EFFECTS OF SUPPORT, COUNSELLING AND THERAPY BEFORE AND AFTER THE LOSS: CAN WE REALLY HELP BEREAVED PEOPLE?

————— Henk Schut and Margaret Stroebe are affiliated with the Department of Clinical and Health Psychology, Utrecht University. Correspondence concerning this article should be addressed to Henk Schut, Utrecht University, Dept. Clinical and Health Psychology, P.O.Box 80140, 3508 TC Utrecht, The Netherlands. Email: h.schut@uu.nl EFFECTS OF SUPPORT, COUNSELLING AND THERAPY BEFORE AND AFTER THE LOSS: CAN WE REALLY HELP BEREAVED PEOPLE?


Introduction
The death of a loved one can occur in a peaceful, timely manner with ample opportunity to say goodbye, or it can follow a violent, untimely death that comes without any forewarning or possibility for preparation. Whatever the circumstances, the loss of a loved one is associated with intense suffering and can lead to serious mental and physical health problems (Stroebe, Schut, & Stroebe, 2007). While some emotional reactions can hardly be avoided following the loss of a loved one (e.g., most people feel intense sadness and distress), a key question is how others can protect the bereaved from unduly long-lasting and/or extreme consequences. Is there scientific evidence that intervention is really beneficial to the bereaved?
When death occurs expectedly, as described above, support from the informal network and volunteers or professional counsellors/therapists can potentially be provided for family members facing the incumbent loss of their loved one; when it is unexpected, this is naturally not possible, intervention can in such cases only take place in the post-loss period. In the available scientific literature, considerable attention has been paid to the efficacy of postloss intervention: empirical studies and reviews in the bereavement research literature have focused on this period. Yet, a growing number of researchers have also examined the efficacy of providing support earlier in time, namely, to family members in palliative or end-of-life-care settings. To our knowledge, and somewhat surprisingly, the link specifically between post-and pre-bereavement intervention efficacy has not yet been made in the research literature. In our view, there is much to be gained by examining the results of these two growing bodies of evidence alongside each other, not least because it should ultimately provide a more complete picture of the benefits of care from others to persons suffering the (impending) loss of a loved one. For example, it is possible that effective pre-loss intervention may lower the need for post-loss intervention. Thus, a major interest here is to examine whether the patterns found for the efficacy of post-loss intervention are paralleled by those for the pre-loss provision of help.
To summarise, a number of questions are central. We revisit the questions that have been the focus of bereavement research in this area: Does support from family and friends ameliorate the impact of loss? Does voluntary or professional help reduce the impact of bereavement? Following the arguments presented above, we add a third question that has typically been omitted in reviews of the efficacy of bereavement intervention: Is intervention effective for family members within the context of palliative and end-oflife care? These three questions form the underlying theme of this article, in which we present key empirical findings and raise some challenging issues for researchers and clinicians alike.
Before addressing these questions, we need to be clear about precisely what types of help we cover in discussing benefits for bereaved persons during the post-and pre-loss periods. We look at psychological support from both informal as well as organised or institutional sources, that is, at help from others that aims to ameliorate grief. When considering volunteer/professional sources of intervention, we cover psychological aid, counselling and therapy. We exclude those sources that have not yet been subject to sufficient empirical testing, such as pastoral care and support being offered by funeral directors and general practitioners.
Furthermore, it is important to note that we are not providing a comprehensive review, that is beyond the scope of this article. Rather, we have selected what we consider to be well-designed, more-conclusive studies as illustrations for the patterns of results reported in the scientific literature. In particular, such studies typically included pre-and post measurement (preferably with follow-up data collection). Studies had an experimental as well as a non-intervention control condition, with careful assignment to groups (Schut, Stroebe, van de Bout, & Terheggen, 2001).
In the next section we concentrate on the efficacy of post-loss intervention. First, we address the question whether social support is helpful to the bereaved. Following that, we examine the role and impact of counselling and therapy in bereavement. We then go on to consider pre-loss efficacy research, examining the benefits of intervention in the terminal phase of the loved one's life. Finally, we discuss broader implications of these findings.

Efficacy of support: post-loss studies
Does social support help?
There are two ways in which help from friends and family can facilitate adjustment during bereavement. First, it can protect individuals against the impact of a major stressor. This so-called buffering effect (Cohen & Wills, 1985) states that the availability of social support protects individuals to some extent from the deleterious effects of stressful life events. Considering that bereavement is a stressful situation, this implies that social support is more helpful to bereaved persons than to people who have not suffered a loss. Second, social support can facilitate life in general, regardless of whether or not one is confronted with stressful situations. The latter is also known as the main effect of social support on health. If this is correct, social support is as helpful to the bereaved as it is to the non-bereaved, in other words, it makes life easier in general.
The above two possibilities were first put to the test in a study by Stroebe, Stroebe, Abakoumkin, and Schut in 1996, in which the impact of social support on depression was examined in a sample of widowed men and women compared with married counterparts. Results showed that there was no evidence of a buffering effect: social support helped widowed as much as it did married persons. Those with more support from persons around them reported less depression than those with lower levels of support, regardless of marital status. This seems to suggest that, although social support in general is helpful, the loss of a loved one leads to deficits in support that cannot be compensated by others, providing no evidence for a buffering effect. Rather, these findings may be more consistent with theories that state that such compensation is not possible. Most notably, they seem to be in line with attachment theory claims. Bowlby (1973) argued that an attachment figure is uniquely able to foster general feelings of security and that others cannot simply take over this function. More specifically, Weiss (1973) differentiated social from emotional loneliness. Social loneliness refers to the feeling that there is nobody to count on for support, there is absence of an engaging social network and lack of social embeddedness. Emotional loneliness denotes a sense of utter aloneness and isolation, whether or not others are accessible. Weiss reasoned that the loss of an attachment figure results in emotional loneliness and that social support from friends cannot reduce this type of loneliness.
Further analyses of the Stroebe et al. (1996) data confirmed this. Levels of social loneliness were similar for married and for widowed, but were higher for those low on social support in both marital categories. By contrast, the pattern for emotional loneliness showed that widowed persons had significantly higher levels of this type of loneliness, regardless of whether they were receiving high or low social support. In other words, the widowed were extremely emotionally lonely, compared with the married, and it did not seem that social support from others had an impact on this. Further support for this comes from a study by Guiaux (in progress), which included a large number of participants, had data collected both before and after the loss, and used sophisticated techniques of data analysis. Results showed that help from friends and family positively affected social loneliness, but did not affect emotional loneliness. Together, these studies suggest that emotional loneliness is a core element of grief, and that it is precisely this that cannot be reduced by friends and family. As much as one may want to help, this illustrates the limitations inherent in our efforts to try to help the bereaved.
These results seem disappointing with respect to receiving support. But is it possible that giving support to others is successful in predicting stress-related outcomes among the bereaved? In this context, Brown, Brown, House, and Smith (2008) suggested that bereaved people who provided support to others would show evidence of stress buffering. They examined the role of self-reported helping behaviour on the bereaved helpers' depression rates in a longitudinal investigation, controlling for many potentially-confounding variables (e.g., health and robustness). Giving help to other bereaved people was associated with faster decline in depression for the helper over time. Along similar lines, in another sophisticated, longitudinal study, Li (2007) investigated how volunteer participation among widowed persons has an impact on coping with the death of a spouse. Importantly, participation among widowed persons was compared with that of continually married counterparts, enabling examination of bereavement-specific versus general effects of volunteering. Compared with their continually married counterparts, people who experienced spousal loss reported greater likelihood of pursuing volunteer roles a few years after the death of their spouse. Importantly, volunteering was found to protect against depressive symptoms, suggesting that this type of helping offsets the negative effect of widowhood on well-being.
In conclusion, although it seems evident that help is needed and appreciated, there are limits to the support that others can offer in helping the bereaved. One cannot, it seems, take away the pain of losing the deceased person or in any way "replace" him or her. Ironically, taken together, the studies outlined above also seem to suggest that social support is helpful, but even more to the one who is providing than the one who is receiving it.

Do grief counselling and grief therapy help?
Next we turn from informal to more organised or professional help. Before assessing the state of knowledge about the efficacy of these types of intervention, it is necessary to draw a fundamental distinction, namely, between satisfaction with versus effectiveness of intervention. In both research and clinical settings, these two very different phenomena are often treated as the same thing, causing much confusion about the efficacy of intervention.

Satisfaction with versus effects of intervention
The question whether grief interventions are helpful is likely to be answered differently according to the interests of or type of person to whom one addresses this question. If one asks clients, studies show that the vast majority is satisfied. Gallagher, Tracey, and Millar (2005), for instance, undertook an evaluation of bereavement counselling by clients subsequent to their participation in a service provided by Cruse, a national organisation for bereavement care in the UK and Northern Ireland. Six weeks after the end of counselling, clients in the study reported (strongly or very strongly) that they now felt the loss less intensely (89%); experienced fewer physical symptoms (88%); felt less anxious (81%), found it easier to cope (85%); felt more confident (72%), were more able to relate to others (86%) and were more able to look to the future (80%). Therapists and counsellors may indeed be inclined to endorse this picture of the bulk of clients being satisfied with the help offered. Certainly, these results are impressive in that they indicate high satisfaction with the provision of help.
However, as noted above, it is of crucial importance to differentiate satisfaction with intervention from effects of intervention. Change in distress that takes place during intervention is often attributed to the intervention itself, whereas over time change can take place naturally, and precisely this is the case with a process like grieving, in contrast to more stable conditions such as phobias. For example, finding it easier to cope could be a result simply of the passing of time and may have nothing to do with the intervention. If researchers were asked about the efficacy of intervention, they would be more likely to take this natural change into account. A good design of an efficacy study would include the possibility of differentiating natural change from change attributable to the intervention. To establish whether an intervention is helpful, it would require a comparison between an intervention and a non-intervention control group. The design of such a study would necessitate a pre and post intervention assessment 1 . Although the two kinds of approaches complement each other, it needs to be kept in mind that clients can be satisfied with help that is not effective and vice versa.

Effects of intervention post-loss
During the last decade, several extensive qualitative reviews and metaanalyses have been published, reporting patterns in the efficacy studies of grief therapy and grief counselling (Allumbaugh & Hoyt, 1999;Currier, Holland, & Neimeyer, 2007;Currier, Holland, & Neimeyer, 2010;Currier, Neimeyer, & Berman, 2008;Kato & Mann, 1999;Larson & Hoyt, 2007;Schut et al., 2001). Although these evaluations do not all come to exactly the same conclusions, and the scientific debate about the efficacy of bereavement intervention still continues, the evidence points in the direction that most bereaved people do not need and will not gain from grief therapy or counselling. It seems that the majority of bereaved people is resilient enough to adapt to the loss without the involvement of counsellors and therapists (Bonanno, Wortman, & Nesse, 2004). Outreaching interventions aimed at all bereaved people, regardless of their background or situation, cannot be regarded as beneficial in terms of diminishing grief-related symptoms ). This appears to be true for interventions for adults as well as for children (Currier et al., 2007). The first one to draw this conclusion was actually Parkes (1996), more than a decade ago, when he concluded that there was no evidence that all bereaved people benefit from counselling and that research had shown no benefit to arise from the routine referral of people to counselling for no other reason than that they have suffered a bereavement. Some years have passed since Parkes wrote these words, and there are signs that processes and procedures have changed in bereavement care. Nevertheless, unsolicited and routine help is still being offered to the bereaved.
Intervention programs that are in principle open to all bereaved persons, with the criterion for participation being simply that one has experienced a loss through death, are known as primary preventive interventions 2 . For primary preventive intervention to be helpful it seems at least necessary for clients to initiate help themselves, instead of help being offered to them (Currier et al., 2008;Schut et al., 2001). Interventions targeted at bereaved persons at risk of developing complications, so-called secondary -----preventive interventions, appear sometimes to be modestly effective, but often this improvement is only temporary. When important longer-term effects are analysed, positive effects of the intervention often seem to vaporise. Clear and positive results are established when interventions exclusively target grievers displaying complicated grief in trying to adapt to the loss, known as tertiary preventive interventions. In the latter category, outcomes may even compare favourably with psychotherapies for other disorders (Currier et al., 2008).
How can such rather disappointing results be explained, specifically the results in the first category of interventions for all bereaved people? First of all, as was suggested before, it seems likely that, specifically when help is being offered to the bereaved instead of being asked for, a substantial proportion of bereaved people was not in need of help and was resilient enough to deal with the loss without interference from care givers (cf. Bonanno et al., 2004;Raphael, Minkov, & Dobson, 2001. The fact that they nevertheless accepted help being offered to them does not contradict this. That people may be resilient does, after all, not mean that they are not affected by the loss, and help offered in times of emotional turmoil is likely to be accepted, despite the fact that the person may have been able to deal with their loss by using their natural resources.
Research does indeed suggest that primary preventive intervention that is initiated by the bereaved themselves shows better results (cf. Larson & Hoyt, 2007;2009;, which can be understood as substantiating the above argument. A second explanation may be that help offered may interfere with the natural grieving process. Tudiver, Hilditch, Permaul, and McKendree (1992, p. 180), for instance, suggested that, for their mutual help groups, "focusing widowers' attention on their own and others' grief may have hindered an early (and measurable) recovery from their grief". A study in the Netherlands by de Keijser (1997) also showed that the natural support system of the bereaved tends to be perceived by the bereaved themselves as withdrawing when professional help is accepted. A third possible explanation may be that the intervention offered to the bereaved could be based on questionable assumptions. The majority of interventions for bereaved persons is based on the assumption that confronting the loss, working through grief, is indeed beneficial, although research has shown this not always to be the case (cf. Bonanno, Keltner, Holen, & Horowitz, 1995;Stroebe, 1992, Wortman & Cohen-Silver, 2001.
In conclusion, taken together, the patterns of findings available so far, as summarised above, suggest that intervention is not effective for all bereaved persons. Health care professionals need to identify and provide support for "at risk" persons and those experiencing complications in their grieving process. Nevertheless, we want to emphasise that more research is needed, as illustrated next.

The relevance of risk factors in intervention: an example
Since prevention of pathology is desirable on the one hand, and since, on the other hand, it is equally desirable to target interventions as much as possible toward those who gain from them, preventive interventions for risk groups need more attention (this point will be elaborated on when we come to pre-bereavement intervention). The fact that effects for that category of bereavement care are generally speaking modest and temporary should be regarded as a challenge, both scientifically and clinically. It seems likely that results within this category of grief interventions could be improved if we had better understanding of risk and protection factors. Such factors have been categorised according to the situation and circumstances of death, intrapersonal risk or protective factors, interpersonal or non-personal resources and protective factors, as well as coping styles, strategies and processes (Stroebe et al., 2007). In recent decades many studies have focused on tracing risk factors within these categories, but thus far this has not resulted in a clear picture of moderators and mediators in the grief process (Stroebe et al., 2007). This is likely to be due to the complexity and interaction of these factors in how they influence the level and course of grief manifestations (e.g., between personality and circumstances of death) (Stroebe, Folkman, Hansson, & Schut, 2006). This makes it very difficult to develop valid and reliable screening instruments for use by practitioners.
Yet, not all risk factors are complicated to start with. Let us take a closer look at gender for instance. Research has sufficiently proven that men react more strongly to the loss of their spouse than do women  and there is also ample evidence that women show stronger grief reactions after the loss of a child than men (cf. Murphy, 2008). This suggests a rather strong interaction between gender and relationship to the deceased on grief manifestations. Furthermore, several intervention studies have suggested gender specific effects of grief intervention for persons experiencing difficulty in dealing with their loss. Schut, Stroebe, van den Bout, and de Keijser (1997) found that an emotion focused intervention was more efficacious in the long term for widowers, while a problem focussed approach turned out to be more helpful for widows. This study needs replication, but it may also explain why some results of grief intervention are disappointing. For instance, Walter (1999, p. 182), on the basis of these results, concluded that "In general, in such situations, women like talking about their feelings and men like to stay away from counselling, but overall, the system may be ineffective, because it gives clients what they want rather than what they need.
[…] Bereavement organisations have been offering the wrong things to the wrong people".
Murphy (2008; Murphy, Johnson, Cain, Das Gupta, Dimond, Lohan et al. 1998) also studied gender differences in effects of grief intervention, but focussed their study on parents having lost a child through suicide, homicide or accident. Furthermore, Murphy developed an intervention that combined emotion and problem focussed aspects. Strikingly, Murphy and colleagues found no effects at all of the intervention for fathers. Among mothers, they found a positive outcome for those with relatively high levels of distress before the interventions, while negative results were discovered for mothers with relatively low levels of distress at baseline.
To conclude: even with respect to a risk factor such as gender that should be relatively easy to investigate, we have to conclude that too little attention has thus far been paid to the impact of grief intervention. Yet, the studies mentioned above do suggest that risk factors such as gender are very important to take into consideration in developing bereavement care in the community. However, the studies also seem to suggest that even a simple moderator like gender soon becomes rather complicated, in that apparent levels of distress at baseline and relationship to the deceased seem to play an important interacting role in terms of effects of interventions. Nevertheless, it seems of crucial importance to take these aspects into consideration, because otherwise bereavement organisations could indeed be offering the wrong things to the wrong people as Walter (1999) seems to be warning us.
Effects of intervention before the loss As noted at the outset, reviewers of the efficacy of bereavement intervention, including ourselves, have excluded examination of the effects of professional support on bereaved family members in the period before the loved one died. Perhaps this is not surprising since, historically, end-of-life care and bereavement care were typically the focus of separate scientific investigation. Also, clearly, such provision would not be possible for the subgroup of persons whose loved one dies suddenly or unexpectedly. In our view, though, it is a natural extension to examine whether pre-bereavement intervention is effective for family members, and whether the patterns identified above are actually replicated in the end-of-life phase.
One important distinctive feature of this pre-loss as opposed to post-loss care -that could actually influence the impact of this type of interventionis the fact that there is often continuity in the care provided before and after the loss. Such continuity often implies a certain stability in pre-bereavement and post-bereavement support (e.g., Field, Payne, Relf, & Reid, 2007), possibilities for early risk assessment (e.g., Parkes, 1996;Payne & Relf, 2001) and rather naturally-occurring possibilities for follow-up assessment (Walsh, Foreman, Curry, O'Driscoll, & McCormack, 2008). For such reasons, we cannot assume that the conclusions drawn so far with regard to the effects of post-bereavement intervention are applicable to the provision of support before the loss has occurred. For example, pre-loss intervention may indeed be more generally beneficial to family members, not just for those at high risk of difficulties in adjusting to impending loss. It becomes very important, then, to extend the scope of investigation to examining the efficacy of end-of-life care for family members.
Major reviews of the literature on the efficacy of end-of-life care on the bereaved have been provided by Lorenz, Lynn, Morton, Dy, Mularski, Shugarman et al. (2004), and by Higginson, Finlay, Goodwin, Hood, Edwards, Cook et al. (2003), who conducted a meta-analysis of 26 studies of palliative and hospice care. These reviewers came to similar conclusions. Family members were more satisfied with the quality of care than those who had not received terminal care (there were also substantial positive effects on patients). However, it was concluded that there were no effects with respect to post-loss bereavement outcomes on caregivers and family members. Lorenz et al. (2004) further concluded that results were highly discrepant across studies. The reviewers also had critical remarks to make about the quality of the studies in general (Harding & Higginson, 2003). In one of the best studies reviewed (Ringdal, Jordhoy, Ringdal, & Kaasa, 2001), family members were randomly assigned to receive comprehensive palliative care or conventional oncological care. Follow-up was conducted one year after loss. No differences in bereavement outcome were found between the family members in the comprehensive intervention and the control group.
Similar to the conclusions we drew from the post-bereavement intervention efficacy studies, the pattern here seems to be in line with those we drew concerning primary intervention. Again there is no sound evidence that offering help to family members in the context of end-of-life care in general is effective. Therefore, we have to look at risk groups again. One well-designed randomized controlled trial that examined the impact of family focused grief therapy in palliative care was conducted by Kissane, McKenzie, Bloch, Moskowitz, McKenzie, & O'Neill (2006;Kissane & Lichtenthal, 2008). While the overall impact of family focused grief therapy was modest, significant improvement in distress and depression was found among individuals with high baseline levels of distress.
The above results indeed suggest that a focus on risk groups in the provision of pre-loss intervention may be the best strategy to follow, as was the case for post-loss intervention. However, as Harding and Higginson (2003) stated, more high quality research is needed before firm conclusions can be drawn.

Conclusions
The identification of patterns, and finding similarities in these across the pre-and post-loss investigations of the efficacy of intervention, is hopefully useful to both researchers and clinicians in planning their work. Nevertheless, it is evident that helping the bereaved deal with the consequences of bereavement is not simple. That is not only true in the case of friends and family, but also for volunteers and professionals in the bereavement sector. We all need to be modest in what we think we can accomplish. This does not mean that we cannot help, but we need to be aware that what may be the core element of grief, emotional loneliness, is something a bereaved person needs to slowly and at his or her own pace adjust to, and that help from others is incapable of healing that pain. That does not mean that friends and family cannot help, they can indeed. But it does not seem to help bereaved persons any more than it does non-bereaved.
As for volunteers and professionals providing help to the bereaved, according to the research available at present, they need to be modest too. No evidence has been found that care for the bereaved in general, provided on an institutional level, is effective. This is specifically the case when help is provided routinely and is initiated by care providing agencies instead of the bereaved themselves. Grief interventions for specific target groups show better results, but there is much that needs to be done before this, in itself a promising category of intervention, proves its true value. The primary challenge, and maybe even the key responsibility for further development of the theoretical and empirical basis for this type of intervention, lies with the scientific community, although close collaboration with practitioners will be necessary. Tertiary preventive interventions, advanced therapeutic programs for complicated grief, have convincingly been developed and tested in recent years (cf. Boelen, de Keijser, van den Hout, & van den Bout, 2007;Shear, Frank, Houck, & Reynolds, 2005). The primary focus, therefore, should now be on the development and testing of interventions targeted toward risk groups, thus preventing complications from occurring. This direction urgently needs more than crude examination of differences in change over time in grief manifestations between intervention groups and non-interventions groups. Moderators and mediators need to be taken into account too. Clinicians are typically aware of this, but researchers such as ourselves can do well to remember: It seems clear that one size does not fit all.