Comparison of the experience of caregiving at end of life or in hastened death: a narrative synthesis review

The review question, “What are the experiences of family and friends providing care at home for a person at the end of life or in the context of the patient’s hastened death?” can be broken into clearly defined population, exposure, context, outcome, and study design (PECOS) criteria [23, 24] listed in Table 1.

The review question further may be broken down into a series of subquestions that align with Popay’s steps as follows:

The review included two sets of searches of Medline, CINAHL, Web of Science, and PsycINFO — one for general end-of-life caregiving and one for hastened death. The Boolean search terms are described in Table 2.

For parity between the two sets of data, the searches were limited to studies published between 1998, the year medical aid in dying was legalised in Oregon, the first U.S. jurisdiction to explicitly allow it, and April 2020. Searches were limited to peer-reviewed literature published in English involving human subjects. Additional studies were identified through citation tracking in relevant systematic reviews identified in the search process and in studies selected for inclusion.

Predefined inclusion and exclusion criteria (Table 3) guided title and abstract review of the initial results of each search and were the same for both searches. In addition, full-text searching omitted studies in which caregivers’ experiences could not be separated from those of patients or professionals, or studies in which current and former caregivers’ experiences were interwoven.

All studies selected for full text review were reviewed using the Relevance, Appropriateness, Transparency, and Soundness (RATS) Quality/Appropriateness Appraisal Tool [25] to identify studies with limitations, such as unspecified recruitment or analysis methods that could warrant concern about the validity of the findings.

Using Popay’s [22] narrative synthesis approach, participant narratives and author analysis in all studies in both searches were coded first in NVivo (QSR International) using thematic analysis to identify codes that fit within a priori themes aligned with elements of Hudson’s model (such as appraisal, coping, and event outcome, See Appendix 1: A Priori Codes), and subsequently using inductive codes representing concepts not found in the model (See Fig. 1).

Each code was analysed separately in each data set, employing subcodes where needed to clarify multiple concepts (for example, a favourable resolution could be getting needed services or the patient having a peaceful death). The two pools of studies were then synthesised individually within each theme of Hudson’s model as well as themes constructed outside the model. The two synthesised data sets then were analysed side by side to identify commonalities or differences. Where particularly illustrative, quotes are included.

Of the 34 end-of-life caregiving studies, six were from Australia, seven from Canada, five from the U.S., four from the United Kingdom, 10 from elsewhere in Europe, one from Japan, and one from New Zealand (Table 4). Among the 12 hastened death studies, five were from the U.S., four from the Netherlands, two from Switzerland, and one from Canada (Table 5). Across both sets, cancer was a frequent cause of patient death, along with motor neurone diseases. Tables 4-5 list studies included in the syntheses; findings from studies are presented according to components of Hudson’s model in Tables 6-10.

In Hudson’s model, the process of appraisal, coping, and resolution begins with identification of an event. Caregiving at end of life was both a single overarching event and the sum of many smaller events. However, for end-of-life caregivers, events focused on changes in patients’ needs, whereas in hastened death, events primarily followed a predictable pattern of planning, preparation, orchestrating the death, and tying up loose ends. The results of the synthesis are presented in the context of Hudson’s model from appraisal through outcome, followed by influencing factors, and lastly by inductive themes not represented within the model.

Hudson lists 18 variables that can influence caregivers’ experience (see Appendix 1 for definitions). Although each is distinct and based on other research or conceptual models, they can be broadly clustered as:

Caregiver age, gender, and socioeconomic status were excluded from this analysis because they were not possible to tease apart in a synthesis of multiple published works.

End-of-life caregiving studies had ability-related codes more often than hastened death studies, possibly because the duration of end-of-life caregiving facilitated learning new skills or gaining confidence in abilities (Table 10). Hastened death caregiving, by contrast, was a one-time process with few steps repeated and little precedent. End-of-life caregivers frequently described exhaustion and mentioned the value of respite, but hastened death caregivers did not, perhaps because of the shorter timeframe or a choice to defer their own needs until after the death.

Many influencing factors could be positive or negative. Social isolation and lack of information were stressful for both end-of-life and hastened death caregivers. Meeting the patient’s wishes was related to satisfaction in both groups, while being unable to meet expectations for care was stressful. Hastened death caregivers, particularly in the U.S. and Canada, often described preparing for the death as communal, and the death itself as sacred or beautiful, whilst those in Switzerland were more likely to describe fear of stigma if the cause of death were widely known.

Most factors identified in Hudson’s model could either enhance caregivers’ experience or create additional challenge. For example, patients who spoke openly about their own decline and impending death relieved caregivers of feeling a need to protect the patient from the truth or hide their own acceptance of the coming death; end-of-life caregivers for patients in denial expressed distress about not wanting to dash the patient’s hopes.

Healthcare professionals are not listed as an influencing factor in Hudson’s model, but their role is a frequent theme in caregiving studies, either as sources of support and information or representing failures of the health system to adequately respond to patient and caregiver needs (Table 11).

Beyond the themes outlined in Hudson’s model, other internal and external factors appear to affect caregivers’ experiences (Table 12). The structure of healthcare, nationally or locally, affects whether homecare services or hospice is available, whether specialised care for conditions such as motor neurone disease is available, and whether patients and caregivers can readily find out about services for which they are eligible. Costs of medication and equipment also can add to caregivers’ burden. For hastened death caregivers, whether hastened death was legal and whether information and support were available affected moral distress and preparedness to facilitate a comfortable death.

Caregivers reported different fundamental motivations for providing care. In addition to cultural norms and a desire for closeness at the end of the patient’s life, some end-of-life caregivers also expressed distrust of the healthcare system, particularly hospitals, as motivation to care for the patient at home. Whilst some caregivers saw the hospital as a fallback solution if the patient’s needs became too great, others saw the potential of sending the patient to the hospital as a sign that they had failed at caregiving.

Finally, grief affects caregiving at the end of life. Anticipatory grief was common among end-of-life caregivers. On the one hand, coming to terms with the patient’s impending death was associated with easier resolution of grief after death. On the other hand, the weight of anticipatory grief could lead caregivers to shut down their emotions or to seek distraction in the form of tasks. In this respect, grief might affect whether caregivers take a problem- or emotion-focused approach to events in either group.

Healthcare professionals play a major role in caregivers’ experience at end of life. Professionals provide knowledge, teach skills, take decision making pressure off the shoulders of caregivers, offer support and validation, and can be a gateway to resources [3, 6, 7]. When healthcare professionals are unavailable, do not fulfill promises to take measures to relieve patient suffering, or do not support caregivers’ assessment that the care is too much to handle, caregivers often report feeling isolated. Meta-analyses of caregiver studies noted that across many studies, caregivers expected health professionals to take responsibility for developing a trusting, supportive relationship with families [3, 6]. For some end-of-life caregivers, the regular presence of hospice staff is a welcome, regular break in caregiving, and its loss is felt after the patient’s death. In hastened death contexts, professionals’ legal ability or personal willingness to discuss the patient’s wishes and options, and provide practical support, contributed to caregivers’ reduced moral distress and increased satisfaction that the patient’s wishes could be achieved. The role of professionals is not highlighted in Hudson’s model but might fall into either social support or information.

Whilst some caregivers reported having their needs anticipated well and addressed, others reported isolation, stress, and in some cases financial strain as the patients’ needs outstripped the support structures available [3]. For example, family caregivers for patients with motor neurone disease in Australia reported difficulty accessing community palliative care services or support adequate to the increasing demands of the disease [29]. These structural issues are distinct from the availability or attitude of individual health professionals and may be more relevant in countries with limited or inequitable healthcare infrastructure as opposed to national health coverage [11]. Categorised unmet needs in studies of palliative care patients receiving services at home included transportation, equipment, caregiving support, and respite, in addition to adequate communication and information from professionals. However, a systematic review of quantative studies of caregiver experience found a lack of consistent, high-quality evidence that specific services and programmes improve caregiver outcomes [13].

Acceptance and anticipation of patient’s death appears related to having less grief before and after death. In hastened death studies, caregivers are actively working toward the patient’s goal of a peaceful death, whereas some end-of-life caregivers are unprepared or surprised by the death. Hudson’s model is not end-of-life-specific, but grief may be a relevant factor for caregivers when death is likely. Broady’s [8] scoping review of caregiver literature notes that anticipatory grief may encompass awareness of both the patient’s impending death and the change in identity, away from caregiver, that will follow.

The analyses reveal similarities and differences between caregivers’ experiences in end of life and hastened death settings. Across studies, caregivers commonly sought closeness with the patient and reported satisfaction at having done their best to care for the patient in a critical time. Hastened death caregivers were more consistently reconciled to the patient’s death and the belief that death was preferable to anticipated suffering. Some but not all end-of-life caregivers reached this conclusion. However, the deliberate nature of hastened death may mean that patients choose likeminded caregivers more deliberately than in situations where caregiving may not be expected to lead to death [14]. Unlike in studies of end-of-life caregivers, physical exhaustion and burden were not commonly reported among hastened death caregivers [14], possibly because their scope of preparing for hastened death is finite.

This review is the first to apply Hudson’s model as a lens for synthesising literature on the experience of caregivers at end of life. As such, the review also evaluates the limits of Hudson’s model and identifies potential refinements, such as the role of healthcare professionals as an influencing factor, that could strengthen it.

A major limitation of synthesising qualitative studies against such a model is that they may use other analytic models that may emphasise different aspects of caregiving. Further, because each study represents data synthesised from participants by the authors, salient aspects of Hudson’s model, such as the appraisal of benign caregiving events, may have fallen out of the earlier published work in favour of events that better aligned with the authors’ own theoretical underpinnings. Applying Hudson’s model against a full set of original data may better illuminate its strengths and weaknesses.

Analysis is further limited by the unknown degree to which studies have accurately represented the experiences of participants; for example, whether caregivers in hastened death shared completely with interviewers their emotional response to the death. Limiting inclusion to English-language publications reduces the comprehensiveness of the hastened death analysis. Research on hastened death in the Netherlands, Belgium, Switzerland and other countries may be published in local languages. Future analyses could take additional steps to secure multilingual sources, including soliciting articles from other scholars in the field.