The varieties of grief experience

Clinical Psychology Review, Vol. 21, No. 5, pp. 705 ± 734, 2001 Copyright D 2001 Elsevier Science Ltd. Printed in the USA. All rights reserved 0272-7358/01/$ ± see front matter

PII S0272-7358(00)00062-3

THE VARIETIES OF GRIEF EXPERIENCE George A. Bonanno Columbia University

Stacey Kaltman The Catholic University of America

ABSTRACT. The bereavement literature has yet to show consensus on a clear definition of normal and abnormal or complicated grief reactions. According to DSM-IV, bereavement is a stressor event that warrants a clinical diagnosis only in extreme cases when other DSM categories of psychopathology (e.g., Major Depression) are evident. In contrast, bereavement theorists have proposed a number of different types of abnormal grief reactions, including those in which grief is masked or delayed. In this article, we review empirical evidence on the longitudinal course, phenomenological features, and possible diagnostic relevance of grief reactions. This evidence was generally consistent with the DSM-IV's view of bereavement and provided little support for more complicated taxonomies. Most bereaved individuals showed moderate disruptions in functioning during the first year after a loss, while more chronic symptoms were evidenced by a relatively small minority. Further, those individuals showing chronic grief reactions can be relatively easily accommodated by existing diagnostic categories. Finally, we found no evidence to support the proposed delayed grief category. We close by suggesting directions for subsequent research. D 2001 Elsevier Science Ltd. All rights reserved. KEY WORDS. Grief, Bereavement, Loss, Stressor. INTRODUCTION GRIEF IS A painful, but unfortunately common experience. Most people at different points in their lives are confronted with the death of a close friend or relative. There are, however, marked individual differences in how intensely and how long people grieve. Some grieve openly and deeply for years, and only slowly return to a semblance of their normal level of functioning. Others suffer intensely, but for a relatively more proscribed period of time. Still others appear to get over their losses Correspondence should be addressed to George A. Bonanno, Ph.D., Department of Counseling and Clinical Psychology, Teachers College, Columbia University, Box 218, 525 West 120th Street, New York, NY 10027, USA. E-mail: [email protected] 705

706

G. A. Bonanno and S. Kaltman

almost immediately, and to move on to new challenges and new relationships with such ease as to raise doubts among their friends and relatives as to whether they may be hiding something or running away from their pain. The extent that grief varies across individuals suggests important questions about what constitutes normal or common grief, and when, if at all, too much or too little grief might be considered abnormal, or even pathological. Unfortunately, the bereavement literature has yet to agree on a clear, empirically defensible definition of grief, or its normal and abnormal course and manifestations (Bonanno, 1998; Hansson, Carpenter, & Fairchild, 1993). The present review was motivated by this deficiency. First, we review competing formulations of normal and complicated grieving found in the Diagnostic and Statistical Manual of Mental Disorders (DSMIV, American Psychiatric Association, 1994) and in the bereavement literature. According to the DSM-IV, bereavement is a stressor that produces relatively normal and expectable distress. DSM-IV does not offer a complicated grief diagnosis, and allows for bereavement-related diagnoses only in extreme cases when existing diagnostic categories (e.g., Major Depression) may be relevant. In contrast, bereavement theorists have argued for the clinical necessity of a complicated grief diagnosis, and have proposed a number of different types of complicated grief. In an effort to reconcile these competing positions, we review the available empirical evidence on grieving. Specifically, we examine three kinds of evidence: longitudinal studies of grief outcome, descriptive studies of the phenomenological features of grieving, and diagnostic studies that focus on the distinction between normal and pathological or complicated variants. Our analysis is generally supportive of the DSM-IV's conception of grief as a normal reaction to an enduring stressor, and reveals little in the way of empirical evidence to support the complex taxonomies of grief outcome developed in the bereavement literature. We summarize the available evidence into a working definition of grief course patterns, and close by suggesting several avenues for future bereavement research. THEORETICAL AND OBSERVATIONAL CLASSIFICATIONS OF GRIEF COURSE DSM-IV: Bereavement as a Stressor The DSM-IV views the death of a close friend or relative as a stressor with generally normative and predictable consequences. According to this approach, bereavement is considered among the V codes and is used diagnostically when the ``focus of clinical attention is a reaction to the death of a loved one'' (American Psychiatric Association, 1994, p. 684). The conceptualization of bereavement in terms of the V code is clearly intended to represent mourning as a normal phenomenon. In addition, DSM-IV emphasizes culturally determined forms of mourning and grief behavior, and that ``the duration and expression of `normal' bereavement vary considerably among different cultural groups'' (American Psychiatric Association, 1994, p. 684). An important limitation of DSM-IV's vagueness on this point, however, is that it does not provide a means of precisely distinguishing between individuals who show common grief reactions from those who do not. As we will show later in this article, this distinction may prove crucial to the understanding of individual difference in long-term grief course. By using a V code, the DSM-IV also explicitly avoids the categorical distinction of complicated versus uncomplicated bereavement. Instead, extreme cases in which

Varieties of Grief Experience

707

grief becomes exceptional or severe are conceptualized using other known categories of psychopathology. For example, Major Depressive Disorder can be diagnosed after an interpersonal loss if the depressive symptoms endure longer than 2 months or are marked by significantly disrupted functioning, suicidal ideation, psychotic symptoms, psychomotor retardation, or extreme feelings of worthlessness. Alternatively, DSM-IV indicates that Posttraumatic Stress Disorder may be diagnosed following the sudden, unexpected death of a family member or close friend in the presence of the characteristic constellation of intrusion, avoidance, and hyperarousal symptoms. Thus, beyond the normal bereavement response, DSM-IV provides for the diagnosis of more severe pathology in exceptional cases, but only in the form of existing diagnostic categories. The Taxonomy of Complicated Grief in the Bereavement Literature In contrast to DSM-IV's conception of bereavement as a normal stressor, clinically oriented bereavement theorists have articulated a relatively complex set of taxonomies to describe aberrant or complicated mourning. One of the earliest expositions of normal versus pathological or complicated grieving was proposed by Parkes (1965). Parkes distinguished normal or typical grief from three forms of atypical grief based upon interviews with patients who had been hospitalized for psychiatric illnesses within 6 months following the death of a parent, spouse, sibling, or child. He identified chronic grief as the most common form of grief in the interviewed sample, defined as an extended variant of typical grief in which symptoms are particularly pronounced: ``The reaction is always prolonged and the general impression is one of deep and pressing sorrow'' (Parkes, 1965, p. 14). In contrast to chronic grief, Parkes defined inhibited grief when a bereaved person evidenced little overt reaction to the loss. Although Parkes observed no examples of inhibited grief in the interviewed sample, he highlighted inhibited grief as an atypical grief reaction that is present primarily in children. Finally, Parkes described delayed grief as occurring when a typical or chronic grief reaction follows a period in which grief is inhibited. Although Parkes made these classifications using unstructured clinical interviews and a nonrepresentative sample of bereaved persons, his taxonomy has proved highly influential and has served as the basis for subsequent categorizations of grief. In a widely cited book, Bowlby (1980) identified two ``disordered variants'' of grief. Chronic mourning was described as ``more intense and disrupting than in healthy mourning'' (p. 147). This classification was based upon studies of bereaved individuals who showed chronic despair 12 months post-loss (Gorer, 1965) and depression, disorganization, and alcoholism two or more years following the loss of a spouse (Glick, Weiss, & Parkes, 1974). Bowlby highlighted a number of factors that were associated with a chronic grief course which included ``the death having been sudden, a delayed response, nightmares connected with the death, quarrels with relatives and others, an attempt to escape the scene; and prior to the bereavement, a history of an unsettled childhood and of having been brought up to bottle up feelings'' (Bowlby, 1980, pp. 149 ± 150). Bowlby also described the prolonged absence of grieving as a disordered variant of grief, noting that while a brief period of numbing is expected following a significant loss, it is not expected to last longer than a few days to a week. Although a bereaved individual may show no overt signs of grieving, Bowlby described ``tell-tale signs that the person has in fact been affected and that his mental equilibrium is disturbed'' (Bowlby, 1980, p. 153). Signals of the prolonged

708

G. A. Bonanno and S. Kaltman

absence of conscious grieving included the avoidance of reminders, physical symptoms, distressing dreams, and appearing overcontrolled. Following Bowlby's work, Belitsky and Jacobs (1986) applied an attachment framework to the grieving process and similarly defined two forms of ``pathological grief.'' They described distorted grief, which is manifested in either a prolonged or unusually intense reaction to loss, and delayed grief in which an overt grief reaction is absent for a long period of time. In a book intended to aid clinicians in the diagnosis and treatment of grief, Worden (1982) delineated abnormal grief reactions into four distinct patterns: chronic, exaggerated, delayed, and masked. Chronic grief was defined solely in terms of the duration of the reaction, in contrast to prior defiitions which were based on both the duration and intensity of grief. Worden described chronic grief as a prolonged reaction that is ``excessive in duration'' and which ``never comes to a satisfactory conclusion'' (p. 59). Exaggerated grief was defined as a reaction to loss that is excessive in intensity. Worden defined delayed grief as an emotional reaction that is not commensurate with the loss but may be experienced more fully at a later time. Finally, masked grief was defined as occurring when a bereaved person experiences symptoms or behaviors that lead to difficulties in functioning but does not recognize these symptoms as being related to the loss. Worden further identified two possible manifestations of this type of reaction in which the grief is masked either as a physical symptom or as a maladaptive behavior. Based on a theoretical position that elucidates the processes that tend to produce different types of grief, Rando (1992) proposed the most elaborate system of complicated grief reactions to date. Rando observed that complicated mourning develops when there is a ``compromise, distortion, or failure of one or more of six 'R' processes of mourning'' (p. 45). The six hypothetical `R' process include: recognizing the loss, reacting to the separation, recollecting and reexperiencing the deceased and the relationship, relinquishing the old attachments to the deceased and the old world, readjusting to move adaptively into the new world without forgetting the old, and reinvesting. In Rando's system, all forms of complicated mourning result from attempts to ``deny, repress, or avoid aspects of the loss, its pain, and the full realization of its implications for the mourner'' and help the unwilling bereaved survivor ``hold onto, and avoid relinquishing, the lost loved one'' (Rando, 1992, p. 45). Rando used her model of ``R'' processes to delineate seven unique syndromes of complicated mourning, which are divided into three categories. Absent mourning, delayed mourning, and inhibited mourning are related to ``problems in expression'' (Rando, 1992, p. 46). Distorted mourning of the extremely angry or guilty types, conflicted mourning, and unanticipated mourning are syndromes with ``skewed aspects'' (Rando, 1992, p. 46). Chronic mourning is a syndrome with a ``problem in ending'' (Rando, 1992, p. 46). In addition to these patterns, Rando observed that complicated mourning can take the form of symptoms, mental or physical disorder, or death. She further highlighted factors which predispose the bereaved to a complicated mourning reaction. High-risk factors associated with the death include: ``a sudden and unanticipated death, especially when it is traumatic, violent, mutilating, or random; death from an overly lengthy illness; the loss of a child; and the mourner's perception of preventability'' (Rando, 1992, p. 47). High-risk factors related to antecedent or subsequent variables include: ``a premorbid relationship with the deceased which has been markedly angry or ambivalent or markedly dependent, the mourner's prior or concurrent mental health problems and/or unaccommodated losses and stresses, and the mourner's perceived lack of social support'' (Rando, 1992, p. 47).

Varieties of Grief Experience

709

When considered together, the various formulations suggest a complex, multifaceted taxonomy of grief reactions that encompasses a range of behaviors and manifestations. Beyond the simple distinction between normal and complicated bereavement, it is assumed that there are a number of different types of aberrant grieving determined primarily by presence or absence of overt signs of grieving, the duration of the grief reaction, and the intensity of the grief reaction over time. The fact that these formulations were based largely on clinical inference or global observational studies, however, raises important questions about their generalizability. Indeed, many of the suggested categories are rife with assumptions about what grieving should be, or how bereaved individuals should feel, and how long they should feel it. For example, Rando's complex system of aberrant grief patterns was based on an equally complex set of assumptions about the processes (``R'' processes) assumed to underlie successful grief resolution. It is worth noting, however, that many of Rando's ``R'' processes suggest a relatively traditional perspective on ``working through'' grief that has become highly controversial. Reviews of the bereavement literature over the past decade have argued, for instance, that there is no evidence to support traditional assumptions that it is necessary to express the pain of loss, that it is necessary to work through the thoughts and memories associated with the loss, or that it is necessary even to relinquish attachments to the deceased (Bonanno, 1998; Bonanno & Kaltman, 1999; Klass, Silverman, & Nickman, 1996; Stroebe & Stroebe, 1987; Wortman & Silver, 1989).

TOWARD AN EMPIRICALLY BASED, WORKING MODEL OF GRIEF The contrast between the DSM-IV's relatively simple conception of grief as a normative stressor and the various complex taxonomies for complicated mourning proposed in the bereavement literature suggests an imperative need to resolve the differences between these approaches. Somewhat surprisingly, however, there have been few attempts to examine the validity of the specific grief outcome categorizations in relation to the available empirical data. In Section 4 of this article, we review the empirical evidence that is available with the aim of elucidating the parameters of common or typical grief, as well as extreme forms of grieving. To achieve this aim, we review three different types of studies. First, we consider longitudinal studies of grief course. Second, we review descriptive studies of the basic dimensions or phenomenological features of grieving. Third, we evaluate diagnostically oriented studies that have focused on the categorical distinction between normal and complicated grief. A working model of grief outcome, based on this analysis, is shown in Fig. 1. Our review is generally supportive of the approach to bereavement taken by the DSM-IV. That is, consistent with the DSM-IV and in marked contrast to the complex taxonomies proposed in the bereavement literature, we find ample evidence that the majority of bereaved individuals show what might be considered a normal or nonpathological form of grief. Specifically (as we review below), depending upon the measures used, between 50% and 85% of the bereaved individuals in these studies appeared to exhibit a common grief pattern consisting of moderate disruptions in cognitive, emotional, physical, or interpersonal functioning during the initial months after a loss. Although some disruptive aspects of grief continue for several

710

G. A. Bonanno and S. Kaltman

FIGURE 1. A working taxonomy of grief reactions. years after the loss, most bereaved individuals returned to normal (baseline) levels of functioning by the end of the first year. In addition, most bereaved individuals experienced positive thoughts and emotions associated with the lost relationship, even in the early months of grieving. Only a relatively smaller subset (approximately 15%) of the bereaved individuals in these studies tended to continue to show serious disruptions in functioning at the 1 ± 2 year point, thereby suggesting some form of chronic grief. In further concordance with DSM-IV, these individuals appeared to experience symptoms similar to individuals suffering from depression and anxiety disorders, and to some extent trauma reactions, and thus were adequately captured by existing diagnostic categories. Importantly, our review also revealed a significant minority of bereaved individuals who showed mininal grief (i.e., little or no overt signs of disrupted functioning) in the early months after a loss. Due to variations in measurement and other uncontrolled factors, the size of this group has varied, ranging from 15% to 50% of the bereaved participants across studies. DSM-IV does not discuss the absence of grief symptoms, as could be expected given its view of bereavement as a stressor event and its primary emphasis on psychopathology. In contrast, the bereavement literature has tended to view individuals showing little or no grief with suspicion, and as having some form of inhibited or masked grief or as prone to develop delayed grief symptoms. Wortman and Silver (1989) were among the first to note the absence of empirical support for the existence of delayed or inhibited grief. A decade later, however, we still find no concrete evidence to support these categories and, rather, we suggest that the absence of intense grief in the early months after a loss should be considered an indicator of adjustment, and not a potentially abnormal or pathological response. LONGITUDINAL STUDIES OF GRIEF COURSE The longitudinal assessment of the psychological and physiological difficulties experienced by bereaved individuals is an optimal strategy for investigating basic

Varieties of Grief Experience

711

patterns of grief course. Unfortunately, there have been relatively few studies of grieving beyond the first year. The longitudinal studies that are available, however, provide important evidence: (a) that grief over the loss of a loved one tends to disrupt psychological and physiological functioning in most but not all bereaved individuals, (b) that a small, but significant minority of bereaved individuals will continue to evidence disrupted functioning for many years after a loss, and (c) that there does not appear to be any concrete justification for the assumption that minimal grief in the early months of bereavement will eventually lead to delayed grief. The earliest studies of bereavement suggested relatively optimistic conclusions about the normal duration of grief. Lindemann (1944) conducted what is generally viewed as the first formal bereavement study and concluded that ``with a period of from four to six weeks, it was ordinarily possible to settle an uncomplicated and undistorted grief reaction'' (p.144). Parkes (1964) later endorsed a similar position, concluding that his preliminary studies of health difficulties after loss provided evidence ``consonant with the traditional picture of grief as a severe but self-limiting affective disorder'' (p. 276). Although these investigators presaged later observations of the time-limited aspects of normal grieving, subsequent studies provided convincing evidence that for some bereaved individuals the death of a spouse or close relative may lead to more protracted disruptions in psychological and physical health. Parkes and Brown (1972) used structured interviews to examine long-term health decrements among a sample of middle-aged conjugally bereaved individuals and a matched married comparison sample. The bereaved sample had higher levels of depressive symptoms 14 months after their loss, but was no longer distinguished from the comparison sample after 2 years of mourning. However, by the second year after their loss, the bereaved participants still reported significantly greater numbers of sick days, hospital admissions, disturbances in sleep and appetite, and greater consumption of alcohol, tobacco, and tranquilizers, compared to their nonbereaved counterparts. Lehman, Wortman, and Williams (1987) used a cross-sectional design to assess bereaved individuals between 4 and 7 years after the sudden and unexpected death of either a spouse or a child. Although the results were strongest for the bereaved spouses, this study provided important evidence for the enduring nature of severe grief reactions. Compared to a matched married sample, the bereaved spouses showed significantly more depression on several different indices, reported higher levels of all types of psychiatric symptoms (e.g., anxiety, somatization, hostility), had increased worry, less optimism, felt that their lives were significantly more negative, and evidenced significantly greater mortality. A similar but less consistent pattern of deficits was observed for the bereaved parents. Although these studies provided convincing evidence that grief symptoms may persist for years after a loss, they did not address the question of individual variation in grief severity or duration. In an early effort to examine individuals differences in grief symptoms and difficulties, Zisook, Devaul, and Click (1982) developed the Expanded Texas Inventory of Grief (ETIG) to measure the ``frequency and time course of present grief-related behaviors and feelings'' (p. 1590). Using a crosssectional design, they administered the ETIG to 211 people who had lost a loved one between 1 month to 22 years previously. Zisook et al. observed that grief-related behaviors and feelings peaked for most participants between 1 ± 2 years following the death and then gradually declined. Nonetheless, there were still a small number of

712

G. A. Bonanno and S. Kaltman

the bereaved participants who endorsed items related to present feelings of grief as late as 10 years after their loss. Subsequently, Lundin (1984) used the ETIG in a cross-sectional, retrospective study of bereaved participants who had lost either a spouse or a child 8 years earlier. To assess outcome, Lundin selected items from the ETIG that satisfied Parkes and Weiss'(1983) proposed criteria that recovery from bereavement involved a decreased preoccupation with grief, the return of the ability to function well in everyday life, and a sense of well being. Using these criteria, 65% of the sample had evidenced recovery at the 8-year mark after their loss. These results, however, point to a significant minority of subjects who were still preoccupied with the death and unable to achieve a complete adjustment in this same lengthy time period. What about delayed grief? Many of the grief course taxonomies reviewed earlier included the delayed appearance of grief symptoms, typically accompanied by the assumption that grief at early points in bereavement had been avoided (Bowlby, 1980; Rando, 1992; Raphael, 1983; Sanders, 1993; Worden, 1982). Like many aspects of bereavement theory, the assumption of delayed grief appears to have persevered despite the relative absence of empirical support (Bonanno & Kaltman, 1999; Wortman & Silver, 1989). There is some evidence that therapists can reliably identify delayed grief from clinical vignettes (Marwit, 1996). However, diagnostic reliability in this study was not surprising because a portion of the case vignettes were explicitly designed to illustrate the delayed grief pattern. Further, none of the case vignettes represented normal or uncomplicated bereavement. Thus, it is impossible to determine the true reliability of the diagnoses. Finally, even if the reliability of the delayed grief category was accepted, there is still a glaring absence of validity data for the category. Only two studies have directly attempted to examine the presumed existence of delayed grief, and neither study was able to provide concrete support for the construct. Middleton, Burnett, Raphael, and Martinek (1996) examined the grief course of bereaved spouses (N = 44), bereaved adult children (N = 40) and bereaved parents (N = 36). All three groups were interviewed about their grief experiences within one month of the loss and again at 10 weeks, 7 months and 13 months of bereavement. Cluster analyses were used to classify participants into the most representative categories of grief course. The results were similar across the different types of bereavement. The most common pattern (43% to 70%) was a moderate level of grief symptoms that declined over time. Relatively smaller number of bereaved participants showed either low levels of grief across all four assessments (17% to 25%), or relatively elevated grief across all four assessments (14% to 31%). Importantly, despite the authors' conviction that delayed grief is a genuine clinical phenomenon, Middleton et al. (1996) concluded that ``no evidence was found for the pattern of response which might be expected for delayed grief'' (p. 169). In our own longitudinal research, we examined the relationship between grief reactions and emotional avoidance among 42 conjugally bereaved individuals at 6, 14, 25, and 60 months of bereavement (Bonanno & Field, 2001; Bonanno, Keltner, Holen, & Horowitz, 1995; Bonanno, Znoj, Siddique, & Horowitz, 1999). To operationally define delayed grief, we first defined severe and mild grief reactions using the 6-month median score on a structured clinical interview for grief-related disruptions in daily functioning (e.g., sleep disruption, weight loss, dysphoric affect). The validity of the mild and severe grief categories was established by their

Varieties of Grief Experience

713

concordance with a number of common symptom measures used in bereavement research, such as the Texas Revised Inventory of Grief (Faschingbauer, 1981), with clinical assessments of mild versus severe grief made independently by psychotherapists specializing in bereavement (Bonanno et al., 1995), and with untrained observers' ratings of the degree that they thought the participants appeared to be suffering from their loss (Keltner & Bonanno, 1997). Using these categories, bereaved participants were assigned to the delayed grief pattern when they showed mild grief at the initial assessment, but severe grief at any of the later assessments. Not a single participant from this sample evidenced delayed grief at 14 or 25 months (Bonanno et al., 1995, 1999). In a recent follow-up study, Bonanno & Field (2001) examined possible delayed symptom elevations at 60 months post-loss using the structured grief interview and additional questionnaire measures of grief, depression, and somatic complaints. Using these multiple outcome measures, 3 of the participants were found to show elevated symptoms on at least one of the scales. According to clinical formulations of delayed grief, these participants should have shown grief avoidance at earlier assessments. It was possible to examine this presumption using data on the emotional processing of the loss. At the 6-month point in bereavement, participants were asked to talk freely about their relationship to the decreased and their feelings about the loss of that relationship. Changes in automatic arousal during the interview were monitored relative to a baseline period. Participants were also asked to report how much negative emotion they experienced during the interview. Combining these measures, the avoidance or dissociation of grief-related distress was operationally defined as a subjective experience of negative emotion that was relatively low in comparison to concurrent increases in autonomic responsivity. In previous studies, this type of affective ± autonomic response dissociation has been linked to reduced awareness of distress and to repressive defensiveness (Asendorpf & Scherer, 1983; Newton & Contrada, 1992; Weinberger & Davidson, 1994; Weinberger, Schwartz, & Davidson, 1979). In addition, the affective ±autonomic dissociation score has shown convergence with clinical ratings of the avoidance of emotional awareness (Bonanno et al., 1995). How did participants showing affective ± autonomic dissociation fare in our bereaved sample? Overall, bereaved participants who showed this type of emotional avoidance had reduced symptom levels even when initial symptoms were statistically controlled. Thus, emotional avoidance in the early months of bereavement predicted relatively healthy adaptation. The same results were also observed for other variables related to emotional processing of the loss. Both the degree that participants expressed negative emotion in their faces when they talked about the loss, and the degree that participants described negative thoughts and emotions when they talked about the loss, were consistently associated with increased grief-related symptoms at later dates. Again, these relationships remained significant after initial grief symptoms were statistically controlled. What about the three participants who showed elevations in grief symptoms for the first time at 60 months? None of these participants had shown unusually low levels of emotional processing in either their facial expressions of negative emotion or their verbal descriptions of negative thoughts and emotions. One of these participants did show clear evidence of affective ±autonomic dissociation at 6 months. However, of the 13 remaining participants in this study who had shown affective ±autonomic dissociation at 6 months, none had elevated grief scores on any of the four measures at any

714

G. A. Bonanno and S. Kaltman

point in the study. Thus, there was not a reliable association between emotional avoidance and either elevated grief or delayed elevations in grief. Still another avenue we explored was the possibility of delayed increases in somatic symptoms. Somewhat unexpectedly, bereaved individuals who showed affective ± autonomic dissociation in the early months of bereavement reported concurrent elevations in somatic complaints. However, at follow-up assessments, these participants no longer had high levels somatic complaints (Bonanno et al., 1995, 1999). There was a small number of participants at each assessment who did show marked increases in somatic complaints, relative to the previous assessment, but there was no systematic relationship between changes in somatic complaints and affective ±autonomic dissociation. Further, the level of change was always within a standard deviation of the sample mean, suggesting normal measurement variance. Finally, affective ± autonomic dissociation was found to be unrelated to other indices of health, such as the frequency of visits to medical professionals. It might be argued that even one participant showing both avoidant behavior and delayed elevations in grief at 60 months lends the category at least a modicum of clinical utility. However, even in this limited application, a relationship between emotional avoidance and delayed elevations in grief was observed for only 1 participant in 14 (or 7%), hardly a solid basis for clinical intervention. Indeed, there are several more plausible, and more parsimonious, explanations for delayed increases in just a few participants, explanations that may also be applied to clinical reports of isolated cases of delayed grief. One such explanation is that other stressors over-and-above the original loss produced the observed increase in symptoms. An even more parsimonious explanation is that elevations in symptom levels for a few participants is simply the result of normal, random measurement error (Nunnally, 1978). Bereavement theorists have assumed that when delayed grief emerges, it is usually ``as fresh and intense as if the loss had just occurred'' (Humphrey & Zimpfer, 1996, p. 152). Thus, delayed grief should be evident on multiple grief indices. A measurement error explanation of delayed grief suggests, however, that participants who show delayed symptom elevations will tend to do only on isolated measures, and will score within the normal range on other, related symptom measures (i.e., the elevated symptoms should occur in a random manner across measurements). The data supported this point. When assessed at 60 months, the three participants showing delayed elevations from the original Bonanno et al. (1995) study each showed an elevation on only one measure (Bonanno & Field, 2001). Further, when measurement error was minimized by weighting each of several grief-related symptom measures by their alpha reliability and then summing the weighted measures into a composite grief score, there were effectively no cases of delayed grief (Bonanno & Field, 2001). When this evidence is considered together with the fact that delayed grief has yet to be demonstrated convincingly in any empirical study (Bonanno & Kaltman, 1999; Wortman & Silver, 1989), it would appear that a diagnostic category for delayed grief is unwarranted. Nonetheless, we echo Bonanno and Field's (2001) conclusion that it may be premature to completely dismiss the possible existence of delayed grief on the basis of only two empirical studies, especially when it is considered that the construct is still strongly endorsed in the bereavement literature. However, given the increasingly important role empirical evidence plays in modern clinical science, we also echo Bonanno and Field's (2001) conclusion that the burden of proof now lies

Varieties of Grief Experience

715

most heavily on the shoulders of those who believe the construct to be valid and clinically useful.

DESCRIPTIVE STUDIES OF THE GRIEVING PROCESS In the next section, we consider more closely the distinction between common, timelimited disruptions in functioning and more severe or chronic grief reactions. We review a relatively large body of research unified by its emphasis on the descriptive or phenomenological features of grieving. This research provides further evidence consistent with DSM-IV's conception of bereavement as a stressor event with relatively normal and predictable consequences. Our review reveals that most bereaved individuals commonly experience four types of disrupted functioning in the first year after their loss: cognitive disorganization, dysphoria, health deficits, and disruptions in social and occupational functioning. There was also considerable evidence of positive experiences during bereavement, which further underscores the normative aspects of the common grief experience. In addition, and similar to the longitudinal data, large numbers of bereaved individuals also tended not to show any major disruptions in functioning, while a relatively smaller subset of bereaved individuals tend to exhibit more extreme and enduring variations on these same disruptions in a manner consistent with chronic grief. Cognitive Disorganization A number of investigators have noted the disorganizing impact of interpersonal losses. These difficulties appear to be largely due to the way interpersonal losses challenge the survivor's personal sense of identity and to the difficulty most bereaved survivors experience comprehending a loved one's death. These types of cognitive alterations have been found in most bereaved people during the initial months after the loss and, although many bereaved people report continued struggles to make sense of a loss even years after its occurrence, gradually decline during the first year. In addition, there appear to be two subsets of bereaved individuals: one group who show no signs of altered cognitive functioning, and one group who evidence more pronounced alterations in cognitive functioning similar to trauma reactions. Confusion and preoccupation. A number of investigators have reported that bereaved individuals often experience difficulty accepting the reality of their loss, as well as an accompanying sense of derealization, disorganization, and preoccupation (Lindemann, 1944; Parkes, 1972; Parkes & Weiss, 1983; Shuchter & Zisook, 1993). Shuchter and Zisook recruited a large sample (N = 350) of conjugally bereaved individuals using death certificates and observed that 2 months after the loss, 70% of the sample found it ``hard to believe'' that their spouses had actually died, and that almost half (49%) continued to do so into the second year of bereavement. However, more extreme cognitive difficulties were only exhibited by a considerably smaller portion of the sample. For instance, about one fifth of the sample reported at 2 months post-loss that they had difficulties concentrating (20%), or making decisions (17%). Likewise, a relatively small subset of the sample reported that they made ``more mistakes that usual'' at work or in other areas of functioning (12% at 7 months, 11% at 13 months). These proportions were greater, however, than a matched, married sample, in which

716

G. A. Bonanno and S. Kaltman

only 1% to 4% endorsed any of these items. Finally, although it has been assumed that many bereaved individuals feel that something is wrong with them, or that they are ``approaching insanity'' (Lindemann, 1944, p. 142), when Shuchter and Zisook (1993) asked participants directly if they felt that something was wrong with their minds, only 5% endorsed this item. Again, however, this was greater than the 1% endorsement of this item by the married counterparts. Similar findings were reported by Horowitz and colleagues (Horowitz et al., 1997). Among a sample of middle-aged, conjugally bereaved adults, the majority (72%) experienced ``unbidden'' memories and images of the deceased at 6 months post-loss. However, fewer participants experienced extreme difficulties concentrating (34%) or perceptual aberrations in which it seemed that the deceased was still alive (24%) or had appeared in a public place (17%). Even smaller subsets of participants evidenced cognitive alterations commonly associated with trauma reactions (DSM-IV; APA, 1994), such as hypervigilance (13%), or a foreshortened sense of the future (12%). Disturbances of identify. A related feature of grief-related cognitive disruptions is the bereaved individual's sense of lost identity or merger of identity with the deceased loved one. Shuchter and Zisook (1993) reported that most (87%) of their conjugal bereavement sample endorsed the item ``a piece of me is missing,'' while large portions (55%) found themselves doing things more like their deceased spouses, or becoming more like their deceased spouses (39%). More extreme forms of identity merger have been noted among severely grieved individuals, including taking on the physical symptoms that the deceased had experienced (Anderson, 1949; Bowlby, 1980; Lindemann, 1944). Although the clinical nature of these reports makes it difficult to gauge the validity of the evidence, Shuchter and Zisook (1993) found that a small subset (10%) of their conjugally bereaved sample experienced the same physical symptoms as their recently deceased spouses. Likewise, Horowitz et al. (1997) found that 14% of their conjugally bereaved sample experienced recurrent thoughts that their own death would follow, or mirror, their spouse's death. These latter findings are especially striking given that they were observed in samples of middleaged, and presumably, healthy individuals. Sense of disrupted future. In the early months after a loss, the majority of bereaved individuals report uncertainty about the future (Horowitz et al., 1997; Shuchter & Zisook, 1993). However, by the end of the first year of bereavement, most bereaved individuals no longer report this experience (Horowitz et al., 1997). Small subsets of bereaved individuals (less than 15%) also tend to experience a sense of hopelessness or perception of foreshortened future (Horowitz et al., 1997; Shuchter & Zisook, 1993). Lehman et al. (1987) found that 4 to 7 years after a sudden loss, conjugally bereaved individuals may still endorse feeling less ``optimistic about the future'' than matched controls. The long-term search for meaning. Another cognitive alteration commonly associated with bereavement is the concerted and enduring search for some way to understand or find meaning in the loss. Schwartzberg and Janoff-Bulman (1994) questioned undergraduates who had experienced the death of a parent within the past 3 years and a group of matched controls on several meaning-oriented dimensions. Compared with their matched counterparts, the bereaved participants in general

Varieties of Grief Experience

717

described the world as less meaningful and believed more strongly that personal events were determined by chance. Further, the more grief the bereaved individuals reported, the less likely they were to perceive the world as meaningful or to perceive themselves as worthy, and the more likely they were to believe that personal events were determined by chance or by powerful others. Lehman et al. (1987) demonstrated convincingly that the search for meaning after a loss is a typically long term if not impossible proposition. As late as 4 to 7 years after a sudden loss of a spouse or child, the vast majority of bereaved individuals continued to talk about the loss, to review memories, thoughts, or mental pictures of the deceased, or to ask themselves the question ``Why me?'' or ``Why my spouse/child?'' Perhaps due to the potentially traumatic nature of unexpected or violent losses, close to half the participants in this study also reported that they continued to relive, or review events leading up to, their loved one's death. Finally, 68% of the bereaved spouses and 59% of the bereaved parents reported that they had not found any meaning or made any sense at all of the loss. Dysphoria Another dimension of bereavement, closely related to disordered cognitive functioning, is prolonged or intensified emotional malaise or dysphoria. Most but not all bereaved individuals experience dysphoric states in the early months after a loss that gradually subside during the initial year of bereavement, while more extreme or enduring forms of dysphoria tend to characterize only the smaller minority of severely grieved individuals. Dysphoric emotion. Myriad clinical accounts of bereavement have associated interpersonal losses with a broad scope of distressing emotions, most often centering around anger, irritability, hostility, sadness, fear, and guilt (Abraham, 1924; Belitsky & Jacobs, 1986; Bowlby, 1980; Cerney & Buskirk, 1991; Kavanagh, 1990; Lazare, 1989; Osterweis, Solomon, & Green, 1984; Raphael, 1983). Shuchter and Zisook (1993) questioned conjugally bereaved participants about their emotional experiences in the first year after the loss. Surprisingly, only a small portion of the participants in this study (less than 15%) endorsed having experienced the emotions commonly linked to grieving, such as anger, guilt, and fear. One explanation for the infrequency of dysphoric emotion in this study may be that retrospective self-report measures are not adequate to accurately record ephemeral emotional experiences. A far greater prevalence of dysphoric emotion during conjugal bereavement was observed, in fact, in a subsequent study that measured facial expressions of emotion as they occurred during a 6-month interview about the lost relationship (Bonanno & Keltner, 1997). Using this immediate, nonverbal assessment, anger and sadness were exhibited by almost two-thirds of the bereaved participants, and contempt and disgust by approximately one third (Bonanno & Keltner, 1997). The common experience of dysphoric emotions during bereavement has also been demonstrated by recent investigations of bereaved individuals' narratives as they discussed their loss. Stein, Folkman, Trabasso, and Christopher-Richards (1997) found that bereaved gay men described on average 13.7 negative emotional states in interviews during the first month after their loss. Bonanno, Mihalecz, and LeJeune (1999) coded emotion themes from bereaved widows' and widowers' narrative descriptions of the lost relationship at the 6-month point in bereavement and found

718

G. A. Bonanno and S. Kaltman

a wide range of dysphoric themes. Distress was the most common theme, evidenced in 39% of the narratives, followed by sadness (36%), contempt (34%), and anger (14%). In addition, the themes of anxiety, shame, guilt, disgust, envy, fear, and jealousy were evidenced in lesser frequencies. Pining or yearning. From an emotional perspective, although grief may evoke a range of different emotional responses, it is nonetheless distinguished from emotion by its enduring nature and by its recruitment of long-term coping efforts (Bonanno, in press; Izard, 1977; Lazarus, 1991). Several investigators have, however, identified more complex and enduring aspects of dysphoria during bereavement. For example, Parkes (1970) concluded that ``the central and pathognomonic feature of grief'' is an intense ``pining'' or yearning for the deceased, such that ``without it grief cannot truly be said to have occurred and when present it is a sure sign of a person grieving'' (p. 451). Parkes and Weiss (1983) later identified a ``high yearning'' group of conjugally bereaved individuals who appeared to yearn for their deceased spouses ``constantly,'' ``frequently'', or ``whenever inactive.'' A team of interviewers also found the high yearning group to have a poorer grief outcome during the second to fourth years of bereavement. Although these findings would suggest that intense yearning for the deceased is found only in complicated or severe grief, Parkes and Weiss (1983) stressed that ``high yearning'' was by far the most frequent categorization of their participants, describing about two thirds of the sample during the initial months after the loss. Other studies have also demonstrated the normative, and timelimited aspect of yearning. Of the 350 widows and widowers questioned by Shuchter and Zisook (1993), 77% reported yearning for their deceased spouses in the first 2 months after the loss, while 58% still yearned for their spouses into the second year of bereavement. Using a structured clinical interview format, Horowitz et al. (1997) found that 58% of their conjugally bereaved sample experienced strong yearning for their lost spouses nearly every day at the 6-month point in bereavement, while this percentage dropped to 35% by 14 months post-loss. Loneliness. A similar dysphoric feature commonly associated with grieving is intense loneliness. Shuchter and Zisook (1993) reported that 59% of the widows and widowers they questioned during the first two months of bereavement experienced loneliness, while 37% felt lonely even when around other people. By the second year of bereavement, these proportions dropped to 39% and 23%, respectively while, in contrast, only 3% of a match married sample endorsed experiencing any form of loneliness. Horowitz et al. (1997) report similar findings, with 59% of their sample experiencing loneliness at 6 months post-loss, and 38% experiencing loneliness at 14 months post-loss. Recently, Stroebe, Stroebe, Abakoumkin, and Schut (1996) assessed two different types of loneliness during bereavement based on a distinction proposed by Weiss (1975). Social loneliness was considered a general form of loneliness defined by the absence of an engaging social network and feelings of boredom and social marginality. In contrast, emotional loneliness involved ``a sense of utter aloneness, whether or not the companionship of others is in fact accessible'' (p. 1242) and suggests a deeper form of inner loneliness similar to the dysphoria observed during bereavement. Consistent with this distinction, emotional loneliness turned out to be the most relevant to the grieving process: Conjugally bereaved participants reported significantly more emotional loneliness than did a matched married comparison sample,

Varieties of Grief Experience

719

while social loneliness was not significantly different between the married and bereaved participants. Health Deficits There is considerable evidence that the stress of interpersonal loss incurs a significant health cost. This has been observed most commonly in the form of increased doctor visits and complaints about general health. As with the other aspects of grieving, health deficits are commonly observed among recently bereaved individuals, but appear to be particularly enduring among a small subset of severely grieved individuals. Health behaviors and complaints. A number of studies have provided clear evidence associating interpersonal loss with increased somatic difficulties, including shortness of breath, palpitations, digestive difficulties, loss of appetite, restlessness, and insomnia (Clayton, 1974; Horowitz et al., 1997; Lindemann, 1944; Parkes, 1970). Several studies offered preliminary evidence for increased mortality and morbidity rates during bereavement (Marris, 1958; Young, Benjamin, & Wallis, 1963). However, several studies had also failed to find bereavement-related health deficits (Clayton, 1982; Heyman & Gianturco, 1973). To investigate these inconsistencies, Thompson, Breckenridge, Gallagher, and Peterson (1984) compared the health of an elderly bereaved sample 2 months into bereavement with an elderly comparison sample and statistically controlled for a number of potentially confounding socioeconomic and demographic variables. With potential confounding variance controlled, the bereaved participants still reported more new or worsened illnesses, more severe illnesses, more new or increased use of medications, poorer perceived health, and poorer health relative to others of the same age. Parkes (1964) examined doctor visits prior to and after the death of a spouse. In the first 6 months of bereavement, doctor visits increased by 60%. Although the number of doctor visits fell during the second and third 6-month periods, participants on the whole still visited their doctors more often than they had prior to the loss. Bereaved individuals have also been found to report significantly greater health complaints, into the second year of bereavement, relative to matched comparison groups (Maddison & Viola, 1968; Parkes & Brown, 1972). In Lehman et al.'s (1987) more lengthy 4 to 7 year study, however, neither parentally bereaved nor conjugally bereaved individuals differed from matched controls in perceived health scores. To explore individual differences in somatic complaints over time, Bonanno et al. (1999) categorized conjugally bereaved individuals as has having high or low levels of somatic symptoms based on the sample median at the 6-month point in bereavement. Using this definition, approximately one third (35%) of the participants had low levels of somatic complaints at each assessment through 25 months, while another third (32%) showed initial elevations in somatic symptoms but dropped to low levels between the first and second year of bereavement. A smaller group of bereaved participants (19%) had varied levels of somatic complaints across assessments, while only a small subgroup, comprising 13% of the sample, reported chronically elevated somatic complaints. Neuroendocrine activity and immune functioning. Several studies have examined the possible role of neuroendocrine activity, such as changes in cortisol or serum

720

G. A. Bonanno and S. Kaltman

growth hormone levels, during bereavement. However, as yet these studies have failed to provide clear evidence that neuroendocrine activity is altered significantly during bereavement, or that it influences the course of grieving (Kim & Jacobs, 1993). In contrast, there is a growing body of evidence to suggest a link between grief and a relatively short-lived compromise in immune functioning. Bartrop, Luckhurst, Lazarus, Kiloh, and Penny (1977) examined cellular immunity in a conjugally bereaved sample at 2 and 6 weeks post-loss and in a matched control sample. The bereaved group showed reduced T-cell responsivity at both assessments, but did not differ from the comparison sample in B-cell responsivity, or in T-cell or B-cell counts. Schleifer, Keller, Camerino, Thornton, and Stein (1983) obtained repeated assessments of immune functioning among 15 men before and after the death of their wives from metastatic breast cancer. In the first month following the loss, over half of the bereaved men showed reduced lymphocyte responsivity. However, there were no differences pre and postloss in absolute T-cell or B-cell counts. Further, lymphocyte responsivity was not different from preloss levels when assessed beyond the first 2 months of bereavement, suggesting that immune suppression due to loss is specific and relatively short-lived. Subsequent studies have refined this conclusion, indicating that the link between bereavement and brief immune suppression may be mediated by depression (Irwin, Daniels, Smith, Bloom, & Weiner 1987; Irwin & Weiner, 1987; Linn, Linn, & Jensen, 1984). For instance, Zisook et al. (1994) examined several different immune functioning variables in a sample of middle-aged conjugally bereaved women and in a sample of matched controls. There were no significant differences between the bereaved women at 2 months post-loss and their married counterparts on any of the immune variables. However, when the bereaved sample was separated into depressed and nondepressed widows using the DSM-III-R, depressed widows had a lower concentration of T-cells, lower NK activity, and a trend toward lower lymphocyte stimulation responses compared to the nondepressed widows. As compelling as the link between grief-related depression and a short-term suppression of immune functioning may be, it has not yet been demonstrated that the depressive aspects of grief exert a longer term influence on immune functioning, or that the short-term suppression effects produce any lasting health consequences (Irwin & Pike, 1993). Additional research will be necessary to address this question further. Mortality. A dramatic aspect of the health consequences of grieving is the association of bereavement with increased mortality, an association that appears to be particularly robust in the early months after a loss. In one of the more impressive demonstrations of the bereavement± mortality effect, Kaprio, Koskenvuo, and Rita (1987) examined longitudinal data on 95,647 conjugally bereaved individuals in Finland during their first 4 years of bereavement. Compared to the average expected mortality rate in Finland, the conjugally bereaved individuals showed an overall 6.5 percent higher likelihood of dying. This increase was smaller (3.2% increase) for deaths by natural causes, e.g., cardiovascular disease. However, deaths by natural causes were almost twice as high as the normal expectancy rate in the first week of bereavement. Deaths by violent causes, e.g., traffic accidents, were even more likely (93% increase) among the bereaved individuals. Finally, bereaved individuals showed considerably more deaths (242% increase) by suicide.

Varieties of Grief Experience

721

Since the extensive literature on the bereavement ± mortality effect has been reviewed in detail elsewhere (Mergenhagen, Lee, & Gove, 1985; Stroebe & Stroebe, 1993; Stroebe, Stroebe, Gergen, & Gergen, 1981; Stroebe & Stroebe, 1987), we will not consider individual studies further in this article. Rather, we reiterate the conclusions of Stroebe and Stroebe (1993). Following an extensive review of a large number of cross-sectional and longitudinal bereavement studies conducted in several different countries, they concluded that despite inconsistencies across studies, there was plentiful evidence that the ``bereaved are indeed at higher risk of dying than are nonbereaved persons'' (p.188), and that the risk of increased mortality extends to all types of bereavement, particularly in the early months after the loss and particularly among younger bereaved individuals. Disrupted Social and Occupational Functioning In addition to its cognitive, emotional, and somatic manifestations, grief has also been associated with disruptions in social and occupational functioning. These difficulties have been observed most commonly in the form of social withdrawal and isolation, or as the inability to fulfill normal social and occupational roles. As with the other features of grief, disruptions in social and occupational functioning are common for most but not all bereaved individuals in the initial months after a loss, and tend to appear in a more extreme and enduring form among a subset of severely grieved individuals. Social Withdrawal and Isolation Lindemann (1944) first reported disturbed interpersonal relations among bereaved individuals. Subsequently, Parkes and Weiss (1983) observed that the majority of the conjugally bereaved individuals they interviewed had withdrawn from participation in social activities in the early months after the loss, but that more than half of their sample showed increased social activity by the end of the first year of bereavement and that the vast majority had returned to their previous normal level of social activity sometime after the second year. Recent studies have more precisely mapped the short-term course of social withdrawal in the normal grieving process. Horowitz et al. (1997) found that 50% of a middle-aged, conjugal bereavement sample felt that they were less emotionally available in significant relationships after 6 months of bereavement than they were before the loss. By the 14-month point, however, the portion of the sample feeling emotionally unavailable had dropped to only 19%. Similarly, Shuchter and Zisook (1993) found that at the 13-month point of bereavement, only relatively small percentages of the conjugally bereaved individuals they interviewed endorsed items such as, ``My feelings are easily hurt'' (12%), ``I feel that people are unfriendly'' (4%), or, ``(I am) very self-conscious with others'' (6%). Negative impact on others. One factor that may contribute to the distinction between normal social withdrawal and more chronic social difficulties is the way grief may impact on others. In other words, the constant expression of pain on the part of severely grieved individuals may drive away potential avenues of social support (Bonanno & Keltner, 1997; Coyne, 1976; Harber & Pennebaker, 1992; Kelly & McKillop, 1996). This possibility received preliminary support in a recent study in which untrained observers were asked to rate their subjective reactions to

722

G. A. Bonanno and S. Kaltman

videotapes of conjugally bereaved individuals as they described their loss to an interviewer (Keltner & Bonanno, 1997). Those bereaved participants who were perceived by the observers as less well adjusted also evoked in the observers greater frustration and less compassion. In a subsequent study (Capps & Bonanno, 2000), another group of untrained judges were shown narrative transcripts of these interviews. The more the bereaved participants described negative thoughts and emotions during the interview, the more the judges reported they would be inclined to avoid the participant. Role disruptions. Bereaved individuals also appear to struggle in their social roles as parents or in their careers (Parkes & Weiss, 1983). Several cross-sectional studies have provided data consistent with this conclusion. Compared to a matched sample of married men, widowers recruited during their first year of bereavement reported greater difficulties in their work roles both outside the home and in the home, greater difficulties managing spare time, and greater difficulties with their family roles (Tudiver, Hilditch, & Permaul, 1991). In another cross-sectional study, Lehman et al. (1987) also found that 4 ± 7 years after the sudden death of a spouse, conjugally bereaved individuals were still less likely to look forward to doing things with others, less confident that they could handle or cope with a serious problem or major change in their life, and rated the general state of their lives more negatively, compared to a matched married group. The conjugally bereaved parents in the Lehman et al. (1987) study showed similar but nonsignificant trends toward the same results. A more precise estimation of the impact of bereavement on work performance and its decrease over time was provided by Shuchter and Zisook (1993) in their prospective study of a large, representative sample of recently conjugally bereaved adults. They found that over a third (36%) of their sample reported dissatisfaction in their work performance at 7 months, and that 28% still endorsed this problem at 13 months post-loss. Although these percentages show that dissatisfaction with work may not be as widespread as other grief-related difficulties, is was still more common among bereaved individuals relative to the 10% of the married comparison sample who reported such difficulties. Difficulties with new relationships. The death of a spouse is commonly associated with temporary difficulties initiating and maintaining new intimate relationships. Horowitz et al. (1997) found that 59% of the conjugally bereaved individuals they interviewed at the 6-month point in bereavement had difficulties developing new intimate relationships. However, by 14 months, only about a third of the sample (32%) reported such difficulties. Schneider, Sledge, Shuchter, and Zisook (1996) reported similar findings. Because these investigators considered a relatively large conjugal bereavement sample (N = 350), they were also able to identify relevant gender differences (W. Stroebe & Stroebe, 1993). Two months after the deaths of their spouses, 55% of the widowers and 46% of the widows had difficulties developing new relationships. By 25 months, however, the portion of widowers who experienced difficulties with new relationships had dropped to 37%, while the proportion for widows had risen significantly to 58%. Widows were also relatively slower to engage in new romantic involvements. By the 2-month point in bereavement, only 3% of the widows were invovled in new romantic relationships and this portion rose only modestly to 19% at 25 months. On the other hand, 12% of the widowers were in new romantic relationships at 2 months, and by 25 months this portion had risen

Varieties of Grief Experience

723

sharply to 61%. Widowers were also more likely than widows to remarry. Younger widows were more likely to develop new romantic involvements than older widows, while for widowers new romantic involvements were associated with income and education level. Finally, but again similar to other facets of grief, there appears to be a more severely grieved subgroup for whom developing new relationships is a particularly exacting task. Regardless of gender, bereaved individuals who were depressed found it more difficult to begin new relationships. Positive Aspects of Bereavement Until recently, positive experiences were thought to be quite rare during bereavement. Further, positive behaviors and responses that would be considered adaptive in most situations have been viewed as indicators of denial or avoidance when they are observed during bereavement (Deutsch, 1937; Parkes & Weiss, 1983; Raphael, 1983; Sanders, 1979). However, recent studies have demonstrated quite clearly that positive cognitive and emotional experiences associated with bereavement are far from infrequent, and that such experiences tend to be associated with a relatively mild or normal grief course (Bonanno & Kaltman, 1999). Positive thoughts, beliefs, and appraisals. Several recent studies have shown that significant interpersonal losses often induce relatively positive changes in identity, sometimes even in the first few months after the loss. Shuchter & Zisook (1993) found that at the 2-month point in bereavement, 42% of the conjugally bereaved respondents they questioned had already reported that they were better people for having gone through the experience of grief. Further, over a third of their 2-month sample (36%) reported that they had begun to enjoy the freedom of being on their own. And these percentages only increased over time. By the second year of bereavement, over half the bereaved sample had endorsed each item. Evidence from recent narrative studies has provided even more striking contrast to the traditional assumption that positive experiences are rare during bereavement. In their study of bereaved gay men's narratives, Stein et al. (1997) coded a number of appraisal variables and found that overall the men in their sample described significantly more positive than negative appraisals in the first month of bereavement. More specifically, the bereaved men reported almost twice as many positive ``belief appraisals'' (e.g., belief in self-growth from past events) than negative belief appraisals (e.g., fear of the future), and three times as many positive goal outcomes (e.g., ``I gave him a massage, and he felt better'') than negative goal outcomes (e.g., ``Although I gave him the medicine, he got worse.''). The proportion of positive appraisals in the bereaved men's narratives was also associated with a greater emphasis on both current and future-oriented goals and plans. Finally, the greater the proportion of positive appraisals in the bereaved men's narratives during the first month of bereavement, the more likely they were to show improved a health outcome 12 months after the loss. In a related study, Capps & Bonanno (2000) coded bereavement narratives for the extent that they included positive and negative thought content (thoughts, beliefs, appraisals, opinions, and other forms of personal information that could not be verified without access to the narrators' internal state) (Stiles, Shuster, & Harrigan, 1992). Thought content was slightly more often positive (29%) than negative (25%). Further, positive thought content at the 6-month point in bereavement was predictive

724

G. A. Bonanno and S. Kaltman

of less interviewer-rated grief two years later. In addition, transcripts that showed a greater prevalence of positive thoughts were rated by naive judges as indicative of better adjustment and reduced suffering. Positive emotions and laughter. It has become increasingly evident that positive emotions serve an important adaptive function in the negotiation of negative life events (Bonanno, in press; Fredrickson & Levenson, 1998; Lazarus, Kanner, & Folkman, 1980). Recent bereavement studies have shown that, like positive thoughts and appraisals, positive emotional experiences play an important role in the grieving process (Bonanno & Kaltman, 1999). Shuchter and Zisook (1993) found that 28% of their conjugally bereaved sample reported experiencing relief as early as 2 months after the loss, while the majority of the sample (82%) reported that they were ``comforted'' by a sense that their spouse was in heaven. Stein et al. (1997) found that, while bereaved gay men described overall more negative emotion states than positive emotion states, positive emotion states were far from infrequent and occurred in 39% of all the emotion states coded. In the Capps & Bonanno (2000) study of conjugal bereavement narratives, mention of emotion in general was relatively infrequent, occurring in only 11% of all narrative units. However, when emotions were mentioned, almost half (49%) were positive emotions and an additional 21% included both positive and negative emotion. Finally, in Bonanno et al.'s (1999) investigation of core emotion themes during conjugal bereavement, positive themes were the most common themes mentioned, including pride in the deceased (82%), love/affection, (81%), and happiness (55%). Conjugally bereaved participants have reported experiencing positive emotions while describing their loss to an interviewer (Bonanno et al., 1995). Bonanno and Keltner (1997) examined the facial expressions of emotion that occurred in these same interviews and found that most of the bereaved participants also showed facial expressions associated with enjoyment (smiling: 63%) and amusement (laughter: 58%). Further, smiling and laughter during the 6-month interviews were predictive of reduced grief into the third year of the loss, and this predictive relationship remained significant even when initial levels of grief and self-reported emotion were statistically controlled. In a subsequent study, Keltner and Bonanno (1997) further explored the adaptive role of laughter during bereavement. Bereaved individuals who showed genuine or Duchenne laughter, which involves the orbicularis occuli muscles around the eyes, at some point during their discussion of the loss reported less anger and more enjoyment during the interview, evidenced the dissociation of distress (affective ±autonomic response dissociation), reported better social relations, and evoked more positive responses from strangers who viewed them on videotape.

DIAGNOSTIC STUDIES OF SEVERE OR COMPLICATED GRIEF The descriptive studies reviewed in Section 5 provide considerable evidence that a small but important minority of bereaved individuals evidence many of the same difficulties seen in common grieving, but to a more extreme and enduring degree, thus suggesting a chronic grief pattern. In this section, we review attempts by bereavement researchers to identify severely grieved individuals using a diagnostic approach. Several studies have attempted to identify a general category of complicated or ``pathological'' grief. However, these studies have tended to include many

Varieties of Grief Experience

725

normative or common aspects of grief among the assumed pathological symptoms and, as a result, have been overinclusive. In contrast, other studies have identified chronic depressive and anxiety disorders and, more recently, grief reactions similar to those associated with Posttraumatic Stress Disorder. Consistent with both the DSM-IV and the descriptive studies reviewed earlier, these disorders have been observed in approximately 15% of the individuals studied. Complicated or Pathological Grief Several investigators have attempted to identify a general category of complicated or ``pathological'' grief. Prigerson et al. (1995) assessed a set of common bereavementrelated symptoms in an elderly sample within the first 6 months of bereavement. Based on a principal components analysis of these data, Prigerson et al. identified a depression factor and a grief-specific factor as distinct consequences of bereavement. Field, Bonanno, Williams, and Horowitz (2000) reached a similar conclusion in a study of middle-aged conjugal bereavement. They used the Texas Revised Inventory of Grief (Faschingbauer, 1981) to assess ``grief-specific'' symptoms, and questionnaire measures of depression, anxiety, and somatic complaints to assess ``general distress'' symptoms. It is noteworthy, however, that the grief-specific symptoms in both these studies were dominated by yearning and preoccupation with thoughts of the deceased, features identified in the descriptive studies reviewed in Section 5 of this article as common for most individuals during the early months of bereavement. Thus, the grief-specific symptoms identified in these studies suggest a normal rather than pathological grieving pattern that can be distinguished from depression and anxiety. Kim and Jacobs (1991) developed a structured interview to assess ``pathological grief'' based on the ``symptoms of separation distress.'' Their list of symptoms included ``crying, sighing, yearning and searching for the deceased, preoccupation with thoughts of the deceased, and functioning with a perceptual set for the deceased'' (p. 258). Although Kim and Jacobs did not report full demographic information, it appears that they interviewed a small sample (N = 25) of conjugally bereaved individuals between 6 and 13 months post-loss. Using this method, a surprisingly large portion (64%) of the sample met criteria for pathological grief. A prevalence rate this high suggests that what Kim and Jacobs have termed ``pathological'' are best interpreted as normative aspects of grieving. Indeed, the symptoms these investigators listed for pathological grieving were, again, similar to those identified earlier as common to most bereaved individuals during the early months of bereavement. Further, when Kim and Jacobs examined participants' scores on the Texas Inventory of Grief (Faschingbauer, Devaul, & Ziskook, 1977), they did not find meaningful differences between their pathological grief group and the remaining ``nonpathological'' bereaved participants. Horowitz et al. (1997) also attempted to identify possible criteria for a diagnostic category of ``complicated grief'' using structured interview data. Based on sensitivity and specificity analyses of the individual interview items, they identified a set of seven complicated grief symptoms. These included unbidden memories or intrusive fantasies related to the lost relationship, emotional pangs, distressing yearning for the deceased, feelings of being alone or personally empty, excessive avoidance of reminders (people, places, or activities associated with the deceased), unusual levels of sleep disturbance, and loss of interest in work, social, or recreational activities.

726

G. A. Bonanno and S. Kaltman

Using a criteria that at least three of these symptoms be observed at a level of severity that interferes with daily functioning, Horowitz et al. identified ``Complicated Grief Disorder'' in 41% of the 14-month sample. Only 21% of these participants evidenced a comorbid diagnosis of Major Depression. However, much like the Kim and Jacobs (1991) data, the high prevalence of Complicated Grief diagnosis in this study appears to have resulted from the inclusion of common grief features among the ``symptoms'' of Complicated Grief. Major Depressive Disorder Prevalence rates more consistent with the descriptive evidence reviewed earlier have been found in studies attempting to identify depressive disorders. In one of the earliest diagnostic studies, conducted prior to the DSM-III, Clayton, Halikas, and Maurice (1972) categorized 20% of a conjugally bereaved sample in the first month of bereavement as having ``definite depression'' and another 15% of the sample as having ``probable depression.'' In a subsequent study, using DSM-III criteria, Jacobs, Hanson, Berkman, Kasl, and Ostfield (1989) concluded that Major Depressive Disorder was evident in 32% of a conjugal bereavement sample at 6 months postloss, and in 27% of another conjugal bereavement sample at 12 months post-loss. In a later study, using DSM-III-R criteria, Zisook and Shuchter (1991) identified major depression in 24% of a conjugal bereavement sample at 2 months, and in 16% of the sample at 13 months. These percentages were higher than the 4% depression prevalence found in a matched sample of married individuals. Recently, Zisook, Paulus, Shuchter, and Judd (1997) reexamined Zisook and Shuchter's (1991) data using DSM-IV criteria and identified major depression in 20% of the bereaved sample at 2 months post-loss and in 1% of the matched comparison sample. They also applied DSM-IV criteria for minor depression (depressed mood or loss of interest and two to four depressive symptoms) to categorize an additional 20% of the 2-month bereaved sample and 2% of the comparison sample. Finally, they applied an even weaker subsyndromal depression category, based on the endorsement of at least 2 DSM-IV depressive symptoms (Judd, Rapaport, Paulus, & Brown, 1994), to categorize 11% of the 2-month bereaved sample and 3% of the comparison sample. Using this encompassing set of depression categories, the remaining 49% of the 2-month bereaved sample and 94% of the comparison sample were categorized as having ``no depression.'' When Zisook et al. examined these data longitudinally, they found that the proportions decreased in the bereaved participants at 13 months post-loss to 12% major depression, 17% minor depression, 10% subsyndromal depression, and 62% no depression. At 25 months post-loss, the proportions were 6% major depression, 13% minor depression, 11% subsyndromal depression, and 70% no depression. Generalized Anxiety Disorder A similar prevalence of anxiety-related symptoms have been identified in a number of bereavement studies (Clayton, Desmarais, & Winokur, 1968; Maddison & Viola, 1968; Parkes, 1964; Sable, 1989; Zisook, Schneider, & Shuchter, 1990), and appear to be a particularly tenacious aspect of chronic grief (Zisook & Shuchter, 1986). Jacobs et al. (1990) identified anxiety disorders in 25% of a 6-month conjugal bereavement sample, and in 44% of a 12-month conjugal bereavement sample. Generalized

Varieties of Grief Experience

727

Anxiety Disorder (GAD) was by far the most common diagnosis. Compared to the community prevalence rate of 9%, GAD was identified in 23% of the 6-month bereaved sample and 39% of the 12-month bereaved sample. Panic Disorder was the next most common. Compared to the less than 1% prevalence in the community sample, Panic Disorder was identified in 6% of the 6-month bereaved sample and 13% of the 12-month bereaved sample. In addition, the Jacobs et al. (1990) data evidenced considerable overlap between chronic anxiety and depression. The majority of depressed participants in the 6month sample (63%) were also diagnosed with an anxiety disorder, and all of the bereaved individuals in the 12-month sample who had Major Depression were also diagnosed with an anxiety disorder. Thus, although depression typically declines during bereavement, it appears that the more anxious, chronically grieved individuals also continue to experience enduring depression. Additional longitudinal data will be needed to examine this conclusion further. Posttraumatic Stress Disorder Most recently, bereavement investigators have attempted to identify the symptoms of Posttraumatic Stress Disorder (PTSD) as a possible response to bereavement. Rynearson and McCreery (1993) examined grief and trauma symptoms in a small sample (N = 18) of individuals who had lost a family member to homicide. Relative to scores obtained in previous studies of natural-death bereavement, the homicidebereavement participants reported more intrusive thoughts on the Impact of Event Scale (Horowitz, Wilner, & Alvarez, 1979) and increased grief-specific symptoms on the Texas Revised Inventory of Grief (Faschingbauer, 1981). In addition, the homicide bereavement participants reported dissociative experiences at a level 3 times that of nonbereaved individuals. In a related study, Green (1997) assessed lifetime diagnoses among women who had experienced sexual assault, or a potentially traumatic loss (accidental death, homicide, or suicide). Although the women in this study were considerably younger (mean age = 19) than in other bereavement studies, the results were comparable. The traumatic loss group had high instances of trauma-related disorders: 16% had Acute Stress Disorder and another 6% had PTSD. In addition, the traumatic loss group had the highest scores on the Dissociative Experiences Scale (Bernstein-Carlson & Putnam, 1986). Interestingly, the proportion of the traumatic loss group that at any point received the Major Depression diagnosis (13%) was lower than in previous bereavement studies, suggesting that traumatic losses tend to produce trauma responses rather than depression responses. However, given the young age of the sample and the fact that the lifetime diagnoses were obtained retrospectively, these results need to be considered with great caution. More recently, several studies have reported associations between PTSD symptoms and the loss of a loved one through violent death. Zisook, Chenstova-Dutton, and Shuchter (1998) examined conjugally bereaved individuals 2 months into bereavement and found that approximately 10% of those whose spouses had died of natural causes met the criterion for PTSD. In contrast, over one-third of the participants whose spouses had died of suicide or accidents met PTSD criteria. In our own recent study (Kaltman & Bonanno, 1999), we found that conjugally bereaved individuals who lost spouses to violent deaths (suicide, accident, homicide) showed more PTSD symptoms at each assessment across 25 months of bereavement, and also tended to

728

G. A. Bonanno and S. Kaltman

show more chronic depressive symptoms than bereaved individuals whose spouses had died of natural causes. Together, these studies suggest that loss due to violent death may be one of the factors that most strongly contribute to the eventual development of chronic grieving.

FUTURE DIRECTIONS IN BEREAVEMENT RESEARCH When considered together, the longitudinal, descriptive and diagnostic studies reviewed in this article were generally supportive of the approach to bereavement adopted by the DSM-IV. That is, the majority of bereaved individuals appear to endure similar types of disruptions in daily functioning, and these experiences when considered together appear to constitute a normal reaction to the stress of interpersonal loss. In further concordance with DSM-IV, and in contrast to the numerous types of complicated grief proposed in the bereavement literature, a small but important subset of bereaved individuals appear to experience more intense and enduring disruptions in functioning that overlap considerably with the existing diagnostic categories of Major Depression, Generalized Anxiety, and Posttraumatic Stress Disorder. The supportive evidence for the use of existing DSM-IV categories to explain chronic grief reactions raises an important question: Do chronically grieved individuals warrant unique treatment considerations compared to other individuals showing the same diagnostic indicators? A number of theorists have proposed using standard interventions for depression to treat chronic depression during bereavement, such as interpersonal psychotherapy (Rounsaville & Chevron, 1982) and ``biologically informed'' psychotherapy (Zisook & Shuchter, 1996, p. 32). Similarly, psychotherapeutic interventions for bereaved individuals considered to be at risk for developing PTSD symptoms (e.g., bereavement after homicide) have been modeled after standard trauma interventions (e.g., Horowitz, 1986; Rynearson, 1996). However, in each of these approaches, the standard interventions were modified to at least some degree to encompass the unique features of bereavement. Unfortunately, there have been few systematic studies to examine whether such modifications are warranted. It will be imperative for future bereavement researchers to address this issue. Bereavement-specific treatment approaches have also been suggested by the proponents of the more complex grief taxonomies. Indeed, these theorists have argued for the necessity of various complicated grief categories based on their perception that the clinical implications derived from the DSM categories are ``unacceptable'' (Rando, 1992, p. 55). By contrast, the conclusion we have advanced in this review, that the various complicated grief taxonomies are unwarranted, carries with it the further proposition that the treatment implications derived from these taxonomies may also be unwarranted. For instance, uncritical acceptance of the delayed or inhibited grief category has led to assumptions that the experience and expression of the emotional pain of the loss is an essential component of treatment (Rando, 1993; Worden, 1996), and that suspected cases of delayed grief need to be referred to an expert grief counselor (Humphrey & Zimpfer, 1996). Given the lack of empirical evidence for these assumptions, it would appear incumbent upon those who espouse these interventions to provide systematic evidence in support of their purported efficacy.

Varieties of Grief Experience

729

Our review also revealed a significant distinction that has not been well represented in either the DSM-IV or the contemporary bereavement literature. The available research suggests that considerable numbers of bereaved individuals experience little or no overt disruptions in daily functioning as a result of their bereavement. In other words, these individuals do not show the common grief symptoms, but neither do they show any indication of abnormal or pathological mourning or any concrete signs of delayed difficulties. One implication of this distinction is that there needs to be greater research on prebereavement factors, and their role in informing the initial reaction to loss. There has been remarkably little research on prebereavement variables, perhaps due in part to the fact that such measurements are notoriously difficult to obtain. Certainly, some of the contextual factors that might inform initial grief can be measured retrospectively (e.g., type of loss, change in financial status, etc.). However, a number of other variables that might potentially serve as predictors of grief response (e.g., coping styles, attachment behavior, the quality of the marital relationship) tend to be confounded with initial grief reactions and, thus, are best measured a priori. One of the few bereavement studies to incorporate prebereavement measurements, conducted by Folkman and her colleagues (Folkman, Chesney, Collette, Boccellari, & Cooke, 1996), is illustrative of this point. Folkman et al. (1996) first recruited a sample of gay men who were caring for their partners, many of whom were expected to die of AIDS in the near future. They measured a number of variables, including life stressors, depression, and coping strategies, and followed the caregivers through their partner's death and into bereavement. Using this design, Folkman et al. were able to control for initial increases in depression after the partner's death, and to show that depression at 7 months post-loss was predicted by prebereavement stressors (daily hassles) and by the prebereavement use of distancing and self-blame to cope with the burden of care giving. More research of this type will greatly enhance our understanding of the initial stages of bereavement, and may provide invaluable information to inform strategies for early clinical intervention. Although much of the recent bereavement research has focused on determining the factors that predict grief severity over the long term (for a review of this research, see Bonanno & Kaltman, 1999), there has been relatively little research on the role of the initial grief response. Indeed, if initial grief reactions are considered at all, generally they are viewed in terms of variance that needs to be controlled while predicting later grief. The research we reviewed in this article suggests, however, that intense or pronounced initial grief reactions may be an important indicator of whether individuals go on to develop chronic grief reactions. The available evidence on this point was far from conclusive, however, therefore indicating another important area for future research. A further implication of our review is suggested by the absence of clear support for the concept of delayed grief. In addition to assumptions about treatment, uncritical acceptance of the existence of delayed grief has also unnecessarily limited the scope of bereavement research. For example, delayed grief is widely assumed to arise from the minimization or avoidance of grief-related distress in the early months after a loss. This assumption has tended to preclude investigation of the potentially adaptive value of grief-related avoidance, or of positive experiences such as laughter. It is only recently that these phenomena have been examined empirically and, in contrast to traditional assumptions about the importance of ``working through'' grief, the evidence thus far indicates that adjustment during bereavement is fostered by

730

G. A. Bonanno and S. Kaltman

minimizing the experience and expression of grief-related negative emotions (Bonanno & Keltner, 1997; Bonanno et al., 1995, 1999; Capps & Bonanno, 2000), by fewer attempts to comprehend a loss or one's reactions to the loss (NolenHoeksema, McBride, & Larson, 1997), and by positive emotions and appraisals (Bonanno et al., 1999; Keltner & Bonanno, 1997; Stein et al., 1997). Finally, the evidence reviewed in this article suggests an imperative need for crosscultural and comparative bereavement research. No study has yet compared grief severity across cultures using standardized measured and a prospective, longitudinal design. When such data become available, it will be possible to determine whether the types of grief course suggested in Fig. 1 characterize human bereavement in general, or whether specific cultures might show unique patterns of outcome. By the same token, it may be equally fruitful to begin systematic comparisons of grief severity across different types of loss events, such as the loss of a job (Harvey & Miller, 1998), or symbolic losses, such as the loss of a child to mental illness or intellectual disability. It is our hope that our review will stimulate further longitudinal research of these important points, and that a greater understanding of the course of normal and complicated grief reactions will be forthcoming in the not so distant future. AcknowledgmentsÐThe authors thank Diane Arnkoff for her many useful insights during the development of this manuscript. REFERENCES Abraham, K. (1924). A short study of the development of the libido; viewed in the light of mental disorders. In: K. Abraham (Ed.), Selected papers on psychoanalysis ( pp. 418 ± 501). London: Hogarth Press. American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author. Anderson, C. (1949). Aspects of pathological grief and mourning. Internal Journal of Psychoanalysis, 30, 48 ± 55. Asendorpf, J. B., & Scherer, K. R. (1983). The discrepant repressor: differentiation between low anxiety, high anxiety, and repression of anxiety by autonomic ± facial ± verbal patterns of behavior. Journal of Personality and Social Psychology, 45, 1334 ± 1346. Bartrop, R. W., Luckhurst, E., Lazarus, L., Kiloh, L. G., & Penny, R. (1977). Depressed lymphocyte function after bereavement. Lancet, 1, 834 ± 836. Belitsky, R., & Jacobs, S. (1986). Bereavement, attachment theory, and mental disorders. Psychiatric Annals, 16, 276 ± 280. Bernstein-Carlson, E. M., & Putnam, F. W. (1986). Development, reliability, and validity of a dissociation scale. Journal of Nervous and Mental Disease, 174, 727 ± 735. Bonanno, G. A. (1998). The concept of ``working through'' loss: a critical evaluation of the cultural, historical, and empirical evidence. In: A. Maercker, M. Schuetzwohl, & Z. Solomon (Eds.), Posttraumatic stress disorder: vulnerability and resilience in the life-span ( pp. 221 ± 247). GoÈttingen, Germany: Hogrefe and Huber. Bonanno, G. A. (in press). Grief and emotion: comparing the grief work and social ± functional perspectives. In: M. Stroebe, W. Stroebe, R. O. Hansson, & H. Schut (Eds.), New handbook of bereavement: consciousness, coping, and care. Washington, DC: American Psychological Association. Bonanno, G. A., & Field, N. P. (2001). Predicting bereavement outcome across five years: I. The question of delayed grief. American Behavioral Scientist, 44, 798 ± 816. Bonanno, G. A., & Kaltman, S. (1999). Toward an integrative perspective on bereavement. Psychological Bulletin, 125, 760 ± 776. Bonanno, G. A., & Keltner, D. (1997). Facial expressions of emotion and the course of bereavement. Journal of Abnormal Psychology, 106, 126 ± 137. Bonanno, G. A., Keltner, D., Holen, A., & Horowitz, M. J. (1995). When avoiding unpleasant emotion

Varieties of Grief Experience

731

might not be such a bad thing: verbal ± autonomic response dissociation and midlife conjugal bereavement. Journal of Personality and Social Psychology, 46, 975 ± 989. Bonanno, G. A., Mihalecz, M. C., & LeJeune, J. T. (1999). The core themes of conjugal loss. Motivation and Emotion, 23, 175 ± 201. Bonanno, G. A., Znoj, H. J., Siddique, H., & Horowitz, M. J. (1999). Verbal ± autonomic response dissociation and adaptation to midlife conjugal loss: a follow-up at 25 months. Cognitive Therapy and Research, 23, 605 ± 624. Bowlby, J. (1980). Loss: sadness and depression (attachment and loss), vol. 3. New York: Basic Books. Capps, L. & Bonanno, G. A. (2000). Narrating bereavement: thematic and grammatical predictors of adjustment to loss. Discourse Processes, 30, 1 ± 26. Cerney, M. W., & Buskirk, J. R. (1991). Anger: the hidden part of grief. Bulletin of the Menninger Clinic, 55, 228 ± 237. Clayton, P. J. (1974). Mortality and morbidity in the first year of bereavement. Archives of General Psychiatry, 30, 747 ± 750. Clayton, P. J. (1982). Bereavement. In: E. S. Paykel (Ed.), Handbook of affective disorders ( pp. 403 ± 415). London: Livingstone. Clayton, P. J., Desmarais, L., & Winokur, G. (1968). A study of normal bereavement. American Journal of Psychiatry, 125, 168 ± 178. Clayton, P. J., Halikas, J. A., & Maurice, W. L. (1972). The depression of widowhood. British Journal of Psychiatry, 120, 71 ± 76. Coyne, J. C. (1976). Depression and the response of others. Journal of Abnormal Psychology, 85, 186 ± 193. Deutsch, H. (1937). Absence of grief. Psychoanalytic Quarterly, 6, 12 ± 22. Faschingbauer, T. R. (1981). The Texas revised inventory of grief manual. Houston, TX: Honeycomb. Faschingbauer, T. R., Devaul, R. D., & Zisook, S. (1977). Development of the Texas inventory of grief. American Journal of Psychiatry, 134, 696 ± 698. Field, N. P., Bonanno, G. A., Williams, P., & Horowitz, M. J. (2000). Appraisals of blame in adjustment to conjugal bereavement. Cognitive Therapy and Research, 24, 549 ± 568. Folkman, S., Chesney, M., Collette, L., Boccellari, A., & Cooke, M. (1996). Postbereavement depressive mood and its prebereavement predictors in HIV+ and HIV gay men. Journal of Personality and Social Psychology, 70, 336 ± 348. Fredrickson, B. L., & Levenson, R. W. (1998). Positive emotions speed recovery from the cardiovascular sequelae of negative emotions. Cognition and emotion, 12, 191 ± 220. Glick, I. O., Weiss, R. S., & Parkes, C. M. (1974). The first year of bereavement. New York: Wiley-Interscience. Gorer, G. (1965). Death, grief, and mourning in contemporary Britain. London: Tavistock Publications. Green, B. L. (1997). Traumatic loss: conceptual issues and new research findings. Keynote address. 5th International Conference on Grief and Bereavement, Washington, DC. Hansson, R. O., Carpenter, B. N., & Fairchild, S. K. (1993). Measurement issues in bereavement. In: M. S. Stroebe, W. Stroebe, & R. O. Hansson (Eds.), Handbook of bereavement: theory, research and intervention ( pp. 62 ± 76). Cambridge, England: Cambridge Univ. Press. Harber, K. D., & Pennebaker, J. W. (1992). Overcoming traumatic memories. In: S. A. Christianson (Ed.), The handbook of emotion and memory ( pp. 359 ± 388). Hillsdale, NJ: Erlbaum. Harvey, J. H., & Miller, E. D. (1998). Toward a psychology of loss. Psychological Science, 9, 429 ± 434. Heyman, D. K., & Gianturco, D. T. (1973). Long term adaptation by the elderly in bereavement. Journal of Gerontology, 28, 359 ± 362. Horowitz, M. J. (1986). Stress response syndromes. Northvale, NJ: Aronson. Horowitz, M. J., Siegel, B., Holen, A., Bonanno, G. A., Milbrath, C., & Stinson, C. H. (1997). Diagnostic criteria for complicated grief disorder. American Journal of Psychiatry, 154, 904 ± 910. Horowitz, M. J., Wilner, N., & Alvarez, M. A. (1979). Impact of event scale: a measure of subjective distress. Psychosomatic Medicine, 41, 209 ± 218. Humphrey, G. M., & Zimpfer, D. G. (1996). Counseling for grief and bereavement. Thousand Oaks, CA: Sage Publications. Irwin, M. R., Daniels, M., Smith, T. L., Bloom, E., & Weiner, H. (1987). Impaired natural killer cell activity during bereavement. Brain, Behavior, and Immunity, 1, 98 ± 104. Irwin, M. R., & Pike, J. (1993). Bereavement, depressive symptoms, and immune function. In: M. S. Stroebe, W. Stroebe, & R. O. Hansson (Eds.), Handbook of bereavement: theory, research, and intervention ( pp. 160 ± 171). Cambridge, England: Cambridge Univ. Press. Irwin, M. R., & Weiner, H. (1987). Depressive symptoms and immune function during bereavement. In: S. Zisook (Ed.), Biopsychosocial aspects of grief ( pp. 159 ± 174). Washington, DC: American Psychiatric Press.

732

G. A. Bonanno and S. Kaltman

Izard, C. (1977). Human emotions. New York: Plenum. Jacobs, S., Hansen, F., Berkman, L., Kasl, S., & Ostfeld, A. (1989). Depressions of bereavement. Comprehensive Psychiatry, 30, 218 ± 224. Jacobs, S., Hansen, F., Kasl, S., Ostfeld, A., Berkman, L., & Kim, K. (1990). Anxiety disorders during acute bereavement: risk and risk factors. Journal of Clinical Psychiatry, 51, 269 ± 274. Judd, L. L., Rapaport, M. H., Paulus, M. P., & Brown, J. L. (1994). Subsyndromal symptomatic depression (SSD): a new mood disorder? Journal of Clinical Psychiatry, 55, 18 ± 28. Kaltman, S., & Bonanno, G. A. (1999). Trauma and bereavement: examining the role of sudden and violent deaths (submitted for publication). Kaprio, J., Koskenvuo, M., & Rita, H. (1987). Mortality after bereavement: a prospective study of 95,647 bereaved persons. American Journal of Public Health, 77, 283 ± 287. Kavanagh, D. G. (1990). Towards a cognitive ± behavioral intervention for adult grief reactions. British Journal of Psychiatry, 157, 373 ± 383. Kelly, A. E., & McKillop, K. J. (1996). Consequences of revealing personal secrets. Psychological Bulletin, 120, 450 ± 465. Keltner, D., & Bonanno, G. A. (1997). A study of laughter and dissociation: distinct correlates of laughter and smiling during bereavement. Journal of Personality and Social Psychology, 73, 687 ± 702. Kim, K., & Jacobs, S. (1991). Pathologic grief and its relationship to other psychiatric disorders. Journal of Affective Disorders, 21, 257 ± 263. Kim, K., & Jacobs, S. (1993). Bereavement and neuroendocrine changes. In: M. S. Stroebe, W. Stroebe, & R. O. Hansson (Eds.), Handbook of bereavement: theory, research and intervention ( pp. 143 ± 159). Cambridge, England: Cambridge Univ. Press. Klass, D., Silverman P. R., & Nickman S. L. (Eds.) (1996). Continuing bonds: new understandings of grief. Washington, DC: Taylor & Francis. Lazare, A. (1989). Bereavement and unresolved grief. In: A. Lazare (Ed.), Outpatient psychiatry: diagnosis and treatment (2nd ed.,) pp. 381 ± 397). Baltimore, MD: Williams & Wilkins. Lazarus, R. S. (1991). Emotion and adaptation. New York: Oxford Univ. Press. Lazarus, R. S., Kanner, A. D., & Folkman, S. (1980). Emotions: a cognitive ± phenomenological analysis. In: R. Plutchik, & H. Kellerman (Eds.), Theories of emotion. Emotions: theory, research and experience vol. 1. pp. 189 ± 217). New York: Academic Press. Lehman, D. R., Wortman, C. B., & Williams, A. F. (1987). Long-term effects of losing a spouse or child in a motor vehicle crash. Journal of Personality and Social Psychology, 52, 218 ± 231. Lindemann, E. (1944). Symptomatology and management of acute grief. American Journal of Psychiatry, 101, 1141 ± 1148. Linn, M. W., Linn, B. S., & Jensen, J. (1984). Stressful events, dysphoric mood, and immune responsiveness. Psychological Reports, 54, 219 ± 222. Lundin, T. (1984). Long-term outcome of bereavement. British Journal of Psychiatry, 145, 428 ± 434. Maddison, D., & Viola, A. (1968). The health of widows in the year following bereavement. Journal of Psychosomatic Research, 12, 297 ± 306. Marris, P. (1958). Widows and their families. London: Routledge and Kegan Paul. Marwit, S. J. (1996). Reliability of diagnosis complicated grief: a preliminary investigation. Journal of Consulting and Clinical Psychology, 64, 563 ± 568. Mergenhagen, P. M., Lee, B. A., & Gove, W. R. (1985). Till death do us part: recent changes in the relationship between marital status and mortality. Sociology and Social Research, 70, 53 ± 56. Middleton, W., Burnett, P., Raphael, B., & Martinek, N. (1996). The bereavement response: a cluster analysis. British Journal of Psychiatry, 169, 167 ± 171. Newton, T. L., & Contrada, R. J. (1992). Repressive coping and verbal ± autonomic response dissociation: the influence of social context. Journal of Personality and Social Psychology, 62, 159 ± 167. Nolen-Hoesksema, S., McBride, A., & Larson, J. (1997). Rumination and psychological distress among bereaved partners. Journal of Personality and Social Psychology, 72, 855 ± 862. Nunnally, J. C. (1978). Psychometric theory (2nd ed.). New York: McGraw-Hill. Osterweis, M., Solomon F., & Green F. (Eds.) (1984). Bereavement: reactions, consequences, and care. Washington, DC: National Academy Press. Parkes, C. M. (1964). Effects of bereavement on physical and mental health: a study of the medical records of widows. British Medical Journal, 2, 274 ± 279. Parkes, C. M. (1965). Bereavement and mental illness. British Journal of Medical Psychology, 38, 1 ± 26. Parkes, C. M. (1970). The first year of bereavement: a longitudinal study of the reaction of London widows to the death of their husbands. Psychiatry, 33, 444 ± 467.

Varieties of Grief Experience

733

Parkes, C. M. (1972). Bereavement: studies of grief in adult life. New York: International Universities Press. Parkes, C. M., & Brown, R. J. (1972). Health after bereavement: a controlled study of young Boston widows and widowers. Psychosomatic Medicine, 34, 449 ± 461. Parkes, C. M., & Weiss, R. S. (1983). Recovery from bereavement. New York: Basic Books. Prigerson, H. G., Frank, E., Kasl, S. V., Reynolds III, C. F., Anderson, B., Zubenko, G. S., Houck, P. R., George, C. J., & Kupfer, D. J. (1995). Complicated grief and bereavement-related depression as distinct disorders: preliminary empirical validation in elderly bereaved spouses. American Journal of Psychiatry, 152, 22 ± 30. Rando, T. (1992). The increasing prevalence of complicated mourning: the onslaught is just beginning. Omega, 26, 43 ± 59. Rando, T. (1993). Treatment of complicated mourning. Champaign, IL: Research Press. Raphael, B. (1983). The anatomy of bereavement. New York: Basic Books. Rounsaville, B. J., & Chevron, E. (1982). Interpersonal psychotherapy: clinical applications. In: A. J. Rush (Ed.), Short-term psychotherapies for depression: behavioral, interpersonal, cognitive, and psychodynamic approaches ( pp. 107 ± 142). New York: Guilford. Rynearson, E. K. (1996). Psychotherapy of bereavement after homicide: be offensive. In Session: Psychotherapy in Practice, 2, 47 ± 58. Rynearson, E. K., & McCreery, J. M. (1993). Bereavement after homicide: a synergism of trauma and loss. American Journal of Psychiatry, 150, 258 ± 261. Sable, P. (1989). Attachment, anxiety, and loss of a husband. American Journal of Orthopsychiatry, 59, 550 ± 556. Sanders, C. M. (1979). The use of the MMPI in assessing bereavement outcome. In: C. S. Newmark (Ed.), MMPI: clinical and research trends ( pp. 223 ± 247). New York: Praeger. Sanders, C. M. (1993). Risk factors in bereavement outcome. In: M. S. Stroebe, W. Stroebe, & R. O. Hansson (Eds.), Handbook of bereavement: theory, research, and intervention ( pp. 255 ± 270). Cambridge, England: Cambridge Univ. Press. Schleifer, S. J., Keller, S. E., Camerino, M., Thornton, J. C., & Stein, M. (1983). Suppression of lymphocyte stimulation following bereavement. Journal of the American Medical Association, 250, 374 ± 377. Schneider, D. S., Sledge, P. A., Shuchter, S. R., & Zisook, S. (1996). Dating and remarriage over the first two years of widowhood. Annals of Clinical Psychiatry, 8, 51 ± 57. Schwartzberg, S. S., & Janoff-Bulman, R. (1994). Grief and the search for meaning: exploring the assumptive worlds of bereaved college students. Journal of Social and Clinical Psychology, 10, 270 ± 288. Shuchter, S. R., & Zisook, S. (1993). The course of normal grief. In: M. S. Stroebe, W. Stroebe, & R. O. Hansson (Eds.), Handbook of bereavement: theory, research, and intervention ( pp. 23 ± 43). Cambridge, England: Cambridge Univ. Press. Stein, N. L., Folkman, S., Trabasso, T., & Christopher-Richards, A. (1997). Appraisal and goal processes as predictors of well being in bereaved care-givers. Journal of Personality and Social Psychology, 72, 863 ± 871. Stiles, W. B., Shuster, P. L., & Harrigan, J. A. (1992). Disclosure and anxiety: a test of the fever model. Journal of Personality and Social Psychology, 63, 980 ± 988. Stroebe, M. S., & Stroebe, W. (1993). The mortality of bereavement: a review. In: M. S. Stroebe, W. Stroebe, & R. O. Hansson (Eds.), Handbook of bereavement ( pp. 175 ± 195). New York: Cambridge Univ. Press. Stroebe, M. S., Stroebe, W., Gergen, K. J., & Gergen, M. (1981). The broken heart: reality or myth? Omega, 12, 87 ± 105. Stroebe, W., & Stroebe, M. S. (1987). Bereavement and health. Cambridge, England: Cambridge Univ. Press. Stroebe, W., & Stroebe, M. S. (1993). Determinants of adjustment to bereavement in younger widows and widowers. In: M. S. Stroebe, W. Stroebe, & R. O. Hansson (Eds.), Handbook of bereavement ( pp. 208 ± 226). New York: Cambridge Univ. Press. Stroebe, W., Stroebe, M. S., Abakoumkin, G., & Schut, H. (1996). The role of loneliness and social support in adjustment to loss: a test of attachment versus stress theory. Journal of Personality and Social Psychology, 70, 1241 ± 1249. Thompson, L. W., Breckenridge, J. N., Gallagher, D., & Peterson, J. A. (1984). Effects of bereavement on self-perceptions of physical health in elderly widows and widowers. Journal of Gerontology, 39, 309 ± 314. Tudiver, F., Hilditch, J., & Permaul, J. A. (1991). A comparison of psychosocial characteristics of new widowers and married men. Family Medicine, 23, 501 ± 505. Weinberger, D. A., & Davidson, M. N. (1994). Styles of inhibiting emotional expression: distinguishing repressive coping from impression management. Journal of Personality, 62, 587 ± 613. Weinberger, D. A., Schwartz, G. E., & Davidson, J. R. (1979). Low-anxious and repressive coping styles:

734

G. A. Bonanno and S. Kaltman

psychometric patterns of behavioral and physiological responses to stress. Journal of Abnormal Psychology, 88, 369 ± 380. Weiss, (1975). Marital separations. New York: Basic Books. Worden, J. W. (1982). Grief counseling and grief therapy: a handbook for the mental health practitioner. New York: Springer. Worden, J. W. (1996). Tasks and mediators of mounting: a guide for the mental health practitioner. In Session: Psychotherapy in Practice, 2, 73 ± 80. Wortman, C. B., & Silver, R. C. (1989). The myth of coping with loss. Journal of Consulting and Clinical, 57, 349 ± 357. Young, M., Benjamin, B., & Wallis, C. (1963). Mortality of widowers. Lancet, 2, 454. Zisook, S., Chentsova-Dutton, Y., & Shuchter, S. R. (1998). PTSD following bereavement. Annals of Clinical Psychiatry, 10, 157 ± 163. Zisook, S., Devaul, R. A., & Click, M. A. (1982). Measuring symptoms of grief and bereavement. American Journal of Psychiatry, 139, 1590 ± 1593. Zisook, S., Paulus, M., Shuchter, S. R., & Judd, L. L. (1997). The many faces of depression following spousal bereavement. Journal of Affective Disorders, 45, 85 ± 95. Zisook, S., Schneider, D., & Shuchter, S. R. (1990). Anxiety and bereavement. Psychiatric Medicine, 8, 83 ± 96. Zisook, S., & Shuchter, S. R. (1986). The first four years of widowhood. Psychiatric Annals, 16, 288. Zisook, S., & Shuchter, S. R. (1991). Depression through the first year after the death of a spouse. American Journal of Psychiatry, 148, 1346 ± 1352. Zisook, S., & Shuchter, S. R. (1996). Psychotherapy for depression in spousal bereavement. In Session: Psychotherapy in Practice, 2, 31 ± 46. Zisook, S., Shuchter, S. R., Irwin, M., Darko, D. F., Sledge, P., & Resovshy, K. (1994). Bereavement, depression, and immune function. Psychiatric Research, 52, 1 ± 10.