Herein with report the first Austrian study on some aspects of the end of life of prostate cancer patients in a near population-based setting within a highly socialized, equal access health care system. Most patients (50%) had combination hormone therapy (ADT plus NHT), whereas 1/3 had ADT monotherapy. Younger men were more likely to receive NHT. Prescriptions for most analgesics increased during the final year of life. Prescriptions generally decreased with age. An age-dependency was also noticed for hospitalisation rates with 6 admissions (median: 46.5 days) in the youngest age cohort (< 70yrs) compared to 3 (median: 32 days) in the oldest age cohort (> 85yrs). Approximately 2/3 died in the hospital with highest rates in younger men.
In a similar nationwide study in Sweden, Lycken et al identified 8 326 men who died from prostate cancer from 2009–2012 and assessed their oncological history, sociodemographic status and medication history in the last three years of life (8). Medication history showed similar trends to our cohort: ADT was almost unchanged at 89% while prescriptions for analgesics (NSAID 14–19%, Paracetamol 35–62% and opioids from 30–72%), antidepressants (13–22%), anxiolytics (9–27%) and sedative-hypnotics (21–33%) increased. Older men were less likely to receive ADT (OR: 0.64), NSAID (OR: 0.28), opioids (OR: 0.56), glucocorticoids (OR: 0.23) and anxiolytics (OR: 0.75) whereas they were more likely to receive paracetamol (OR: 1.2). Additional to these findings, our study analysed prescription of NHT and cannabinoids. Changes in NHT-prescription were frequent in younger patients, with initiation in 27% and discontinuation in 47% at EOL. Cannabinoids were prescribed in 15.3% of men < 70yrs compared to 0.8% >85yrs. Medical cannabis is primarily used for its analgesic effects, but also has positive effects on nausea, depression, anxiety and insomnia (14, 15). Current literature reveals conflicting data regarding the use of cannabis in elderly patients, yet suggests cautious prescription because of medication interactions and altered pharmacodynamics but also regarding potential negative effects on cognitive and mental function (16–18). Pain medication was similarly age-dependent in the study by Lycken et al and our analysis, but the results should be interpreted with caution. Lycken et al. also included oncological characteristics and men > 80 were more likely to have high risk PC (32–33%) compared to men < 70 (18%). Locally advanced disease was distributed almost evenly, but younger men had higher rates of advanced disease with distant metastases (55% vs. 29–30%) (8). This could possibly bias the results with younger men needing more pain medication due to advanced cancer stage whereas older men might be suffering from less painful, non-cancer conditions. Rates of analgetic use in our study could be caused by similar differences. Additionally, other possible reasons for lower analgesic prescriptions in older men such as inability to communicate pain, less sensitivity to pain or less access to the healthcare system cannot be accounted for. Since the aim of our study was to show resource use in the last year of life of prostate cancer patients in general, we conclude this does not limit our findings.
Westgeest et al. investigated factors associated with and the effect of high-intensity care in the last six months of patients with castration resistant prostate cancer from the Dutch CAPRI-registry (19). This study observed a decreasing proportion of new treatment initiations from the last 3–6 months (28% of patients had new treatment initiation) to the EOL period (20%) as well as shift from taxane-based therapy to NHT as the treatment choice (10% vs 10% of all treatment initiations with taxane vs NHT at 6 − 3 months to 16–26% <3 months). Patients in whom LPD was initiated had worse tumour characteristics, were younger and had better performance scores. This complicates the identification of the start of EOL-phase and limitation to palliation. Clinicians might still consider LPD to avoid cancer-related complications or symptoms and to preserve quality of life, but also have to consider drug related adverse events. Westgeest et al. demonstrated that patients in whom LPD was initiated had higher rates of hospital admission compared to patients without LPD (≥2 admissions 40% vs 19%), opioid use (38% vs 16%), CRPC (40% vs 17%)- and LPD (14% vs 1%)-related complications. In general, these data fit well with our findings.
Our study design did not allow comparison with a control group. Still, analysis of place of death (patients with cancer diagnosis or any diagnosis) was also performed by Baumgartner J in a nationwide austrian analysis. Data were collected for 2019 and therefore offer a Pre-Covid-19 comparator. Place of death widely differed between federal states, in Vienna 71.9% of patients with cancer diagnosis died in a hospital compared to 57.8% with any diagnosis (20, 21). These results fit with our findings (66.1%). In a multinational European study by Ko et al. interviewed general practitioners for their knowledge on cancer patients preferences regarding their place of death (22). General practitioners stated, that most patient preferred to die at home, followed by palliative care units/hospice, care facilities, and hospitals in descending order (71–90% vs 2.5–13% vs 1.2–11% vs 0.8–6.2%). Actual hospitalized death rates ranged from 17–38%, rate of patients who died at home ranged from 35–61%. The British QUALYCARE study analysed cancer patient preferences and factors associated with not dying at home in a retrospective case-control study. Most patients preferred to die at home. Odds for dying at home were 75% lower, when hospital admission was 15–28 days in the last three months and even 91% lower with a cumulative admission duration > 28 days. Analyzation of possible co-founders revealed no difference in pain scores (died at home vs. did not die at home), whereas patients who died at home felt more at peace. At the same time, intensity of grief in relatives was lower when patients died at home. Regression analysis revealed four factors to be associated with place of death: patient wish, relative’s wish, receipt of home palliative care and receipt of home district/community nursing (23). The findings by Ko et al. are in contrast to our study, which revealed a high proportion of patients who died in the hospital (17–38% vs 66%). Although our study had different cut-offs for hospital admission rates and duration, the findings by Gomes et al. in the QUALYCARE study still offer possible explanations for our findings regarding the place of death. Despite recent efforts in Vienna towards ambulatory palliative terminal care, access is still limited. Community nursing also is a common, yet limited resource in our health care system and often community nurses see themselves unable to care for patients with progressing diseases or disability. Since intensified care or admission to nursing-homes is not accessible on a day to day basis, patients are regularly admitted to hospitals to bridge the time between application and availability of care resources. This shows an aspect that could need improvement, e.g. adjusting care levels more precautionary rather than reactionary to acute events.
Oosterveld-Vlug et al. aimed to relate treatment aims (e.g. palliative, curative, mixed) to hospitalisation rates and death in the hospital in oncological patients (24). Three months prior death, PC patients had the highest proportion of palliative treatment only compared to other malignancies (PC: 70%; lung cancer: 51%; colorectal cancer: 55%; breast cancer: 47%; pancreatic cancer 56%). At the same time, PC-patient had the lowest admission rates (≥1) and died rarely in the hospital (18.5% and 3.1%). This study, however, provides no information on clinical stage and treatment regimens, but it supports the benefit of palliative treatment decisions to preserve patient preferences at the EOL-phase.
There are several limitations of our study such as the retrospective character and the missing oncological/co-morbidity data. Initially, all patients with PC diagnosis were included. In a pre-analysis we experienced frequent coding-discrepancies (e.g. patients with history of PC and curative treatment who had no evidence of recurrence for 1–2 decades were still coded for PC as contributor to death). Therefore, we limited the population for the current study to those who received ADT or NHT in the final year of life. Still, discrimination between men who died with or from prostate cancer was not possible. Chemotherapy, radiotherapy as well as patients refusing any therapy were not accessible via the insurance-provider database. Additionally, it is possible that some men, despite having social insurance, decided to admit to hospitals in the private sector or to collect medication with private prescriptions. We estimate this would only have small effect on our analysis. Despite these limitations, our database contains a fairly large cohort in a real-life setting with complete history of drug prescriptions and hospitalisations in the last year of life.
Conclusion
We report herein the first Austrian study to assess medication and hospitalisation use in the last year of life of PC patients. More than 1 100 patients were included to this near population-based analysis. Resource use increased during the end of life and was highest in younger men. Hospitalization rates were high and 2/3 died in the hospital, both showing a clear age dependency with higher rates, duration and death in the hospital for younger men. Given the limited literature on this topic, these data add to the existing literature, potentially improving knowledge and management of PC-patients during their final period of life.
