People with schizophrenia face an increased risk of premature death from chronic diseases and injury. This study describes the trajectory of acute care health service use in the last year of life for people with schizophrenia and how this varied with receipt of community-based specialist palliative care and morbidity burden.
A population-based retrospective matched cohort study of people who died from 01/01/2009 to 31/12/2013 with and without schizophrenia in Western Australia. Hospital inpatient, emergency department, death and community-based care data collections were linked at the person level. Rates of emergency department presentations and hospital admissions over the last year of life were estimated.
Of the 63508 decedents, 1196 (1.9%) had a lifetime history of schizophrenia. After adjusting for confounders and averaging over the last year of life there was no difference in the overall rate of ED presentation between decedents with schizophrenia and the matched cohort (HR 1.09; 95%CI 0.99–1.19). However, amongst the subset of decedents with cancer, choking or intentional self-harm recorded on their death certificate, those with schizophrenia presented to ED more often. Males with schizophrenia had the highest rates of emergency department use in the last year of life. Rates of hospital admission for decedents with schizophrenia were on average half (HR 0.53, 95%CI 0.44–0.65) that of the matched cohort although this varied by cause of death. Of all decedents with cancer, 27.5% of people with schizophrenia accessed community-based specialist palliative care compared to 40.4% of the matched cohort (p<0.001). Rates of hospital admissions for decedents with schizophrenia increased 50% (95% CI: 10%-110%) when enrolled in specialist palliative care.
In the last year of life, people with schizophrenia were less likely to be admitted to hospital and access community-based speciality palliative care, but more likely to attend emergency departments if male. Community-based specialist palliative care was associated with increased rates of hospital admissions.
Citation: Spilsbury K, Rosenwax L, Brameld K, Kelly B, Arendts G (2018) Morbidity burden and community-based palliative care are associated with rates of hospital use by people with schizophrenia in the last year of life: A population-based matched cohort study. PLoS ONE 13(11): e0208220. https://doi.org/10.1371/journal.pone.0208220
Editor: Andrea Gruneir, University of Alberta, CANADA
Received: June 14, 2018; Accepted: November 14, 2018; Published: November 29, 2018
Copyright: © 2018 Spilsbury et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: Data in this study was accessed through the WA Data Linkage Branch based in the Western Australian Department of Health. Release of the de-identified data used in this study to researchers requires both ethical approval from both the Department of Health Human Research Ethics Committee and the institutional Human Research Ethics Committee where the research will be conducted. In addition, data custodian approvals may be required. Data requests may be sent to Department of Health WA Human Research Ethics Committee (EC00422), P: 9222 4278 E: firstname.lastname@example.org.
Funding: This work was supported by the National Health and Medical Research Council of Australia through a project grant #1084890 (https://www.nhmrc.gov.au/grants-funding) awarded to LR. The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
The average life expectancy of people living with schizophrenia in developed countries is around 20 years less than the general population [1, 2] although this varies by age of schizophrenia onset and gender.  This has been mostly attributed to premature death from cardiovascular disease, respiratory diseases, cancers and injury . Factors associated with these excess early deaths include adverse side-effects of some antipsychotic medications ; under diagnosis of metabolic syndrome ; reduced rates of cancer screening  resulting in more advanced stages of cancer at diagnosis; high-risk lifestyle behaviours such as cigarette smoking [4, 7], alcohol and drug use ; and high levels of other medical comorbidity.  A review of cancer care in people with schizophrenia reported they experienced longer delays from diagnosis to treatment, were less likely to undergo surgery but had greater 30 day mortality post-surgery and received fewer chemotherapy and radiotherapy sessions. 
The episodic positive symptoms of schizophrenia such as delusions, thought disorder and hallucinations and the negative symptoms of apathy, social withdrawal, reduced self-care and cognitive dysfunction may act as barriers to accessing appropriate health care. [10, 11] A Canadian study reported that in the last six months of life, people with schizophrenia had more visits to general practitioners and psychiatrists but fewer visits to medical specialists, 27% less hospitalisations and were less likely to receive palliative care compared to a matched cohort.  A study from the United States reports that having any pre-existing psychiatric illness was associated with less acute care hospitalisations and intensive care but higher rates of ED presentations in the last 30 days of life and a greater likelihood of dying in a nursing home.  In contrast, a study of end-of-life care for mostly male US war veterans dying with cancer found those with schizophrenia received comparable care with similar proportions of patients who underwent surgical treatment for cancer, enrolled in hospice care and had advance directives and resuscitation orders in place, although a lower proportion of veterans with schizophrenia initiated chemotherapy. 
Community-based specialist palliative care delivered in the home or place of residence in the last year of life is associated with reduced presentations to ED [15, 16], reduced admissions to hospital [17, 18], shorter lengths of hospital stays [19, 20] and reduced hospital costs.  While most studies have focused on people dying with cancer, the evidence supporting the benefit of specialist palliative care in the setting of life-limiting non-cancer conditions is growing. [22, 23]
The aim of this study was to describe the trajectory of acute care health service use in the last year of life for people with schizophrenia and how this varied with morbidity burden, cause of death and access to community-based specialist palliative and non-palliative care. We hypothesised that people with schizophrenia would show low rates of community-based specialist palliative care in the last year of life, but that barriers to accessing acute care health services would be reduced in those who did.
This was a population-based retrospective matched cohort study of acute health care service and community-based health care service use over the last year of life in people who died from 1 January 2009 to 31 December 2013 with or without a history of schizophrenia and aged 20 years and older in Western Australia (WA). Decedents with a lifetime history of schizophrenia, schizophrenia-like psychosis or schizoaffective disorders were eligible for inclusion and are referred to collectively as the schizophrenia cohort. Decedents in the schizophrenia cohort were identified from coded administrative data by searching all death registrations in WA from 2009 to 2013 and person-linked hospital admissions (1979 to 2013), emergency department (ED) presentations (2005 to 2013) and mental health registrations (1970 to 2013) using International Classification of Diseases (ICD) Version 9  code 295 and ICD-10-AM codes F20, F21, F231, F232 and F25.  Data linkage and de-identified data extraction from the WA Data Linkage System was performed by the Data Linkage Branch at the WA Department of Health. In the absence of a unique personal identifier in Australia, the WA Data Linkage System uses probabilistic matching based on name and other identifiers.  Ethical approval to conduct this retrospective study on anonymized data with a waiver of the requirement to obtain consent was provided by the Human Research Ethics Committees at the WA Department of Health and Curtin University.
Matching for population-based comparison cohort
Coarsened exact matching (CEM) was used to identify a matched comparison cohort of decedents without a history of schizophrenia. Matching was performed to create a similar balance in age groups, sex, partner status, indigenous status, residential location, relative social disadvantage and country of birth in both the schizophrenia and comparison cohorts.  Observations were weighted according to the size of the matching strata with unmatched population controls excluded. Matching and statistical analyses were adjusted for indigenous status, but we do report separately for this population subgroup.
Causes of death
The Australian Bureau of Statistics uses ICD-10 to conduct multiple cause coding of all death certificates in Australia. The underlying cause of death is defined as “the disease or injury which initiated the train of morbid events leading directly to death…”, while other morbid conditions are classified as intermediate/intervening causes of death or contributory causes.  For the purposes of analysis we created cause of death groups from the reported leading causes of death overall in Australia ; most common causes of death in the cohort of decedents with schizophrenia; and less common causes of death that were over represented in the schizophrenia cohort compared to the matched cohort (S1 Table).
Social and demographic variables
Marital status was classified as partnered (married or de-facto) or not/unknown at time of hospital admission, ED presentation and at time of death. Each decedent’s address was geocoded and used to assign accessibility categories based on the Australian ARIA+ index that takes into account road distance measurements to the nearest Service Centres and population size.  Socioeconomic status was estimated using quintiles of the Index of Relative Disadvantage (IRSD) which estimates the average disadvantage of small geographic areas.  Manual geocoding was also used to identify the type of residence at ED presentation, hospital admission and at death for decedents with schizophrenia and were classified as independent living (e.g. private residence); residential aged care facility (RACF); other care facility; no fixed address (e.g. homeless); and unknown/not stated.
Morbidity was defined as the presence of one or more of the 31 Elixhauser  conditions recorded in the primary or other 21 diagnostic fields during in-patient hospital stays over the last five years of life.  Morbidity burden was recorded as a temporal variable over the last year of life. For example, a decedent who had first mention of cancer on hospital admission at six months before death would have been recorded as having zero morbidity burden for the first six months of the last year of life before increasing to one after the cancer-related admission. We included psychoses in the calculation of total morbidity burden because this allowed us to differentiate health service use in decedents with schizophrenia before and after first mention of hospital admission for psychosis.
Outcome measure: Emergency department presentations
Emergency department (ED) presentations were assigned daily per person and only the first ED visit per day was included in the time-to-event analysis. The ED presentations as part of inpatient hospital transfers were excluded. Available ED data included the triage category, hospital admission status, specific diagnosis and presenting symptom. Coded presenting symptom data and ICD-10 coded diagnosis data were only available for metropolitan based hospitals (approximately 70% of ED presentations). Comparative analyses involving coded ED presenting symptom and diagnosis data excluded non-coded ED presentations from the estimation process.
Outcome measure: Hospital admissions
Hospital data were provided at the patient level in episodes of care. Multiple episodes of care were brought together into a single hospital admission when they included a statistical discharge (e.g. change of hospital care type but still in hospital) or transfer to another hospital. The principal diagnosis for the hospital admission was taken from the earliest episode of care when multiple episodes of care were present.
The Western Australian Home and Community Care (HACC) Program is a Commonwealth and State Government funded program that provides support services to frail older people and younger people with disabilities and their carers for a nominal fee. Services provided included basic nursing care (e.g. wound dressings), personal care (e.g. showering), domestic assistance (e.g. shopping, house cleaning), assistance with maintaining social support, client care coordination, allied health (e.g. podiatry), group activity centres, transport and meals. The types of services accessed by decedents in the last year of life was available, but patterns of service use over the last year of life could not be described due to lack of reliable date information.
Silver Chain WA is the main provider of community-based specialist palliative care in WA, although the most comprehensive service is restricted to major metropolitan areas. A team of palliative care clinicians and nurses, allied health professionals and volunteers provide home nursing care, counselling, respite options, practical support and links to other services with the aim of enabling people with life-limiting illness to remain at home. A palliative nurse consultancy service is available to residential care facilities where client care is managed by registered nurses. A palliative care rural telephone advisory service provides specialist advice to local rural service providers 24 hours per day. Study data included enrolment and discharge dates into each Silver Chain service and the number and length of time of home visits. A client may have had multiple periods of enrolment if their condition was of a relapsing and remitting nature. Life-limiting conditions considered amenable to palliative care for this study were cancers, heart failure, renal failure, liver failure, chronic obstructive pulmonary disorder, motor neurone disease, Parkinson’s disease and Alzheimer’s disease as described previously. 
Strata-weighted chi-square tests were used to assess equality of proportions. Differences between means were assessed with design-adjusted Wald tests. Multiple failure time-to-event analyses were performed to investigate the association of schizophrenia with acute care health service use in the last year of life. The multiple failure outcome events of interest were ED presentations and hospital admission any time in the one-year follow-up period. As cause of death is strongly associated with health service use, separate time-to-event analyses of the association of schizophrenia with health service use were performed for each cause of death. Decedents were excluded from the risk pool during periods of hospital stays.
A weighted kernel-density estimate of the hazard was used to graph unadjusted rates of ED presentations and hospital admissions over the last year of life. Time-to-event analyses were performed using flexible parametric Royston-Parmar models.  Variables that demonstrated temporal changes were treated as time-varying covariates including state of hospitalisation, number of comorbid conditions, partner status, residence in areas of relative disadvantage and level of accessibility to services. Standard errors were adjusted using the clustered sandwich estimator to account for correlated time between failure events. Matching strata weights were applied in all estimation procedures involving the matched comparison cohort. Variables used in the matching process were also included in regression analyses for full adjustment. The Sidak correction for multiple testing for each cause of death, assuming an alpha of 0.05 and 30 simultaneous comparisons, estimated a p-value of <0.0017 would demonstrate evidence of an association. Stata Statistical Software: Release 14 (Stata Corp, College Station, TX) was used.
Of the 63508 deaths in WA, 1196 (1.9%) of decedents had a lifetime history of schizophrenia. Decedents with schizophrenia were younger, lived in urban areas of more socioeconomic disadvantage, were more likely born in Australia and to have no partner at time of death compared to the general population of decedents (Table 1). A matched comparison cohort of 39265 decedents without a history of schizophrenia matched into 511 strata by age at death, sex, partner status at earliest record, indigenous status, residence in a major metropolitan area or not, socioeconomic disadvantage and country of birth was identified, each with a weighting relative to strata size. Two female decedents with schizophrenia could not be matched and were excluded from comparative analysis.
Amongst decedents with schizophrenia, the demographic profile varied by sex. The average age at first schizophrenia record was 28.2 (standard deviation (SD) 8.1) years for males and 30.1 (SD 8.4) years for females (p = 0.099). A greater proportion of female decedents with schizophrenia (n = 265, 46.7%) were living in a residential aged care facility at time of death compared to males (n = 136, 22.4%), whereas a greater proportion of males were homeless or without a fixed address at time of death (n = 10, 1.6%) compared to females (<5, 0.2%). More male decedents with schizophrenia (n = 433, 70%) were born in Australia compared to females (n = 350, 61%). A greater proportion of female decedents with schizophrenia lived in urban areas (n = 457; 80% versus n = 447; 72%) while a greater proportion of males with schizophrenia died aged less than 30 years (n = 43; 7% versus n = 7; 1%).
Causes of death
The most frequent principal causes of death in the schizophrenia cohort were cancers, ischaemic heart disease, intentional self-harm, chronic lower respiratory disease, dementias and accidental poisoning (Table 2). One third of decedents with schizophrenia had cancer recorded on their death certificate compared to more than half of all decedents in the matched cohorts. The exception was for breast cancer, where no difference between the cohorts was noted. The proportion of deaths in the schizophrenia cohort from diseases of the circulatory system were the same as that for the matched cohort. Amongst decedents with schizophrenia, a greater proportion of males had an underlying cause of death from intentional self-harm (n = 86, 14%) or accidental poisoning (n = 49, 8%) compared to females (n = 23, 4% and n = 15, 3% respectively, p<0.001).
The average total morbidity burden at the beginning of the last year of life was similar in the schizophrenia cohort and matched controls but the type of conditions varied (Table 3). Decedents with a lifetime history of schizophrenia had a higher proportion of uncomplicated diabetes, chronic pulmonary disease, neurological disorders, depression, hypothyroidism, obesity, alcohol and drug abuse and psychoses recorded in the five years prior to the last year of life relative to the matched cohort. The matched cohort had a significantly higher proportion of decedents with a history of cancer, circulatory disorders and liver disease at the start of the last year of life.
ED presentations over the last year of life
ED presentations were more frequent for male decedents with schizophrenia compared to matched male controls (Table 4). Decedents with a history of schizophrenia tended towards less urgent ED presentations which were more likely to occur for social or behavioural symptoms or related to alcohol use relative to the matched comparison cohorts, with the difference more marked in males. A greater proportion of decedents with schizophrenia were transported to ED by police or correctional services compared to the matched cohort.
The rate of ED presentations was not constant over the last five years of life (Fig 1A). The unadjusted rate of ED visits over the last year of life was consistently higher in the schizophrenia cohort compared to matched cohort until the last two months of life when the rate of ED presentation of the matched cohort increased rapidly. When stratified by sex, most of the increased rate of ED presentations in schizophrenia appear to be driven by males.
The average yearly (hazard) rate of ED presentations per decedent over the last year of life for the schizophrenia and matched cohorts and stratified by sex, estimated from A) an unadjusted kernel-weighted estimator of the hazard and B) an adjusted flexible parametric proportional hazards model. Values in B) predicted for Australian born non-Indigenous decedents living in a metropolitan area of average disadvantage without a partner, and aged 50 years or older at death. Sex and comorbidity were entered as interaction terms.
After adjusting for confounders and averaging over the last year of life there was no difference in the rates of ED presentation between decedents with schizophrenia and the matched cohort (HR 1.09; 95%CI 0.99–1.19). However, testing for interaction terms indicated that the degree of association was strongly modified by gender and by the number of comorbid conditions (Fig 1B). Amongst male decedents with no comorbid condition, the rate of ED use was greatest in males with schizophrenia, whereas amongst female decedents with no comorbidity, the rate of ED use was highest amongst females in the matched cohort. With two comorbid conditions, the rate of ED presentations was lower for both female and male decedents with schizophrenia compared to their gender matched cohorts and by four comorbid conditions, there was no significant differences between the sexes or cohort in rate of ED presentations.
We investigated whether the rates of ED presentation between decedents in the schizophrenia and matched cohorts varied by causes of death (Table 5). Of all decedents who had cancer, choking on food/solids or intentional self-harm mentioned anywhere on their death certificate, decedents with schizophrenia tended towards higher rates of ED use in the last year of life compared to decedents from the matched cohort. Amongst decedents who died from intentional self-harm, 40.1% (n = 107) of ED presentations for people with schizophrenia were because of social/behavioural or drug/alcohol problems, compared to 26.9% (n = 806) of ED presentations for people without schizophrenia (p<0.001). Of all decedents who had ischaemic heart disease or respiratory disease recorded as a morbid event on their death certificate, there was no difference in the rates of ED use in the last year of life between those with schizophrenia and those in the matched cohort, except perhaps a trend of lower ED use in decedents with schizophrenia who died with influenza or pneumonia.
In general, decedents with a history of schizophrenia demonstrated less frequent hospital admissions compared to matched comparison decedents, but they tended towards staying in hospital longer once admitted (Table 4). Relatively fewer females but not males with schizophrenia were admitted to hospital at least once compared to their gender matched cohort. Hospital admissions by decedents with schizophrenia were more often for psychogeriatric care, maintenance care and rehabilitation but less often for acute care or palliative care. Of decedents with schizophrenia who were admitted to hospital, more were emergency admissions, involved more days of psychiatric care, involved more intensive care (males only) or ventilator support (females only).
The average unadjusted rate of hospital admissions for the schizophrenia cohort was consistently lower than the matched comparison cohort over the last year of life (Fig 2A) a trend that was consistent for both males and females. After adjusting for confounders, the rates of hospital admission over the last year of life for decedents with schizophrenia remained consistently lower for both females and males than their gender matched cohort (Average HR 0.53; 95%CI 0.44–0.65).
The average yearly (hazard) rate of hospital admissions per decedent over the last year of life for the schizophrenia and matched cohorts and stratified by sex estimated from A) an unadjusted kernel-weighted estimator of the hazard and B) a flexible parametric proportional hazards model adjusted for age at death, comorbidity, socioeconomic status, accessibility index, sex, partner status, indigenous status and country of birth. Hazard rates and 95%CI in B are those predicted for Australian born non-Indigenous decedents living in a metropolitan area of average disadvantage without a partner, aged 50 years or older at death and with three comorbid conditions.
Rates of hospital admission generally showed a lower trend in decedents with schizophrenia for almost all causes of death, although limited statistical power was present in less common causes of death. Overall, decedents with schizophrenia who died from cancer had 40% lower rates of hospital admission in the last year of life compared to the matched cohort who also died from cancer, except for breast cancer where admission rates were similar. Lower rates of hospital admissions in the last year of life were observed for decedents with schizophrenia who had infections (influenza, pneumonia or bacterial sepsis), diseases of the circulatory system (ischaemic heart disease, heart failure and cerebrovascular disease), liver disease and pneumonitis recorded on their death certificate compared to decedents in the matched cohort with the same conditions. There were no causes of death where rates of hospital admission by decedents with schizophrenia was higher than the matched cohort, except a non-significant increase for those who died from intentional self-harm.
Community-based care in the last year of life
Just under 7% of all decedents with schizophrenia accessed community-based specialist palliative care in the last year of life compared to 16% of decedents in the matched cohort (Table 6). When restricted to decedents who died from conditions considered amenable to palliative care around 12% of decedents with schizophrenia and 25% of the matched cohort received community-based specialist palliative care. Decedents with cancer accessed the greatest proportion of community-based specialist palliative care in both cohorts.
The relative rate of ED presentation and hospital admissions during periods of time of receiving community-based specialist palliative care compared to periods of time not receiving this care were investigated on the subset of decedents with conditions amenable to palliative care (Table 7). Amongst decedents with schizophrenia, days enrolled in community-based specialist palliative care was associated with no change in rate of ED presentations and a 50% increase in rate of hospital admissions. Amongst decedents in the matched cohort, periods of time enrolled in the service was associated with reduced rates of ED admissions but no difference in rate of hospital admissions.
Around 42% of decedents in both the schizophrenia cohort and matched cohort were registered as clients of community-based (non-palliative) care services in last year of life. Decedents with schizophrenia were more likely to have received centre-based day care, counselling for both client and carer, social support, personal care and client management services compared to the matched cohort.
We report that health service use in the last of life for people with schizophrenia living in Western Australia is very different to that experienced by people without schizophrenia. People with schizophrenia died from different causes of death, had unique patterns of ED use, were half as likely to enrol in community-based specialist palliative care and were half as likely to be admitted to hospital in the last year of life compared to people without schizophrenia who had a similar sociodemographic profile.
Causes of death
The leading causes of death for people with schizophrenia were cancers, diseases of the circulatory system, intentional self-harm and accidental poisoning (drug overdoses) compared to an age-matched cohort, patterns similar to that reported in Canada , United States  and Sweden . Deaths from epilepsy were overrepresented in decedents with schizophrenia. Temporal lobe epilepsy has long been associated with a schizophrenia-like psychosis although understanding of the neuropathological and temporal mechanisms remain unclear.  Just over 2% of all decedents with schizophrenia had Parkinson’s disease recorded as the underlying or a contributory cause of death. Coexistence of schizophrenia with Parkinson’s disease is generally considered a rare condition with the hyperactive dopamine transmission observed in schizophrenia considered oppositional to the dopaminergic deficiency underlying Parkinson’s disease.  It is possible that drug-induced Parkinsonism, the result of anti-psychotic medication, was misclassified as Parkinson’s disease in most cases in our study.
The over representation of death from pneumonitis due to solids/liquids and choking in decedents with schizophrenia are not unexpected as disorders of swallowing are common and result from both the illness and the medication used to treat psychotic disorders . An increased number of deaths with intestinal obstruction were observed in decedents with schizophrenia, possibly associated with gastrointestinal hypomotility induced by some antipsychotic agents, such as clozapine. 
ED use in last year of life
The association of schizophrenia with rate of ED use in the last year of life was complex and varied with sex and comorbidity. Within the subset of decedents without a history of comorbidity, males with schizophrenia were the highest users of ED. The greater proportion of ED presentations involving police or corrective services and symptoms related to alcohol and drug abuse and behavioural problems in males with schizophrenia reflects common psychiatric comorbidities in this group. Decedents with schizophrenia who died of intentional self-harm presented to ED 50% more often in the last year of life than decedents without schizophrenia who also died from self-harm. More ED presentations for social/behavioural and drug and alcohol problems in people with schizophrenia who died from suicide suggests less stable schizophrenia and consequences of psychosis. However, as decedents with schizophrenia who died with cancer also had higher rates of ED presentation compared to the matched cohort, reasons for increased ED use are likely multifactorial and include patient, provider and system processes. People with schizophrenia and cancer may find it difficult to communicate medical history and health care providers may be more likely to attribute medical symptoms to the schizophrenia , potentially leading to repeat ED presentations.
In the subset of decedents who were dying with conditions amenable to palliative care, we observed that periods of time receiving community-based specialist palliative care was associated with reduced ED presentations for the matched cohort in the last year of life, as reported previously in Western Australia.  However, this was not the case in decedents with schizophrenia where no change in the rates of ED presentation during periods receiving and not receiving community-based specialist palliative care were observed. We speculate that contact with a specialist palliative care change was likely to have influenced the motivators of ED presentation in people with schizophrenia, however our data lacked the clinical detail and statistical power to identify if any subtle changes in the reasons for ED presentations during times with and without palliative care had occurred.
Hospital use in the last year of life
Overall the rate of hospital admission in the last year of life was 50% reduced in decedents with schizophrenia, larger than the 27% reduction in rate of hospital separations by people with schizophrenia in the last six months of life reported in a Canadian study.  This difference could partly be explained by methodology as we excluded person-time at risk while hospitalised.
Reduced rates of hospital admission for people with schizophrenia were evident across most causes of death with the strongest evidence observed for infections, cancers, most circulatory system diseases, dementias, depression, alcohol abuse and liver disease. The only causes of death where rates of hospital admission were similar between the cohorts were for breast cancer and intentional self-harm. While the average length of hospital stay was longer and involved more days of psychiatric care for decedents with schizophrenia, reduced rates of hospitalisation likely represent the same complex interaction of patient, individual clinician and healthcare system factors that contribute to disparate outcomes in other marginalised groups.
In the subset of decedents with schizophrenia who also had conditions amenable to palliative care, being enrolled in community-based specialist palliative care was associated with a 50% increase in rates of hospital admission. This finding suggests that receiving home-based specialist palliative care is reducing barriers to accessing hospital care. However, our data were insufficiently detailed to determine whether this increase reflected appropriate hospitalizations or instead reflected reduced access to appropriate care elsewhere in the community. Further work is required to address this issue.
Access to community-based care
We found evidence that people with schizophrenia who are dying with conditions amenable to palliative care were half as likely to receive specialist palliative care in the home compared to the matched cohort. This is similar to findings from New Zealand  and Canada.  Yet, we found that decedents with schizophrenia were just as likely to have accessed non-palliative community-based services in the last year of life as decedents without schizophrenia. Studies from countries with universal health care like Australia have also reported regular and increased access to GP services by people with schizophrenia. [12, 43] This suggests the barriers inhibiting receipt of palliative care for people with schizophrenia are at least partly health system driven rather than patient driven. Considering the vulnerability of people with schizophrenia to poorer general health outcomes and lower life expectancy, efforts to remedy this are essential.
Study strengths and limitations
Strengths of this study were the population based source of participants and record linkage across multiple different datasets. Limitations of the study included lack of clinical information and certainty related to schizophrenia diagnosis, particularly when coded in emergency department data or as a comorbid condition in hospital records. Our cohort of decedents with schizophrenia were selected from administrative records dating back to 1970 so misclassification of older decedents with schizophrenia who only had hospital records prior to this time was possible. It is also possible that comorbidity was either reported less frequently on medical records or was underdiagnosed for people with schizophrenia possibly leading to some differential bias. Lack of accurate date of service data for the community-based non-palliative care limited interpretation. Our methodological approach involved multiple regression models, thus interpretation of findings raises the issue of multiple comparisons. We did not make statistical adjustment to account for multiple testing in this study based on the view that readers will take the design of the study, the associated effect sizes, support from other published studies and our conservative interpretation into account. We also note that for less frequent failure events and small cause of death categories, the statistical power was only sufficient to detect very large differences between the two cohorts and that false positive findings were also possible.
In the last year of life, people with schizophrenia were less likely to be admitted to hospital, more likely to attend ED if male, less likely to access community-based speciality palliative care and had a different comorbidity and cause of death burden compared to matched decedents with a similar sociodemographic profile. Community-based specialist palliative care was associated with increased rates of hospital admission in decedents with schizophrenia. Our findings suggest that accessing community-based specialised palliative care can provide an alternative path to hospital care for persons with schizophrenia dying from conditions amenable to palliative care during the last year of life.
The authors wish to thank the staff at the Western Australian Data Linkage Branch and the custodians of the Hospital Morbidity Data Collection, the Emergency Department Data Collection, the Mental Health Information System, Silver Chain WA, Home and Community Care and the Register of Births, Deaths and Marriages for access to and linkage of their data.
- 1. Laursen TM. Life expectancy among persons with schizophrenia or bipolar affective disorder. Schizophr Res. 2011;131(1–3):101–4. pmid:21741216.
- 2. Tiihonen J, Lonnqvist J, Wahlbeck K, Klaukka T, Niskanen L, Tanskanen A, et al. 11-year follow-up of mortality in patients with schizophrenia: a population-based cohort study (FIN11 study). Lancet. 2009;374(9690):620–7. pmid:19595447.
- 3. Leng CH, Chou MH, Lin SH, Yang YK, Wang JD. Estimation of life expectancy, loss-of-life expectancy, and lifetime healthcare expenditures for schizophrenia in Taiwan. Schizophr Res. 2016;171(1–3):97–102. pmid:26811230.
- 4. Olfson M, Gerhard T, Huang C, Crystal S, Stroup TS. Premature Mortality Among Adults With Schizophrenia in the United States. JAMA Psychiatry. 2015;72(12):1172–81. pmid:26509694.
- 5. Holt RI, Abdelrahman T, Hirsch M, Dhesi Z, George T, Blincoe T, et al. The prevalence of undiagnosed metabolic abnormalities in people with serious mental illness. J Psychopharmacol. 2010;24(6):867–73. pmid:19304868.
- 6. Xiong GL, Bermudes RA, Torres SN, Hales RE. Use of cancer-screening services among persons with serious mental illness in Sacramento County. Psychiatr Serv. 2008;59(8):929–32. pmid:18678693.
- 7. Dickerson F, Stallings CR, Origoni AE, Vaughan C, Khushalani S, Schroeder J, et al. Cigarette smoking among persons with schizophrenia or bipolar disorder in routine clinical settings, 1999–2011. Psychiatr Serv. 2013;64(1):44–50. pmid:23280457.
- 8. Lambert TJ, Velakoulis D, Pantelis C. Medical comorbidity in schizophrenia. Med J Aust. 2003;178 Suppl:S67–70. pmid:12720526.
- 9. Irwin KE, Henderson DC, Knight HP, Pirl WF. Cancer care for individuals with schizophrenia. Cancer. 2014;120(3):323–34. pmid:24151022.
- 10. Hudson TJ, Owen RR, Thrush CR, Han X, Pyne JM, Thapa P, et al. A pilot study of barriers to medication adherence in schizophrenia. J Clin Psychiatry. 2004;65(2):211–6. pmid:15003075.
- 11. Muir-Cochrane E. Medical co-morbidity risk factors and barriers to care for people with schizophrenia. Journal of psychiatric and mental health nursing. 2006;13(4):447–52. pmid:16867129.
- 12. Chochinov HM, Martens PJ, Prior HJ, Kredentser MS. Comparative health care use patterns of people with schizophrenia near the end of life: a population-based study in Manitoba, Canada. Schizophr Res. 2012;141(2–3):241–6. pmid:22910402.
- 13. Lavin K, Davydow DS, Downey L, Engelberg RA, Dunlap B, Sibley J, et al. Effect of Psychiatric Illness on Acute Care Utilization at End of Life From Serious Medical Illness. Journal of pain and symptom management. 2017;54(2):176–85 e1. pmid:28495487.
- 14. Ganzini L, Socherman R, Duckart J, Shores M. End-of-life care for veterans with schizophrenia and cancer. Psychiatr Serv. 2010;61(7):725–8. pmid:20592010.
- 15. Spilsbury K, Rosenwax L, Arendts G, Semmens JB. The Association of Community-Based Palliative Care With Reduced Emergency Department Visits in the Last Year of Life Varies by Patient Factors. Annals of emergency medicine. 2017;69(4):416–25. pmid:28169049.
- 16. Sutradhar R, Barbera L, Seow HY. Palliative homecare is associated with reduced high- and low-acuity emergency department visits at the end of life: A population-based cohort study of cancer decedents. Palliative medicine. 2016. pmid:27507635.
- 17. Pouliot K, Weisse CS, Pratt DS, DiSorbo P. First-Year Analysis of a New, Home-Based Palliative Care Program Offered Jointly by a Community Hospital and Local Visiting Nurse Service. Am J Hosp Palliat Care. 2015. pmid:26656032.
- 18. Spilsbury K, Rosenwax L, Arendts G, Semmens JB. The impact of community-based palliative care on acute hospital use in the last year of life is modified by time to death, age and underlying cause of death. A population-based retrospective cohort study. PloS one. 2017;12(9):e0185275. pmid:28934324; PubMed Central PMCID: PMCPMC5608395.
- 19. Chapman M, Johnston N, Lovell C, Forbat L, Liu WM. Avoiding costly hospitalisation at end of life: findings from a specialist palliative care pilot in residential care for older adults. BMJ Support Palliat Care. 2016. pmid:27496356.
- 20. Miccinesi G, Crocetti E, Morino P, Fallai M, Piazza M, Cavallini V, et al. Palliative home care reduces time spent in hospital wards: a population-based study in the Tuscany Region, Italy. Cancer Causes Control. 2003;14(10):971–7. pmid:14750536.
- 21. Smith S, Brick A, O'Hara S, Normand C. Evidence on the cost and cost-effectiveness of palliative care: a literature review. Palliative medicine. 2014;28(2):130–50. pmid:23838378.
- 22. Ferroni E, Avossa F, Figoli F, Cancian M, De Chirico C, Pinato E, et al. Intensity of Integrated Primary and Specialist Home-Based Palliative Care for Chronic Diseases in Northeast Italy and Its Impact on End-of-Life Hospital Access. Journal of palliative medicine. 2016;19(12):1260–6. pmid:27697009.
- 23. Sahlen KG, Boman K, Brannstrom M. A cost-effectiveness study of person-centered integrated heart failure and palliative home care: Based on a randomized controlled trial. Palliative medicine. 2016;30(3):296–302. pmid:26603186.
- 24. National Coding Centre. International Classification of Diseases, 9th Revision Australian Clinical Modifications (ICD9-CM). Sydney: National Coding Centre, University of Sydney; 1995.
- 25. National Centre for Classification in Health. International Classification of Diseases, 10th Revision, Australian Modification (ICD-10-AM). Sydney: National Centre for Classification in Health; 2000.
- 26. Holman CD, Bass AJ, Rouse IL, Hobbs MS. Population-based linkage of health records in Western Australia: development of a health services research linked database. Aust N Z J Public Health. 1999;23(5):453–9. pmid:10575763.
- 27. Blackwell M, Iacus S, King G, Porro G. cem: Coarsened exact matching in Stata. Stata Journal. 2009;9(4):524–46. WOS:000273272200002.
- 28. Pink B. Information Paper: Cause of Death Certification, Australia. Canberra: Australian Bureau of Statistics, Commonwealth of Australia; 2008.
- 29. Australian Bureau of Statistics. Causes of death, Australia 2011. Canberra: Australian Bureau of Statistics; 2013.
- 30. Glover J, Tennant S. Remote Areas Statistical Geography in Australia: Notes on the Accessibility/Remoteness Index for Australia (ARIA + version).2003 07/07/2017; Working Paper Series No. 9. Available from: http://phidu.torrens.edu.au/publications-1999-to-2004/remote-areas-statistical-geography-in-australia-notes-on-the-accessibility-remoteness-index-for-australia.
- 31. Pink B. Information Paper: An Introduction to Socio-Economic Indexes for Areas (SEIFA) 2006. Canbera: Australia Bureau of Statistics (ABS); 2008.
- 32. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998;36(1):8–27. Epub 1998/02/07. pmid:9431328.
- 33. Quan H, Sundararajan V, Halfon P, Fong A, Burnand B, Luthi JC, et al. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care. 2005;43(11):1130–9. pmid:16224307.
- 34. Rosenwax LK, McNamara B, Blackmore AM, Holman CD. Estimating the size of a potential palliative care population. Palliative medicine. 2005;19(7):556–62. pmid:16295289.
- 35. Royston P, Parmar MK. Flexible parametric proportional-hazards and proportional-odds models for censored survival data, with application to prognostic modelling and estimation of treatment effects. Stat Med. 2002;21(15):2175–97. pmid:12210632.
- 36. Martens PJ, Chochinov HM, Prior HJ. Where and how people with schizophrenia die: a population-based, matched cohort study in Manitoba, Canada. J Clin Psychiatry. 2013;74(6):e551–7. Epub 2013/07/12. pmid:23842025.
- 37. Crump C, Winkleby MA, Sundquist K, Sundquist J. Comorbidities and mortality in persons with schizophrenia: a Swedish national cohort study. Am J Psychiatry. 2013;170(3):324–33. Epub 2013/01/16. pmid:23318474.
- 38. Kandratavicius L, Hallak JE, Leite JP. What are the similarities and differences between schizophrenia and schizophrenia-like psychosis of epilepsy? A neuropathological approach to the understanding of schizophrenia spectrum and epilepsy. Epilepsy & behavior: E&B. 2014;38:143–7. pmid:24508393.
- 39. Lan CC, Su TP, Chen YS, Bai YM. Treatment dilemma in comorbidity of schizophrenia and idiopathic Parkinson's disease. Gen Hosp Psychiatry. 2011;33(4):411 e3–5. Epub 2011/07/19. pmid:21762841.
- 40. Kulkarni DP, Kamath VD, Stewart JT. Swallowing Disorders in Schizophrenia. Dysphagia. 2017;32(4):467–71. Epub 2017/04/28. pmid:28447217.
- 41. Every-Palmer S, Ellis PM. Clozapine-Induced Gastrointestinal Hypomotility: A 22-Year Bi-National Pharmacovigilance Study of Serious or Fatal 'Slow Gut' Reactions, and Comparison with International Drug Safety Advice. CNS Drugs. 2017;31(8):699–709. Epub 2017/06/18. pmid:28623627; PubMed Central PMCID: PMCPMC5533872.
- 42. Butler H, O'Brien AJ. Access to specialist palliative care services by people with severe and persistent mental illness: A retrospective cohort study. Int J Ment Health Nurs. 2017. Epub 2017/07/12. pmid:28692186.
- 43. Hetlevik O, Solheim M, Gjesdal S. Use of GP services by patients with schizophrenia: a national cross-sectional register-based study. BMC Health Serv Res. 2015;15:66. Epub 2015/04/18. pmid:25884721; PubMed Central PMCID: PMCPMC4339084.