Article Text
PDF
Abstract
Background Good end-of-life care is essential to ensure dignity and comfort in death. To our knowledge, there has not been a national population-based study in England of community prescribing of all drugs used in end-of-life care for patients with cancer.
Methods 57 632 people who died from malignant cancer in their own home or in a care home in 2017 in England were included in this study. National routinely collected data were used to examine community prescriptions dispensed for drugs for symptom control and anticipatory prescribing by key sociodemographic factors in the last 4 months of life.
Results 94% of people who died received drugs to control their symptoms and 65% received anticipatory prescribing. Prescribing increased for the symptom control drug group (53% to 75%) and the anticipatory prescribing group (4% to 52%) over the 4-month period to death.
Conclusions Most individuals who died of cancer in their own home or a care home were dispensed drugs commonly used to control symptoms at the end of life, as recommended by best-practice guidance. Lower prescribing activity was found for those who died in a care home, highlighting a potential need for improved end-of-life service planning.
- terminal care
- cancer
Data availability statement
Data are available on reasonable request. Data held by Public Health England can be requested using the Office for Data Release: https://www.gov.uk/government/publications/accessing-public-health-england-data/about-the-phe-odr-and-accessing-data.
Statistics from Altmetric.com
Key messages
What was already known?
Current guidelines recommend a combination of pain relief and symptom control drugs at the end of life.
Previous research has focused on opioid use at the end of life, which is used to manage pain, one of five symptoms considered in guidelines.
What are the new findings?
A high proportion of patients with cancer were dispensed medications to manage symptoms at the end of life, as recommended by best-practice guidance.
Increased prescribing of drugs for symptom control and anticipatory prescribing with approaching time to death.
What is their significance?
Symptoms may be better managed at home compared with those who died in a nursing home.
Suggested responses include targeted research of nursing home prescribing activity with a focus on training, service planning and delivery.
Introduction
The Department of Health’s End of Life Care Strategy (2008)1 advocates identifying people approaching the end of life (who will be in the last 6–12 months of life) because good care during this ‘end-of-life’ period is essential to give the best chance of ensuring dignity and comfort in death. In England, poor communication, planning and pain management have been found to cause needless suffering at the end of life.2 Utilisation and duration of palliative care in people with advanced cancer has been found to be significantly associated with: being less likely to die in hospital; being more likely to receive an opioid; and avoidance of late chemotherapy.3 These measures are accepted as indicating better end-of-life care.
Current guidelines from the National Institute of Health and Care Excellence (NICE)4 and best-practice guidance on pain and symptom control in adults with life-limiting illness5 recommend availability of pain relief and symptom control drugs in patients approaching the end of life and in the last days of life. Internationally, there have been several publications on opioid pain relief use in end-of-life care,6–8 but a limited number of studies have investigated other recommended drugs.9–11 Pain is one of five common symptoms likely to occur in the last days of life, the other four being: agitation and anxiety; breathlessness; nausea and vomiting; and noisy respiratory secretions.4 To improve our understanding of care in people approaching the end of life, it is essential to understand the prescribing of drugs used commonly for symptoms in life-limiting illness and in the last days of life.
Anticipatory prescribing is designed to ensure that there is a supply of drugs in the patient’s home, available to an attending clinician for use after an appropriate clinical assessment. With 64% of patients with cancer preferring to die in their own home,12 anticipatory prescribing is designed to enable prompt symptom relief at home at any time, based on the premise that the occurrence of acute events as patients approach the end of life can often be predicted.13
There has been a lack of population-based data available on the community-based delivery of end-of-life care in England. A partnership between the National Health Service Business Services Authority and Public Health England has provided data on prescriptions dispensed in the community in England.14 These administrative data record the drug dispensed, without clinical indication. The objective of this study was to use these data to investigate, for the first time, the community prescribing activity in the last 4 months of life among people who died from cancer in England. Cancer is a common cause of death; lung cancer was the fifth most common cause of death in England in 2018.15 The availability of rich cancer registration data permit the detailed investigation of end-of-life care among those who have died of this common cause. We examined prescribing activity of both drugs for symptom control and anticipatory prescribing by key sociodemographic factors.
Materials and methods
Study population
Individuals who died from cancer (International Classification of Diseases version 10 (ICD-10) C00–C97; excluding non-melanoma skin cancer (C44) in 2017 in either their own home or a care home were identified using mortality data from the Office for National Statistics (ONS) linked to the National Cancer Registration Dataset.16 Individuals were restricted to those diagnosed with cancer in England after 1994, thereby not solely relying on cause of death information. For those diagnosed with more than one cancer, information on the most recent cancer diagnosed was selected for analysis to avoid duplication. Individuals with two cancers diagnosed on the same day were excluded (online supplemental figure A1). The cohort included individuals who died in 2017 with an underlying cause of death of cancer, indicating the cancer was responsible for the train of events leading directly to death.17 No exclusions were made on site of the cancer recorded in the cancer registration data compared with the death certificate.
Supplemental material
Individuals were linked to data of prescriptions dispensed in community pharmacies in England between April 2015 and September 2019.14 The date of death was compared with the dispense date of the prescription to create a 4-month timeline up to the date of death in which to study prescribing for end-of-life care. The month, not full date, of prescription was considered for the time series.
Identification of place of death
The National End of Life Care Intelligence Network (NEOLCIN) identifies the location of death of an individual using detailed mortality data.18 The granularity of mortality data available within the National Cancer Registration Dataset differed from this; therefore, an algorithm based on the full postcode of death was developed which applied the NEOLCIN methodology using the best available data (online supplemental figure A1).
This study included individuals who died in either their own home or a care home. Care homes include residential and nursing homes and range from independent sheltered accommodation to full nursing homes. Nursing homes in England are care homes with at least one qualified nurse on duty to provide nursing care. Testing determined that the algorithm depicted in online supplemental figure A1 assigned 45% of cancer deaths, which is consistent with the NEOLCIN figure of 44%.
Identification of prescribing for palliation
Drugs used for symptom control and drugs used for anticipatory prescribing were identified from the British National Formulary (BNF) version 6814 by a Consultant in Palliative Medicine (KF, online supplemental table A1) consistent with best-practice guidance on symptom control for patients with life-limiting illness.5 Both symptom control and anticipatory prescribing categories covered five symptoms: anxiety, delirium and agitation; breathlessness; nausea and vomiting; noisy respiratory secretions; and pain. Drugs for anxiety, delirium and agitation were considered together since patients often have both a benzodiazepine and haloperidol and/or levomepromazine written up, and care home staff and district nurses in the community must decide which to administer. Prescribing data only identifies that a drug was dispensed, not its indication, so opioids for pain and breathlessness could not be differentiated. Drugs for anticipatory prescribing were restricted to those indicated for the five end-of-life symptoms administered by injection, since most patients are unable to swallow once these drugs are indicated.
Statistical methodology
Descriptive statistics were used to investigate the variation in drugs dispensed to people who died in 2017 in their own home or in a care home. Figures are presented by location of death, drug, sociodemographic and tumour characteristics (sex, age, ethnicity, time since cancer diagnosis, stage at cancer diagnosis, cancer site and income domain of the Index of Multiple Deprivation) and month prior to death. The income deprivation quintile was derived by linking each patient’s postcode at diagnosis (at the Lower Super Output Area geography) to the income domain of the Index of Multiple Deprivation 2015.19 Cancer sites were defined using ICD-10 codes as per online supplemental table A3.20
Statistical significance was defined as two-sided. Likelihood ratio tests following logistic regression were used to assess differences between people who died in a care home and those who died in their own home, and within each sociodemographic characteristic. Two sample t-tests were used to test differences in the proportion of individuals prescribed the drug groups between people who died in a care home and those who died in their own home. The significance of the trend for the time trends was estimated using a generalised estimating equation21 followed by Wald tests. A likelihood ratio test following linear regression was used to test the trend in the date distribution by location of death. All analyses were carried out with Stata software.22
Results
Patient characteristics
In 2017, 57 632 people died from cancer in their own home (70%) or in a care home (30%) (table 1), with the majority dying within 5 years from diagnosis (70%). For context, the own home and care home deaths accounted for 44% of all deaths, with 36% occurring in hospital (online supplemental figure A1). The distribution of sociodemographic and tumour factors varied significantly (2p<0.001) among those individuals who died in their own home as compared with those who died in a care home for all factors assessed. The mean and median age at death of individuals in the cohort were 76.8 and 78, respectively, with the care home cohort dying at an older age (median age=84) compared with the own home cohort (median age=76).
Individual characteristics of all people who died from cancer in England during 2017 in either a care home or their own home
Overall findings
Overall, 94% of people received drugs to control their symptoms and 65% received anticipatory prescribing (table 2). Over the 4-month period prior to death, prescribing increased from 53% to 75% for the symptom control drug group, and from 4% to 52% for anticipatory prescribing (2p<0.001) (figure 1, panel 1). Drugs for pain/breathlessness were most frequently prescribed within both drug groups (table 2).
Drugs prescribed in the 4-month period prior to death to all people who died from cancer in England during 2017 in either a care home or their own home
Time trend of prescribing activity among all people who died from cancer in England during 2017 in either a care home or their own home, by overall drug groups and drugs prescribed in the 4-month period prior to death. Panel 1 presents the overall drug groups; panel 2 presents drugs within the symptom control drug group; and panel 3 presents drugs within anticipatory prescribing.
Prescribing of specific drugs for end-of-life care
The largest increase in prescribing for symptom control drugs over the 4 months prior to death was seen for drugs for pain/breathlessness (48% to 70%, 2p<0.001) and nausea and vomiting (11% to 26%, 2p<0.001) (figure 1, panel 2). The marked trend of drugs for pain/breathlessness is reflected in trends for anticipatory prescribing (3% to 45%, 2p<0.001, figure 1, panel 3).
Variation in prescribing of drugs was observed by individual and tumour characteristics, with highly significant differences partly driven by the large cohort size (table 3). More women received symptom control drugs to control nausea than men (51% vs 44%, 2p<0.001) and there was significant heterogeneity by age in the proportion of people receiving symptom control drugs to control anxiety: ranging from 38% for those aged less than 60 to 20% for those aged 90 and older (2p<0.001, table 3).
Variation in prescribing activity by individual and tumour characteristics among all people who died from cancer in England during 2017 in either a care home or their own home, by subgroup of drugs prescribed in the 4-month period prior to death
Significant heterogeneity was also observed by cancer site, and this was observed for all drug groups. Specific examples include symptom control drugs for nausea most often prescribed for individuals who died from oesophageal and pancreatic cancer (60%, heterogeneity across all cancer sites: 2p<0.001); and anticipatory drugs for anxiety most often prescribed to individuals with pancreatic cancer (61%, heterogeneity across all cancer sites: 2p<0.001). Differences in prescribing of drugs were also observed by deprivation quintile, with a lower proportion of those living in the most deprived areas receiving drugs to manage all symptoms compared with those living in the least deprived areas. The exception is drugs for the symptom control of pain/breathlessness, where the proportion was consistently 90%–91% for individuals living in all deprivation quintiles.
Prescribing for individuals who died in their own home as compared with a care home
Anticipatory prescribing was most common for people who died in their own home (68%) compared with those who died in a care home (59%) (2p<0.001) (table 2). The largest absolute difference in prescribing by location was seen for symptom control drugs for nausea and vomiting (50% vs 41%, 2p<0.001) and anticipatory prescribing of drugs for pain/breathlessness (62% vs 53%, 2p<0.001).
Overall, the proportion of individuals receiving both drugs for symptom control and anticipatory prescribing increased significantly from 3 months before death to the same month as death for those who died in a care home and their own home (figure 2). To calculate this, the generalised estimating equations assessed the month-by-month increases. This showed that the increasing temporal trend in prescribing of drugs for symptom control and anticipatory prescribing was highly significant at each time point for those who died in their own home (2p<0.001 for each month comparison). However, for the individuals in a care home, anticipatory prescribing increased significantly every month (2p<0.001 for each month comparison) yet prescribing of drugs for symptom control stabilised in the same month as death (2p=0.05) as compared with the month before.
Time trend of prescribing activity among all people who died from cancer in England during 2017 in either a care home or their own home, by location of death. Panel 1 is for drugs used for symptom control and panel 2 is for anticipatory prescribing.
The temporal trends by location diverged from 1 month before death. Amongst individuals in care homes, 69% received drugs for symptom control compared with 74% of those who died in their own home (figure 2). The difference was larger for those receiving anticipatory prescribing (19% vs 23%). When focusing on prescribing in the same month as death, this gap widened for those who received anticipatory prescribing (44% vs 55%), whereas it remained almost stable for the symptom control drug group (72% vs 77%). The differences in prescribing by location of death for both symptom control and anticipatory prescribing were maintained in people aged 80 and older (online supplemental figure A2).
Prescribing of opioids in the last 2 months of life
During the last 2 months of life, 77% of individuals received any opioid, with the largest proportion receiving morphine (64%, figure 3). More people who died in their own home received any opioid (80%) compared with those who died in a care home (72%) and this difference was reflected in the prescribing of morphine (67% and 58%, respectively).
Prescribing of opioids in the last 2 months of life among all people who died from cancer in England during 2017 in either a care home or their own home. Panel 1 shows all people and panel 2 shows prescribing broken down by location of death.
Discussion
Our population-based study aimed to understand the variation in prescribing of drugs commonly dispensed to individuals with cancer in the last 4 months of life. Real-world data showed a large proportion of individuals were dispensed drugs commonly used to control symptoms in life-limiting illness and in the last days of life, as recommended by best-practice guidance. Increased prescribing of drugs for symptom control and anticipatory prescribing with approaching time to death was shown.
Findings in the context of other studies
Studies have shown that at the end of life, the most prevalent symptoms are fatigue, difficulty undertaking daily activities (eg, needing help to eat or bathe), feeling unwell and breathlessness.23–25 Our study shows that drugs for pain/breathlessness are the most frequently prescribed to individuals with cancer in the last months of life. This is consistent with other studies showing that pain relief is frequently prescribed during end-of-life care to individuals with cancer.8 26 Behaviour of treating clinicians can play an important role in opioid prescribing27 which may be linked to the variation found in this study, although we are unable to explore this using the data available. A subanalysis on opioid prescribing during the final 2 months of life demonstrated that morphine is most highly prescribed compared with other opioids, suggesting that morphine is still prescribed in preference,28 which is consistent with previous findings.26
Register studies on the prevalence of symptoms suggest that patients have pain (59%–69%), anxiety (33%–44%), respiratory rattle (42%), confusion (22%) and nausea (10%–11%)29 30 as they approach the end of life; these figures are for patients being cared for in nursing or residential homes. Patients receiving palliative care have lower symptom burdens, with 29% being moderately to severely distressed by fatigue, and 22% by pain and breathing problems.31 It is notable, however, that patients in the community are more likely to be distressed by symptoms other than breathlessness. Distressing breathlessness is difficult to cope with at home and patients are often admitted to hospital, but with other symptoms it has been suggested that reduced symptom control is the choice people make to die at home.31 Our findings are therefore encouraging in that 59% of patients in care homes and 68% in their own homes were prescribed anticipatory medication. However, it is possible that more patients will have had pain and respiratory secretions than had prescriptions for the relevant medications, particularly in care homes.
Access to, and longer duration of, palliative care has been shown to be associated with reductions in hospital admissions and increased prescribing of opioid analgesics.3 Our study shows most individuals were dispensed drugs for pain control, and end-of-life symptoms more generally, within the community. Several surveys have highlighted unmet needs with regard to end-of-life care in the community such as insufficient access to pain relief and delays in administering analgesics.32 Concerns have also been raised about uncoordinated and discontinuous care, failure to implement appropriate anticipatory care and poor availability of pain relief out of hours.33–36 Our findings provide a positive message in England regarding dispensed medications, although studies should continue to investigate these concerns.
Clinical and policy implications
A higher proportion of individuals who died in their own home were dispensed drugs, suggesting that symptoms may be better managed compared with those who died in a care home. It was hypothesised that this may be due to the individuals who died while resident in a care home being older than those who died while resident in their own home. However, a supplementary analysis restricting to those aged 80 years and older found the differences in prescribing by location of death were still statistically significant for both anticipatory prescribing and drugs for symptom control. The lower rates of prescribing in the care home cohort may reflect the fact that palliative care services are variable throughout the UK, with challenges in information sharing between General Practitioner (GP), care home, hospital and pharmacy.34 These challenges may be exacerbated when there are several GP practices or community pharmacies linked to one community establishment, as has been found in nursing homes.37
Further research on the provision of end-of-life care to residents within care homes in the UK is required, particularly into the process of transferring and sustaining end-of-life care policy in nursing homes specifically.37 Nursing home staff have been found to have variable confidence in the appropriate use of medicines for end-of-life care, for example, syringe pumps were used only when specialist palliative care staff were involved.37 In nursing homes, the use of end-of-life care medication has traditionally been requested and prescribed on an individual basis during each resident’s last days of life and anticipatory prescribing has varied.38 GPs can be reluctant to provide anticipatory prescribing if, at the time of their visit, the resident is symptom-free, although this also applies for individuals in their own home. Even when prescribed, staff can experience difficulties obtaining the medication out of hours.39 It has been shown that the implementation of the Gold Standards Framework in Care Homes Programme resulted in a change in staff behaviour, and increases were seen in the use of integrated care pathways and anticipatory prescribing.39 The level of care at home has been found to influence patient care and outcomes40 and must be considered when planning end-of-life care.
Strengths and limitations
This study is the first to investigate national community prescribing in England of drugs that manage and alleviate the most common symptoms experienced by individuals with cancer in the last few months of life. The availability of national population-based data permits the robust investigation of prescribing activity.
This study focuses on individuals who died before the COVID-19 pandemic, and therefore will not include changes in practice that have occurred. However, this study can be used as a baseline to assess the impact of the pandemic. The changes in practice which occurred have been reported and have identified positive changes in practice such as use of otherwise unused medication.41 These findings can hopefully be combined with those of Antunes et al to help improve anticipatory prescribing for patients in the community.
The proportion of individuals prescribed drugs in the same month as death may be an underestimate due to the non-censoring of people who died earlier in the month. This is because the full prescription date is not available in the dataset.14 However, as an average time series was created using a full year of data, we consider the impact of this to be limited. The prescribing data is a record that the drugs were dispensed and therefore does not mean that the individual took the drug. There is missing prescribing data due to a lack of patient identifier; however, the impact of this is thought to be small due to the low proportion.14 The prescribing data only captures community-based activity, and therefore, these findings cannot be extrapolated to those under active hospital care.
The community prescribing data did not permit complete investigation of treatment for breathlessness. Individuals may receive oxygen therapy only, which is provided by specific providers on receipt of a Home Oxygen Order Form, rather than a community prescription. The lack of information on indication also did not permit differentiation of opioids for management of pain as compared with breathlessness. No indication information restricted the classification of levomepromazine as an antiemetic, despite evidence that it can also be prescribed for agitation.
Conclusion
We found most individuals in this study were dispensed drugs commonly used to control symptoms in life-limiting illness and in the last days of life, as recommended by best-practice guidance. This study has highlighted significant differences in prescribing by location of death, with symptoms more likely to be recognised earlier and appropriate drugs dispensed in people who died in their own home compared with those who died in a care home. Suggested responses include targeted research of nursing home prescribing activity with a focus on education, service planning and delivery.
Data availability statement
Data are available on reasonable request. Data held by Public Health England can be requested using the Office for Data Release: https://www.gov.uk/government/publications/accessing-public-health-england-data/about-the-phe-odr-and-accessing-data.
Ethics statements
Patient consent for publication
Acknowledgments
Data for this study are based on pseudonymised patient-level information collected by the NHS Prescriptions Services, part of NHS Business Services Authority, for reimbursement purposes. The authors would like to thank colleagues at the NHS Business Services Authority. The authors would also like to thank colleagues in the National Disease Registration Service, specifically Carolynn Gildea for her statistical advice and Dr Cong Chen, Dr Harriet Fisher and Jack Anderson for their quality assurance of the code. Data for this study are based on patient-level information collected by the NHS, as part of the care and support of patients with cancers. The data are collated, maintained and quality assured by the National Disease Registration Service, which is part of Public Health England (PHE).
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Contributors Conception and design of the work: KEH, GE, LH, JV, KF. Analysis of the data: KEH, GE. Clinical interpretation: JV, KF. All authors made substantial contributions to the interpretation of the findings. All authors contributed to drafting the manuscript or revising it critically for important intellectual content and approved the final version submitted. All authors have agreed to be accountable for all aspects of the work.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.