Background
Most patients with life limiting diseases suffer from multiple co-occurring symptoms, like pain, fatigue, insomnia and depression, [
1,
2] with a potentially negative impact on functional and cognitive status, [
3] and quality of life [
4,
5]. Often, these symptoms and problems don’t occur isolated but in groups or symptom clusters which are defined as two or more concurrent symptoms in stable groups, distinct from other clusters [
2,
6]. The study of the prevalence and stability of symptom clusters is clinically essential to develop intervention strategies [
7] and to maintain and improve patients’ quality of life, as compared to single symptoms, symptom clusters worsen patients’ outcomes [
2].
Current studies focusing on symptom clusters of oncological patient groups with different life-limiting illness, [
4,
5,
7‐
11] identified multiple symptom clusters, like anxiety-depression, nausea-vomiting, nausea-appetite loss and fatigue- dyspnea-drowsiness-pain [
12]. Some studies evaluated symptom clusters in patients with advanced non-oncological diseases and also identified various symptom clusters, e.g. a study on patients with chronic kidney disease found two clusters: first, weakness, mouth problems, poor mobility, difficulty sleeping, feeling anxious, and feeling depressed, and second, nausea, vomiting, and diarrhoea [
13]. Another United Kingdom multicentre, cross-sectional study developed a symptom cluster model for patients with chronic obstructive pulmonary disease and identified three clusters, one respiratory related, one psychological and one cough-insomnia related symptom cluster [
14]. A German study on symptom clusters in inpatient palliative care settings found no significant differences, with five clusters for oncological and non-oncological patients, respectively. One cluster identified was nausea and vomiting, another was anxiety, tension and depression and slight variations in the distribution of the other clusters in symptoms and problems like weakness, tiredness, loss of appetite, and assistance with activities of daily living [
15]. Possibly, the pattern and progression of symptoms differ within diagnosis groups with different stages of disease [
16].
While most studies focused on inpatient specialist palliative care (SPC) settings, a substantial number of patients receive SPC in the home care setting [
17]. Therefore, the focus of the present study is on patients who were cared for by a specialist palliative home care (SPHC) team. To preserve patients’ autonomy and quality of life at the end of life through comprehensive support in pain management and symptom control is an important responsibility of SPHC teams [
18]. Pro-active symptom management is especially important in community based palliative care, as the care system can be easily destabilized in the home care setting [
19], which may result in unnecessary hospital admissions.
Emerging evidence suggests that very diverse symptom clusters occur in patients with a variety of chronic, oncological and non-oncological diagnoses, [
2] however, little is known about diagnosis-related differences in the prevalence of symptoms and problems, and the occurring symptom clusters of patients receiving SPHC. Identifying symptom clusters can support care planning, improve quality of care and facilitate better outcomes [
2]. Therefore, the aim of this study was to describe the prevalence of physical symptom burden and psychosocial problems of adult patients in SPHC, and to evaluate diagnosis-related symptom clusters.
Discussion
We present a national study of differences in symptom burden and psychosocial problems between oncological and non-oncological patients in SPHC, and their diagnosis-related symptom and problem clusters. In addition to the symptom and problem clusters already described for the inpatient setting, this is, to our knowledge, the first attempt to examine symptom and problem clusters in patients with advanced illnesses in the home care setting in Germany. The most prevalent symptom burden was weakness and poor mobility in both groups. While the oncological patient group had a higher number of burdensome symptoms, the burden related to symptoms and problems of non-oncological patients were higher.
We identified two symptom clusters for episodes of non-oncological patients: a psychosocial cluster and a physical functioning cluster. For episodes of oncological patients, three clusters were identified: a psychosocial cluster, a physical functioning cluster, and a communicational /practical cluster. These clusters represent the dimensions of palliative care as an adequate explanatory model and include the different levels of physical symptom burden and psychosocial concerns in the unit of care and, additionally, care organization. The main difference between symptom clusters of non-oncological and oncological episodes was the existence of the communicational/practical problem cluster in the latter. This does not mean that practical problems are not prevalent or do not occur in the non-oncological episodes, however, they did not have strong correlations with other symptoms or problems that form a cluster, in our data. Oncological patients with this cluster were younger and had a higher functional status compared to all oncological episodes. Psychosocial clusters of oncological and non-oncological episodes only differed in their cross loadings: while in non-oncological episodes ‘drowsiness’ often clusters with psychosocial symptoms, in oncological episodes they occurred with practical problems. Non-oncological patients had shorter episodes of care, with a significantly higher number ending with death. We can therefore assume that they were admitted to SPHC later, as seen in other studies, [
31‐
33] in contrast to their usual longer disease trajectories, [
34] but we do not know at what point patients died. Complementary to this, Just et al. (2021) identified a significantly reduced survival time of non-oncological patients in SPHC, with performance status and age as the most important predictors of low life expectancy [
31]. On average, these patients were older, and supporting services such as nursing services may have already been involved, possibly explaining the less prevalent practical matters of these patients.
Most other studies in patients with advanced cancer identified four common symptom clusters: anxiety–depression, nausea–vomiting, nausea–appetite loss, and fatigue–dyspnea-drowsiness–pain [
12]. In contrast to this, as well as studies of symptom clusters of non-oncological patients [
13‐
15], we could only identify the anxiety-depression cluster embedded in a broader psychosocial cluster. There may be different reasons, e.g. patient sample, assessment tools and statistical methods, as mentioned in Barsevick et al. (2006) [
16]. While almost all identified studies focused on patients in inpatient settings or outpatient clinics, our data describe patients in the community setting, indicating high overall symptom burden and patients being close to death [
17,
35]. Correspondingly, the physical symptom cluster includes mostly symptoms associated with physical decline and illness progression. In contrast to other studies [
10,
25,
36‐
39], pain and breathlessness as leading symptoms in palliative care did not occur in the identified symptom and problem clusters, neither in the oncological nor in the non-oncological patient group. Although both symptoms are common and frequently occurring symptoms in patients with advanced disease [
40], and were also prevalent in our data, they did not correlate with other symptoms in this analysis. This is possibly due to a discrepancy between clinically perceived and statistically identified symptom clusters [
36].
Regarding the assessment tools, most studies calculated clusters based on assessments of symptom severity or distress [
25,
38,
41,
42] e.g., using the Edmonton Symptom Assessment System [
43] (ESAS). So far, no study applied proxy-reported assessment tools that address symptom burden. The methodological impact on cluster occurrence and composition should be considered [
36]. Some severity assessments might not capture the whole range of symptoms that palliative care patients experience, which may result in underidentification of clusters [
12]. In contrast, IPOS contains 17 items including a broad range of aspects of palliative care, and covers patients’ burden of main symptoms, family distress as well as existential, spiritual and practical concerns. This allows identification of symptom clusters that are related to family distress and practical problems and cover all dimensions of SPC [
24]. Overall, our results for oncological episodes are in line with the main results of the IPOS validation study of Murtagh et al. (2019), which indicated three main subscales. Nevertheless, in contrast to Murtagh et al., our physical clusters display the symptoms that are associated with the progression of the disease and the proximity to death/dying process and not all physical symptoms. This may indicate that our patient sample differed from the sample in the validation study with regards to disease progression. Also, the communicational/practical cluster is missing in the non-oncological episodes. This may be due to the small number of non-oncological patients (15%) included in the validation study, leading to an underrepresentation of this group. Furthermore, no setting-specific differences were taken into account, which could explain specific clusters for patients in the community setting in our results [
24].
Moreover, different methods for the statistical calculation of symptom clusters are used in various other studies, most commonly principal component analysis, EFA, and hierarchical cluster analysis, which tend to identify different clusters [
12,
41,
44,
45]. Only anxiety and depression seem to occur within the same cluster regardless of statistical analysis [
25]. In order to develop robust cluster models, a CFA is useful because it is not only based on visual evaluation of rotated factor loadings as in EFA, but on a goodness-of-fit test which quantifies how well the data actually fit the hypothesized model [
46].
Family distress was identified as a relevant problem which occurs simultaneously with other psychosocial problems. Caregivers are an integral part of the unit of care but also provide most of the care at home. They take a high level of responsibility regarding practical and medical assistance, as well as emotional support. Therefore, this topic and its associated problems like patient anxiety and depression are of key importance for SPHC teams, and it is crucial that reliable, well-coordinated professional support is provided to caregivers [
47].
Overall, to provide good and adequate care to people who are at the end of their life, it is necessary to understand how their symptoms might progress and to develop care plans accordingly. Especially in non-oncological patients, the trajectories are often more individual with unpredictable exacerbations, [
34] and may differ from oncological patients with respect to symptom-related therapies [
33]. The results demonstrate that symptom burden is perceived differently by various patient groups. SPHC teams must therefore take into account and anticipate the individual needs of patients reflected in the different manifestations of the symptom clusters. The psychosocial burden regarding both patient and family was similar in both groups, but more associated with physical issues in the non-oncological patients and with practical problems in the oncological group, which indicates the high relevance of coordination of care in a multidisciplinary SPHC team.
Strengths and limitations
The main strength of this analysis lies in the prospectively collected national data set of oncological and non-oncological patients in SPHC using proxy assessments. Although there is evidence that some of the experienced symptoms and problems are perceived differently by patients and professionals, [
24,
48] the use of proxy assessments allowed us to include all patients admitted to SPHC without exclusion due to physical decline, cognitive impairment, or inability to consent. Data reliability was maximized through SPHC staff training in application of the assessments, regular feedback, and plausibility checks during data collection, which reduced data bias and missing data.
However, the presented symptom and problem clusters have been developed based on SPHC data, hence cannot be transferred to other, especially inpatient settings, without further evaluation. Further limitations are the restriction to complete cases in the factor analyses, where not assessable items were treated like missing values. Although there were only few missing values, there were many symptoms and problems that could not be assessed, especially in non-oncological episodes and regarding psychosocial problems. For these values, we cannot reliably determine the reasons why they could not be assessed and if they occur at random, and therefore an imputation of values was not feasible. Furthermore, there were considerably more episodes from patients with oncological diseases in total. Due to the small proportion of non-oncological episodes, we did not distinguish between different diseases or disease groups. Although episodes of non-oncological patients were underrepresented, this does not impact the overall aim of achieving a better understanding of the differences between these groups in SPHC. We revealed diagnosis-related differences in the prevalence of symptom burden and psychosocial problems and the corresponding clusters, but we did not aim to explore their consistency [
42] or stability [
12] over time. Especially for non-oncological patients, further analysis is needed regarding their symptom and problem trajectories in SPHC. It should also be examined if symptom clusters worsen outcomes (e.g., unwanted hospitalizations, symptom relief, caregiver satisfaction) or impact the resources required in SPHC.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.