Findings and comparison with the literature
In care-dependent people initially receiving home care, 57.5% died within 3.5 years. Overall, 36.9% died in hospital and in-hospital deaths were found most often in those still receiving home care as well as care in semi-residential arrangements at the time of death. People who died in hospital were younger and had lower care dependency compared to all other analysed care settings. More than half of home care receiving people moved to another (care-) setting before death (44.0% either to long- or short-term care in nursing home, 4.4% to semi-residential arrangements, plus 3.5% to shared housing arrangements).
Actual place of death
Nearly 37% of all deaths took place in hospitals, which is the most common place of death in our care receiving cohort as well as in the total German population [20–22] and in those of most other countries [23–25].
After the hospital, the own home was found as the second most common place of death with 26.6%, which was also found by Dasch et al. (21.3%) for 2017 although their representative German cohort of all persons dying was in mean 77.6 years old, almost 10 years younger than our care receiving cohort [22]. Moreover, Herbst et al. analysed two random samples of German death certificates from 2007 and from 2017. They showed that while in 2007 home also was the second most frequent place of death (26.1%), it slid to third place in 2017 with 19.8% [20]. Taking results of published studies together, the likelihood to die at home has been decreasing since recent years [20, 22, 26]. In a review of 1998, Grande et al. already investigated the relation between patient characteristics and home deaths [27]. They found out that improved access to home care is likely to increase home deaths for older people [27]. Especially palliative home care and hospice care are associated with fewer hospitalisations and more home deaths [28]. But there are several more potential factors influencing death at home, for example patients functional status, their preferences, living with relatives, and extended family support [29].
The present cohort showed that 23.5% died in nursing homes, which is a little higher than found in the younger-aged representative sample by Dasch et al. (20.4%) but lower than in the random sample of Herbst et al. with 27.1% [20, 30]. The older the people, the higher the probability to die in a nursing home instead of a hospital [31]. Also in our study nursing home residents receiving long-term care were the oldest group.
Shared housing arrangements were not considered in previous studies investigating place of death, even if this setting can be seen as an increasingly used, familiar care alternative for long-term nursing homes in Germany [15, 33, 34]. To the author’s knowledge there is also no data on the frequency of transitions to short-term care before death as well as nursing home as place of death for short-term care recipients. Studies based on German death certificate data cannot include care information, because they are not routinely covered in these documents. Our study shows that both settings are of relevance and should be included in future studies in EOL care. Quantification and deeper understanding of all possible care transitions at the end of life are important to estimate the relevance and trends for place of death from a public health perspective. Even if the hospice as place of death is still rare and unfortunately not covered in the present data, it has been shown to increase as place of death in Germany in recent years [22].
Hospital death by care setting
As in the present cohort, the proportion of in-hospital death in older, care receiving people seems to be smaller compared to the general population [25]. The availability of formal versus informal care seems to influence hospital death rates. Kaspers et al. found out that older people receiving informal care were more likely to die in hospital than people receiving formal home care or institutional care [35]. Klinkenberg et al. also showed for Dutch people who only received informal care in their last three months of life that the odds of dying in a hospital was much higher compared to those who received a combination of formal and informal home care [36]. In the present cohort, the proportion of in-hospital death was also highest in people receiving home care (44.7%), where the largest proportion of informal care can certainly be found. The proportion of in-hospital deaths was lower in our group of nursing home residents receiving long-term care. Although this proportion goes in line with previous German analyses [37, 38], it is, however, internationally compared somewhat higher [39]. Nevertheless, the proportion of in-hospital deaths among nursing-home residents internationally varies markedly even within countries with an overall median of 22.6% [39].
There are other possible factors influencing the risk of dying in hospital for elderly, care receiving people like the care level, age and sex. In our cohort, the younger the people and the lower the care level, the higher was the proportion of in-hospital death. The same was found by Menec et al. and other previous studies [10, 39, 40]. This could possibly partly explain, why men in our cohort were a little more likely to die in hospital than women. However, there is increasing evidence of “real” sex-specific differences in burdensome interventions like transitions of care or invasive procedures during EOL and future studies should put more emphasis on sex-specific analyses [41].
Moving to other care settings before death
Our result regarding the frequent transition from home to other care settings before death indicates that home care cannot always be maintained until the EOL, although most patients wish so [5–8]. It was already shown in international studies that the frequency of care setting transitions of elderly people increases near to death [42]. For example, Kaspers et al. also found that nearly half of their 55–85 years old home-living sample in the Netherlands was transferred between care settings one or more times in the last 3 months of life, mostly from home to hospital [35]. Care setting transitions at the EOL are seen as increasingly problematic, also because of potential medication and care errors, disrupting care teams, and a loss of care information [20, 27, 31], even if these transitions have the potential to be a relief for family caregivers [23]. Looking at the German situation, it also should be mentioned that structures in outpatient palliative care have been introduced just within the last 15 years and the growing number of general and specialist outpatient palliative care services (AAPV and SAPV) provides more possibilities of outpatient palliative care since the last years [43], which can strengthen the quality of care at home at the patient’s EOL. Therewith unwanted care transitions can also be prevented. However, the care at home should not automatically be equated with the best care [9, 44] because institutionalised palliative care like hospice care and in-hospital palliative care can improve the quality of dying and death [45, 46]. Overall, care decisions always should be weighed individually to enable appropriate and timely care setting transitions in accordance with individualised EOL care needs [47].
There are different indicators already mentioned being associated with a risk of care transitions. Perrels et al. showed in their population-based, retrospective cohort study in the UK, that people with severe cognitive impairment were the most likely group to move to other care settings [42]. Our results also show that the people with dementia more often died in another care setting than home. Stiefler et al. analysed predictors of admission to nursing home in care dependent people based on longitudinal secondary data and also found dementia, cognitive impairment, cancer of the brain and higher age as risk factors, which goes in line with our results [48].
Strengths and Limitations
The strengths of this study are its real-world character, its large sample size which allowed us to stratify the analyses by sex, age and other variables. Furthermore, we had valid information on care setting pathways and place of death. Just like the strengths, the limitations are based to the nature of the administrative data from LTCI funds. The data were not captured for the purpose of scientific research and further information that could influence the placement and dying in different care settings (clinical data, socioeconomic status, marital status or family support, respectively) were not available. The same applies to further information related to the specific care recipient’s institution, e.g. staffing ratios or the nursing home’s ownership. For the ones living in semi-residential arrangements we were not able to differentiate between the care-recipients died at home or during day or night care, respectively. Besides, our data did not contain palliative care units and hospices as places of death. However, their joint proportion on places of death in Germany is 11% [22]. Another limitation relates to the fact that data for this study were only obtained from one health insurance fund. Since the DAK-Gesundheit insures more women and a population with a generally poorer health status [49], our results cannot be extrapolated to the entire care receiving population in Germany. Nevertheless, the DAK-Gesundheit is with 5.6 million insured persons one of Germany's largest health insurance funds [13].