Introduction
Lung cancer remains one of the leading causes of cancer-related mortality and has historically been associated with a poor prognosis and limited survival (Siegel et al. Reference Siegel, Giaquinto and Jemal2024). Lung cancer patients often experience a high burden of symptoms, treatment-related side effects, and significant psychological stress – all of which severely impact quality of life (QoL) (Tjong et al. Reference Tjong, Doherty and Tan2021). Incorporating palliative care into standard oncology care has proven beneficial in the treatment of these patients, with studies validating its impact on psychosocial support, management of symptoms, and improvement of QoL (Temel et al. Reference Temel, Greer and Muzikansky2010; Hall et al. Reference Hall, Singhal and Dickerson2019; O’Sullivan et al. Reference O’Sullivan, Conroy and Power2022). While new treatments, such as immune checkpoint inhibitors (ICI), have considerably improved survival rates and the prognosis of patients diagnosed with advanced lung cancer, most patients still die from cancer-specific causes (Chaft et al. Reference Chaft, Rimner and Weder2021). Because of this, prioritizing QoL in clinical decision-making and supportive care strategies remains important (Kelly et al. Reference Kelly, O’Brien and Hannigan2022; O’Sullivan et al. Reference O’Sullivan, Conroy and Power2022).
Studies have found that when discussing palliative care, including end-of-life care, patients consider place of death (POD) an important issue that highly impacts QoL in the final stages of life (Starr et al. Reference Starr, Ulrich and Corey2019; Fereidouni et al. Reference Fereidouni, Rassouli and Salesi2021). Among patients and families with life-threatening illnesses, home was found to be the most preferred POD, as it was associated with greater dignity and autonomy as well as the possibility of being surrounded by loved ones (Pinto et al. Reference Pinto, Lopes and de Sousa2024). Similar findings have been documented in research involving cancer patients (Gomes et al. Reference Gomes, Higginson and Calanzani2012; Higginson et al. Reference Higginson, Daveson and Morrison2017; Gutiérrez-Sánchez et al. Reference Gutiérrez-Sánchez, Gómez-García and Roselló2021). In addition, passing away in a hospital department other than the palliative care unit (PCU) was the least preferred option and associated with the poorest QoL at the end of life (Higginson et al. Reference Higginson, Daveson and Morrison2017). These findings highlight the importance of exploring factors related to POD in real-world lung cancer patients, as knowledge of this could help identify patients at greater risk of dying in a hospital ward and not at home. This could be valuable information for clinicians when determining the timing of palliative care initiation. Given the widespread use of ICI in the treatment of lung cancer, it is relevant to examine the clinical characteristics of patients and how these differ according to POD in a real-world cohort of lung cancer patients treated with ICI.
Based on data from the prospective cohort Lung Cancer, Immunotherapy, Frailty, Effect (LIFE), this study investigated factors related to POD (home, hospice/PCU, or hospital) among patients with advanced non-small cell lung cancer (NSCLC) treated with ICI. The specific objectives were to I) identify patient characteristics by POD, focusing on clinical characteristics and baseline health-related QoL (HRQoL), and II) determine the association between changes in HRQoL during treatment with ICI and POD.
Methods
Study design
This study was conducted using prospectively collected data from the LIFE cohort (ClinicalTrials.gov identifier: NC T03870464). The cohort was initially established to describe lung cancer patients receiving ICI and to investigate the impact of treatment on QoL (Bjørnhart et al. Reference Bjørnhart, Hansen and Asmussen2022, Reference Bjørnhart, Kristiansen and Asmussen2023a, Reference Bjørnhart, Mouritzen and Kristiansen2023b; Meklenborg et al. Reference Meklenborg, Møller and Möller2025). Thus, the objectives of this study are not defined in the original protocol. When the database linked to the LIFE cohort was created, information on death and POD was included to enable a subsequent analysis of factors influencing POD. The database was updated prior to this study, and all updates were included in the analysis.
The study cohort included patients with advanced NSCLC initiating ICI treatment between April 1, 2018, and April 30, 2021, at the Department of Oncology, Odense University Hospital. Patients were recruited in an outpatient setting at their first meeting at the oncology department or when treatment changed to ICI in second- or third-line treatment. Patients were asked to participate if they met the following inclusion criteria: age >18 years, stage IV NSCLC or recurrent NSCLC, squamous or non-squamous histology, candidate for treatment with ICI (PD-1/ -L1 targeting agents), and able to understand information material in Danish and give written consent. All patients received ICI, either Nivolumab, Pembrolizumab, or Atezolizumab, as monotherapy according to recommendations from the Danish Medicines Council at the particular time of treatment initiation (DMC 2024). Treatment was given for a maximum of 2 years or until disease progression or unacceptable toxicity.
Patient characteristics
Relevant patient information, including age, sex, Eastern Cooperative Oncology Group performance status (PS), Charlson Comorbidity Index (CCI) (Charlson et al. Reference Charlson, Carrozzino and Guidi2022), polypharmacy (yes/no, defined by the use of 5 or more medications), metastatic sites (confirmed by computed tomography or magnetic resonance imaging), histology, PD-L1 expression level, and treatment line was gathered from the electronic patient records and registered in a REDCap database.
QoL assessments
Participants were asked to complete QoL questionnaires at baseline (1 week before receiving the first dose of ICI) and at predefined times after treatment initiation. Follow-up QoL questionnaires were sent to patients at 9 weeks, 18 weeks, 6 months, 1 year, and 2 years during treatment with ICI (see Figure 1).
Study flowchart. ICI = immune checkpoint inhibitor; HRQoL = Health-related Quality of Life. *Mostly due to language barriers, 1an incorrect version of the QoL questionnaire was mistakenly sent to 2 patients.
Figure 1 Long description
The flowchart outlines the study process for ICI treatment. It begins with ′Planned for ICI as monotherapy within enrollment period (n=191).′ Next, ′Eligible for inclusion + signed written informed consent (n=166).′ Some participants ′Declined to participate (n=15) or did not meet inclusion criteria (n=10).′ The process continues with ′Start of ICI treatment,′ followed by ′Baseline HRQoL and symptom registration (n=164).′ Follow-up HRQoL assessments occur at ′9 weeks (n=78), 18 weeks (n=55), 6 months (n=26), 1 year (n=7), 2 years (n=4).′ The treatment concludes with ′End of ICI.′ ′Registered place of death and included in analysis (n=121).′ Some participants are ′Excluded from analysis: Unknown/other place of death (n=11) or alive at most recent follow-up (November 2024) (n=34).′.
This study used the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ-C30) (Aaronson et al. Reference Aaronson, Ahmedzai and Bergman1993) and EuroQol 5 Dimensions 5 Levels (EQ-5D-5 L) questionnaire (EuroQol Research Foundation 2019), as these are validated questionnaires for patient self-reports when assessing HRQoL in cancer patients (Boutros et al. Reference Boutros, Bruzzone and Tanda2021). The cancer-specific EORTC QLQ-C30 questionnaire is a 30-item instrument used to measure three aspects defining QoL: symptoms, daily functioning, and global health status/QoL (GHS/QoL). Questions related to symptoms or daily functioning are measured on a 4-point Likert scale, and the two questions related to GHS/QoL are measured on a 7-point scale. The responses are used to calculate a standardized score from 0 to 100 for each scale. A higher score signals a higher burden of symptoms, a higher level of function, and a better GHS/QoL, respectively (Aaronson et al. Reference Aaronson, Ahmedzai and Bergman1993). The generic EQ-5D-5L questionnaire was used to measure the level of severity of problems regarding mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. Based on patient responses, an individual EQ-5D index score is derived, representing the overall health status of the patient: 1 indicating complete health, 0 indicating as bad as being dead, and <0 indicating worse than death. In the EQ-5D-5L questionnaire, patients were also asked to rate their current health status on a visual analog scale (EQ-VAS), with 0 indicating the worst imaginable health and 100 indicating the best imaginable health.
Place of death
Information on POD was retrieved from electronic patient records. It could be registered as one of the following: home, PCU, oncology inpatient unit, another hospital unit, hospice, other, or unknown. To increase the statistical power, patients with PCU as POD were combined with patients who had hospice as POD and labeled “hospice/PCU.” This was considered acceptable, as both settings focus on specialized palliative care and symptom management. For the same reasons, patients with oncology inpatient unit or another hospital unit were combined and labeled as “hospital” since settings were similar, and admission indications were comparable. Given the extent of patient information documented in the electronic patient records, a small number of patients were expected to have other or unknown as POD. They were excluded from the analysis, as the inclusion of this data would introduce considerable statistical uncertainty.
Statistical analysis
Statistical analysis was performed using Stata version 18. Fisher’s exact test was used in the analysis of categorical variables, and the Kruskal–Wallis test was used to test for differences in patient characteristics between the groups when data variables were continuous or ordinal but not normally distributed. A linear mixed-effects regression model was performed on the data concerning changes in HRQoL, symptom scores, and functioning scales. Missing measurements were handled implicitly by the linear mixed-effects regression model, which included all available measurements from each patient. A 2-sided p-value < 0.05 was considered statistically significant.
Results
A total of 166 NSCLC patients were included in the study. Of these, 121 patients were included in the analysis, as 11 had unknown or other POD, and 34 patients were alive at the time of the most recent follow-up (Figure 1). The 121 patients included in the analysis were categorized according to POD (home, hospice/PCU, or hospital). Baseline patient characteristics along with p-values for comparisons are shown in Table 1.
Baseline characteristics of non-small cell lung cancer patients according to place of death
Table 1 Long description
Baseline characteristics are reported for three groups by place of death: Home with 44 patients, Hospice or palliative care unit with 46 patients, and Hospital with 31 patients, with a p value column. Age mean with range was 68.8 with 53 to 82 at home, 65.7 with 35 to 81 in hospice or palliative care unit, and 67.2 with 47 to 82 in hospital, and age over 70 was 22 patients 50 percent, 18 patients 39.1 percent, and 10 patients 32.3 percent. Sex counts were male 28 and female 16 at home, male 17 and female 29 in hospice or palliative care unit, and male 16 and female 15 in hospital. ECOG performance status counts were 0: 12, 8, and 11; 1: 28, 31, and 15; and 2: 4, 7, and 5 for home, hospice or palliative care unit, and hospital. Charlson Comorbidity Index categories were under 3: 32, 37, and 19 and over 3: 12, 9, and 12, and the Charlson Comorbidity Index mean with range was 1.7 with 0 to 6, 1.4 with 0 to 4, and 2.4 with 0 to 9. Polypharmacy yes was 18, 24, and 18; metastatic sites were brain 12, 15, and 6, liver 6, 7, and 1, and bones 14, 19, and 9. Histology was squamous 10, 5, and 7 and non squamous 34, 41, and 24; PD L1 categories were 50 percent or higher 28, 30, and 23, 1 percent to under 50 percent 12, 13, and 6, and under 1 percent 4, 3, and 2. ICI treatment line was 1: 29, 31, and 17 and greater than 1: 15, 15, and 14; EQ 5D index score median with interquartile range was 0.79 with 0.25, 0.74 with 0.20, and 0.78 with 0.29, and EQ VAS score median with interquartile range was 70.0 with 30.0, 60.0 with 30.0, and 77.5 with 30.0. GHS or QoL median with interquartile range was 58.3 with 25.0 at home, 50.0 with 33.3 in hospice or palliative care unit, and 75.0 with 50.0 in hospital.
CCI = Charlson Comorbidity Index; ECOG-PS = ECOG performance status; IQR = interquartile range; p-value < 0.05 was considered statistically significant.
Forty-four patients (36.4%) died at home, 46 patients (38.0%) in a hospice/PCU, and 31 (25.6%) while admitted to a hospital. The distribution of sex significantly differed between the patient groups (p = 0.041). Post-hoc pairwise comparisons using Fisher’s exact test showed the largest difference between the home and hospice/PCU groups (unadjusted p = 0.020). However, no pairwise comparisons remained statistically significant after Holm correction for multiple testing. Age, PS, lung cancer histology, metastatic sites, PD-L1 expression, treatment line, and polypharmacy did not differ between the groups. Patients who died in a hospital had a higher baseline CCI of 2.4, though it did not significantly differ from patients dying at home or in a hospice/PCU, with scores of 1.7 and 1.4, respectively.
When comparing baseline GHS/QoL, patients who died in a hospital had the highest median score of 75.5, whereas those who died at home or in a hospice/PCU had scores of 58.3 and 50.0, respectively. A statistically significant difference was observed between the groups (p = 0.034). When analyzing the results of the EQ-5D-index and EQ-VAS baseline scores, no statistically significant difference was found.
Response rates to the HRQoL questionnaires differed across the POD groups, ranging from 57% to 74% after 9 weeks of ICI treatment, and declined over time in all groups (Table 2). When comparing QoL scores within the groups, the results show that patients with home or hospice/PCU as POD had significantly improved self-reported QoL during treatment with ICI (p < 0.05) (Figure 2). Patients with hospital as POD, having a higher baseline score, experienced no significant change in EQ-5D-index, EQ-VAS, and GHS/QoL at any timepoint during ICI treatment. QoL scores did not differ significantly between the groups during follow-up, as reflected by the overlapping confidence intervals.
HRQoL at baseline and predefined timepoints during immunotherapy according to place of death.
Figure 2 Long description
EQ 5D index score over time. X-axis: Baseline, 9 weeks, 18 weeks, 26 weeks, 52 weeks, 104 weeks. Y-axis: 0.0 to 1.0. Three lines represent home, palliative/hospice and other hospital settings. Significant improvements noted in home and hospice groups at various timepoints. Graph 2 shows EQ 5D visual analogue scale. X-axis: same as Graph 1. Y-axis: 0 to 100. Home and hospice groups show significant increases at specific timepoints. Graph 3 shows global health status. X-axis: same as Graph 1. Y-axis: 0 to 100. Home group shows notable improvement over time. Legend indicates line styles for each group. Significant changes marked with specific symbols. Overall, home and hospice groups show consistent improvement across measures, while hospital group remains stable.
Response rates to HRQoL questionnaires at predefined timepoints during treatment with ICI according to place of death. N = number of respondents (% of total n in groups)
Table 2 Long description
Response counts and percentages are listed at 9 weeks, 18 weeks, 6 months, 1 year, and 2 years for three groups: Home with 44 people, Hospice or palliative care unit with 46 people, and Hospital with 31 people. At 9 weeks, responses were 29 people at home, 66 percent; 26 people in hospice, 57 percent; and 23 people in hospital, 74 percent. At 18 weeks, responses were 24 people at home, 55 percent; 14 people in hospice, 30 percent; and 17 people in hospital, 55 percent. At 6 months, responses were 16 people at home, 36 percent; 11 people in hospice, 24 percent; and 8 people in hospital, 26 percent. At 1 year, responses were 7 people at home, 16 percent; 3 people in hospice, 6.5 percent; and 5 people in hospital, 16 percent. At 2 years, responses were 1 person at home, 2.3 percent; 0 people in hospice, 0 percent; and 3 people in hospital, 9.7 percent.
HRQoL = health-related QoL; ICI = immune checkpoint inhibitors.
When comparing symptoms and daily functions at baseline and throughout treatment with ICI, no statistically significant differences between the groups were observed, as the overlapping confidence intervals suggest (Figure 3).
EORTC Q-30 symptom scales and functional scales according to place of death.
Figure 3 Long description
The composite image contains multiple line graphs displaying EORTC QLQ-C30 symptoms scales and functional scales according to place of death. Each graph plots scores on the vertical axis ranging from 0 to 100, where higher scores indicate better functioning for functional scales and greater symptom burden for symptom scales. The horizontal axis represents four time points: Baseline, Week 6, Week 12 and Week 24. Three lines appear in each graph corresponding to the three groups identified in the legend: Home, Palliative/hospice and Other hospital. Each line includes visible error bars representing confidence intervals at each time point. The graphs are arranged in two columns across multiple rows. The functional scale graphs include physical functioning, role functioning, cognitive functioning, emotional functioning and social functioning. The symptom scale graphs include fatigue, nausea and vomiting, pain, dyspnoea, insomnia, appetite loss, constipation, diarrhoea and financial difficulties. Across most graphs, the three lines follow closely overlapping paths from Baseline through Week 24, with confidence intervals that overlap at most time points. In several symptom scale graphs, a visible upward or downward shift is observed at Week 24 for the Palliative/hospice group compared to the Home and Other hospital groups, representing the largest visible divergence across the time series. In the functional scale graphs, scores for all three groups remain relatively stable from Baseline through Week 12, with minor fluctuations visible at Week 24. The Palliative/hospice line shows the most visible deviation at the final time point in several panels, while the Home and Other hospital lines remain closer together throughout. The confidence intervals are widest at Week 24 across most graphs, reflecting smaller sample sizes at that time point.
Discussion
The results of this study suggest that, among patients with advanced NSCLC treated with ICI, sex and self-reported GHS/QoL before treatment initiation might be associated with POD. In this cohort, a higher percentage of patients who died at home were male, and a higher percentage of patients who died in a hospice/PCU were female. This difference could be due to gender-associated preferences or family dynamics, causing women not to prefer end-of-life care at home, as they do not want to burden their families, especially if widowed (Lee and Jang Reference Lee and Jang2018). However, due to the descriptive design of this study, a comprehensive explanation of this finding cannot be provided.
GHS/QoL scores at baseline were highest in the group of patients with hospital as POD, and the scores significantly differed between the groups. This was a surprising finding, since dying in a hospital is generally associated with a lower QoL (Higginson et al. Reference Higginson, Daveson and Morrison2017). This could indicate that reasons for hospitalization and in-hospital death are often caused by acute complications and consequently not shown in the responses to QoL questionnaires earlier in their cancer trajectory (Pinto et al. Reference Pinto, Lopes and de Sousa2024). Therefore, this finding more likely reflects the fact that a higher QoL is not necessarily associated with a reduced likelihood of dying in a hospital setting.
When evaluating QoL during treatment, the results of this study showed no change in QoL in patients who ended up dying in a hospital, but improved QoL in both patients dying at home or in a hospice/PCU. While the reasons for these findings are unknown, the patients dying at home or in a hospice/PCU may have had a better response to treatment, thereby reducing symptoms and improving QoL. Although not statistically significant, the patients dying at home or in a hospice/PCU had less comorbidity, less polypharmacy, and a relatively higher proportion received ICI as first-line treatment, which is usually considered to be associated with a better prognosis (Chen et al. Reference Chen, Li and Zhou2022; Suazo-Zepeda et al. Reference Suazo-Zepeda, Maas and Vinke2024). It is also possible that patients with home or hospice/PCU as POD had an earlier encounter with the specialized palliative care team than the patients dying in a hospital, since early palliative care in lung cancer patients has been shown to decrease symptom burden, improve QoL, and lower the odds of receiving end-of-life care in a hospital (Starr et al. Reference Starr, Ulrich and Corey2019; Kelly et al. Reference Kelly, O’Brien and Hannigan2022; O’Sullivan et al. Reference O’Sullivan, Conroy and Power2022). However, this study did not register encounters with the specialized palliative care team, which limits the ability to explain the change in QoL during treatment.
The results of this study are generally consistent with findings from other studies examining patient characteristics of lung cancer patients based on POD, although the literature on this subject is sparse. Equivalent to our study, a Japanese retrospective study by Hamamoto et al. (Reference Hamamoto, Ibe and Kodama2020) did not find age, PS, and histology to be associated with POD. In contrast to the findings of our study, Hamamoto et al. also found that sex was not associated with POD, however, there may be some cultural differences between the two studies. A retrospective UK study by O’Dowd et al. (Reference O’Dowd, McKeever and Baldwin2016) found that cancer histology, CCI, and PS ≤ 3 were not associated with POD. Only high age (>85), male sex, or PS >3 was found to be associated with an increased likelihood of dying in a hospital (O’Dowd et al. Reference O’Dowd, McKeever and Baldwin2016). In another study, a higher level of physical function was associated with a greater likelihood of dying at home, whereas patients in need of more physical care were at greater risk of dying in a hospital (Higginson et al. Reference Higginson, Daveson and Morrison2017). Although anticipated, this was not found in our study when examining baseline PS or level of symptom burden and daily function measured on the EORTC QLQ-C30. One possible explanation could be that most patients eligible for inclusion in our study had a PS of 0 or 1, which is required to be a candidate for treatment with ICI in Denmark (DMC 2024). Consequently, the study population consisted of patients with a relatively high functional status, making comparisons with existing literature challenging.
Although the patients in our study had a good PS, the generally poor prognosis associated with lung cancer underscores the relevance of this study (Siegel et al. Reference Siegel, Giaquinto and Jemal2024). The data used for this study are clinically valid, as the cohort consists of real-world advanced lung cancer patients receiving ICI, a treatment often used in the modern field of oncology (Alexander et al. Reference Alexander, Kim and Cheng2020). The included patients were gathered from a general oncology department in Denmark, and although it was a single-center cohort of a rather homogenous population, few exclusion criteria were applied to minimize selection bias and ensure generalizability. A unique feature and strength of this study is the use of prospectively gathered clinical data and QoL information. Patient-reported QoL reflects the dynamics of self-perceived well-being when battling advanced lung cancer, and such data cannot be gathered retrospectively without introducing recall bias. The integration of self-assessed QoL into this line of research is of significant clinical relevance as well as warranted (Hall et al. Reference Hall, Singhal and Dickerson2019).
Although most of the results of this study align with those of previous studies concerning POD in lung cancer patients, it is important to be aware of the potential for bias. Religious, socioeconomic, and familial factors are not investigated in this study, but these potential confounders may influence preferred and actual POD in this patient group (Hamamoto et al. Reference Hamamoto, Ibe and Kodama2020; Pinto et al. Reference Pinto, Lopes and de Sousa2024). In this study, home was considered to be the most preferred POD, but individual, illness-related, and environmental factors may cause patients to prefer hospice/PCU or hospital as a place for end-of-life care and death (Costa et al. Reference Costa, Earle and Esplen2016; Pinto et al. Reference Pinto, Lopes and de Sousa2024). Preferred POD may also be a dynamic variable that changes throughout the illness trajectory (Fereidouni et al. Reference Fereidouni, Rassouli and Salesi2021; Pinto et al. Reference Pinto, Lopes and de Sousa2024). To address this, these aspects need to be considered in the design of future research.
Another potential bias not accounted for is the missing information on specialized palliative care encounters in the study population. In this study, specialized palliative care ought to be considered a confounder, as it influences both HRQoL and POD. If specialized palliative care encounters had been registered and adjusted for in the analysis of HRQoL, a greater difference in HRQoL between the groups would more likely be found. Specialized palliative care is more often assigned to those with the greatest disease burden, thereby reducing the variability in HRQoL across patients. This needs to be considered when interpreting the findings of this study.
An additional limitation of this study is the relatively small number of patients included in the subgroup analyses, especially for the late follow-up visits. This increases the risk of introducing type 2 errors. This also influences the degree of statistical uncertainty that is reflected in the relatively wide confidence intervals. Furthermore, only univariate analyses were performed in this study. Consequently, the observed associations may be influenced by confounding, and it cannot be determined whether the identified factors are independently associated with POD. Future studies with larger cohorts should evaluate these associations using multivariable models.
Conclusion
Among advanced NSCLC patients treated with ICI, a higher percentage of those dying at home were male, and a higher percentage of those dying at a hospice/PCU were female. Surprisingly, baseline GHS/QoL was higher in the group of patients dying at a hospital. No other baseline characteristics significantly differed between patients when divided into groups based on POD. Furthermore, this study reports that patients who died at home or in a hospice/PCU had improved QoL during treatment with ICI. In contrast, patients with hospital as POD had a higher baseline QoL but no significant change in QoL at any time throughout the treatment period. Overall, the findings of this study suggest that neither the investigated patient characteristics nor the change in QoL during treatment with ICI are primary determinants of POD. Although the findings suggest that sex and baseline QoL scores are associated with POD, the exploratory nature of this study limits its ability to determine how these factors should be incorporated into clinical decision-making regarding palliative care initiation. Ensuring a dignified death by honoring preferences on place of death is important, making continued research on this subject clinically relevant.
Data availability statement
The data used in this study are available from the corresponding author upon reasonable request.
Acknowledgments
The authors would like to acknowledge the patients for their participation and the clinical staff for their assistance with patient recruitment and data collection. OPEN (Odense Patient Data Explorative Network, Odense University Hospital) assisted with data handling and storage.
Author contributions
The study was designed by A.R., B.B., and L.A. Data were collected by A.R., B.B., L.A., and I.M. Data analysis was performed by B.B., S.M., and P.M. The first draft of the manuscript was written by A.R. and revised by B.B., L.A., I.M., and P.M. All authors read and approved the final manuscript.
Funding
B.B. acknowledges support for research salary from the Danish Neye Foundation. B.B. and A.R. acknowledge support from the Danish AgeCare organization.
Competing interests
All authors declare no conflicts of interest.
Ethical approval
The study was conducted in accordance with the Declaration of Helsinki and approved by the regional research ethics committee of Southern Denmark.
Ethics and consents
Written informed consent was obtained from all patients before participation. Data were stored properly in a REDCap database at Odense Patient Explorative Network.
Open access
