End‐of‐life care of nursing home residents: A survey among general practitioners in northwestern Germany

To describe general practitioners' (GPs) perspectives on end‐of‐life care of nursing home residents.


Introduction
With an aging population, in many Western countries, nursing homes are becoming an increasingly important place in which many older people spend their last days of life and eventually die. In Germany, for instance, the proportion of older people dying in nursing homes increased to 20% in recent years. 1 Similar trends have been identified in other countries. 2,3 Previous research showed that the hospitalization rates of nursing home residents (NHR) are the highest near the end of life, 4 although a large proportion of such hospitalizations are deemed inappropriate, are distressing for patients and often result in negative health outcomes. 5 In many cases, treating residents in their familiar environment of the nursing home might more adequately correspond to residents' wishes. 6 Nevertheless, in Germany, approximately 30% of NHR die in hospital instead of the nursing home. 7 Advance care planning (ACP) or the presence of explicit advance directives (AD) might contribute to a reduction in the rate of hospitalization, and have been shown to impact the quality of end-of-life care. 8,9 However, in Germany, ACP, which aims to define a person's wishes regarding future medical treatment decisions in the case of physical or mental deterioration, 10 is still uncommon in nursing homes. Furthermore, AD are completed only by a minority of residents and are often difficult to interpret in complex situations. 11 The available international evidence also suggests that inappropriate hospital admissions at the end of life could be reduced by palliative care consultations. 12 In Germany, patients have free choice of physicians in outpatient care. 13 The majority have their own general practitioner (GP), who usually continues providing medical care after the patient has entered a nursing home. However, palliative medicine is not an obligatory part of the training pathway for GPs in Germany. 14 Almost all NHR are regularly visited by their GP. 15 GPs are often involved in the decision whether or not to transfer a NHR to a hospital, and play an important role in end-of-life care. 16,17 Their perspectives can therefore be considered essential in examining the end-of-life care of NHR, but, to the best of our knowledge, research on this is rare.
The aim of the present study was to assess GPs' perspectives on end-of-life care and hospitalization among NHR, and to identify areas for improvement.

Study design
the Association of Statutory Health Insurance Physicians (including general internists working in primary care; approximately n = 5500) in the German federal states of Bremen and Lower Saxony. In August 2018, GPs were invited by postal letter to complete a paperbased questionnaire. A reminder was sent after 2 weeks. All data were collected anonymously.
Several strategies shown by a Cochrane review to increase response to postal surveys were applied. 18 Those include a short questionnaire, follow-up contact, providing a second copy of the questionnaire at follow up, personalized letters, prepaid return envelopes and academic origin of the study.

Questionnaire
A multidisciplinary research team developed the four-page questionnaire on medical care provision in nursing homes, hospital admissions, emergency department visits and end-of-life care.
On the topic of end-of-life care, we first asked for the estimated proportion of NHR who have completed an AD. Furthermore, the GPs were asked to judge how often these directives include information regarding hospital transfers at the end of life and how often these directives are not taken into account. The second part contained five questions addressing current practices and potential deficits of end-of-life care on a 5-point Likert scale ranging from "0 = strongly disagree" to "4 = strongly agree". GPs were also asked to estimate the proportion of NHR dying in hospital, and whether this proportion has changed over the past 10 years ("decreased", "remained unchanged", "increased"). Another question regarded the overall rating of end-of-life care in nursing homes ("rather poor", "rather good"), and if rated as "rather poor", GPs were invited to state what they considered the most important measure to improve end-of-life care (free text).
Furthermore, the following characteristics of the GPs were requested: age, sex, number of years working as a GP, practice type (single practice, group practice/medical care center) and location (rural ≤20 000, semi-urban >20 000-≤ 100 000, urban >100 000 inhabitants), number of nursing homes visited, number of residents cared for and whether the GP had a qualification in palliative medicine. In Germany, to obtain a qualification in palliative medicine as a subspecialty, physicians have to attend a 12-month training program in palliative care or 120 h of case seminars in addition to a theory course. 19 The questionnaire was pretested and discussed with GPs before data collection was started.

Statistical analysis
Descriptive measures were computed. Frequencies were calculated for categorical data, and continuous data are presented as mean with standard deviation (SD). Responses relating to the perspectives on current practices and deficits of end-of-life care were compared between GPs with and without a qualification in palliative medicine using χ 2 -tests. Free-text answers were summarized to inductively generated categories based on the given answers (measures for improvement). This was carried out independently by two authors and any disagreement was resolved by discussion. The categories were presented as frequencies. We did not exclude questionnaires due to missing values. However, each analysis was restricted to individuals with no missing values for the respective question (i.e. the number of included questionnaires differs).
Multivariable logistic regression was used to estimate odds ratios (OR) and 95% confidence intervals (CI) to identify factors associated with rating end-of-life care as "rather poor". The logistic regression model included sex, years working as a GP, practice type, practice location, number of residents cared for and qualification in palliative medicine. Respondents' age and number of nursing homes visited were not included, as they correlate closely with years working as a GP and the number of NHR cared for, respectively. A statistically significant difference was stated for Pvalues <0.05.
All analyses were carried out using SAS for Windows version 9.4 (SAS Institute, Cary, NC, USA).
This study received a waiver from the local medical ethics committee of the Carl von Ossietzky University of Oldenburg (no. 2018-080).

Respondents
Of 1121 mailed questionnaires, 375 were returned (response rate 33.5%). The mean age of the participants was 54.4 years, and 57.6% were men ( Table 1). Most participants worked in group practices or medical care centers (67.0%), and more than half were located in rural areas (52.3%). A total of 24.3% of GPs had a qualification in palliative medicine. The mean number of GPs' patients living in nursing homes was 46.8 (SD 43.5, range 0-360), and the respondents were visiting a mean of 4.1 (SD 2.2, range 0-15) nursing homes. Advance directives and hospitalization at the end of life The mean proportion of NHR dying in hospital was estimated to be 43.1% (SD 20.1%). When asked how the proportion of inhospital deaths changed over the past 10 years, 31.4% of GPs answered that this proportion decreased, 26.2% answered that it remained unchanged, whereas 42.4% thought there was an increase of in-hospital deaths (Table S1). GPs estimated the average proportion of NHR with an AD to be 36.8% (SD 20.7%). When asked how often these directives include information regarding hospital transfers at the end of life, the reported mean proportion was 31.9% (SD 28.5%). Respondents estimated the proportion of AD that were not taken into account in correspondence with the patient's stated wishes to be 40.3% (SD 27.0%; Table S1).

End-of-life care
Overall, 71.1% answered that NHR at the end of life are treated in hospital too often. A much smaller proportion agreed that residents should be enrolled more frequently in specialized outpatient palliative care (SAPV) (40.9%). Most physicians (81.8%) agreed that nursing staff should be better trained in end-of-life care. More than half of the respondents (56.5%) thought that GPs are generally well trained to care for residents at the end of life (Fig. 1). A total of 41.9% of respondents agreed that GPs should be available for end-of-life care outside of office hours, whereas 39.5% disagreed.
Overall, more than half of all respondents (53.8%) rated end-oflife care of NHR as "rather poor" (Table S1). In the multivariable regression, we found that GPs who had a qualification in palliative medicine were significantly more likely to rate end-of-life care as "rather poor" (OR 1.89, 95% CI 1.10-3.23; Table 2). A statistically significant association was also found for being female and working as a GP for <10 years, and a more negative rating of end-of-life care. All other characteristics were not statistically significant.

Measures for improvement
Of those 193 GPs who rated end-of-life care of NHR as "rather poor", 292 free-text responses regarding the most important measures for improvement were received. From these, we constructed 16 categories. The three measures stated as most important were a higher nurse staffing ratio (33.6%), as well as better qualification for nursing staff including more training in palliative care (30.5%), followed by further strengthening palliative care structures (11.3%). Other aspects mentioned included, for example, better communication and increased adoption of AD (Table S2).

Comparison by qualification in palliative medicine
As already shown by the multivariable regression, GPs with a qualification in palliative medicine were more likely to rate endof-life care of NHR as "rather poor" (64.0% vs 50.5%). They also considered more frequently the proportion of hospitalizations at the end of life as too high (80.2% vs 68.2%). More GPs with a qualification in palliative medicine agreed that more residents Figure 1 Perspectives of general practitioners on the status quo and deficits of end-of-life care. "Agreed" and "strongly agreed" are presented together as agreeing.
should be enrolled in specialized outpatient palliative care, and that GPs should also be available to provide end-of-life care outside office hours (Table 3).
There were only minor differences between GPs with palliative medicine qualifications and those without such qualifications in their reporting of the proportion of NHR dying in hospital (with a mean of 41.6% vs 43.6%), as well as of residents having an AD (with a mean of 37.7% vs 36.6%).

Discussion
Our survey among GPs in northwestern Germany shows some variation in the perspectives of end-of-life care depending on a qualification in palliative medicine. Overall, more than half of respondents rated end-of-life care in nursing homes as "rather poor". Approximately three-quarters agreed that residents are treated too often in hospital at the end of life. For both of these questions, GPs with a qualification in palliative medicine tended to rate the care of NHR at the end of life more negatively. GPs considered the most important measures to improve end-of-life care to be a higher staff-to-resident ratio, as well as more highlyskilled nursing staff. Furthermore, AD are frequently regarded as not adequately valid in terms of end-of-life hospitalizations.
The proportion of NHR who had completed an AD was estimated to be 36.7%. This corresponds to the findings of another recent study interviewing residents of five nursing homes in southern Germany. 6 Although two other German studies reported considerably lower proportions, 11,20 a further recent study of primary care patients aged ≥85 years including community-dwelling older adults and NHR suggested a much higher dissemination of AD at almost 70%. 21 However, it was shown that most existing AD are inadequate or not sufficiently informative for decisionmaking in specific scenarios. 11 This supports our finding that GPs estimated just 30% of all AD include information regarding hospital admissions at the end of life. The present study also suggests a high proportion of non-compliance with wishes stated in AD in German nursing homes. This issue has previously been raised in the literature. 11 Taking the results of the present study together, only approximately every 10th resident seems to have an AD with valid information on preferences about end-of-life hospitalizations.
In order to increase the availability of valid AD and therefore improve adherence to residents' wishes, ACP might be of great benefit. It has been shown that ACP leads to increased adoption of AD in nursing homes. 20 Also, some of the GPs in our survey mentioned strengthening the residents' wishes and the presence of AD as the most important measure to improve end-of-life care. However, it is important to bear in mind that ACP is inherently a process of discussing in advance a person's preferences regarding future treatment decisions in the case of physical or mental deterioration. 10 As dementia is an important reason for transitions to a nursing home and more than half of residents suffer from dementia, it is clear that this communication process should begin before institutionalization. 22 Overall, the majority of GPs (71%) agreed that NHR at the end of life are hospitalized too often. In Germany, almost 30% of NHR die in hospital. 4 In the present study, GPs perceived this proportion as even higher and indicated this as a major problem. When compared with other Western countries, end-of-life hospitalizations of NHR occur more frequently in Germany. 23 For instance, a Dutch study found that just 1.5% of NHR with various

P-value
Residents are too often treated in hospital at the end of life (n = 374) † Strongly disagree/disagree (0-1) 7.7% 13.4% 0.0864 Neither agree nor disagree (2) 12.1% 18.4% Agree/strongly agree (3)(4) 80.2% 68.2% Residents should be enrolled more frequently in specialized outpatient palliative care (SAPV) (n = 374) † Strongly disagree/ disagree (0-1) 20.9% 37.1% 0.0161 Neither agree nor disagree (2) 29.7% 24.7% Agree/strongly agree (3)(4) 49.5% 38.2% Nursing staff should be better trained in end-of-life care (n = 374) † Strongly disagree/disagree (0-1) 1.1% 6.7% 0.0199 Neither agree nor disagree (2) 7.7% 14.5% Agree/strongly agree (3)(4) 91 stages of dementia were hospitalized in the last week of life. 24 Although <5% had an AD, nearly all had a palliative care goal on the day of death, which means that in the Netherlands, physicians often withhold potentially burdensome treatments for those residents, irrespective of an AD. However, in the Netherlands, specially trained elderly care physicians are employed by nursing homes, with approximately one full-time physician per 100 residents and palliative care is a focus of their training. 25 These findings suggest that variations in end-of-life hospitalizations between countries might be explained not only by the presence of AD or palliative care goals, but also by the respective structures (e.g. availability and training of physicians) in the different healthcare systems, as well as by family and cultural preferences. However, further investigation is required on this question. 23 Approximately 40% of the respondents in our survey thought that GPs should be available for end-of-life care outside office hours. This is in line with another survey carried out in Lower Saxony that reported that 45% of GPs indicated they are available around the clock to their palliative patients. 26 In the Netherlands, this proportion is even higher, with 60% and 86% of GPs available outside office hours for their terminally ill patients. 27,28 Interestingly, GPs with a qualification in palliative medicine were more likely to agree with this statement in our survey, indicating a higher awareness of this problem.
GPs with a qualification in palliative medicine were also more likely to rate end-of-life care as "rather poor". This underlines that these physicians are more critical, which is in line with findings of Reyniers et al. showing that GPs with formal palliative care training more frequently labeled terminal hospital admissions as potentially avoidable. 29 To ensure adequate end-of-life care, NHR therefore might benefit from further integration of palliative care into GP training schemes. Furthermore, in the present study, almost half of all respondents thought that GPs are not well trained for end-of-life care, and an even higher proportion thought the same for nurses. This is underlined by a survey across six European countries, which concluded that the knowledge of nursing staff on palliative care needs to be improved. 30 Overall, better education on palliative care seems to be essential to improve end-of-life care.
A strength is that we were able to analyze nearly 400 GPs, which was also our originally calculated sample size. It might be suggested that a response of 33.5% could point to selection bias, and that GPs with an interest in palliative care might have been overrepresented. At 24%, the proportion of GPs qualified in palliative medicine was comparatively high in the present study. In contrast, our response is quite comparable and in the upper range of other recently published surveys of German GPs. Furthermore, compared with the gross sample, respondents did not differ with respect to sex (with 60.1% vs 57.6% men). We also applied several strategies shown by a Cochrane review to increase response to postal surveys. 18 Some of the present results (e.g. on the proportion of NHR having AD) have to be interpreted with caution, as they are solely based on GPs' own estimation and more valid methods for estimation might exist. However, our study aim was to assess GPs' views on end-of-life care of NHR.
Overall, approximately seven out of 10 GPs considered that NHR are too often hospitalized during end of life, and more than half rated end-of-life care as "rather poor". GPs with a qualification in palliative medicine were even more critical. The fact that end-of-life hospitalizations of NHR are more common in Germany than in other Western countries, where palliative care is more widely available, underlines that healthcare professionals should critically assess when palliative approaches are required. To do so, GPs and nursing staff require more training in palliative care. In addition, ACP can help to ensure that residents' wishes regarding their end of life are respected.