Abstract
Introduction
Parkinson’s disease (PD) is the second most common neurodegenerative condition in older people and currently affects 2% to 3% of the population aged above 65 years (Poewe et al., 2017). Globally, rates of PD are set to double by 2030 as life expectancy increases (Dorsey et al., 2007). Later stage PD is associated with poor quality of life for both the person with PD and their carer, and the mean annual costs of PD care per patient in the United Kingdom were estimated to be £5,993 in 1998, although current estimates are likely to be considerably higher (Findley et al., 2003; Hand, Oates, Gray, & Walker, 2018). Slowed motor functioning and nonmotor symptoms can make independent living exceptionally challenging, especially during the later stages of the disease. As a result, PD patients rely more heavily on hospital services than people without PD and 10% to 15% of all people with PD live in a care home (Porter, Henry, Gray, & Walker, 2010).
Although it is well known that people with PD are often admitted to a care home after a hospital admission, and that carer inability to cope is a common reason for care home admission, the period leading up to care home admission is not well understood (Aarsland, Larsen, Tandberg, & Laake, 2000; McDonnell et al., 2014; Walker et al., 2014). Whether care home admission is preceded by a steady deterioration in symptom load and steady increase in carer burden or a more sudden change is not clear. In the United Kingdom, it has been suggested that a crisis period, due to a single event, or a series of events in a short time period, often precipitates care home placement in PD (McDonnell et al., 2014). If many care home placements are a response to a period of crisis, this suggests that our ability to anticipate the need for care home placement is poor. Furthermore, if there is no means of identifying that the crisis period is approaching, the decision for care home placement may be rushed or hospital stay unduly extended, resulting in suboptimal outcomes for the person with PD, their family, and health and social care providers. While this theory of a crisis period immediately preceding care home placement is backed up by anecdotal clinical experience, little published research evidence exists to support it.
One way to identify when a patient and their family are struggling to cope in the community is to look at their use of hospital services. We know from previous studies that PD patients use emergency hospital services more than the general population and that hospital admission is associated with a greater probability of subsequent care home placement (Gerlach, Winogrodzka, & Weber, 2011; Walker et al., 2014). What we do not know is how hospital usage changes in the lead-up to care home placement.
We conducted a retrospective longitudinal study of a representative cohort of community dwelling PD patients in late stage disease. We investigated the number and duration of hospital admissions leading up to care home placement and compared it with admissions in those who remained living in the community or died over a three and a half year time frame. We also looked at whether the reason for admission changed in the 120 days prior to care home placement in those patients who were admitted to a care home.
Method
Ethics and Consent
This study described here is nested within the longitudinal Northumbria Care Needs Project that looks at changing care needs in people with moderate to advanced PD over 10 years. Ethical approval for the Northumbria Care Needs Project was granted by the Newcastle and North Tyneside 1 research ethics committee (ref: 14/NE/1093). Informed consent was not required for this nested study as it involved a review of existing clinical data.
Setting, Participants, and Inclusion/Exclusion Criteria
The Northumbria Healthcare NHS Foundation Trust PD service covers two local government areas in the North East of England: North Tyneside and Northumberland. Patients are admitted to the service upon diagnosis (or occasionally from another service after diagnosis) and reviewed in clinic or at home according to clinical need. Discharge from the service occurs only if the patient dies or moves to another service. On January 1, 2015, the service had 1,386 registered patients. Based on previous prevalence studies, the vast majority of people with PD in the catchment are under the service (Porter, Macfarlane, Unwin, & Walker, 2006; Walker, Hand, Jones, Wood, & Gray, 2010).
Included in the study were all patients known to the service with a diagnosis of idiopathic Parkinson’s disease (IPD) or Parkinson’s disease with dementia (PDD) of Hoehn and Yahr Stages III to V and those with PD-plus syndromes including corticobasal degeneration (CBD), multisystem atrophy (MSA), and progressive supranuclear palsy (PSP). The Hoehn and Yahr staging system for PD is as follows: I (unilateral symptoms only, minimal functional impairment), II (bilateral or midline affected, no balance impairment), III (bilateral, balance impairment, independent in activities of daily living), IV (disability in activities of daily living, able to walk and stand unassisted), and V (confined to bed or wheelchair unless aided; Hoehn & Yahr, 1967). Furthermore, to be included, patients needed to be alive, diagnosed (or be displaying symptoms consistent with their later confirmed disease) and living independently at home on January 1, 2015. We chose to include those with atypical PD syndromes because they are managed in a similar way and present similar challenges upon hospital admission and care home placement to those with IPD (Walker et al., 2014). Patients were excluded from the study if they had died or were resident in a residential or nursing care home on January 1, 2015.
Data Collection
Data were collected from clinical databases held by Northumbria Healthcare NHS Foundation Trust and medical notes (e.g., discharge summaries) by J.K., A.H., L.O., S.M., J.Z., M.P., M.A., and Z.W. during January to October 2017.
Demographics and disease characteristics
Data for age, sex, disease stage, disease duration, diagnosis, and place of abode on January 1, 2015 were collected. Anyone living in a residential or nursing care home was defined as living in a care home. Anyone living in their own home, a relative’s home, or any form of sheltered accommodation was defined as living in their own home. Although some admissions to respite care can result in subsequent admission to a care home, anyone who was in respite care on January 1, 2015, with the expectation they would return to their own home, was classified as living at home, as they had not yet entered a care home permanently.
Care home placement and death
The two key outcomes of interest were care home placement or death from January 1, 2015 to June 30, 2017. Dates of care home admission or death were recorded. As anyone admitted to a care home was deemed to have exited the study, deaths subsequent to care home placement were not recorded.
Hospital admissions
Data on hospital admissions were collected for a three and a half year period from January 1, 2014 to June 30, 2017. This period included data from 1 year prior to the start of data collection relating to deaths and care home admissions to allow meaningful data to be available for those who died or were admitted to a care home soon after January 1, 2015. Data collected included total number of admissions, dates of admission, and durations of admission. Reasons for hospital admission were collected in those placed in a care home during the study period. Admission was defined as an unplanned overnight hospital stay. Admissions for elective procedures (e.g., colonoscopy) were not included. To determine the cause of admission, discharge summaries and medical notes were consulted. All reasons for admission were reviewed by a junior doctor (J.K.) and a consultant geriatrician (R.W.) and a consensus reached on primary cause of admission. Previous studies (Low et al., 2015; Temlett & Thompson, 2006; Woodford & Walker, 2005) have found pneumonia, urinary tract infection, injury (e.g., hip fracture), PD-related problems (e.g., motor and nonmotor symptoms, drug noncompliance and overdose), orthostatic hypotension, cardiovascular diseases (e.g., acute coronary syndrome, arrhythmias, and heart failure), and gastrointestinal complaints (including gastroenteritis) to be the most common causes for hospital admission in patients with PD. We used these categories to group reasons for admission. Where the reason for admission did not fit one of these categories it was labeled “other.” Where a nonspecific diagnosis, such as fall or delirium, was the primary reason, the patient’s notes were consulted further to find a precipitant that could explain admission.
Data Analysis and Statistical Methods
The data were quantitative in nature and collected at a nominal, ordinal, and interval/ratio level. Data were analyzed using standard statistical software, IBM SPSS (version 24; IBM, Armonk, NY, USA) and R (version 3.4.2; R Core Team, R Foundation for Statistical Computing, Vienna, Austria). R is an Open Source environment for statistical computing, which is based on the S computing language. In primary analysis, participants were grouped according to the reason for their exit from the study, namely: care home placement, death, or continued living in their own home at the study end point (June 30, 2017).
Age, disease duration, total bed days, and total hospital admissions data were normally distributed and so were summarized in terms of mean and standard deviation and ANOVA used to assess significance. Categorical data were summarized using frequency, with chi-square tests used to draw inference.
The “cmprsk” package in R was used to conduct competing risks regression (command: crr) based on the subdistribution hazard model developed by Fine and Gray (1999). Competing risks regression can be used to investigate survival when a person is at risk of more than one mutually exclusive event. The event of interest need not be death and the word survival is used in a broad sense. The method is a modification of Cox proportional hazards regression and considers the number of occurrences of each event, the time to its occurrence, and the role of covariates (Haller, Schmidt, & Ulm, 2013). Two models were developed to identify the independent role of total occupied bed days (Model 1) and number of hospital admissions (Model 2) in the 120-day period immediately prior to study exit with regard to subsequent care home placement (the dependent variable) in the presence of the competing risk of death. The estimated hazard ratio can be interpreted as the increased hazard associated with a one-unit change in the independent variable (i.e., one additional bed day or admission). Adjustment was made for age, sex, disease duration, and Hoehn and Yahr stage (Stage III used as reference category).
Results
Population Characteristics
A total of 286 patients met the criteria for inclusion. Between January 1, 2015 and June 30, 2017, 21 (7.3%) patients were admitted to a care home, 81 (28.3%) died, and 184 (64.3%) remained living in the community. Demographic and clinical data for these groups are presented in Table 1.
Demographics, Disease Features, and Hospital Admission in Those Who Were Admitted to a Care Home, Died, or Had Neither of These Outcomes During the Study Period.
Note. PD = Parkinson’s disease.
Hospital Admissions in the Lead-Up to Care Home Placement
The percentage of each group admitted to hospital per year in the study and the numbers of bed days per person per year in the study for each group are shown in Figure 1a and 1b, respectively. In Figure 2a and 2b, the same data are split into 30-day periods leading up to the study exit point (time zero). The number of admissions and bed days rose noticeably in the last 120 days prior to care home placement or death. In the care home placement group, the rise in bed days was most pronounced between 30 and 120 days prior to care home placement. The average number of hospital admissions in the last 120 days prior the study exit point was similar for those who died and those who were admitted to a care home (Table 1). However, in the same period, patients admitted to a care home spent twice as long in hospital as those who died and more than 10 times as long in hospital as those who remained living at home in the community (Table 1).
Overall (a) number of hospital admissions and (b) number of bed days.
(a) Number of hospital admissions and (b) number of bed days per 30-day period.
In competing risks regression Model 1, number of occupied bed days in the last 120 days prior to study exit was significantly associated with subsequent care home placement (subdistribution hazard ratio: 1.03, 95% confidence interval [CI] = [1.02, 1.05]; p < .001), together with disease stage (subdistribution hazard ratio for Hoehn and Yahr Stage V: 5.58, 95% CI = [1.12, 27.80]; p = .036), after adjusting for age, sex, and disease duration. However, in Model 2, total number of hospital admissions in the last 120 days was not significantly associated with subsequent care home placement (subdistribution hazard ratio: 1.55, 95% CI = [0.95, 2.53]; p = .079), although disease stage was (subdistribution hazard ratio for Hoehn and Yahr Stage V: 6.72, 95% CI = [1.12, 40.21]; p = .037). This suggests that total length of stay, rather than the number of independent admissions, is most strongly associated with subsequent care home placement, even after adjusting for demographic and disease characteristics.
Cause of Admission
Causes of admission for those admitted to a care home are summarized in Table 2. Although the number of admissions is too small to allow meaningful inferential analysis and firm conclusions to be drawn, it is notable that in the last 120 days before care home placement pneumonia (often with a secondary delirium) is by far the most common primary reason for admission. Prior to this period, PD-related problems are the major cause of admission.
Causes of Hospital Admission in Those Admitted to a Care Home in the 120 Days Prior to Care Home Placement and in the Period Greater Than 120 Days.
Note. PD = Parkinson’s disease.
Injury includes fall with fracture.
PD-related motor includes falls with no other underlying cause or associated injury.
PD-related psychiatric includes hallucinations and delirium with no other underlying cause.
PD drug compliance includes medication noncompliance and overdose.
Discussion
To our knowledge, this is the first study to identify that care home placement in people with PD is frequently preceded by an increase in the number and duration of hospital admissions. Previous research has shown that care home placement is more likely to occur after hospital admission, although the fact that the number of admissions and the number of bed days start to increase around 120 days prior to admission is of note (Walker et al., 2014; Woodford & Walker, 2005). This suggests that, rather than care home placement being preceded by a single event, there is an increase in hospital bed days over a relatively short period of time, representing a period of crisis where independent living is becoming unsustainable.
While many hospital admissions will be unavoidable, regardless of whether someone is living independently at home or in a care home, some admissions, such as those due to falls, may be prevented in a well designed and well managed care home. Perhaps more importantly, many hospital bed days in the United Kingdom occur once a patient is medically fit for discharge, but before a suitable care home, with an available room, can be found. These excess bed days are clearly avoidable. Previous work has shown that hospital admissions and bed days in hospital go down after care home placement (Walker et al., 2014). Although reduced admissions to hospital after placement may be partly due to changes in long-term care goals upon care home placement, it is possible that the extra burden placed on hospital services prior to care home placement could be prevented if the need for care home placement was identified early and a more anticipatory approach adopted. In the United Kingdom, given the financial pressure on hospital services and the limited availability of care home beds in many regions, the need for such planning is of particular importance.
Previous research suggests that, along with disease features such as hallucinations and cognitive decline, carer strain and carer inability to cope are important precipitants of care home placement (Aarsland et al., 2000; Walker et al., 2014). One possible explanation for our findings is that repeated hospital admissions are likely to increase carer strain as extra care is required as the person with PD recovers. In the United Kingdom, although health and social care needs will be reviewed on hospital discharge, additional care input will often be for a limited period and may not be free at the point of delivery, depending on personal circumstances. This may further increase carer strain, leading to a self-perpetuating cycle, at the end of which care home placement is inevitable. Monitoring patients for changes in key symptoms and carers for increasing strain may offer a practical means of anticipatory care planning and avoiding such crisis periods.
Profile of Those Admitted to Care Home
It is unsurprising that those patients admitted to a care home over the course of the study were older and had more severe disease; previous studies have observed the same trend (Walker et al., 2014). Interestingly, in the group of patients that died without care home placement, PDD and PD-plus conditions were particularly common. The reason why this was not seen in the care home group is likely due, in part, to the small numbers in this group. Another possible explanation is that, due to the more complex nature of their diagnosis, these patients may receive a greater level of community-based care input, which potentially protects against care home placement. Alternatively, as, in the United Kingdom, it often takes many months for the need for care home placement to be recognized and a placement organized, it may be that these patients deteriorate at such a rate that death occurs after the need has been recognized, but before care home placement can be arranged.
Causes of Hospital Admission Prior to Care Home Placement
Detailed inferential analysis of our data on cause of admission has not been conducted due to the relatively small numbers and the fact that not all admissions were independent of each other. The primary reasons for hospital admission in the group placed in a care home are broadly similar to the causes identified in previous studies on PD patients in general (Temlett & Thompson, 2006; Woodford & Walker, 2005). It is interesting to note that the primary reason for admission in the last 120 days is pneumonia while prior to that it is PD motor symptoms. Furthermore, more admissions in the last 120 days are associated with delirium. Although our study numbers are not large enough to draw any concrete conclusions, our findings fit with current understanding of PD progression. Aspiration pneumonia, urinary tract infections, falls, and delirium are all common in the latter stages of PD (Low et al., 2015). These complications are likely to be harder to manage in the community and thus could be a precipitant of care home placement.
Study Strengths
Our study investigated a well-defined, representative, community-dwelling population of people with IPD or atypical PD syndromes. Although one might argue that the inclusion of atypical PD syndromes increases the heterogeneity of our study population, we chose to include this patient group because they are managed in a similar way and present similar challenges to health and social care services as those with IPD. We relied as little as possible on coded data, which is created by nonmedical staff, and instead used an experienced physician and a nurse consultant to stage those with IPD and used discharge summaries written by the clinician responsible for the patient’s care to determine cause of hospital admission.
Limitations
Limitations include the small number of patients who went into care homes; this precludes in-depth analysis of the characteristics of admissions. To have collected larger numbers of patients in the care home group, we would have had to continue data collection over many more years, by which time a large proportion of the cohort would have died. Our decision to report our findings at two and a half years follow-up was a pragmatic one. Our definition of admission as one that crosses midnight means that admissions lasting less than 24 hr were included or excluded depending on what time the patient presented. This meant that some included admissions were only for a few hours (if a patient presented late at night). However, this potential bias is counterbalanced by admissions of longer duration that were excluded because they did not cross midnight. Finally, our data are U.K. specific. However, the situation with regard to pressure on care services and the need for anticipatory care planning in PD will have applicability to many settings.
Conclusion
Our findings provide evidence that PD patients who are placed in a care home have an increased need for emergency hospital services in the lead-up to placement. Investigating the underlying precipitants of care home placement may allow us to avoid crisis events leading to unplanned hospital admission and so minimize distress to people with PD, and their families. This will also help avoid unnecessarily long hospital stays while care options are reviewed and a suitable care home, with an available room, is identified. Future work should look to identify key factors associated with subsequent care home placement, and include markers for symptom load, carer strain, and use of care services.
Footnotes
Acknowledgements
The authors thank Aishling Foley, Doori Oh, and Lydia Trendall (all Newcastle University medical students) who assisted in data collection and input.
Authors’ Note
S.M., J.Z., M.P., M.A., and Z.W. completed this study in part fulfillment of the requirements of a medical degree at Newcastle University.
Author Contributions
This study was conceived, organized, and managed by W.K.G., A.H., and R.W. Data collection was by J.K., L.O., A.H., S.M., J.Z., M.P., M.A., and Z.W. Statistical analysis and writing of the first draft of the paper were done by J.K. and W.K.G. All listed authors were involved in the preparation, review, and critique of the final manuscript.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The larger Northumbria Care Needs Project was part funded by a grant from Parkinson’s UK (Ref: G-1310).
