none was a spouse. ⁴⁸

Frost and Clayton ²³ evaluated 344 psychiatric inpatients for bereavement in the six months preceding admission. These patients were matched with nonpsychiatric hospitalized patients. In each group, 2 percent reported the death of a first-degree relative. Three psychiatric patients (less than 1 percent) had experienced the death of a spouse within six months of the current admission to the hospital; there were no deaths of spouses in the control group.

Many of the bereaved subjects in all these studies reported marked increases in their consumption of pills, alcohol, and tobacco, which could indicate psychiatric morbidity, and some reported initial use of these substances following bereavement.

Parkes ⁵³ observed that sedative drugs were prescribed seven times more frequently for widows under the age of 65 during the first six months of bereavement than in the period preceding the death. For widows aged 65 and older, there was no significant increase in the prescription of sedatives.

In their later study, Parkes and Brown ⁵⁶ observed that 25 percent (17) of the 68 widows in Boston under the age of 45 years reported an increase in the consumption of tranquilizers, alcohol, and tobacco during the 13-month period after bereavement. Twenty-eight percent of bereaved subjects reported an increase in smoking, compared with 9 percent of controls. Increased consumption of alcohol was reported by 28 percent of the bereaved women versus 3 percent of the controls. A first use of tranquilizers was reported by 26 percent of the widows and 4 percent of the controls. There was no statistically significant difference between bereaved subjects and controls in the use of sleeping pills.

Maddison and Viola ⁴⁴ reported a marked increase in both sedative (tranquilizers) and hypnotic (sleeping pills) drug intake in their sample of 374 widows aged 45-60 years old, compared with controls, during the 13 months after bereavement. They also noted increased alcohol consumption and tobacco use among the widows versus the controls.

There was a small but significant difference in the consumption of sleeping pills between the 90 widows and widowers (average age 61 years) and the matched controls in Clayton's ⁹ study. There were no differences between the two groups, however, in consumption of tranquilizers or other medicine taken for general health.

In a study of recently widowed persons over the age of 55 and a control group of married individuals, Thompson et al. ⁷⁰ report both increased and new medicine use among the widowed. Fifty-four percent of the medicines used by the widowed were analgesics, sedatives, sleep medication, or antidepressants.

Self-Reported Symptoms and Health Status. Many investigators have asked the bereaved about physical and other symptoms following bereavement. These include crying, changes in sleep patterns or appetite, difficulty in breathing, sighing, palpitations, inability to concentrate, and a host of other signs of distress. Unfortunately there are only a few studies that assessed self-reported symptoms before and after bereavement. Self-reports of symptoms and of perceived changes in health status are likely to be exaggerated by the general distress of bereavement such that these reports are not ideal measures of actual health status. They may be better indicators of general distress, especially if repeated measures are done over time.

Crisp and Priest ^18,61 administered a brief self-rating inventory intended to cover a full range of neurotic symptoms to samples of bereaved and controls between the ages of 40 and 65 who were registered with a group practice in southwest London. Although their numbers were small, they found minimal differences between the bereaved widows and the controls. They commented that the bereaved subjects on the whole withstood the stress in a '' robust way."

Heyman and Gianturco, ³³ as part of a continuing Duke Center Longitudinal Study for Aging, reported on the reactions to widowhood in the elderly. They found little difference in before-and-after scores of either sex on health, leisure activity, financial security, or ratings for anxiety or hypochondriasis. There was a small increase in depression in the women after bereavement, but it was felt that this depression was mild and that all the widows kept active contact with their friends despite depression.

Parkes ⁵⁴ studied a group of London widows, average age 49 years, who had visited their doctors during the first month of bereavement. At 13 months, six widows (27 percent) reported that their health was definitely worse. In the later study, in Boston, Parkes and Brown ⁵⁶ found that in terms of self-reported physical symptoms, widows and their controls were not significantly different, but widowers reported more severe symptoms and more anxiety than controls. No excess of psychosomatic illness was found. There were no differences in general health or in "health worries."

In a retrospective mail survey of 375 widows aged 40-60 (132 in Boston, Massachusetts, and 243 in Sydney, Australia), Maddison and Viola ⁴⁴ found significant differences between widows and matched controls in reports of several symptoms and complaints. Among widows considered to be at high risk, Raphael ⁶² found marked deterioration in health in the first year following bereavement as evidenced by numerous physical symptoms, diminution of work capacity, weight loss, and health-damaging behaviors.

Parkes and Weiss ⁵⁷ report marked differences between well-matched young widowed spouses and married controls in self-reported symptoms, reflecting the somatic effects of anxiety. There were no significant differences in the worsening of chronic symptoms, perceived general health, or health worries between the groups. However, Thompson et al. ⁷⁰ found that recently widowed elderly individuals reported wors ening of existing illness as well as new illnesses more frequently than controls.

These prospective studies document less physical distress than had been expected. The literature suggests that the young widowed have more anxiety symptoms than their married counterparts. Except for those already ill, older men and women reported little change in their general health status following bereavement.

Specific Medical Disorders. The hypothesis that bereavement predisposes to, precipitates, or exacerbates medical illness and thus increases morbidity has been proposed for several disorders, including acute closed-angle glaucoma, cancer, cardiovascular disorders, Cushing's disease, disseminated lupus erythematosus, idiopathic glossodynia, pernicious anemia, pneumonia, rheumatoid arthritis, thyrotoxicosis, tuberculosis, and ulcerative colitis (see Klerman and Izen ³⁸ for a detailed review of this literature).

The evidence linking each of these diseases to bereavement and grief is meager. Not only is much of the evidence based on clinical case reports of small numbers of patients, but many studies use "loss," "stress," or "depression" broadly defined as the condition preceding the disorder. Bereavement as a specific stressor and the ''depression of grief" as distinct from other types of depression are often not distinguished, thus rendering it difficult to draw definitive conclusions about the association between bereavement and specific diseases. Adequate testing of the hypothesized associations requires prospective studies with large samples.

Nonetheless, an extensive literature in the psychosomatic tradition suggests that bereavement is a contributing factor to somatic disease in individuals who are already predisposed to that disease because of genetic susceptibility, physiologic responsiveness, or preexisting psychologic susceptibility.

For example, studies of hyperthyroidism (thyrotoxicosis or Graves' disease) in children ^45,46 and in adults ³⁷ suggest that traumatic events, especially loss, may activate the disease in susceptible individuals. In the case study reports of children with hyperthyroidism, loss of a parent by divorce, separation, or death was found to be a common antecedent to depression, which in turn may activate the disease in those children who are genetically predisposed or physiologically vulnerable. In adults who are already excessively anxious or depressed, a "normal life stress," such as bereavement, can contribute to the development of hyperthyroidism. However, as Kleinschmidt et al. ³⁷ caution, affective states and disorders are unlikely to cause the disease without some genetic or physiologic vulnerability as well.

Parental loss has also been proposed as a precipitator of diabetes in children ⁴⁰ but probably only in genetically or physiologically predisposed individuals. There is also some evidence of an association between psychologic trauma and exacerbation of diabetes in some adult patients (see, for example, Grant et al. ²⁶ and Treuting ⁷¹). In many discussions of this issue, however, bereavement as a specific stressor is not separated out from other stressors, nor is it clear what makes some individuals vulnerable to an altered course of their disease when they are under stress.

Bereavement has been implicated as a contributor to many different kinds of cancer. As with studies of other diseases, most of the cancer studies are retrospective and often do not distinguish among various losses that precede the disease. Schmale and Iker ⁶⁵ reported an association between loss and several cancers, especially cancer of the cervix. Several studies by Greene ^27-29 suggest a high incidence of leukemia and lymphoma among individuals with a recent loss (actual or psychic). Men who are separated from their mothers or mother figures appear to be particularly vulnerable. Once again, however, these findings remain inconclusive because of lack of control groups, small sample sizes, and a failure to distinguish actual bereavement from other losses. ⁴⁷ Although the link between bereavement and the development of various types of cancer has not been confirmed by rigorous epidemiologic studies, there is currently great interest among researchers in exploring the immunologic mechanisms that could contribute to the development of cancer. These are discussed in Chapter 6.

The most extensive evidence of a link between disease in a specific organ system and bereavement exists for the cardiovascular system. Sudden cardiac arrhythmias, myocardial infarction, and congestive heart failure are the most frequently mentioned conditions in that system. Studies have shown that patients with congestive heart failure ⁸ and with essential hypertension ⁷⁶ are particularly prone to exacerbation of their condition in response to threatened or actual loss of human relationships.

In a study of 170 sudden and rapid deaths during psychologic stress, Engel ²¹ found that 39 percent of the women and 11 percent of the men died immediately following the death of someone close and another 23 percent of women and 20 percent of men died within 16 days of such a death. Although Engel did not have access to clinical data for all the people in his sample, he suggests (based on what he did have and on the literature) that most of these sudden deaths were due to cardiac arrest in individuals with preexisting cardiovascular disease. As Parkes and Weiss ⁵⁷ conclude:

It certainly seems unlikely that bereavement causes arteriosclerosis since that condition takes many years to develop, but it seems very likely that a person who already has arteriosclerosis affecting his or her heart is at special risk after bereavement. The added burden of bereavement on that heart may be sufficient to produce a myocardial infarction and to reduce the person's chances of surviving an infarction should one occur.

Beyond case reports there are two additional sources of data regarding specific morbidity following bereavement that lend credence to the hypothesized association. Several of the studies discussed in the section on mortality examined specific causes of death. For example, Helsing et al. ³⁰ found that widowed women who died had a higher percentage of deaths from cirrhosis of the liver than would have been expected. In men there were increased death rates from infectious diseases, accidents, and suicides, but no increase from cardiovascular disease. Kraus and Lilienfeld ³⁹ found for the young widowed who had not remarried greatly increased mortality rates for several kinds of cardiovascular and infectious diseases. And in diagnosis-specific health care use studies, Parkes ⁵² found that increased rates of physician visits in the first six months following bereavement were due largely to increased consultations for vascular and articular conditions, especially osteoarthritis in a widowed population under 65 years of age.

Aside from psychosomatic mechanisms involving physiologic changes, Jacobs and Ostfeld ³⁶ suggest that excess mortality, and presumably morbidity, in the recently bereaved may be mediated by behavioral changes that compromise health maintenance or chronic disease management. Increased alcohol consumption and cigarette smoking, common behavioral changes in the bereaved, may exacerbate or precipitate illness. Excess mortality, especially among bereaved men, is explained in large part by deaths from suicide, cirrhosis, and cardiac arrest. All three conditions have clinical antecedents (depression, alcoholism, and cardiovascular disease) that could be detected before or very shortly after bereavement, thus identifying three high-risk groups for whom early intervention might be useful.

Health Care Utilization. Use of health services can be an indicator of medical morbidity or a measure of the burden, including costs, of bereavement on the health care delivery system. The second is straightforward. As a proxy for actual medical disease, however, the use of health services, especially physician visits, is imperfect; it is well established in the health services research literature that a substantial proportion of visits are precipitated by psychosocial concerns and nonorganic symptoms rather than by diagnosable illness.

Given that bereavement often results in significant distress, depression-like symptoms, and increased use of drugs, and is sometimes associated with increased morbidity and mortality, significant increases in the use of health services would be expected. Yet most studies conducted in the United States show no increase in physician visits or hospitalization following bereavement. ^14,56 On the average, Americans make almost five physician visits each year. ⁵⁰ Among the bereaved it appears that although some visits are related to their reactions to loss, the actual number of visits does not increase. Most of the evidence from England, ^52,54 however, and one U.S. study of elderly widows in a prepaid health plan ⁷⁶ do show an increase in physician visits for some people following bereavement. This discrepancy suggests that payment method may have a powerful effect on people's decisions to visit a physician, with fee-for-service systems inhibiting health care utilization.

Possible Consequences Beyond the First Year

The form that grief takes, its outward expressions, and the length of the recovery process all are influenced by the social and cultural context within which bereavement occurs (see Chapter 8). Although most studies report a decrease in the reported distress and other manifestations of grieving by the end of the first year, persistent experiences of distress are frequently observed. Many people who are grieving report "it is always with you." The significance of these reactions is not clear. True delayed grief—in the sense of physiologic disruptions, social withdrawal, and persistent sadness and yearning that emerge only after a period of absent grief—is so rare that little research has been conducted on it. Yet, distortions of personality and alterations in the quality of social functioning that are related to the distress associated with memories of the dead person have been observed by many clinicians, particularly for patients in psychotherapy.

A number of observers have proposed that failure to cope adequately during the usual bereavement period predisposes a person to later psychiatric and medical problems. Although there are no systematic data available to support or refute this hypothesis, its proponents put forth two alternative explanations of the process. The first predicts that later health problems are a consequence of the duration and intensity of distress—the more prolonged and intense the distress, the greater the burden on an individual's adaptive capacity.

In contrast, many psychiatric and mental health professionals believe that the grieving process is adaptive and that "failure to grieve" or an interruption of the grieving process leaves an individual vulnerable to later illness. This hypothesis is usually attributed to Erich Lindemann in his writings on the survivors of the Coconut Grove Disaster ⁴² and is a view held widely in mental health circles. Mental health professionals generally encourage emotional expression during the bereavement process. Although some believe in prescribing medications to facilitate grieving, others are hesitant to recommend the use of drugs, including tranquilizers and antidepressants, lest the adaptive function of grieving be suppressed. ¹⁹ These issues are discussed further in Chapters 3, 5, and 10.

As discussed earlier in this chapter there is some evidence suggesting that mortality rates remain high for certain categories of bereft individuals beyond the first year—perhaps into the sixth year after their loss. There is also some indication of an increase in medical illnesses in the second and third years following bereavement, but adequate studies have not been performed to verify this hypothesis.

Characteristics of Bereaved Individuals

Some studies have reported that men do more poorly than women following conjugal bereavement. ^25,56,60,72 Other studies ^4,12,67 have not. If premature death is the outcome studied, certainly men do worse; if remarriage is the outcome, men do better. The physical and psychologic outcomes are unclear because most prospective studies have not had large enough samples of men to draw any conclusions about their relative risk of illness.

It is generally held that bereavement reactions are more intense and have more enduring consequences for younger people, particularly for children but also for adolescents and young adults. Older individuals appear to experience fewer, less intense consequences, perhaps because experiencing the death of someone close, family, or friends is common over the age of 60. Another possibility is that older individuals already have passed through the period of highest risk for psychiatric problems, such as alcoholism, depression, and anxiety disorders.

It seems clear that poor prior physical or mental health is a risk factor. Those who are physically ill before a bereavement are more likely to be ill after it too, with more physician visits and perhaps even with greater risk of death in the first year. ⁷⁶ Those with a previous psychiatric disorder or with a history of misuse of drugs will be at risk psychiatrically.

Most researchers and clinicians in this field would hypothesize that personality variables probably affect outcome; unfortunately only one study has examined this systematically. Vachon and her colleagues ⁷³ used Cattel's 16 Personality Factor Questionnaire ⁷ to test whether specific aspects of personality were related to level of distress and adjustment to bereavement. They found that widows with enduring high distress scores were characterized as emotionally less stable, more apprehensive and worrying, and highly anxious. Widows with low distress scores were more likely to be emotionally stable, mature, conscientious, moralistic, conservative, controlled, and socially precise. These and other assessments were done both six months and two years after the bereavement. Ideally, these personality variables should be tested before the terminal illness or when a spouse first becomes ill, as distress itself could confound the personality inventories.

In addition to the previously mentioned high suicide rate among alcoholics shortly after bereavement, alcoholics are more likely than nonalcoholics to be doing poorly one year following bereavement. When psychiatric hospitalization occurs, the diagnosis is likely to be alcoholism. ^49,64

The association between socioeconomic status and bereavement outcomes has not been adequately studied. Whether individuals with low incomes do any worse following bereavement than the more well-to-do is not known. A frequent concomitant of bereavement, however, especially for widows, is financial difficulty brought on by the loss of the major wage earner and perhaps by large medical bills as well. Recent work by Vachon et al. ⁷³ has shown that poor outcome in middle-aged widows was associated with financial problems.

Relationship to the Deceased

Each kinship relationship has particular difficulties associated with it following bereavement (see Chapter 4). In addition to kinship, the nature of the relationship with the deceased has been hypothesized to influence outcomes. The literature on conjugal bereavement is replete with data indicating that individuals who had highly ambivalent relationships with their spouses do worse following bereavement than people whose relationships did not have these characteristics (see Chapters 3 and 4). There also is evidence to suggest that spouses who are unable to function independently do poorly following bereavement, ⁵⁷ although there is some controversy about the meaning and predictive value of this variable.

Nature of the Death

Sudden death is frequently hypothesized to be more traumatic for the survivors and to lead to poorer outcomes than deaths that have been anticipated. There are marked variations in the criteria used to assess suddenness of death. Clayton et al. ¹⁵ defined it as an illness of five days or shorter duration. Parkes ⁵⁵ classified survivors as having "short" preparation if they had less than two weeks' warning that a spouse's condition was likely to prove fatal, or less than three days' warning that death was imminent. Gerber et al. ²⁵ defined "acute illness death" as one occurring without warning and prior knowledge of the condition, or a death after a medical condition of less than two months' duration with the absence of multiple attacks and hospitalization.

Contrary to commonly held views, most of the research literature indicates that sudden death, however defined, does not produce more disturbed survivors. In Parkes' data ⁵⁵ the young widowed who experienced a sudden death had a poorer outcome, as did widows in the study by Vachon et al. ⁷³ However, studies by Clayton et al. ¹⁵ and by Fulton and Gottesman ²⁴ did not find this to be true. In a study by Gerber et al. ²⁵ of older men and women, there was a negative correlation between length of illness and outcome; that is, the longer the terminal illness, the more likely there was to be a poor outcome for the surviving spouse.

The time course of the loss is a risk factor that deserves further attention. Common wisdom holds that time to say goodbye and to express love will facilitate grieving by lessening later feelings of anger and guilt. However, it may be that the moment of death is always a surprise no matter how much warning there has been. Perhaps a very lengthy terminal illness produces its own stresses and strains that complicate bereavement. Perhaps suddenness of death interacts with age of the deceased or age of the survivors in ways that have not yet been uncovered. As discussed in Chapter 10, suddenness of death may have important implications for the design of strategies to assist the bereaved.

A number of studies have been undertaken of suicide's impact on family members and other survivors. Given the high rate of assortative mating (i.e., people with similar characteristics tending to marry each other; depressed people are particularly likely to marry other depressed people), the impact of death by suicide is likely to be associated with a propensity for psychiatric illness in the spouse, particularly for alcoholism and depression. In general, as discussed in Chapters 4 and 5, death by suicide renders survivors vulnerable to increased psychologic distress and, especially in the case of children, may leave them vulnerable to suicide as well.

Risk Factors Appearing After the Death

All bereavement studies report that among people already using alcohol, drugs, or cigarettes, consumption of these substances increased after a death. Some people, however, begin using these substances following bereavement. Even without a chronic dependency developing, increased use of these substances might lead to deterioration in health and well-being. Certain individual symptoms during the early bereavement period may also predict poor outcome. These include suicidal thoughts (particularly after the first month), psychomotor retardation, and morbid guilt.

As discussed in several later chapters, there is mounting evidence to suggest that social support has a positive effect on general health status and may serve as a protective factor to buffer or modify the impact of adversity and stressors, not only on the mental health of an individual, but also upon his or her physical health. ⁵ Perceived lack of social support is one of the most common risk factors cited in the bereavement literature. The perception by the recently bereaved that there is no one to talk to or lean on appears to be a reliable predictor of poor outcome. ⁷³

Research testing the magnitude of these hypothesized risk factors is difficult because of lack of agreement on relevant outcome measures. Different risk factors are likely to be involved in different outcomes; until these conceptual and methodological problems are resolved, risk factor studies and predictive studies will be seriously handicapped.