• Open Access

Reported experiences of bereavement support in Western Australia: a pilot study

Authors

  • Samar M. Aoun,

    Corresponding author
    1. School of Nursing and Midwifery, Faculty of Health Sciences, Curtin University, Western Australia
    • Correspondence to: Professor Samar M. Aoun, School of Nursing and Midwifery, Faculty of Health Sciences, Curtin University, GPO Box U1987, Perth, WA 6845; e-mail: s.aoun@curtin.edu.au

    Search for more papers by this author
  • Lauren J. Breen,

    1. School of Psychology and Speech Pathology, Faculty of Health Sciences, Curtin University, Western Australia
    Search for more papers by this author
  • Bruce Rumbold,

    1. Palliative Care Unit, Department of Public Health, La Trobe University, Victoria
    Search for more papers by this author
  • Denise Howting

    1. School of Nursing and Midwifery, Faculty of Health Sciences, Curtin University, Western Australia
    2. School of Psychology and Speech Pathology, Faculty of Health Sciences, Curtin University, Western Australia
    Search for more papers by this author

  • The authors have stated they have no conflict of interest.

Abstract

Objective: This article describes the pilot testing of a community survey to ascertain the experiences and needs of people who were bereaved 6–24 months before the survey. The pilot study aimed to assess the feasibility and acceptability of the survey and test the theoretical public health model for bereavement support.

Methods: A postal survey was used to collect information from clients of three funeral providers in Western Australia in 2012.

Results: The findings confirmed the feasibility and acceptability of the survey questions. The analysis of the demographic characteristics, experience of bereavement and satisfaction with support revealed differential needs that align with the expectation of low, moderate and high risk, as articulated in the public health model.

Conclusions: The data provided tentative empirical support for the public health model of bereavement support. This is the first empirical test of this model nationally and internationally.

Implications: Considering the lack of evidence to guide development and allocation of bereavement programs in Australia, a larger survey will enable us to determine how the support needs of each of the three groups of bereaved people should be serviced. This is important for cost-effective and equitable resource allocation.

Bereavement is a significant life stressor with outcomes across four domains – emotional (e.g. sadness, anger, guilt), physical (e.g. fatigue, agitation, pain), behavioural (e.g. sleep and appetite disturbance, absentmindedness) and cognitive (e.g. disbelief, confusion, hallucinations). Bereaved persons may also face secondary losses, including social isolation and stigma,1 financial loss,2 and changes in roles and responsibilities.3 Such disruptions can trigger existential or spiritual issues that present as holistic losses: a fractured sense of identity, the fragmentation of community, or loss of meaning and purpose in living.4 Additionally, bereavement is associated with an increased risk of mortality, which is greater for widowed than married, single or divorced people and greater again for widowers than widows, with risk generally peaking in the first few months following bereavement.5

Palliative care provides the most comprehensive strategy for bereavement support in many communities, as the philosophy of palliative care emphasises support for the patient during illness and support for family carers before and after the patient's death.

In Australia, the peak body for palliative care, Palliative Care Australia (PCA),6 outlined several standards of palliative care. Standard 8 asserts, “Formal mechanisms are in place to ensure that the patient, their caregiver/s and family have access to bereavement care, information and support services” (p. 33) and that this support should be based on need. Recent clinical practice guidelines for bereavement support in palliative care also highlighted the importance of providing support according to the assessment of need.7 A recent audit of Australian palliative care services established bereavement care as the highest priority for improvement. The report based on this audit's findings and the latest national palliative care strategy recommended a national evidence-based approach to bereavement support.7,8

Despite these guidelines, recent surveys demonstrated that, while almost all Australian palliative care services offer bereavement supports and services to families of deceased patients, these services and supportive strategies may be only loosely linked with assessments of risk or need.9,10 For instance, one survey of Australian palliative care services determined that 95% (of 236 services) offered some form of bereavement support.10 The authors of this study concluded that there was a “lack of clear evidence to guide development and allocation of bereavement programs in palliative care” (p. 230). The most common types of support were a telephone call (offered by 86%), memorial service, (66%), letter (55%), anniversary card (53%), group sessions (31%), information package (5%), and informal gatherings (4%). About two-thirds of the services reported engaging in some form of bereavement risk assessment at the time of death, with two-thirds relying on multidisciplinary team opinion. More than half used a formal tool to assess bereavement risk and about half relied on a single staff member's opinion (some services reported using more than one method).

In practice, the majority of bereaved people manage their grief with the support of family, friends and neighbours. It is only a small proportion, about 10 to 20%, who experience persistent psychiatric difficulties, including Prolonged Grief Disorder (PGD), previously known as complicated grief.11,12 This group exhibits higher levels of distress and is most likely to benefit from targeted psychological interventions,13,14 whereas such interventions may be of marginal benefit, or even counter-productive, for others.15 Given this, we question the approach of some palliative care services, in Australia and overseas, that attempt to provide blanket or undifferentiated support to bereaved clients. This is neither an effective nor economical use of their limited resources,16,17 and can compound health professionals’ stress and burnout.18,19

Conceptual framework and rationale

Our research team developed the conceptual framework of the public health approach to bereavement support,16 summarised in Table 1. The public health literature typically identifies three levels of intervention that target different populations:

Level 1 – universal (for the whole population of interest, i.e. all bereaved people);

Level 2 – selective (for groups at risk of developing more complex needs, e.g. complications of bereavement)

Level 3 –indicated (for people showing signs of disorder, e.g. PGD).

Table 1. A three-tiered public health model of bereavement care.16
Level of public health interventionsLevels of risk or needType of supportSupport provided byTarget population and level of support neededProportion of bereaved*
  1. * Example of Sobell House Hospice in UK, 1989–2002

Universal1Information about bereavement and relevant supportsFamily and friends (information supplied by health and social care professionals)

All bereaved (normal grief)

Low need

54%
Selective or Targeted2Non-specialist supportTrained volunteers, mutual-help groups, community supportsThose at-risk of developing complex needs Moderate need33%
Indicated3Professional specialist interventionsMental health services, bereavement services, or psychotherapy

Those with complex needs

High need

9%

These levels are congruent with those proposed for preventive intervention for bereavement care – primary (for all bereaved people), secondary (for people at-risk of complications of bereavement), and tertiary (for people with complicated bereavement)20 and the call for applying a public health perspective to end-of-life care services.21

The National Institute for Clinical Excellence (NICE)22 in the United Kingdom proposed a similar three-tiered approach to bereavement in palliative care according to the needs of carers and families and approximated the proportions in each tier. Based on cancer or expected deaths, the model advocates that all the bereaved people should have access to information about bereavement and relevant available supports (first tier). The information would be delivered by the palliative care service involved in the care of the patient and family, with much of the support coming from the bereaved person's social networks, including family and friends. Just over one-third would need more formal opportunities to consider their loss (second tier). The support for this tier would be provided by non-specialist social and therapeutic support such as volunteer bereavement workers, bereavement mutual-help groups, and faith-based and other community groups. This middle-tiered proportion is reflected in the UK Sobell House's analysis of retrospective data on risk assessment and type of support provided for its bereaved carers (1989–2002).23 Face-to-face consultations with palliative care services identified this group as causing challenges for services where timely and adequate support could potentially reduce the proportion of the population experiencing prolonged grief.24 The third tier is made up of the 10–12% who would need specialist intervention – such as counselling, mental health services, bereavement services or psychotherapy – to supplement the first two tiers or because these levels of support are not available to them. Thus, there are about 45% of bereaved people in the second and third tier who would benefit from either targeted or indicated interventions.

To develop an evidence base for interventions that can target the level of risk and need of the bereaved population in Australia, we need to map the baseline situation through a survey. An understanding of the experience and needs of bereaved people, whether they have or have not used services, is essential to guide any work in this field.

This article describes the pilot testing of such a survey aimed to ascertain the experiences of people who were bereaved 6–24 months before the survey, identify their perceived needs and where in the community they have found support, and make recommendations for service provision based on the level of need. We chose this time period as six months post-bereavement is the earliest time period required for diagnosis of PGD while 24 months is not likely to compromise the accuracy of recalled information.25 The objectives of the pilot study were to assess:

  • the feasibility and acceptability of the survey by the bereaved community
  • the appropriateness of the recruitment strategy
  • the support for the theoretical public health model of bereavement care.

Methods

A postal survey was used to collect information from clients of three funeral providers in metropolitan and rural Western Australia, 6–24 months after the death of their family member. Ethics approval was granted by the Human Research Ethics Committee of Curtin University.

Instrument

A questionnaire was developed to obtain demographic information; the supports people accessed; supports they would have liked to have been able to access; their needs and whether they were met. The questionnaire had eight sections with a total of 80 predominantly closed questions aimed at generating a large set of population-based data amenable to quantitative analysis in a future survey with a larger sample size.

  • Section 1 – the carer or closest person to the deceased
  • Section 2 – their deceased loved one
  • Section 3 – their experience caring for a loved one with a terminal or life-limiting illness (expected death)
  • Section 4 – the bereavement support they received from palliative care providers
  • Section 5 – the overall support they received after their loved one died
  • Section 6 – their experiences with the funeral provider
  • Section 7 – exploring their current feelings about the loss of their loved one
  • Section 8 – further comments and an option for those who wish to participate in an in-depth interview to leave their details so they are contacted in 2–3 months’ time.

The questionnaire was developed in consultation with a reference group of representatives of the funeral industry, bereavement counsellors, palliative care services, primary care and community-based services. To encourage responses from people with accessibility impairments and from cultural and linguistically diverse backgrounds, we included a statement at the front of the questionnaire that if anyone needed assistance completing it, they could call our office number.

The questionnaire included a validated risk assessment screening measure for PGD, the PG-13 in section 7.12 Compared with other tools reviewed, the PG-13 is short, easy to self-administer, has a theoretical basis and aligns with the criteria proposed for inclusion in the forthcoming World Health Organization's International Classification of Diseases (ICD-11).12,26 The PG-13 measures responses to separation social/functional impairment, and cognitive, emotional and behavioural symptoms over a period of not less than six months since bereavement. All five criteria must be met to indicate the presence of PGD: event (bereavement); separation distress; duration (i.e. >6 months); cognitive, emotional and behavioural symptoms; and social/occupational impairment. The score range is 11–55 and a score of 36 or more is a clinical indicator of PGD. It should be noted that PGD was not included in the latest edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM5); bereavement was removed as an exclusion criterion for both Major Depressive Disorder and Adjustment Disorder and a new subtype of Persistent Complex Bereavement Disorder was included in section III for further study.27 The practical implications of these changes have yet to emerge, but they would seem to increase the likelihood of clinical treatment for people in low to moderate need of such interventions.

Participants and procedure

The pilot study was undertaken from August to October 2012 in Western Australia. The funeral providers were reached through their professional association (Australian Funeral Directors Association [AFDA]). Three providers volunteered to take part in the pilot study; the others decided to wait for the results as they were apprehensive about the effect the research might have on their clients. One funeral provider was a large corporate that conducts about 5,000 funerals a year in the metropolitan area of Perth. The smaller metropolitan provider conducts about 800 funerals a year and the rural provider about 1,000 per year.

The criteria for inclusion in the study were that the potential participants were bereaved for at least six months. In selecting clients for the pilot study, funeral providers reported being hesitant about approaching people who had experienced a traumatic death or death of a child. The research team agreed to exclude this group in order to accommodate this request.

The three funeral providers mailed the pilot questionnaire to previous clients. A total of 90 study packs were delivered to the three providers (30 to each), containing a cover letter addressed from the service provider to the family, information and consent forms, the questionnaire, a feedback form on the questionnaire (for participants to indicate if they found any of the questions confusing, upsetting, difficult or irrelevant), a list of support services for the family to use in case the participant became distressed while completing the questionnaire, and a reply paid envelope. The funeral providers selected the participants from their databases, attached names and address labels on the envelopes and mailed the study packs.

Analysis

The quantitative data were analysed using IBM SPSS (version 20). The participants’ PG-13 scores were calculated to determine the proportion meeting diagnostic criteria. Descriptive statistics were calculated for the demographic variables and responses relating to caring and supports and services received. The data for each level were scrutinised and collapsed into narrative vignettes to simulate real events while maintaining the participants’ anonymity. The narrative vignettes were based on the predominant variables for each level of risk.

Results

Response rate and instrument feasibility and acceptability

Twenty-three questionnaires and feedback forms were returned completed and one was returned indicating that the person was no longer at this address. The response rate was 26% (21% in the rural area and 27% in the metropolitan area). On average, the questionnaire took 30 minutes to complete and most participants had no problems completing the survey and did not find the questions upsetting. A few of the questions will require minor fine tuning in the future, and the order of several questions will need to be changed for the questionnaire to have a better flow when the larger survey occurs.

Demographic characteristics of respondents

The family carers or next of kin (referred to as carers in this article) who responded to the survey (Table 2) were bereaved for a median of 12 months (7–25 months). Sixty-one per cent of carers were female; mean age of 60.52 years (SD 11:40); 44% were married and 39% widowed; 48% were the spouse of the deceased, and 48% were the son or daughter of the deceased; 57% were Australian and 39% from another English-speaking background; 44% were Christian (44%), 17% had other religion and 39% had no religion; a third had finished high school, a third had a diploma or trade qualification and nearly a third had a university degree; 57% were currently employed and 43% retired. Sixty-four per cent had diagnosed medical conditions with co-morbidities; the most reported of these were high blood pressure, depression, arthritis and anxiety.

Table 2. Summary of carers’ demographics, n=23.
 n(%)
Gender 
  Female14(61)
  Male9(39)
Age (years) 
  mean±SD60.5±11.40
  median[range]62[31–73]
Marital status 
  Married10(44)
  Widowed9(39)
  Never married/single4(17)
Relationship to the deceased 
  Spouse11(48)
  Offspring11(48)
  Parent1(4)
Language/cultural background 
  Australian13(57)
  Other English speaking8(35)
  Non English speaking2(8)
Religious background 
  Christian10(44)
  No religion9(39)
  Other religion4(17)
Highest education level 
  High school8(35)
  Diploma/trade qualification8(35)
  Undergraduate degree7(30)
Employment status 
  Employed13(57)
  Retired9(39)
  Other1(4)
Comorbidities 
  at least ONE14(64)
  Hypertension7(30)
  Depression6(26)
  Arthritis5(22)
  Anxiety3(13)
Bereavement length (months) 
  mean±SD13.8±5.17
  median[range]12[7–25]

Current feelings about the loss (PG-13)

Of the 23 individuals who responded, only one met all five criteria. Two respondents (8.7%) met four criteria, 10 (43.5%) met three criteria, two (8.7%) met two criteria and seven (30.4%) met one criterion (one did not complete this section). All met the event criterion (bereavement), 56.5% the separation criterion, 52.2% the duration criterion (experiencing symptoms at least daily and after six months from loss), 30.4% the social/functional impairment criterion and 4.3% the cognitive, emotional and behavioural impairment criterion.

Bereaved participants were grouped into three categories of risk:

  1. Those with low level of risk (meeting up to two PGD criteria): 41% with scores between 11 and 26 (mean=14.6 SD=4.6).
  2. Those with moderate level of risk (meeting up to four PGD criteria): 55% with scores between 18 and 32 (mean=24.9 SD=4.3).
  3. Those with high level of risk (meeting five criteria): 4% with a score of 49 (only one person in this pilot study).

The analysis of the demographic characteristics, experience and impact of caring and bereavement, and satisfaction with support received from a variety of services according to the three bereavement levels, tentatively revealed different characteristics of the three groups of bereaved people, allowing for the small sample size. In group 1 (n=9), bereaved people were younger and mainly caring for their dying parents; in group 2 (n=12), bereaved people were older and mainly caring for their dying spouses after a reasonable period of care and support; and group 3 (n=1), this bereaved person cared for his dying spouse during a very short period of illness. Details of the predominant variables of the three profiles are presented in Table 3.

Table 3. Predominant profile of each level of risk for bereavement support
VariableLevel 1Level 2Level 3
Carer/next of kin characteristics
SexFemaleFemaleMale
Age
Median (range)59 (31–66 years)64 (42–73 years)65 years
 
Marital statusMarriedWidowedWidowed
EducationUniversity education  
 Diploma/certificate/ trade qualificationHigh school 
EmploymentPaid employmentPaid employment/ retiredRetired
Diagnosed medical conditionsNoYesNo
Time since bereavement13.5 (8–25) months11.5 (7–21) months17 months
Relationship to deceasedDaughterSpouseHusband
Satisfaction with supportReceived as much support as wanted; main source of support was family/friendsReceived as much support as wanted from both palliative care service and family/friendsReceived as much support as wanted from health and community services; did not receive as much as wanted from palliative care
Deceased characteristics
Age
Median (range)76 (65–85) years66 (56–74) years67 years
SexFemaleMaleFemale
Place of residenceNursing homeWith bereavedWith bereaved
Place of deathNursing homeHospital (median 19 days last hospital admission); preferred place of death of bereavedHome; preferred place of death of bereaved and deceased
Type of deathExpected (cancer)Expected (cancer)Expected (cancer)
Duration of illness Median (range)1.8 (0.3–7) years2 (0.1–15) years3 months
Duration of care Median (range)11 (3–24) months10 (1–180) months3 months
Period received palliative care Median (range)6 (2–18) months1 (0.1–12) months7 days
Since bereavement
Physical healthStayed the sameStayed the sameGot a bit worse
Mental healthStayed the sameStayed the same/got a bit worseGot a bit worse
Financial situationImprovedStayed the sameGot a bit worse
PGD criteria metNone (other than event of bereavement)Separation distress; Duration more than 6 months; Some had social/occupational impairmentSeparation distress; Duration more than 6 months; Cognitive, emotional, behavioural symptoms; Social/occupational impairment
PG-13 mean score (95% CI)

14.6

(10.99–18.12)

24.9

(22.21–27.62)

49

Vignettes illustrating the profile of the three levels of risk

The following summary vignettes illustrate the profile in each bereavement level: Level 1 – the parental carer; level 2 – the widow; level 3 – the complicated griever. The three blended vignettes were drawn from the 41% categorised as low risk and the 55% categorised as moderate risk. There was only one person considered high risk according to the diagnostic criteria for PGD (4%).

Low Risk: The Parental Carer

Marie is a university-educated, married woman in her 50s in full-time paid employment. Her 75-year-old mother, Olive, died in a nursing home just over a year ago. Olive's death was expected – she had been diagnosed with cancer two years prior – and Marie was Olive's primary family carer in her last year of life. Olive received palliative care for six months. Since Olive's death, Marie's physical and mental health has stayed the same and her financial situation has improved, due to no longer paying for the nursing home. Overall, she received as much support as she wanted from any health and community service. Following bereavement, Marie did not experience separation distress; cognitive, emotional or behavioural symptoms; or social/occupational impairment. Marie's PGD score is 15.

Moderate Risk: The Widow

Lorraine is in her early 60s and lives with some medical conditions. Before she retired, she worked as a clerical assistant. Just over a year ago, her 66-year-old husband, Alan, died from cancer; he had been diagnosed three years earlier. Lorraine had been his primary carer for his last 17 months. Alan received palliative care in the last month of his life. He died in hospital, which was Lorraine's preferred place of death for him.

Lorraine received as much support as she wanted from palliative care and other services. Since Alan's death, Lorraine's physical health and financial situation have stayed the same but her mental health is a little worse. Lorraine initially experienced separation distress but this subsided in the first few months. She did not report cognitive, emotional of behavioural symptoms but did experience some social/occupational impairment. Lorraine's PGD score is 25.

High Risk: The Prolonged Griever

Bill is a 65-year-old retiree with a high school education and no diagnosed medical conditions. His 67-year-old wife, Iris, died three months after being diagnosed with cancer. Bill was Iris's primary carer and provided day-to-day, hands-on care during those three months. Iris received palliative care for seven days. She died at home, which was Iris and Bill's preferred place for her death. Bill received as much support as he wanted from health and community services. He received some support from palliative care but not as much as he would have liked. Since Iris's death, Bill's physical and mental health and financial situation have all got a bit worse. He reflected that he would have wanted professional support. Bill has experienced separation distress for more than six months; cognitive, emotional and behavioural symptoms; and social/occupational impairment. Bill's PGD score is 49.

Discussion

The findings of this pilot study and extensive participant feedback support the feasibility and acceptability of the survey questions and provide a good indication that the objectives of the larger population-based study can be achieved, in terms of describing the demographic characteristics of bereaved people, experience and impact of caring and bereavement, and satisfaction with support received from a variety of services. It also provided information as to the participants’ perceived needs. Importantly, the data provided tentative empirical support for the public health model of bereavement support.

This is the first empirical test of this model nationally and internationally. The three vignettes illustrate how the level of risk of PGD, as measured by the PG-13, i.e. low, moderate and high risk, is congruent with the participants’ expressions of met and unmet support needs. For instance, risk may be related to duration of palliative care, ranging from an average of six months (low risk), one month (moderate risk) and one week (high risk). The analysis of the demographic characteristics, experience and impact of caring and bereavement, and satisfaction with support received from a variety of services revealed differential experiences and needs that align with the expectation of low, moderate and high risk, as articulated in the public health model, despite the small sample size.

Interestingly, the PG-13 scores of our respondents tended to cluster around 15 for those at level 1 and 25 for those at level 2. (There was only one respondent at level 3, with a score well in excess of that required for a diagnosis of PGD.) This suggests a tentative finding that the low and moderate risk categories, levels 1 and 2, may be distinguished fairly clearly on the basis of PG-13 scores as well as the sources from which support needs were met. Additionally, the vignettes provide a context for the scores: low scores typically arose from losses that were expected and less disruptive to daily routines (e.g. the death of an aged parent); moderate scores typically arose from deaths that were both expected and everyday-disruptive (e.g. the death of an ageing and ill partner); and deaths that are both unexpected or resisted and disruptive may lead to high scores on the PG-13. A much larger sample is needed to verify such a finding, but at the practice level it would be helpful if bereaved people's risk of PGD could be assessed easily and responded to appropriately in accordance with the proposed three service response categories.

In addition to revealing that there might be a difference in type of response between the three levels, the data show differences in degrees of support need across the three levels. The sources of support listed by respondents indicate that, for the most part, those in level 1 were satisfied with support provided through everyday interactions. Those at level 2 were also satisfied with the level of support, but most of them were also linked with some sort of bereavement follow-up program, through palliative care, for example. The sole level 3 respondent considered the support he received to be inadequate; his needs necessitate targeted mental health interventions in addition to other forms of support that were effective for people at levels 1 and 2. There is merit in our hypothesis that the support needs of bereaved people at different levels can be met using different combinations of strategies. Again, it needs further exploration with a larger sample, but it is a finding suggesting the need for flexible and targeted bereavement services, many of them informal and within local communities as well as formal offerings of health services.

Limitations and relevance to the larger study

The participants’ profiles and experiences may not be representative of those of the general bereaved community, as there was no random selection for this pilot group. It was a convenience sample chosen by the funeral providers. However, despite this, the profile of respondents seems similar to that of bereaved people in the Sobell House Hospice in UK.23

To be cautious, the providers selected clients who had an expected death, which explains the over-representation of death due to cancer. Hence, the vignettes are a representation of those who had an expected death and received palliative care. The wider group of providers was reassured by the results of the pilot study and understood our concerns about the selection bias. They agreed that for the larger survey, the questionnaires would be mailed to all their clients who were bereaved between six and 24 months from any cause of death. The pilot study appears to have given confidence to the industry partners to be inclusive rather than selective.

The response rate was lower (26%) than we anticipated from initial discussions with one of the funeral providers, who reported a 50% response rate for their clients’ satisfaction surveys. However, the response rate is comparable to what is expected from anonymous population-based postal surveys with no follow-up. It is possible that participants who were not too distressed were comfortable completing the survey, but those who were distressed found the idea of completing the survey upsetting and did not complete it. As such, risks, needs and scores of PGD could be under-estimated. The length of the survey (mean and median of 30 minutes) may have been burdensome, as respondents were asked to also complete a feedback sheet on each section of the questionnaire for the pilot study. This additional requirement will not be part of the larger study and thus we could expect a higher response rate.

The survey was constructed to provide information on the population-based experiences of bereavement, including the extent of the alignment of bereavement risk and service need. The validated PG-13 means that we will also be able to determine a population rate of PGD in the larger survey. Analysis of the larger data set will comprise descriptive statistics of the bereaved population, comparison of data between those who received and those who did not receive palliative care and/or other services, and correlations and hierarchical regressions to explore the individual, environmental and situational factors that are associated with the complications of bereavement (as measured by the PG-13). However, such analyses were not possible with the small sample of the pilot study.

While assistance was offered to complete the surveys, no-one took up the offer. Therefore, we need to acknowledge that the pilot study is limited in its assessment of the acceptability of the survey and recruitment methods for those for whom English is not the first language, and particularly those who do not have a good command of the English language, and that the categorisation of bereavement may not be equally applicable. The larger study would be better placed to examine cultural and religious differences in bereavement.

Much of what we know about bereavement care has emerged from research conducted in palliative care settings. In Australia, there are 130,000 deaths per year and 100,000 of these are expected, but only about 30% come to the attention of palliative care services.28 One retrospective cohort study in Western Australia showed that only about 60% of the population of people who died of deaths amenable to palliative care received palliative care services.29 Therefore, we know less about bereavement experiences and needs of people who did not use palliative care services following an expected death, and considerably less about the bereavement experiences of those for whom the death was unexpected. This remains the case following this pilot study, where all three vignettes involved a diagnosis of cancer and access to palliative care.

A much larger and a more representative sample is necessary to confirm the tentative trends of needs aligning with risks. As 76% of deaths are expected in Australia,28 and 70% were expected deaths in this pilot study, a public health approach to bereavement services in palliative care offers the foundation for determining the types of bereavement services and supports necessary for carers and families, in line with national and international palliative care policy.

Conclusion

This paper outlines a pilot study providing tentative empirical support for the public health model of bereavement support and provides the basis for a larger, population-based study of the experiences and support needs of bereaved people, the first of its kind nationally and internationally. Appropriate supports and services will ultimately reduce the risk of PGD, through reducing the risk of unmet support needs. Currently, there is a lack of clear evidence to guide development and allocation of bereavement programs in palliative care, including programs to develop community capacity. A larger survey will enable us to fill this gap and to determine how the support needs of each of the three groups of bereaved people should be serviced. This is of utmost importance for cost-effective and equitable resource allocation, and for understanding the contribution the community at large makes to bereavement support.

Acknowledgements

The research team acknowledges the financial support of the Palliative Care Network, WA Health Department, and the cooperation of three funeral providers who assisted with the recruitment of participants. We are grateful to the bereaved families who agreed to complete the survey, especially considering their difficult circumstances. Many thanks for Dr Moira O'Connor's contribution to the project and the reference group members who guided the development of this questionnaire. The second author is supported by the Australian Research Council (DE120101640).

Ancillary