Article Text
PDF
Abstract
Objectives To determine incidence of death in residential care facilities for people with disability in Ireland, primary cause of death, associations of facility characteristics and deaths, and to compare characteristics of deaths reported as expected and unexpected.
Design Descriptive cross-sectional study.
Setting All residential care facilities for people with disability operational in Ireland in 2019 and 2020 (n=1356).
Participants n=9483 beds.
Main outcome measures All expected and unexpected deaths notified to the social services regulator. Cause of death as reported by the facility.
Results 395 death notifications were received in 2019 (n=189) and 2020 (n=206). 45% (n=178) were for unexpected deaths. Incidence of death per 1000 beds per year was 20.83 for all, 11.44 for expected and 9.39 for unexpected deaths. Respiratory disease was the most common cause of death, accounting for 38% (n=151) of all deaths. In adjusted negative binomial regression analysis, congregated settings versus non-congregated (incidence rate ratio (95% CI): 2.59 (1.80 to 3.73)) and higher bed numbers (highest vs lowest quartile) (4.02 (2.19 to 7.40)) were positively associated with mortality. There was also a positive n-shaped association with category of nursing staff-to-resident ratio when compared with zero nurses. Emergency services were contacted for 6% of expected deaths. Of the deaths reported as unexpected, 29% were receiving palliative care and 10.8% had a terminal illness.
Conclusion Although incidence of death was low, residents of congregated and larger settings had higher incidence of death than residents of other settings. This should be a consideration for practice and policy. Given the high contribution of respiratory diseases to deaths and the potential avoidability of these, there is a need for improved respiratory health management in this population. Nearly half of all deaths were reported as unexpected; however, overlap in the characteristics of expected and unexpected deaths highlights the need for clearer definitions.
- health policy
- quality in health care
- public health
Data availability statement
Data are available in a public, open access repository. Data used in this study are from the Database of Statutory Notifications from Social Care in Ireland. The Irish Database of Statutory Notifications from Social Care is a national data collection of the statutory notifications received and generated by the Health Information and Quality Authority. An open access version of this database has been published for research and public use and is available at https://www.hiqa.ie/sites/default/files/2021-02/Database-of-statutory-notifications-from-social-care-in-Ireland-2013-2020.csv; this contains desensitised and pseudonymised facility-level data. The unredacted database is not available publicly but access to these data is possible following data custodian approval.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
STRENGTHS AND LIMITATIONS OF THIS STUDY
National-level data were used, comprehensive for all active residential care facilities in Ireland during the study period.
Cause of death ascertainment was from the cause reported by the person in charge of the residential care facility and not from death certificates.
There was a possibility of over-reporting of unexpected deaths which possibly should have been more accurately reported as expected deaths. This limits the conclusions that can be drawn from the analysis of association of service characteristics and expected and unexpected deaths.
Lack of demographic data of residents who died limited the ability to account for confounding; however, proxies were used for age and resident dependency level.
A novel framework for evaluating whether a death is expected or unexpected was developed and used for this study, shedding light on a difficult area to measure for the first time.
Introduction
People with disability experience poor health outcomes and have higher all-cause mortality rates than the general population.1 2 While adults with intellectual disabilities are now living longer,3 they still die earlier4 5 than people without intellectual disability.6 Such differences are identified as health inequalities, and improving life expectancy for this population is a key international priority.7–9 Although there has been an increase in research examining mortality in populations with disability, the focus of recent inquiry has concentrated on populations with intellectual disability and there has been limited focus on the persons’ living arrangements, that is, living in the general population or in care settings, prior to their death.10 11
In Ireland, Doyle et al9 recently outlined the importance of examining the living arrangements of people with intellectual disabilities prior to death. In their study, which included 4006 deaths over 16 years, the odds of dying in congregated settings (where 10 or more people with a disability live together in a single living unit or any number of residents are placed in accommodation that is campus based) in Ireland were four times higher than those living with family carers. Even after accounting for the severity of intellectual disability, the setting a person lived in was a significant contributor to death. In the UK, recent studies have examined the mortality of people with intellectual disability in community-based and residential settings12–15 to determine the factors that may influence death. These studies identified that while deaths were often not anticipated and uncommon, they occurred at a younger age and were most commonly caused by respiratory disease. These studies have highlighted the need to undertake descriptive epidemiological studies to understand how people are dying, where they are dying and if they are dying unexpectedly to inform the development of improved interventions. This is particularly important given the various, often atypical living situations that many people with a disability live in.16 It is also important to include those with disabilities other than intellectual, who also live in these settings.
There is a small body of research that has specifically examined the primary cause of death of people exclusively living in residential care facilities or identified if deaths were expected or unexpected in these settings (online supplemental figure 1 and online supplemental table 1). Nevertheless, it is well documented that mortality for people with an intellectual disability living in residential settings is higher than the general population or among individuals with intellectual disability living elsewhere.14 17 18 Hosking et al reported that the mortality rate among individuals with an intellectual disability living in communal or shared living arrangements in England was four times higher than the general population.17 Equally, a prospective longitudinal cohort study of people with Down syndrome identified that at follow-up, persons living in institutions were more likely to have died.19 Durvasula et al18 also reported disproportionate rates for individuals with intellectual disability living in residential settings and highlighted that 50% of deaths in their study occurred in institutional settings, despite this cohort accounting for only 25% of the study population. Similarly, the mortality rate reported by Merrick20 was 9.9 per 1000 population for individuals with an intellectual disability living in residential settings in comparison with a mortality rate of 6.6 per 1000 population in the general population. Within residential settings, the cause of death for individuals with intellectual disability appears to be similar in nature, despite heterogeneity in study methods and population demographics. For example, respiratory diseases are reported as a significant cause of death in the majority of studies,14 17 19–23 with the percentages reported ranging from 18.8%17 to just under 50%.21 23 Circulatory diseases also account for between 7%19 and 35% of deaths,20 with cancers reported to range from 6%19 20 to 20%24 of deaths. However, these studies are focused on people with intellectual disability and do not account for the case mix in residential care services. Furthermore, there is a paucity of comprehensive national-level data from any country.
Supplemental material
Supplemental material
Over the last decade, some studies have also drawn attention to the concept of unexpected death in this area. Two separate UK studies have identified that unexpected death for people with intellectual disability who live in social care settings can range from 43% to 64% of all deaths.12 25 Nevertheless, the evidence base regarding unexpected death in this population from an international perspective is underdeveloped and largely absent. From an Irish viewpoint, a similar pattern exists. One reason for this potentially lies in what constitutes an ‘unexpected death’. For example, in Ireland, within the Coroners Amendment Act 2019,26 and regulations made under the Health Act 2007 which govern residential care facilities for people with a disability in Ireland, no explicit definition of ‘unexpected death’ is offered. Rather, responsibility is placed on the person in charge of the residential care facility to ‘use judgement’ when reporting whether the death was unexpected or not.27 Subsequently, the evidence base regarding unexpected death in Ireland for people with a disability is vacant. Therefore, it is critically important to understand the epidemiology of death in this population as such evidence may underpin the development of targeted interventions and reduce the mortality inequality that this population experiences.
Aim of study
The aims of this study were to (1) to identify the incidence of expected and unexpected death of people with disability living in residential care facilities in Ireland; (2) categorise the reported primary cause of death of people with disability using International Classification of Diseases 11th Revision (ICD-11) chapter headings;28 (3) examine the association of characteristics of residential care facilities and deaths; and (4) examine the characteristics of deaths reported as expected and unexpected.
Methods
Setting
This descriptive cross-sectional study was undertaken in Ireland and incorporated all residential care facilities for people with a disability that were operational during 2019 and 2020.
Population
At the end of 2020, there were 1356 residential disability care facilities registered by the regulator to provide a service to circa 9000 people with a disability.29 By law, all residential facilities must be inspected and registered by the regulator to be allowed to operate in Ireland. As residential care facilities may have de-registered or registered during the study time frame, the number of facilities registered and their registered bed numbers on the last day of 2020 were used as the total sampling frame (n=1356). Residential care facilities in Ireland for people with a disability incorporate both accommodation and care and support services on a short-term and long-term basis, whether or not it is their sole place of residence.
Patient and public involvement
In 2021, phone interviews were conducted with residents in residential care facilities for people with disability (n=7) during the development of the Database of Statutory Notifications from Social Care in Ireland (the database used for this analysis).30 31 Residents were asked for their opinions on publishing the data open access and using it for research purposes. Residents were also asked about the type of questions they would like researchers to ask of the data, who dissemination of the findings should target and dissemination formats that they considered appropriate for residents. Resident input was used to inform the prioritisation of topics for analyses using the database and to inform dissemination strategies and formats, including for this analysis.
Data
The 2007 Health Act32 provides the legislative framework for the submission of statutory notifications for residential care facilities. Within these facilities, all deaths, both expected and unexpected, are notified to the regulator. Expected deaths must be reported on a quarterly basis and unexpected deaths must be reported within 3 working days of the event.27
Data used in this study are from the Database of Statutory Notifications from Social Care in Ireland and are publicly available33 (hereafter ‘the database’). This contains data on all statutory notifications received by the regulator from residential care facilities for older persons and for people with disability from 2013 onwards.30 These analyses were limited to facilities for people with disability.
Total, unexpected and expected deaths
A total of 1356 residential disability care facilities were active on 31 December 2020. All notifications of expected and unexpected deaths reported to the regulator with a date of death in the years 2019 and 2020 were included. A 2-year time frame was selected to allow for adequate sample size for analyses. Deaths were coded as per the notification submitted to the regulator by the person in charge (expected or unexpected). As these notifications reported the event of death only, no personal demographic data of the person who died were available.
Cause of death
The reported primary cause of death was extracted from the notifications and coded using the ICD-11 chapter headings.34 The ICD-11 chapter headings is a tabular list organised into chapters according to body system or condition, with diagnosis codes listed alphanumerically in each chapter. As the statutory notifications do not require the certified cause of death to be reported as registered on a death certificate, for the purpose of this research, chapter headings according to body system are only reported on. This process was undertaken independently by two researchers who are also health professionals (MJM and AMO). First, the reported cause of death was extracted independently from the notification submitted by the registered provider. This information was then independently entered into the ICD-11 website for Mortality and Morbidity Statistics (V.2021-05).35 Where the SARS-CoV-2 was diagnosed and identified as the cause of death, this was coded under its own ICD-11 chapter heading, RA01.0, following guidance at that time for the emergency use of ICD codes for COVID-19.36 Second, when the reported cause of death was identified in the web browser, the researchers then identified the monohierarchical chapter heading according to body system of which each reported cause of death was classified within. This was then used to classify the cause of death. This process was undertaken for all deaths and an inter-rater reliability agreement of 98.8% was achieved between MJM and AMO. Third, the coding and classification of cause of death were verified for accuracy by a third researcher (LB) who randomly reviewed 50% of the cause of both expected and unexpected deaths. No discrepancies or errors were identified by LB.
Service characteristics
Service-level data available in the database were used to examine characteristics that were associated with deaths in residential care facilities. These were: care facility profile (provision for adults, children, mixed), sex (single sex, mixed), provider type (voluntary/privately operated or public health service operated in which services are provided directly by the state which includes the state employing the staff), provider funding type (Section 38 (these are typically voluntary agencies that are funded to provide a defined level of service on behalf of the government), Section 39 (these are typically voluntary agencies that are grant aided by the government), private, public health service), provider size (small, medium, large), intellectual disability service (yes, no), congregated setting (congregated vs non-congregated), size of service (number of registered beds), distance to nearest acute hospital (km), staff number and nursing staff number. Staff-to-resident ratio and nursing staff-to-resident ratio were calculated as staff number/number of registered beds at residential care facility level. Residential care facilities were determined to be a congregated setting when they had 10 or more beds in a single unit or were a campus-based setting, irrespective of the number of beds. In Ireland, the term ‘congregated setting’ is used as opposed to ‘institutional setting’ in national policy; however, for international comparison, these definitions may be viewed as synonymous particularly where settings have increased bed numbers.
Free-text data
Free-text data from the notifications were also extracted. These outlined the description of the circumstances leading up to the death and any medical intervention(s) that was made available. This narrative also included risk comments, inspector comments, actions taken, additional details and reasons why the cause of death was unexpected.
Statistical methods and analysis
Characteristics of services were described for all services and for services that submitted a notification as n, %, mean±SD and median (IQR), as appropriate. The number of total deaths and number and per cent contribution to total, of expected and unexpected deaths were calculated for the 2-year period and by year. The number and per cent of services that submitted a notification of death were calculated. Incidence of total, expected and unexpected deaths per 1000 beds per year was estimated as (deaths for the 2-year period/registered bed numbers)×1000)/2. The mean number of deaths reported by services by year was estimated and the difference estimated using a t-test.
Deaths were categorised by primary cause of death and expressed as n (% contribution to all, unexpected and expected deaths) and incidence per 1000 beds, for the total study period. The association between service characteristics and total deaths in the 2 years in residential care facilities was evaluated using negative binomial regression for each service characteristic separately. Three models were constructed: model 1—unadjusted; model 2—adjusted for bed numbers (as a proxy for number of residents), care facility profile (as a proxy for age of residents) and nurse-to-resident ratio (as a proxy to account for differences in dependency levels of residents that impact both the structure of a residential care facility and the risk of mortality); and model 3—additionally mutually adjusted for the other service characteristics examined. Bed numbers and nursing staff-to-resident ratio were collapsed into quartiles and examined as size of service and nursing staff/resident ratio category to reduce collinearity with the linear bed numbers and nurse-to-resident ratio included as covariates.
All exposures were examined for interaction with each other by inclusion of each in model 3. No interactions were significant (p≥0.324) and no interaction terms were included in the final models. An a priori agreed analysis for interaction of expected and unexpected deaths was conducted. All interactions were significant (p<0.001). As such, analyses were additionally conducted, stratified by expected and unexpected deaths, using model 3 as described above.
Goodness of fit for all models was tested using a likelihood-ratio test of alpha=0. The probability of chibar2 being greater or equal to 0 was <0.001 for all models.
The database was complete for services and notifications. Continuous variables included in the regression analysis were imputed to account for missingness, with missing data points for continuous variables replaced with the respective mean. Using means to account for missingness was considered appropriate as the distributions of variables were normal. Regression analyses were repeated in complete cases only to examine the effect of the imputation.
Statistical analyses were performed in R software (V.4.1.2) and Stata (StataCorp, V.16).
Qualitative methods and analysis
A categorical content analysis was conducted to describe characteristics of deaths notified to the regulator as expected or unexpected and to generate an understanding of similarities and differences between these reported notifications. We applied a deductive analysis approach using a de novo predefined coding framework (online supplemental table 2) to analyse these notifications through categorical content analysis. The framework consisted of 13 categories and rules which were informed by previous research conducted on unexpected deaths in adults with intellectual disability,12 the National Policy for Pronouncement of Expected Death by Registered Nurses,37 the Coroner’s Act38 and the WHO’s definition of sudden death.39 This literature informed expected or unexpected death indicators within the coding framework. For example, the category of a person in receipt of palliative care was informed by the National Policy for Pronouncement of Expected Death by Registered Nurses37 that considers the provision of palliative care as an indicator of an expected death. Following the conceptualisation of the framework, the free text from each notification was coded according to the pre-established coding rules (online supplemental table 2). Upon completion of the coding, the categories were grouped into two main concepts: interventions and care plans, and trajectory of illness before death.
Supplemental material
Results
Incidence of death
Two hundred twenty-five (16.6% of all residential care facilities) notified the regulator of a death during the 2-year period. This was consistent across years (2019, 112 residential care facilities; 2020, 113 residential care facilities). Residential care facilities that submitted notifications were more likely to be congregated settings, provide for adults, be funded under Section 38 of the Health Act 2004,40 have higher numbers of registered beds, have higher staff numbers and nursing staff numbers than residential care facilities that did not submit a notification (p<0.05) (table 1). In total, 395 notifications of deaths were received by the regulator in 2019 (n=189) and 2020 (n=206). Of these, 45% (n=178) were unexpected deaths. There was no difference in the mean number of deaths reported by year by service (2019, 1.9; 2020, 1.6, p=0.251). The % contribution of expected and unexpected deaths to total deaths did not differ across the 2 years (2019, 42% unexpected; 2020, 48% unexpected; Χ2=1.5599, p=0.212). Incidence of death per 1000 beds per year was—deaths: 20.83, expected deaths: 11.44, unexpected deaths: 9.39 (table 2).
Description of service characteristics of residential care facilities for people with disability in Ireland, 2019–2020
Cause of death as notified by residential care facilities for people with disability in Ireland, 2019–2020, n notifications=395
Primary cause of death
The most frequent primary cause of death reported in the notifications were diseases of the respiratory system. These accounted for 38% (n=151) of all deaths (table 2). This was followed by neoplasms (19%, n=75) and diseases of the circulatory system (11%, n=42). The leading causes of expected deaths were diseases of the respiratory system (41%, n=89), neoplasms (31%, n=67), and mental and behavioural disorders (eg, dementia) (10%, n=18). The leading causes of unexpected deaths were diseases of the respiratory system (35%, n=62) and diseases of the circulatory system (18%, n=32).
Association of characteristics of residential care facilities and death
In unadjusted analysis (model 1) of service characteristics and all deaths, congregated settings, providing for mixed sex, being a statutory provider, being funded by the public health service, having higher bed numbers, and having higher staff-to-resident ratio and nursing staff-to-resident ratio were associated with higher incidence of death (table 3). Providing for children only and being funded under Section 39 of the Health Act were associated with lower incidence of death. After adjustment for bed numbers, nursing staff-to-resident ratio and care facility profile (model 2), the associations with sex provision, provider type, funding type and staff-to-resident ratio were attenuated to null. After further mutual adjustment for other service characteristics (model 3), the positive association of congregated setting versus non-congregated (incidence rate ratio (95% CI): 2.59 (1.80 to 3.73)) and higher bed numbers (highest vs lowest quartile) (4.02 (2.19 to 7.40)) remained significant. There was also a positive non-linear association with category of nursing staff-to-resident ratio when compared with zero nurses. The second highest quartile (0.5 to 0.99 nursing staff:resident ratio) was associated with the highest risk (4.35 (2.21 to 8.55)).
Association of service characteristics and mortality in residential care facilities for people with disability in Ireland, n centres=1356
When stratified by expected and unexpected deaths, using model 3, congregated setting and higher bed numbers had a larger positive association with expected than unexpected deaths, and a positive association of nursing staff-to-resident ratio and mortality was only significant for expected deaths (table 4). Being a statutory provider was also positively associated with unexpected deaths.
Association of service characteristics with expected and unexpected mortality in residential care facilities for people with disability in Ireland, n centres=1305
Repetition of analyses in complete cases did not materially change results (online supplemental table 3).
Supplemental material
Characteristics of expected and unexpected deaths from the categorical content analysis
Concept 1: interventions and care plans
Categories relating to interventions and care plans were identified in all notifications received (table 5). A total of 48.1% of all deaths reported were for residents in receipt of palliative care. There was the presence of categories in expected deaths that would not be anticipated, for example, emergency services were called for 6.0% of reported expected deaths and 28.7% of reported unexpected deaths were for residents in receipt of palliative care.
Categorical content analysis of characteristics of deaths from free text included in statutory notifications from residential care facilities for people with disability in Ireland, 2019–2020, n notifications=395
Concept 2: trajectory of illness before death
Categories relating to the trajectory of illness before death were identified in all notifications (table 5). For length of illness, there was a spread across all categories of length for both expected and unexpected. Longer illnesses were reported more as expected deaths and shorter as unexpected, although some deaths from sudden illnesses were reported as expected deaths and some deaths from terminal illnesses were reported as unexpected deaths.
A total of 21.2% of expected deaths and 59.6% of unexpected deaths reported hospital admission prior to death. Expected deaths tended to have had longer stays in hospital with zero deaths within 3 days reported.
Discussion
This study provides national population-based evidence about the mortality of people with disability who live in residential care facilities in Ireland. As far as we are aware, this is the only nationally representative study to date that includes the deaths of all people with a disability who live in residential care facilities. In addition, in the absence of a legal definition in Ireland, this study provides a novel framework for evaluating the characteristics of expected and unexpected deaths. These make important contributions to this area given the proportion of people with disabilities who live in residential care facilities internationally.
Similar to Todd et al’s study in the UK,14 our study found that death is not a common event in disability services with just over 16% of residential care facilities reporting a death during 2019 and 2020. Although the incidence of death may be considered low, we report an association of risk of death and living in a congregated setting, higher bed numbers and nursing staff-to-resident ratio (second highest quartile (0.5 to 0.99 nursing staff:residents)) when compared with zero nurses. The association of congregated setting and higher bed numbers and death was larger for expected than unexpected deaths, and the association of nursing staff-to-resident ratio and mortality was only significant for expected deaths. However, due to the overlap in the characteristics of expected and unexpected deaths (which were identified through the categorical content analysis), there is a potential that incorrect classification by the service provider may impact these results and should be interpreted with caution. Other studies have reported similar findings regarding the association of congregated settings and death.9 14 19 Although it may be argued that people with more severe disabilities are more likely to live in residential care facilities, our analyses accounted for this by adjusting for nursing staff-to-resident ratio. Even after adjusting directly for severity of disability, another Irish study9 had similar findings insofar as the severity of disability did not account for increased deaths in such settings. This has important implications for policy in Ireland and in other countries where there is a bias towards institutional and other forms of residential care for people with a disability.29 41 This finding also signals an important avenue for future research given that larger datasets, of which much of the current mortality data are drawn, largely do not always identify the place of residence. In terms of the mechanisms by which these service characteristics are associated with mortality, when thinking about patient/resident outcomes, determinants are oftentimes separated into process measures and organisational structures.42 Organisational structures can impact on the quality of care provided through culture, management structures and physical environment impacting on processes and in turn on quality of care.42 As such, service characteristics are unlikely to impact directly on mortality, but they are a modifiable characteristic that can, through processes, impact.
Ireland’s legislative framework does not set out any minimum requirements of staff that must be on duty nor specifies staffing ratios. As provided in regulations made under the Health Act 2007, staffing requirements are largely determined based on the number and profile of the people living in these residential care facilities and their assessed needs. Regarding staff ratios, however, the findings of our study show an association of nurse-to-resident ratio and expected death, a finding tentatively expected as the increasing complexity of resident needs may be associated with increased nurse-to-resident ratios. Nevertheless, while unexpected deaths were not significant, our results indicate the existence of an n-shaped relationship between nursing staff-to-resident ratio and unexpected death. This may imply that unexpected deaths may be theoretically reduced by increasing nurse-to-resident ratios. An important caveat exists given the non-significance; however, this is a very valuable area for further research to consider given the important role nurses play in reducing health inequalities that this population experiences.43 For example, there is extensive literature that highlights the discrimination this population faces when accessing healthcare,44–48 and it is theoretically possible that nurses in these environments act as a broker with specialist knowledge between disparate health and disability systems49 50 and as a result play an important role in reducing health inequalities that contribute to mortality.
Primary cause of death
Respiratory disease was the largest cause of death in this study, as well as circulatory system diseases. While there were no differences observed between intellectual disability versus non-intellectual disability services in this study, the results match the experience of death in the intellectual disability cohort more broadly as diseases of the respiratory system accounted for 38% of all deaths,6 8 11 18 21 51 while neoplasms and diseases of the circulatory system accounted for 19% and 11% of deaths, respectively.4 10 24 Given the high levels of dysphagia in this population,52 deaths from respiratory disease are potentially predictable given the serious health complications associated with swallowing difficulties that this population experiences. Although our study illustrated that people with a disability die from the same conditions as the general Irish population, the cause of death has a different pattern insofar as neoplasms, disease of the circulatory and respiratory systems are the leading causes, in this order, for the general Irish population.53
Nearly half of all deaths reported were unexpected and it has been reported that such deaths are higher in people with an intellectual disability,25 particularly those with epilepsy.54 55 There are two other potential reasons for this: first, deaths in intellectual disability services are rare. In our study, only 16.6% (n=225) of residential care facilities notified a death over the 2-year period. Similar results have been found in the UK.14 Consequently, this uncommon event is more likely to hasten residential care facilities reporting to the regulator as soon as possible and therefore use the 3-working day notification which then classifies a death as unexpected. This is reinforced by the findings from our categorical content analysis that 28.6% of all unexpected deaths were currently receiving palliative care. Second, there is no definition of what constitutes an unexpected death in Irish legislation,26 56 and while the regulator cites an unexpected death as one that was not anticipated or occurred earlier than expected,57 in the absence of a clear legal definition, there is likely to be ambiguity. While a recent UK study from Bernal et al12 in this area distinguishes deaths derived from guidance from Blackmore et al,58 the Irish context does not permit similar comparison due to reporting mechanisms. Nonetheless, a clear definition in this area is essential to help understand unexpected deaths and facilitate the instigation of appropriate levels of inquiry.59 In comparison with the general Irish population where nearly 45% of all deaths occur in hospital, the categorical content analysis indicated that 38% of all deaths occurred in hospital in our study of which nearly 60% were unexpected. This finding is similar to another UK study which found that 56.4% of all deaths of people with an intellectual disability occurred in hospital; however, it is consistent with the findings that of people who die unexpectedly, they are more likely to die in hospital.14
Deaths from respiratory diseases are likely to reflect underlying vulnerabilities in this population and while it is the leading cause of death in this study, other studies have reported that deaths from cardiovascular disease are now the most common cause of death in this population.17 20 24 60 In this context, many studies have drawn data from multiple sources that include death certification, national registers and local registers. Consequently, as our study concentrates solely on residential care, this may signal an area that requires immediate public health action to mitigate potentially avoidable deaths from respiratory disease in these environments.11 51 Older studies that use residential samples have had similar findings to our study and this may suggest that health improvements in this area have been slow.22 23 61 The evidence is largely consistent regarding increased rates of mortality in congregated settings.14 19 24 While it is not clear if congregated settings are a cause or contributor to mortality, it is evident that there are poorer outcomes in such environments. This underlines the critical need to move to smaller-scale living environments for people with a disability.62
In terms of unexpected deaths, the evidence in this area is scarce, and while both Heslop et al25 and Bernal et al12 had higher incidence of unexpected death in their studies (43% and 64%), findings from our categorical content analysis indicate that in approximately 27% (n=51) of all unexpected death notifications submitted, the person who died was receiving palliative care. This raises the question of if such deaths are genuinely unexpected. From this perspective, it is important that death is anticipated and there is a need to develop measures that identify frailty63 and predict death64 to help improve the end-of-life experiences of this population.12
Strengths and limitations
The primary strength of this study is the nationally representative data that include all deaths of persons with a disability who live in residential care facilities in Ireland, in a 2-year period. As it is a legal duty65 to report each death to the regulator, this is a complete dataset. A second strength is the development and use of a novel framework for evaluating whether a death is expected or unexpected; this sheds light on a difficult area to measure for the first time and has been developed to be reusable by others.
Limitations of the study warrant consideration. There was potential for over-reporting of unexpected deaths. Given the infrequency of death that occurs in residential care facilities, facilities may report a death as unexpected where it may be more correctly classified as expected. Cause of death ascertainment was from the cause reported by the person in charge of the residential care facility and not from death certificates. In the absence of death certificates, we relied on researchers extracting and coding deaths as per the ICD-11 monohierarchical chapter heading according to body system. While this is useful in terms of higher-order coding, if deaths were coded at the underlying cause to fully distinguish and assign the external or pathological causation of death, this would offer a greater insight into the cause of deaths of people with a disability who live in residential services in Ireland. Additionally, the researchers did not receive any formal training for coding purposes and this limitation also needs to be factored in when considering the reliability of the findings. The absence of demographic data of the residents who died to account for associations in the analysis, particularly severity of disability, is an important limitation to acknowledge. While our analysis did use proxies of nursing staff-to-resident ratio for resident dependencies and care facility profile for age, this cannot fully mitigate the absence of these demographic data. In particular, given the known association of age and mortality risk, the implications of care facility profile (under 18 years, over 18 years and any age), to account for age, should be factored into the interpretation of the findings, as it may be that congregated settings have an older population than other settings. That said, we have no evidence to suggest that they do. Nevertheless, this study is representative of the entire population of people with a disability who live in residential care facilities and this significantly enhances the evidence in this regard. As the study period includes the years 2019 and 2020, years where SARS-CoV-2 was not and was in circulation, respectively, there is a possibility that excess mortality from COVID-19 may have impacted on the analyses. We considered this carefully in the design of the analysis and as COVID-19 reported deaths were low (n=18) and there was no difference in mean total deaths reported by services between the 2 years (2019, 1.9; 2020, 1.6; p=0.251), we did not take any actions such as including fixed or random effects in models or exclusions from cause of death analyses.
Implications
Results from this study show the importance of examining the role that congregated settings have on the deaths of people with disability. While this study has not investigated the mechanisms of how congregated settings are associated with death, future research should examine the role the place of residence has on mortality. Additionally, there is a need to develop and implement evidence-informed strategies to reduce respiratory diseases. This is a significant health inequity that people with a disability experience. Given the high degree of overlap in characteristics of expected and unexpected death in this study, there is a need to standardise the definition and undertake further descriptive epidemiological studies to determine the trajectories of death in this population and determine how people with a disability are dying in residential care facilities.
Conclusion
Although incidence of death was low, residents of congregated settings and larger settings had higher incidence of death than residents of other settings, and this should be a consideration for practice and policy. Given the high contribution of respiratory diseases to deaths and the potential avoidability of these, there is a need for improved respiratory health management in this population. Nearly half of all deaths were reported as unexpected; however, overlap in the characteristics of expected and unexpected deaths highlights the need for clearer definitions.
Data availability statement
Data are available in a public, open access repository. Data used in this study are from the Database of Statutory Notifications from Social Care in Ireland. The Irish Database of Statutory Notifications from Social Care is a national data collection of the statutory notifications received and generated by the Health Information and Quality Authority. An open access version of this database has been published for research and public use and is available at https://www.hiqa.ie/sites/default/files/2021-02/Database-of-statutory-notifications-from-social-care-in-Ireland-2013-2020.csv; this contains desensitised and pseudonymised facility-level data. The unredacted database is not available publicly but access to these data is possible following data custodian approval.
Ethics statements
Patient consent for publication
Ethics approval
Ethical approval was not required for this analysis as it is a secondary analysis of notification of events and does not include identifiable personal data; the included data are about events not people. Patient and public engagement was undertaken as part of the wider research project on statutory notifications in Ireland. Ethical approval was sought for this as it involved a vulnerable population group. Ethical approval to engage with people with disability living in residential care facilities for patient/public involvement purposes was granted by the Research Ethics Subcommittee of the Daughters of Charity Disability Support Service Ethics Committee on 26 January 2021.
Acknowledgments
The authors would like to thank the residents in the residential care facilities for people with disability who provided input on the development of the Database of Statutory Notifications from Social Care in Ireland and research questions important to them, thus enabling and informing this work.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
MJM and AMO are joint first authors.
Contributors MJM and LMK conceived and designed the analysis. MJM and AMO performed the literature search, undertook analysis and drafted the manuscript. LMK undertook the statistical analysis. PD prepared the data. MJM, AMO, PD, AD, LB, CG and LMK contributed to methodology development and provided input and critical comment throughout and agreed the final manuscript. LMK as guarantor accepts full responsibility for the work and/or the conduct of the study, had access to the data, and controlled the decision to publish.
Funding The database used for this research was developed and made available through a project funded by the Health Research Board’s Secondary Data Analysis Project Grant (SDAP-2019-005) and the Health Information and Quality Authority. This analysis did not receive any specific funding.
Competing interests None declared.
Patient and public involvement Patients and/or the public were involved in the design, or conduct, or reporting, or dissemination plans of this research. Refer to the Methods section for further details.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.