Course of spousal bereavement in later life

The overall purpose of this chapter is to provide an overview of the major findings from our research over the past 11 years on spousal bereavement among older adults in the United States. We began in 1980 with a longitudinal study funded by the National Institute on Aging (NIA) designed to describe, from a multidisciplinary perspective, the process of adjustment that follows the death of a spouse, examine factors that influence the observed outcomes, and identify potential focuses and strategies for intervention. A sample of 192 recently bereaved spouses and a control group of 104 currently married adults over the age of 50 participated in the study. Findings from this project were used to develop a second study, also funded by NIA, to examine the effectiveness of selfhelp groups in facilitating the bereavement adjustment process. Another sample of 339 recently bereaved spouses participated in this intervention study, with 241 assigned to self-help groups and 98 assigned to a control condition.

The course of spousal bereavement in later life DALE A. LUND, MICHAEL S. CASERTA, AND MARGARET F. DIMOND The overall purpose of this chaptcr is to provide an overview of the major findings from our research over the past 11 years on spousal bereavement among older adults in the United States. We began in 1980 with a longitudinal study funded by the National Institute on Aging (NIA) designed to describe, from a multidisciplinary perspective, the process of adjustment that follows the death of a spouse, examine factors that influence the observed outcomes, and identify potential focuses and strategies for intervention. A sample of 192 recently bereaved spouses and a control group of 104 currently married adults over the age of 50 participated in the study. Findings from this project were used to develop a second study, also funded by NIA, to examine the effectiveness of self­help groups in facilitating the bereavement adjustment process. Another sample of 339 recently bereaved spouses participated in this intervention study, with 241 assigned to self-help groups and 98 assigned to a control condition. In the late 1970s, the National Institute on Aging recognized that little or no empirical research on bereavement had been completed that specifically focused on older adults. Although some studies had included older adults in their samples, there was no systematic attempt to learn more about their bereavement experiences until NIA established bereavement and aging as a priority for research funding. Our first study was one of three that the institute initially supported. Each project had its own unique focus, ques­tions, and measures, but they were similar in their purpose and prospective longitudinal designs. Larry Thompson, Dolores Gallagher, and their colleagues (1989) began their study in Los Angeles in 1979; we followed in Salt Lake City in 1980; and Martin Faletti, Jeanne Gibbs, and colleagues began in Miami in 1981. Although the untimely deaths of Gibbs and Faletti limited the analysis and dissemination of their data, these three studies, and The studies described in this chapter were funded by the National Institute on Aging (R01 AG02193 and R01 AG06244). ‘ ‘ 240 others that followed, have helped to fill the void in our knowledge about late life spousal bereavement. Selected findings from these three studies and six other research projects in the United States have been compiled in the book Older Bereaved Spouses: Research with Practical Applications (Lund, 1989a). These studies were important because they not only provided infor­mation specific to the bereavement experiences of older adults but used improved research designs and more representative samples. For example, these studies were among the first to use prospective longitudinal designs with early assessments and nonbereaved control groups. Bereaved spouses in these studies were interviewed between four and six times over a period of 2 to 2.5 years. In the study by Faletti and his associates (1989), participants were interviewed as early as 14 days following the death. The use of nonbereaved controls allowed estimations of how much impact bereavement had on the well-being of the survivors in the absence of making the ideal pre- and postbereavement comparisons for each person. Also, the community-based sampling procedures provided for the inclusion of both men and women in the same sample and allowed for greater heterogeneity. Although each study had its own sampling limitations, collectively they covered a relatively broad cross section of bereaved elders in the United States. The Miami sample included a sizable portion ofjews (57%), the Salt Lake City sample was unique in the high percentage of Mormons (72%), and the Los Angeles sample had considerable religious and ethnic diversity. We were particularly pleased that the Salt Lake City study was similar to and conducted concurrently with the other two projects because we were concerned about the generalizability of our findings. Although we did not find Mormons to differ from others in our study sample (Lund, Caserta, & Dimond, 1988), it is quite valuable to know that most of our findings are strikingly similar to those from Miami and Los Angeles, as well as from subsequent studies in Riverside, California (Schuster & Butler, 1989), rural Nebraska (Van Zandt, Mou, & Abbot, 1989), several midwestern cities (Gass, 1989), and Cleveland, Ohio (Kitson & Roach, 1989). Understanding this broader context within which our research has occurred is critical to the value and usefulness of our findings. Although our samples were unique in the proportion of those who were of the Mormon religion, we have concluded that older Mormons experience spousal bereavement in much the same manner as older adults across the United States (Lund, 1989b). Their difficulties, needs, and likelihood of making successful adjustments are similar to others who are adjusting to the death of a spouse after many years of marriage. The highlights of our research findings presented here should be interpreted in the context of this broader base of recent research. Spousal bereavement in la te r life 241 Methods Because we summarize findings from two separate studies, it is necessary to provide a brief description of the sampling and data collection procedures for each project. The following descriptions contain only the most essential features because detailed accounts already have been published elsewhere. For example, the most complete description of the methodology used in our first longitudinal study (Lund, Caserta, & Dimond, 1989a) and the second self-help group project (Lund, Caserta, Dimond, & Shaffer, 1989c; Lund, Red burn, Juretich, & Caserta, 1989d) can be found in Older Bereaved Spouses (Lund, 1989a). Many of the same measures were used in both studies, so we combined the brief descriptions of them into a single section. 2 4 2 D. A. LUND, M. S. CASERTA, M. DIMOND Longitudinal descriptive study s Recently bereaved spouses aged 50 and over were identified through the use of local newspaper obituaries in order to complete the first interview or questionnaire as early as three weeks following the spouse's death. Official mortality data obtained laler from the Utah State Department of Health Statistics revealed that this procedure missed only 9% of the actual deaths for those in the same age category. No significant gender, age, or socio­economic differences were found between those who had a published obituary and those who did not. All potential bereaved participants were randomly assigned to either a home interview group (N = 104) or a mailed questionnaire group (N = 88) in order to test for an interview effect. No major interviewer effect was observed (Caserta, Lund, & Dimond, 1985), so the two samples were combined into one sample for most of the statistical analyses. A total of 192 bereaved people participated in the study. Nonbereaved older adults were identified through the use of public voter registry data and were selected on the basis of sex, age, and socioeconomic area of residence. In order to reduce the number of matching procedures, a matched nonbereaved person was selected only for each of the 104 bereaved respondents in the interview group. The refusal rate for this sample was 50 percent and the major reasons for refusal were busyness and lack of interest. The first two questionnaires for each of the nonbereaved partici­pants were delivered by a research assistant according to the same pro­cedures as those used for the mailed questionnaire group of bereaved participants. In both samples, the respondents completed them without the assistance of an interviewer and returned them by mail. All of the 192 bereaved persons were asked to complete questionnaires atsix times during the first 2 years of bereavement: 3 to 4 weeks, 2 months, 6 months, 1 year, 18 months, and 2 years after the death. With the exception of the demographic variables, all six questionnaires were essentially the same. The 104 nonbereaved controls completed similar questionnaires at the same six times. The mean age for the sample of bereaved elders at the beginning of the study was 67.6 years (SD = 8.2), with a range of 50 to 93 years, and they had been married an average of 39 years. The majority were female (74%), white (97%), had graduated from high school (78%), and were Mormon (73%). The mean age of the nonbereaved participants was 66.3 years (SD = 7.8), with a range of 50 to 88 years, and they had been married an average of 38 years. The majority were female (73%), white (98%), had graduated from high school (77%), and were Mormon (70%). Spousal bereavement in la te r life 243 Self-help group study This study was designed to investigate the effectiveness of self-help groups in facilitating bereavement adjustments and more specifically the roles of group leadership and duration by comparing self-help groups led by widows versus professionals and short- versus long-term formats. All groups met weekly for 2 months, but the long-term groups continued to meet once a month for an additional 10 months. Because we are still in the process of analyzing the data regarding the effects of the intervention, we summarize in this chapter the findings about predictors of adjustment. Recently bereaved spouses over the age of 50 were identified through newspaper obituaries and were randomly assigned into one of the four intervention condition groups or the control group. Questionnaires were hand-delivered 2 months after the death event to those who agreed to participate. Three hundred and thirty-nine (241 assigned to intervention condition groups and 98 to the control group) returned the first question­naires. The second measures were obtained immediately after the com­pletion of the eight weekly self-help meetings (roughly 4-8 months' bereaved) and the third questionnaires followed the completion of the long­term groups (roughly 14-17 months' bereaved). The fourth and final questionnaire was obtained at 2 years following the death. The control group also was assessed at the same four time periods but received no intervention. The sample consisted of 242 women (71.4%) and 97 men who ranged in age from 50 to 89 years (M = 67.5, SD = 8.5), were fairly well educated (85% high school graduates), were primarily Caucasian (98.8%), and were generally not employed outside the home (76.3%).244 D. A. LUND, M. S. CASERTA, M. D1MOND Measurement instruments used in the two studies In addition to the typical sociodemographic characteristics of age, gender, education, number of years married, employment status, religious member­ship, race, and income, we included several already existing standardized scales, several inventories and single-item indicators that we developed specifically for the studies, and open-ended questions to generate qualitative data about bereavement adjustments. Again, we only present a brief review of these instruments because more detailed descriptions arc available elsewhere. The major measures that we refer to as the indicators of bereavement adjustments, outcomes, and well-being included a standarized scale of life satisfaction (Neugarten, Havighurst, & Tobin, 1961), two depression scales (Yesavage et al., 1983; Zung, 1965), and the Revised Texas Inventory of Grief (Faschingbauer, 1981). Also, we developed an inventory of bereavement - related feelings and behaviors with subscales measuring emotional shock, helplessness/avoidance, psychological strength, anger/ guilt/confusion, and grief resolution behaviors (Caserta et al., 1985) - and used several single-item measures of health, coping, stress, and loneliness (Caserta, Lund, & Dimond, 1990; Johnson, Lund, & Dimond, 1986). Numerous open-ended questions were asked concerning their greatest dif­ficulties, advice to others, prospects about remarriage, experiences with support networks, health functioning, and social activities. Some measures were used as both predictor and outcome variables, depending on the research questions and appropriate analyses. The fol­lowing measures, however, primarily served as predictor variables. These included the Twenty Statements Test, which assesses various dimensions of a person's self-concept (Kuhn & McPartland, 1954; Spitzer, Couch, & Stratton, 1971); Rosenberg's self-esteem scale (1965); a single-item rating of self-esteem (Johnson et al., 1986); a scale of social support that we developed to assess both structural and qualitative dimensions (Dimond, Lund, & Caserta, 1987; Lund, Caserta, Van Pelt, & Gass, 1990); a detailed skill survey of 94 tasks of daily living; and a competencies scale we adapted to assess social, instrumental, and resource identification skills (Lund et al., 1989c). , _ . Results Highlights of our findings from these two research projects, with an emphasis on the first longitudinal study, are presented in two parts. First are the findings that describe what we believe are the most essential features about the course of spousal bereavement over time. This infor­mation helps to describe the process of adjustment. Second is a summary ofSpousal bereavement in la te r life 245 what we have learned about the factors that influence the course of adjust­ment. We learned that some variables were far more important than we expected in predicting bereavement outcomes, and conversely, some factors were less influential than anticipated. Also, we believe that it is as im­portant to know what was not significant as it is to know what was significant. This is particularly important when we expect a particular finding and learn that the data do not support the expectation. Descriptive findings First, our data indicate that there was considerable diversity in the adjust­ment process both among and within bereaved spouses. Nearly every outcome measure that we used revealed a broad range of responses at each time period among the research participants. Some individuals appeared to manage the stressful situation very well even experiencing personal growth and learning new skills (Lund et al., 1989c). Others, however, were quite devastated by their loss and had great difficulty managing their personal lives for several years (Lund et al., 1985). Some described themselves as being socially active, independent, helpful to others, involved with hobbies, and motivated to make the best of a very difficult situation, whereas others reported being despondent, angry, miserable, and sick of living (Lund, Caserta, Dimond, & Gray, 1986b). This diversity in bereavement reactions also was found within the indi­vidual. It was common to find a person simultaneously experiencing a full range of feelings and behaviors. For example, it was not unusual for the bereaved to feel angry, guilty, and lonely, yet at the same time feel personal strength and pride in how he or she was coping (Lund, Caserta, & Dimond, 1986a). A 70-year-old woman described herself at length as being busy, enjoying many different activities, and doing things with other people, but her final self-descriptive comment was that she still feels lonely. Second, our findings are clear in identifying loneliness as the single greatest difficulty for older bereaved spouses (Lund, 1989b). Although other researchers have reported on the problem of loneliness (Barrett & Schneweis, 1980; Lopata, 1973a; Parkes, 1972a/1987; Carey, 1979), we learned that in the opinions of the bereaved spouses, it was the most common difficulty and it persisted through the first 2 years of adjustment. Unfortunately, loneliness can be very painful and difficult to resolve. Loneliness cannot be managed simply by surrounding oneself with others, as indicated by the statement of a 69-year-old woman who said, "I am lonely but not alone." In addition to the pervasive problem of loneliness is the lack of skill among older bereaved spouses to deal effectively with a relatively unique set of problems of daily living. Older men were found to be deficient in apredictable set of skills, including cooking, shopping, and housecleaning, and older women lacked skills in doing home repairs and managing legal affairs and financial records (Lund et al., 1986a). These deficiencies were sources of difficulty because the tasks were performed primarily by the deceased spouses. Nearly 73% indicated that these deficiencies had made their coping more difficult. Similarly, those who had learned some of these new skills during bereavement reported feeling better about themselves, more independent, and better able to get along with others, and over 30% said that they benefited directly by being better able to deal with their grief (Lund et al., 1989c). Other common problems of bereavement included difficulties with depression (Lund et al., 1986a), family relationships and friendships (Lund et al., 1990), emotions of grief (Lund et al., 1985), physical health (Caserta et al., 1990), personal identity (Lund et al., 1986b), and living according to one's own expectations versus those of others. Although we have not yet reported on the numerous examples of older bereaved spouses who struggled with living their lives according to the expectations of others, this is a problem of considerable importance. This problem usually manifests itself during decision-making times wherein the bereaved spouse must choose between what they want to do and what they believe others would want them to do, or even what they imagine their deceased spouse would expect. We found this dilemma in a variety of situations, including decisions about forming new relationships, dating, sexual practices, traveling, making major purchases, selling homes and personal belongings, and establishing new patterns of social activities. We discovered that some bereaved spouses agonized over the decision to keep or give away a pet that they had never liked. They felt anxious and guilty for wanting to get rid of a dog or cat that was so important to their deceased spouse. They were troubled over what others might think about them, and in some instances this meant that their deceased spouse would be so upset that they decided to keep the pet (Lund, Johnson, Baraki, & Dimond, 1984). Without adding more detail, we believe that it is essential to report that, for many, there is great difficulty in making bereavement adjustments while trying to live according to the expectations of others, including the expectations of the deceased. A third general conclusion about the bereavement process of older spouses is the high degree of resiliency, resourcefulness, and adaptability that they exhibited. Although 72% of our research participants reported the death of their spouses to be the most stressful event they had ever experi­enced, and the self-reported levels of stress were quite high for the entire sample, we found that there were numerous indicators of personal strength, social support, and resourcefulness that helped them cope quite well with the many difficulties they faced. We estimated that 18% of our sample was 246 D. A. LUND, M. S. CASERTA, M. DIMOND having major difficulties 2 years after the death (Lund et al., 1985). The positive aspect of this finding is that about 82% of the bereaved spouses were managing with a good degree of success. Life satisfaction, perceived health, self-esteem, and social support remained relatively high (Lund et al., 1989a; Lund et al., 1985; Lund et al., 1990). Also, our findings that Wortman and Silver (1989) used in their publication on the myths of coping with loss were consistent with data from other studies in confirming that resiliency and even immunity are more common patterns of adjust­ment than previously reported. In other words, many bereaved persons were resourceful enough to find effective ways of managing their grief and making satisfying adjustments. Further documenting this personal resourcefulness were the findings that only 12% of the participants in our second study reported seeking professional help for bereavement difficulties (Caserta & Lund, 1992) and about 72% had learned new skills during the first 2 years of bereavement (Lund et al., 1989c). Although bereavement after many years of marriage is a highly stressful situation, we also need to acknowledge the considerable capability of older adults in meeting many of the demands of a markedly changed life-style. Fourth, because we could not find support for stages of bereavement adjustment, we believe it is more accurate to describe the process like a roller coaster of many ups and downs with gradual improvement over time. We were unable to identify specific times, events, or markers that would parallel stages of adjustment (Lund et al., 1986a). In support of the notion of gradual improvement, our data consistently revealed that the early months following the loss were the most difficult (Lund et al., 1989a, 1989b) and that these early indicators were relatively good predictors of longer term adjustment (Lund et al., 1985; Johnson, Lund, & Dimond, 1986). In both our studies most of the outcome measures showed gradual improvement over the 2-year period. An additional brief telephone survey at 4 to 5 years after the death confirmed that the bereaved spouses in our first study were still improving in terms of their self-reports of stress, coping, and health (Burks, Lund, Gregg, & Bluhm, 1988). With 2- to 6-month intervals between interviews it was difficult to observe the many ups and downs of the bereavement process, particularly with the use of standardized quantitative scales. Fortunately, we included questions that allowed the respondents opportunities to describe in their own words their feelings, difficulties, successes, and advice to others. Transcripts from these parts of the interviews revealed the transitory nature of the process. Typical is a comment by a 70-year-old male who said, "It comes on gradual and reaches a peak until you break down and cry. You get a little better for awhile and it goes up again." Similarly, a 68-year-old woman replied, "These periods of grief come and go. It might strike at any time and it might leave at any time. But they come and go." The ups and S pousal bereavement in la te r life 247 downs of the roller coaster can easily be precipitated by mistakenly seeing the face of the deceased in a crowd, hearing an old familiar song on the radio, driving by the hospital where the spouse was treated, or conversely, feeling proud for successfully balancing a checkbook for the first time, meeting a new friend at a club, or simply mastering the use of a microwave oven. The roller-coaster metaphor is far from ideal to describe the complexities of the bereavement process, but it does help to emphasize both the ups and downs and gradual improvement that we have found in our research. For some, the ride may never end, but they learn to "live with it" or "adjust to it." Others are more active in meeting the challenges and find ways to manage the ups and downs of the process. The next set of findings examines the factors that influence the degree of success in managing the unfavorable consequences and producing more favorable outcomes. , 248 D. A. LUND, M. S. CASERTA, M. DIMOND Findings about the predictors of bereavement adjustments ' ' Two of our findings that describe features about the process of bereavement are particularly relevant to conducting research on the predictors of adjust­ment. Because of the many relatively spontaneous ups and downs during the course of bereavement, it is imperative to use longitudinal designs. Using repeated measures over time enables more accurate assessments, consideration of the transitory nature of bereavement, and specification of cause-and-eff'ect relationships. Also, bceausc of the great diversity in how bereavement can affect each individual and the variability among different people, it is critical that investigators include multiple indicators of adjust­ment. Using only one or two mesures will not reflect the complexity or multidimensional effects of bereavement. As we described in the Methods section of this chapter, we used an array of quantitative and qualitative indicators of physical, emotional, psycho­logical, and social well-being. Although some of our findings about pre­dictors of adjustment were specific to only certain indicators, the following summary avoids some of the lesser details in order to highlight the more important and practical findings. For those who wish to examine more closely some of the specifies, we have included citations for further reading. Our research findings could be characterized as much by what wc have found not to be important as what we found to be predictive of bereavement adjustments. In other words, we have been surprised by the lack of significant effects for some variables we expected to influence the course of adjustment. We have, therefore, organized this section of findings into three parts. First is a summary of the factors we found to be relatively unimportant as predictors of adjustment, followed by variables thatwere moderately influential, and then wc review and discuss the five best predictors. Relatively unimportant predictors. We begin with a cautious reminder that although the factors presented in this category were found to have little or no direct influence on the course of bereavement adjustments, we do not recommend that they be dismissed from future consideration for research or practice. Our samples included only bereaved spouses and those age 50 years and over. We suspect that in other bereavement contexts they might be more influential than what we found. In this category of nonpredictive variables were nearly all of the socio­demographic characteristics, including age, gender, education, income, religious membership and religiosity, and other factors such as perceived health status, marital happiness, and pet ownership. Because our samples were relatively homogeneous in terms of race and ethnicity, we are unable to address the degree of influence they might have on the course of bereavement. Participants in our studies ranged in age from 50 to 93 years. Although we found those over the age of 75 years to be at the greatest risk for experiencing declines in the size and perceived closeness of their social support networks (Lund et al., 1990), age did not predict any of the major indicators of adjustment (Lund et al., 1986a; Dimond et al., 1987). Surprisingly, years of education, income, and income adequacy did not exert a clear influence on bereavement. The only exception was that women, in the higher income categories showed slightly better early adjustments (Caserta, Lund, & Dimond, 1989). The number of years of formal education may not be the best way to measure the broad range of potential benefits from education because it ignores the specifics of curriculum, effort, per­formance, and other issues of quality and practicality. The measures of income and income adequacy are similarly limited because they do not account for how they were used and whether or not they were shared with others. We carefully examined the influence of both gender and several aspects of religion and concluded that they were not very predictivc of our global indicators of adjustment. With respect to gender, we learned that older males and females differed in their lack of specific skills of daily living, but they were similar in their emotional, psychological, social, and health adjustments (Lund et al., 1986a). Also, contrary to our expectations, males were as likely as females to participate in both of our studies - even the second project, which required participation in self-help group meetings (Lund et al., 1989b; Lund et al., 1989d). Our focus on religion was necessitated by the high proportion of Mormons who lived in our study area. Our samples included about 70% to Spousal bereavement in la te r life 249 73% who were Mormon. Without elaborating on all of the unique features of Mormonism, it was possible that their large family size, emphasis on family life (Erickscn, 1975), and belief in an afterlife (Backman, 1970) could affect the course of bereavement. As expected, the Mormons in our samples were found to have more people in their support networks and more frequent contact with them, were more active in their religion, and more religious (Lund et al., 1988). These features, however, did not result in any major differences in the well-being of the study participants during the 2 years of bereavement. Religious activity did have a positive association with more favorable outcomes, but there was evidence in other analyses that led us to conclude that it was not necessarily the religious nature of the activity but the fact that social activity in general has a positive influence on bereavement adjustments. The remaining three variables that we found not to be predictive of adjustment were pet ownership, marital happiness, and health status. Wc included pet ownership as a study variable after learning about a variety of benefits from animal-human interactions (Fogle, 1981). Becausc some pets were actually owned more by the deceased spouse rather than the surviving spouse and pets require carc during periods of time when the bereaved were already occupied by legal, emotional, and social demands on their time, pets were not very helpful during the first year of bereavement (Lund et al., 1984). Also unexpected was the finding that the degree of reported marital happiness with the deceased was unrelated to the course of adjustment. One could argue that the loss of a happy marital relationship would be a greater loss than an unhappy relationship or, conversely, that a long happy relationship would provide a valuable set of memories that could sustain one during the difficult times of bereavement. The complexities of the marital relationship merit closer examination in future studies, but our analyses revealed that the degree of happiness in the marriage did not affect the coursc of bereavement. Some bereaved respondents reported that they missed the arguments with their deceased spouses even though they rated their marriages as being unhappy. We examined self-reported health status as both an independent and dependent variable and found considerable stability in the measure over time (Lund et al., 1989a) and that it did not have much effect on overall coping and adjustment during bereavement (Lund et al., 1985). We should caution, however, that the bereavement context influences the factors that older adults take into account in making their perceived health ratings and that these self-reports should be understood within the specific context. Moderately important predictors. All three of the factors we found to be moderately related to the well-being of the older adults during bereavement 250 D. A. LUND, M. S. CASERTA, M. DIMOND underscore the importance of social relationships with others. Positive associations revealed advantages to those who remarried, had supportive relationships with others, and were active in their religious participation. Self-reports of religious activity remained quite stable through the first 2 years of bereavement for both Mormons and non-Mormons (Lund et al., 1988). Although the statistical associations of religious activity on grief resolution behaviors were modest, we found additional support for the benefits of keeping busy during bereavement when we analyzed more of the qualitative data through content analyses. Many of the respondents recommended keeping busy and socially active when asked what specific advice they would give to others in similar circumstances. Those who gave this advice showed greater gains in their perceived coping ability (Caserta, Van Pelt, & Lund, 1989). Only 10% of the respondents in our first study had remarried within 4 to 5 years after the death of their spouses (80% men). We concluded that those who eventually remarried were more positive in the early interviews about their consideration of remarriage and likely viewed remarriage as a way to help them adjust to the loss and maintain their well-being (Burks et al., 1988). We would certainly not suggest remarriage as a coping strategy for those who express considerable reluctance or opposition, but we did find the remarried persons to have greater reduction in their stress levels and greater improvement in the measures of life satisfaction and resolution of grief. We cxpected features of the social support networks to be among the strongest predictors of adjustment and were surprised to learn that only a few characteristics had a moderate impact. Generally, social supports were quite positive and stable over time, with slight increases in the reliance on friends as opposed to relatives and decreases in the ease of contacting persons in the support network (Lund et al., 1990). When we closely examined the role of social support on bereavement outcomes, we found that stable support was a buffer of stress 2 years later (Duran, Turner, & Lund, 1989). More important, however, were the qualitative aspects of a support network (e.g., perceived closeness, self-expression, contact, shared confidence, and mutual helping) in leading to lower depression and more positive ratings of coping, health, and life satisfaction (Dimond et al., 1987). These findings about the benefits of social support networks need to be tempered with numerous qualitative examples about negative effects of support persons. Our interviewers heard detailed accounts of disappoint­ment, frustration, anger, and sadness expressed by the bereaved because some friends and family members were judgmental, avoidant, inconsiderate, pushy, and demanding. A common complaint of the bereaved was that they wished they could tell others what they had expressed to the interviewers. Spousal bereavement in la te r life 251 We concluded that there is considerable room for improvement in the way support persons can be of help to the bereaved. The best predictors. Contrary to the moderate predictors of bereave­ment adjustment primarily consisting of factors related to social relation­ships and support rcccivcd from others, the strongest predictors were personal resources unique to each person. Experiencing the death of a spouse in later life requires a wide range of adjustments to meet the demands of a radically different social environment and lifestyle. We have seen repeated evidence that supports the need for bereaved adults to take charge of their life-styles and situations if they are to preserve and some­times enhance their well-being. Taking control requires motivation, pride, skill, flexibility, some help from others, and the passage of time. The five best predictors of spousal bereavement that have emerged from our studies emphasize these requirements. First, as we discussed earlier, a common pattern of adjustment is one with many spontaneous ups and downs but with subtle and gradual im­provement. When we entered amount of time since the death event as a variable in a variety of statistical procedures, it was consistently one of the most influential factors. W7ith only few exceptions, the bereaved spouses had more positive well-being scores at each subsequent data collection period. This gradual improvement was evident in both of our studies (Lund et al., 1986a, 1989a, 1989b). We do not believe that simply the passage of time by itself brings about succcssful adjustments. Time is important because the many difficulties require time to emerge, be identified, develop strategies, and achieve some degree of successful adjustment or resolution. A passive strategy of waiting for time to heal a wound will not lead to satisfactory bereavement adjustments. Second, also related to time, is the finding that initial or early bereavement adjustments were good predictors of longer term well-being. Early success or extreme difficulty appears to set the direction for future growth and devel­opment or an unhappy and dissatisfying life-style. In one of our most important publications we reported that difficulties in the first month of bereavement as evidenced by intense negative emotions (e.g., desire to die, frequent crying) were associated with poor coping after 2 years (Lund et al., 1985). These findings and others from our research document the possibility and value of identifying early in the course of bereavement those who are most likely to have long-term difficulties and need greater assistance from others. Our limited resources should be appropriately targeted to those who are at greatest risk and implemented early. In simple terms, good use of time and getting off to a good start will lead to more satisfying adjustments, but more is required to ensure positive well­being. Another good predictor of adjustment was a variable that deals with 252 D. A. LUND, M. S. CASERTA, M. DIMOND both a personal trait and the existence of a relationship with another person(s). Bereaved individuals who were communicative with others about , their thoughts and feelings were more likely to have positive adjustments later (Lund ct al., 1985). Again, self-expression that occurs early in the process was most advantageous. The willingness and ability to express oneself about very sensitive, emotional, and personal issues is not equally shared. Also, not all bereaved persons have available to them a confidant who will actually listen to these expressions. The final two predictors had the strongest impact on the course of adjustment and affected nearly every aspect of well-being that we measured. Positive self-esteem and personal competencies in managing the tasks of daily life resulted in the most favorable long-term bereavement outcomes (Lund et al., 1989c; Lund, 1989b). We believe that those who had high regard for themselves were more likely to be dissatisfied with not coping well, feel that they deserve better, be motivated and skilled to take control of the situation, and persist until they have more favorable outcomes. Conversely, those who had little regard for themselves were more likely to believe that they deserved feeling depressed and being overwhelmed by their grief. With low self-esteem there is little motivation, confidence, and skill to change the circumstanccs. During the course of the interviews in our research projects, many of the bereaved made statements about the need "to find your own way through the mess," and that "others cannot do it for you." These statements, although somewhat simplistic, are meaningful because they call attention to the importance of not being passive but taking charge and doing something active about the situation. Self-esteem provides part of the foundation by motivating action necessary for more successful and satisfying adjustments. Positive self-esteem coupled with the ability to use one's personal resources and skills can help bring about successful adjustments to the demands of the new environment. Particularly for those who have been married for many years and were socialized into accepting very narrow gender roles, spousal bereavement clearly revealed the lack of specific skills. As we reported earlier, the lack of skills in managing one's daily life further complicates the bereavement process. We discovered that being competent in social, interpersonal, instrumental, and resource identification skills lead' to more favorable bereavement adjustments (Lund et al., 1989c). Competencies and self-esteem are also interrelated and have mutually complementary effects. We found those who had high levels of competency had similar high levels of self-esteem and, conversely, those with low self­esteem had low levels of competency. In terms of the course of bereavement we suggest that one of the best ways to enhance self-esteem is to learn new v skills. Fifty-eight percent of our respondents reported feeling better about themselves as a result of learning new skills (Lund et al., 1989c). With Spousal bereavement in laler life 2 5 3 improved self-images the bereaved were often further motivated to work at making other adjustments. We highly recommend that the reciprocal rela­tionship between self-esteem and personal competencies be acknowledged and integrated into intervention efforts. After 11 years of research, perhaps our most important findings are those regarding the critical roles of internal coping resources, self-esteem and competencies, in making successful adjustments to the death of a spouse in later life. Future research is needed to delineate more clearly the cause-and- elfect relationships among these internal coping resources and bereavement outcomes. We highly recommend intervention studies with longitudinal designs because they offer the most appropriate tests of these relationships. 2 5 4 D. A. LUND, M. S. CASERTA, M. DIMOND