Health Care Utilization in the Last Year of Life in Parkinson Disease and Other Neurodegenerative Movement Disorders

Abstract

Background and Objectives

Neurodegenerative movement disorders are rising in prevalence and are associated with high health care utilization. Generally, health care resources are disproportionately expended in the last year of life. Health care utilization by those with neurodegenerative movement disorders in the last year of life is not well-understood. The goal of this study was to assess the utilization of acute care in the last year of life among individuals with neurodegenerative movement disorders and determine whether outpatient neurology or palliative care affected acute care utilization and place of death.

Methods

We conducted a retrospective cross-sectional study including health system administrative data in Alberta, Canada, from 2011 to 2017. Administrative data were used to determine place of death and quantify emergency department (ED) visits, hospitalizations, intensive care unit admissions, and outpatient generalist and specialist visits. Diagnoses were classified by 10th revision of the International Classification of Diseases codes. Stata 16v was used for statistical analyses.

Results

Among 1439 individuals (60% male), Parkinson disease (n = 1226), progressive supranuclear palsy (n = 78), multiple system atrophy (n = 47), and Huntington disease (n = 58) were the most common diagnoses. The most frequent place of death was in hospital (45.9%), followed by long-term care (36.3%), home (7.9%), and residential hospice (4.0%). Most (64.2%) had >1 ED visit, and 14.4% had >3 emergency department visits. Fifty-five percent had >1 hospitalization, and 23.3% spent >30 days in hospital. Few (2.6%) were admitted to ICU. Only 37.2% and 8.8% accessed outpatient neurologist and specialist palliative care services, respectively. Multivariate logistic regression found the odds of dying at home was higher for those who received outpatient palliative consultation (OR, 2.49, 95% confidence interval [CI], 1.48–4.21, p < 0.001) and were with a longer duration of home care support (OR, 1.0007, 95% CI, 1.0004–1.0009, p < 0.001).

Discussion

There are high rates of in-hospital death and acute care utilization in the year before death among those with neurodegenerative movement disorders. Most did not access specialist palliative or neurologic care in the last year of life. Outpatient palliative care and home care services were associated with increased odds of dying at home. Our results indicate the need for further research into the causes, costs, and potential modifiers to inform public health planning.

Parkinson disease (PD) is the fastest growing cause of death and disability among neurologic disorders.¹ As the global population ages and individuals with neurodegenerative disorders survive longer, demand for neurologic care and other health care resources will continue to rise.^1,2

Health care resources are disproportionately expended as the end of life approaches.^3,4 In the United States (US), care for 6% of Medicare beneficiaries who die each year accounts for nearly 30% of Medicare expenditures.⁵ PD is associated with high rates of health care utilization compared with matched controls.⁶ How health care resources are used at the end of life in PD and other neurodegenerative movement disorders is unknown.

As the prevalence of neurodegenerative movement disorders increases, understanding and meeting the health care needs of this population will be of critical public health importance. Using retrospective health system administrative data, this study aimed to describe health care utilization in the last year of life for all individuals who died of neurodegenerative movement disorders in Alberta, Canada, over the 7-year period from 2011 to 2017. We investigated whether outpatient neurologist or specialist palliative care involvement affected place of death, ED visits, and hospitalizations.

Methods

Standard Protocol Approvals, Registrations, and Patient Consents

This study was approved by the University of Calgary Conjoint Health and University of Alberta Research Ethics Boards (REB20-0880 and Pro00076456, respectively). Approvals were obtained with a waiver of consent considering the nature of the study design involving retrospective collection of administrative data after death.

Study Design and Population

We conducted a retrospective cohort study using linked data from multiple administrative databases containing health care use and sociodemographic data for all deaths in Alberta, Canada, between 2011 and 2017. We included adult individuals (18 years and older) who died of neurodegenerative movement disorders based on the recorded underlying cause of death code from the Vital Statistics Registry using the 10th revision of the International Classification of Diseases (ICD-10). The ICD-10 codes used to define the study population are listed in eTable 1, http://links.lww.com/CPJ/A381. Two pediatric individuals, with diagnoses of Huntington disease and dystonia not otherwise specified, were excluded.

Data Sources

Data were drawn from administrative health care databases using personal health numbers, name, and date of birth to link across databases. Data on homecare utilization was limited to individuals residing in one of 5 administrative zones within the province. Eligible decedents were identified through the Vital Statistics Registry and queried in the physician billing, facility living, palliative care, and acute care databases. Data were accessed, linked, and preprocessed by team members at Alberta Health Services Analytics. This was linked with the palliative care databases of each of the zones in Alberta, which completely capture specialist palliative care consultations, intensive/tertiary palliative care unit admissions, and residential hospices. Community care services use was also recorded, obtaining the information from (1) home care (nonpalliative) and (2) long-term care (LTC)/facility living.

Acute care services data were obtained from (1) the Canadian Institute for Health Information Discharge Abstract Database (this database contains diagnostic and procedure information about all acute care hospitalizations in Canada) and (2) the National Ambulatory Care Reporting System (a database that includes all ambulatory care levels within Canada, including ED, day surgery and medical and surgical day clinics within hospitals, the community and private clinics). The Discharge Abstract Database inpatient diagnostic codes were categorized using the Charlson Comorbidity Index to capture major comorbidities.

Gathered information included neurodegenerative disease category using ICD-10 codes, demographics, covariates of interest (sex, age at death, rural or urban living location), family physician and neurologist care visits (within 365 days before death), palliative care access type (palliative home care, inpatient consultation team, community consultation team), acute care use (all acute care service use within 365 days of death), home care utilization (date of initiation of continuous service before death), and location of death (home, residential hospice, long-term care, designated assisted living, hospital). The exact date of health care delivery was recorded for all health care data and was coded as number of days before death. Health care use within 365 days before death was evaluated. Variables were measured at time intervals between the time of death and 30, 90, and 365 days before death. Early palliative care was defined as beginning ≥90 days before death, a cut point commonly used in the palliative care literature.^7,8

Statistical Analyses

Baseline quantitative variables were analyzed using t-tests and categorical data using Wilcoxon rank sum tests. Wilcoxon signed-rank comparisons for nonparametric data were used to compare groups. One-way ANOVA and the Kruskal-Wallis equality-of-populations rank test were used to compare outcomes by diagnosis. The type I error rates for multiple comparisons were corrected with Holm-Bonferroni. Pairwise correlations and point-biserial correlations were used to analyze bivariable relationships between variables. Multivariable regression logistic models using each place of death as a binary dependent variable were used to model potentially explanatory variables including categorical (access to palliative care, diagnosis, visits to the general physician, visits to the neurologist, rurality) and quantitative variables (age, number of hospitalizations, ED visits, days in hospital, acute care admissions, palliative care visits, general physician visits, number of visits to the neurologist, duration of home care services before death). Linear regression models were used to evaluate the relationship between quantitative variables related to health care utilization. All regression models were tested for collinearity. A p-value of < 0.05 was considered statistically significant. All analyses were performed using Stata 16v.

Data Availability

Data are available on request by qualified investigators. The data set from this study is held securely in coded form at the University of Calgary. While the conditions of our ethics approval prohibit making the data set publicly available, access to an anonymized summary level (aggregate data) may be granted (conditional on permission from the data custodian: Alberta Health Services) on request by emailing the corresponding author. In addition, the entire data set creation plan and underlying analytic code are available on request by emailing veronica.bruno@ucalgary.ca, understanding that the programs may rely on unique coding templates or macros from Alberta Health Services and this study.

Results

Over 7 years (2011–2017), there were 1439 adult deaths (60% male, n = 863) because of neurodegenerative movement disorders in Alberta, Canada. Decedents 80 years and older comprised 63% (n = 907) while 6% (n = 86) were 65 years and younger (Table 1). The most common diagnoses were PD (n = 1226, 85.2%), progressive supranuclear palsy (n = 78, 5.4%), Huntington disease (n = 58, 4.0%), and multiple system atrophy (n = 47, 3.3%). Dementia was the most common major comorbidity, affecting 22.9% of individuals (n = 329), while 69.2% (n = 996) had no major comorbidities.

Table 1.

Characteristics and Place of Death for Patients Who Died of Neurodegenerative Movement Disorders in Alberta From 2011 to 2017

In-hospital deaths accounted for 45.9% of all deaths (n = 661), followed by long-term care (36.3%, n = 523), home (7.9%, n = 114), and residential hospice (4.0%, n = 58) deaths. Place of death was unknown for 5.8% (n = 83).

Acute Care Utilization

In the last year of life, 64.2% of individuals (n = 924) had ≥1 ED visit and 24.3% (n = 350) had ≥3 ED visits. In the 30 days before death, 32.0% (n = 460) had ≥1 ED visit. In the last year of life, 55% of individuals (n = 791) had ≥1 hospitalization; 23.3% (n = 335) were admitted for a cumulative 30 days or more; and 10% (n = 138) had ≥3 separate hospital admissions. Those admitted to the intensive care unit (ICU) in the last year of life comprised 2.6% (n = 38), and 60% (23/38) of ICU admissions occurred in the last 30 days of life (Tables 2 and 3).

Table 2.

Acute and Outpatient Health Care Utilization Rates in the Last Year of Life

Table 3.

Health Care Utilization and Place of Death by Diagnosis

Family Medicine, Neurology, and Specialist Palliative Care Utilization

In the last year of life, 93.5% of individuals (n = 1345) visited a family medicine physician and 86.0% (n = 1238) had ≥3 visits. In the last 30 days of life, 75.7% (n = 1089) saw a family physician (Table 2).

Only 37.2% of individuals (n = 505) had ≥1 outpatient neurology visit in the last year of life (table 2), and among them, 28.0% (n = 150) had a single visit. In the last year of life, 17.4% (n = 250) had ≥1 palliative care consultation. Inpatient palliative care consultations occurred more often than outpatient consultations, with 10.6% (n = 152) having ≥1 inpatient consultation compared with 8.8% (n = 127) having ≥1 outpatient palliative care consultation.

Those who had early palliative care (≥90 days before death) died outside hospital more frequently than those who had late (<90 days before death) or no palliative care involvement (early = 71.6%, 58/81; late = 51.0%, 9/194; no palliative care 53.3%, 621/1164). Those who had late palliative care had an average of 0.62 ED visits in the last 30 days of life, compared with 0.30 visits for early palliative care and 0.39 for no palliative care. Those who received late palliative care spent 12.5 days in hospital in the last 30 days of life compared with 3.4 days and 4.2 days for early and no palliative care, respectively.

Associations With Place of Death

Dying in hospital was correlated with several demographic and health care utilization characteristics. Point-biserial correlations showed that those who died in hospital were more likely to reside in a rural location (r = 0.21, p < 0.001) and have dementia (r = 0.12, p < 0.001). They had more days spent in hospital (r = 0.20, p < 0.001), more hospital admissions (r = 0.26, p < 0.001), more ED visits (r = 0.19, p < 0.001), and more ICU admissions (r = 0.12, p < 0.001) in the last year of life. Those who died in hospital were more likely to have outpatient neurologist (r = 0.08, p = 0.003) and other specialist (r = 0.12, p < 0.001) consultations in the last year of life (Figure). By contrast, family physician visits were inversely associated with death in hospital (r = −0.19, p < 0.001). In adjusted multivariable regression models, the odds of dying in hospital was reduced for those with more family physician visits (OR, 0.99, 95% CI 0.98–0.99, p < 0.001) and was higher for those with more hospitalizations (OR, 1.63, 95% confidence interval [CI], 1.36–1.95, p < 0.001) and ICU admissions (OR, 3.79, 95% CI, 1.60–9.01, p = 0.003), as well as those who resided in rural locations (OR, 2.64, 95% CI, 1.99–3.51, p < 0.001).

Figure. Neurologic and Palliative Outpatient Care and Place of Death in the Last Year of Life.

Place of death for patients who received (A) neurologic outpatient care in the last year of life vs those who received no outpatient neurologic care and (B) those who received palliative outpatient care in the last year of life vs those who received no palliative outpatient care.

Death at home was correlated with more outpatient palliative care team consultations (r = 0.15, p=<0.001) and longer duration of home care support before death (r = 0.16, p < 0.001). Death in residential hospice was correlated with outpatient (r = 0.22, p < 0.001) and inpatient (r = 0.36, p < 0.001) palliative care team consultations, whereas rural residence (r = −0.10, p < 0.001) and admission to LTC (r = −0.11, p < 0.001) were correlated with a reduced likelihood of death in residential hospice. Adjusted regression models confirmed that the odds of dying at home was higher for those who received outpatient palliative care consultation (OR, 2.49, 95% CI, 1.48–4.21, p < 0.001) and those with longer duration of home care support (OR, 1.0007, 95% CI, 1.0004–1.0009, p < 0.001). The odds of dying in residential hospice was higher for those who received an inpatient (OR, 15.79, 95% CI, 8.64–28.84, p < 0.001) or outpatient (OR, 5.73, 95% CI, 3.03–10.89, p < 0.001) palliative care consultation in the last year of life.

Discussion

We retrospectively evaluated health care utilization in the last year of life by individuals with neurodegenerative movement disorders and found high rates of ED visits and hospitalizations. A large percentage (45.9%) of the cohort died in hospital. The rates of specialist palliative and outpatient neurology care were low (8.8% and 37.2%, respectively), and outpatient palliative care and home care services were associated with increased odds of dying at home.

Compared with published data on individuals who died of cancer in Calgary, Alberta, we found that fewer individuals with neurodegenerative movement disorders received specialist palliative care (17% vs 86%).⁷ Those with neurodegenerative movement disorders had lower rates of hospital-based acute care in the last 30 days of life: 6.8% had more than one ED visit vs 11% of those with cancer, and 16.1% spent >14 days in hospital compared with 25% of those with cancer. The rate of death in hospital for those with neurodegenerative movement disorders in our study was 45.9%, compared with 41% of individuals who died of cancer.

We found low rates of specialist palliative care utilization including residential hospice. Palliative care has traditionally been associated with management of cancer-related terminal illnesses, which comprise 20–30% of deaths and receive 80% of palliative care services. Palliative care utilization in the context of noncancer neurologic conditions may be low because of the perceived association with cancer care and lack of patient or provider familiarity with palliative approaches for neurodegenerative conditions, although the symptom burden among individuals with PD is of a similar severity to those with cancer.⁹ In addition, residential hospice admission is often restricted to those with a specified life expectancy (e.g., less than 3 months), and it is difficult to accurately predict survival in noncancer conditions.¹⁰ Furthermore, family physicians may have expertise in palliative care or provide generalist palliative care, and palliative services provided by an individual's family physician would not have been captured in our study. Inpatient palliative care consultations were associated with a higher likelihood of in-hospital death; we hypothesize that attending teams were more likely to consult palliative care service for inpatients at the end of life.

Our study found that 68% of those with neurodegenerative movement disorders received no outpatient neurologist care in the last year of life. Similarly, a retrospective study of US Medicare beneficiaries with PD found that 67% of those residing in LTC did not have outpatient neurologist care.¹¹ Contributors to the low rate of neurologist care in our study are hypothesized to include difficulty accessing neurologic care and neurologists' practice patterns. Barriers to accessing specialty neurologic care may include difficulty attending outpatient appointments because of mobility or transport challenges and an insufficient neurologist workforce to meet population needs. While family physicians and palliative care specialists may conduct home or LTC facility visits, neurologists typically provide outpatient care in a clinic setting. Canada has few neurologists per capita compared with other countries (2.5/100,000 population vs 3.9/100,000 US population),² and follow-up may be limited by high patient volumes per physician.

We found that outpatient neurologist care in the last year of life did not significantly affect acute care utilization and was, surprisingly, associated with greater risk of dying in hospital. This association needs to be explored further because prior research on individuals with PD found that neurologic outpatient care was associated with decreased acute care and skilled nursing facility utilization.^11,12 The apparent lack of association in our study between outpatient neurologist care and acute care utilization may be because of unequal disease severity between those who did and did not have neurologist care in the late phase of disease evaluated in our study and the relatively small number of patients who received outpatient neurologist care. Neurologists' comfort managing advanced neurodegenerative movement disorders and engaging in advance care planning conversations may also influence acute care utilization by their patients at the end of life. A survey of neurologists and neurology trainees in the Netherlands found that advance care planning was rarely discussed with people with PD until significant physical or cognitive decline had already started, and most respondents felt they needed further education in the palliative care of chronic progressive neurologic diseases.¹³

Many individuals wish to die at home. A study of 76 patients with advanced PD in Germany found that 68.0% preferred to die at home compared with hospital (6.6%), hospice (5.3%), or nursing home (2.6).¹⁴ However, for many with neurodegenerative movement disorders, the burden of care in the last weeks or months of life may exceed what can feasibly be provided at home by family and home care services. In-hospital death is an undesired outcome for many individuals and families, as well as from a health system perspective because of costs and limited capacity. Optimizing the use of LTC and hospice care may allow for higher intensity end-of-life care than can be provided at home and prevent ED visits and acute hospitalizations. Integrating specialist neurology and palliative care consultation with existing in-house primary care in LTC facilities, by telehealth or in-person care, may be a strategy to expand access to these services. Multidisciplinary approaches combining neurologic and palliative care can also improve quality of life and disease burden in advanced disorders.¹⁵

This study has limitations. We used the Vital Statistics Registry to identify underlying causes of death by ICD-10 codes. Diagnoses based on the completion of death certificates may have led to underascertainment of neurodegenerative movement disorders contributing to death or misclassification of disorders, and we cannot account for physician accuracy in diagnosis or coding. The administrative data used to identify cases do not provide information on disease severity. Whether individuals were seen in the outpatient setting by subspecialist movement disorders neurologist or general neurologist is unknown. Inpatient consultations by neurology services were not captured. Home care data was only available for a minority of patients residing in one of 5 administrative zones included in our study, limiting our ability to draw conclusions about the impact of home care on place of death and utilization of other health care resources.

There are several relevant features of the Canadian health care system to consider in the interpretation of our results. In Canada, all acute care, all outpatient care, and approximately 50% of long-term care are publicly funded without out-of-pocket expenses to patients. The lower number of neurologists per capita in Canada may have influenced utilization of neurologist services compared with the United States.² In a multinational study of health care utilization and place of death for patients dying with cancer, Canada had a higher proportion of individuals who died in acute care hospitals compared with the United States, the Netherlands, and England.¹⁶ Therefore, the proportion of deaths from neurodegenerative movement disorders that occur in hospital in Canada may exceed that in similar countries. The provision of palliative and home care services is also jurisdiction-specific. In Alberta, the maximum available home care support comprises a total of 90 minutes of care per day plus 6 hours of respite per week. Admission to a residential hospice requires referral by a palliative care physician and an estimated life expectancy of less than 3 months.

Further research is needed to investigate the economic and health policy implications of current health care utilization patterns in those with neurodegenerative movement disorders and to identify strategies to optimize quality of care for those with advanced disease. Comparative studies are needed to investigate whether end-of-life care utilization and place of death in neurodegenerative movement disorders can be explained in part by access to outpatient neurologic care and palliative care services and also whether similar trends exist for other neurologic conditions. Further studies are also needed to better understand patient and family wishes for end-of-life care, to develop patient-centered outcome measures and interventions. The impact of social determinants of health and system bias also needs to be considered.

Our study found high rates of acute care utilization and death in hospital among those with neurodegenerative movement disorders in Alberta, Canada, from 2011 to 2017. Most received no neurologist outpatient care or specialist palliative care in the last year of life. Family physician care and duration of home care services were associated with decreased odds of dying in hospital. The high rates of acute care use and hospital-based death among those with neurodegenerative movement disorders indicate a need for research into the causes, economic costs, and potential modifiers for this care utilization pattern to inform public health planning.

Appendix. Authors

Study Funding

The authors report no targeted funding.

Disclosure

The authors report no disclosures relevant to the manuscript. Full disclosure form information provided by the authors is available with the full text of this article at Neurology.org/cp.

TAKE-HOME POINTS

• → In the year before death, there were high in-hospital death rates and acute care utilization among those with neurodegenerative movement disorders.

• → Most did not access specialist palliative or neurologic care in the last year of life.

• → Outpatient palliative care and home care services increased the odds of dying at home.

• → Our results indicate the need for further research into the causes, costs, and potential modifiers to inform public health planning.