The Confidential Inquiry into premature deaths of people with intellectual disabilities in the UK: a population-based study.

Summary Background The Conﬁ dential Inquiry into premature deaths of people with intellectual disabilities in England was commissioned to provide evidence about contributory factors to avoidable and premature deaths in this population. Methods The population-based Conﬁ dential Inquiry reviewed the deaths of people with intellectual disabilities aged 4 years and older who had been registered with a general practitioner in one of ﬁ ve Primary Care Trust areas of southwest England, who died between June 1, 2010, and May 31, 2012. A network of health, social-care, and voluntary-sector services; community contacts; and statutory agencies notiﬁ ed the Conﬁ dential Inquiry of all deaths of people with intellectual disabilities and provided core data. The Oﬃ ce for National Statistics provided data about the coding of individual cause of death certiﬁ cates. Deaths were described as avoidable (preventable or amenable), according to Oﬃ ce for National Statistics deﬁ nitions. Contributory factors to deaths were identiﬁ ed and quantiﬁ ed by the case investigator, veriﬁ ed by a local review panel meeting, and agreed by the Conﬁ dential Inquiry overview panel. Contributory factors were grouped into four domains: intrinsic to the individual, within the family and environment, care provision, and service provision. The deaths of a comparator group of people without intellectual disabilities but much the same in age, sex, and cause of death and registered at the same general practices as those with intellectual disabilities were also investigated. The Conﬁ dential Inquiry reviewed the deaths of 247 people with intellectual disabilities. Nearly a quarter (22%, 54) of people with intellectual disabilities were younger than 50 years when they died, and the median age at death 64 years (IQR 52–75). The median age at death of male individuals intellectual disabilities was 65 years (IQR 54–76), 13 years than the at male individuals and Wales (78 years). The median age at death of female individuals with intellectual disabilities was 63 years (IQR 54–75), 20 years younger than the median age at death for female individuals in the general population (83 years). Avoidable deaths from causes amenable to change by good quality health care were more common in people with intellectual disabilities (37%, 90 of 244) than in the general population of England and Wales (13%). Contributory factors to premature deaths in a subset of people with intellectual disabilities compared with a comparator group of people without intellectual disabilities included problems in advanced care planning (p=0·0003), adherence to the Mental Capacity Act (p=0·0008), living in inappropriate accommodation (p<0·0001), adjusting care as needs changed (p=0·009), and carers not feeling listened to (p=0·006).


Introduction
Premature deaths of people with intellectual disabilities compared with the general population have been consistently identifi ed since the 1970s. [1][2][3][4][5][6][7][8][9] People with more severe intellectual disabilities have been recognised as having shorter life expectancies than those with mild intellectual disabilities. [2][3][4][5] Predictors of early mortality in this group include limited mobility, 3,10,11 reduced feeding ability, 3,12 incontinence, 3 institutional care, 4 and hearing defi cit 13 -most of which correlate with increasing severity of intellectual disability. Some premature mortality in people with intellectual disabilities might be expected because they often have important comorbidities and associated polypharmacy, 14,15 which can contribute to early death; however, there are other broader determinants of health relating to the environment, provision of care, and access to health-care services that might contribute to premature death. [16][17][18] These broader determinants are increasingly being recognised in national and international policy statements. 19,20 The Confi dential Inquiry into premature deaths of people with intellectual disabilities, commissioned by the Department of Health in England after the Michael Report 21 concluded that there was a high likelihood of avoidable deaths of people with intellectual disabilities, attributable to untreated ill health and shortcomings in the provision of health care. An important aim of the Confi dential Inquiry was to establish how similar or diff erent the circumstances leading to deaths of people with intellectual disabilities were when compared with people without intellectual disabilities. In this Article, we report the fi ndings of the Confi dential Inquiry, with a particular focus on the comparison between avoidable deaths and the contributory factors to premature deaths in people with and without intellectual disabilities.

Study design and population
The study area included fi ve (former) Primary Care Trust areas in southwest England with a population of nearly 1·7 million and a mix of urban and rural communities. The proportion of adults with intellectual disabilities identifi ed by general practitioners (GPs) in the study area was 0·48% (n=6962); children with intellectual disabilities formed 2·5% (n=8543) of the school population.
The Confi dential Inquiry reviewed the deaths of people with intellectual disabilities aged 4 years and older registered with a GP in the study area who died between June 1, 2010, and May 31, 2012. The defi nition of intellectual disabilities was as described by Emerson and Heslop. 22 Severity of intellectual disability was established by professional opinion or descriptors (available from the authors on request). The Confi dential Inquiry investigators established a network to notify them of all deaths of people with intellectual disabilities, which included health, social-care, and voluntary-sector services; community contacts; and statutory agencies. Additional checks were made with GPs, prisons, community groups and leaders, development workers in minority ethnic communities, and services supporting people with intellectual disabilities to ensure that all eligible deaths had been reported.
For a subset of deaths of people with intellectual disabilities, the investigators reviewed the deaths of people without intellectual disabilities as comparator cases. They were registered at the same GP practices as those with intellectual disabilities who had died, and were much the same in age, sex, month of death, and broad category of death. Further contextualisation of the fi ndings was made by reference to national mortality data.
Study approval was obtained from the NHS Research Ethics Committee, local NHS Research and Development teams, and the (former) National Information Governance Board (Section 251 approval).

Procedures
Core data were requested from all agencies providing services or support to the adults with intellectual disabilities who had died. Friends or family members who wished to contribute to the review were interviewed by an experienced specialist intellectual disabilities nurse. Paid carers and health-care and social-care professionals were interviewed by a Confi dential Inquiry investigator, who also reviewed all case notes pertaining to the individual and did a root cause analysis 23 of the death. The median number of informants per case was seven (range one to 15). The UK Offi ce for National Statistics provided data about the coding of individual cause of death certifi cates for all but three cases; the cause of death for these cases were taken from post-mortem reports. All available information was collated into a standardised format and presented at a local review panel meeting to which all involved professionals were invited. The focus of the meeting was to review and discuss the circumstances of the death and any contributing factors, record good practice, and identify lessons learned and recommendations that could be made. Documentation from each case was then anonymised and scrutinised by the Confi dential Inquiry overview panel, an external, multidisciplinary group of health-care and social-care professionals and family carer representatives.
For the deaths of children (<18 years) with intellectual disabilities the local statutory child death review team took the lead in conducting a review of the death, but the Confi dential Inquiry overview panel, which was given full access to the reports of the child death review process, then reviewed each case again. The process of investigating the deaths of the comparator cases was the same as for the deaths of adults with intellectual disabilities.
The Offi ce for National Statistics 24 defi nes avoidable deaths as those that are preventable, amenable, or both. A death is preventable when all or most deaths from that cause (subject to specifi c age limits when appropriate) could be avoided by public health interventions in the broadest sense. A death is amenable when, with the medical knowledge and technology available at the time of death, most deaths from that cause (subject to specifi c age limits when appropriate) could be avoided through good quality health care.
A death was deemed premature if, without a specifi c event that formed part of the pathway that led to death, it was probable (ie, more likely than not) that the person would have continued to live for at least 1 more year. This approach allowed consideration of whether something had (or had not) happened in the care of the person that might have contributed to the death and of additional life-limiting factors (such as lifestyle or comorbidities).
Contributory factors to deaths were identifi ed and quantifi ed by the case investigator, verifi ed by the local review panel meeting, and agreed by the Confi dential Inquiry overview panel. Contributory factors were grouped into four domains: intrinsic to the individual, within the family and environment, care provision, and service provision.
Potential comparator cases were identifi ed from listings of deaths at general practices where the death of a person with intellectual disabilities had been reported. The comparator group was not matched on a 1:1 basis but was weighted (balanced) by month of death, broad cause of death, age at death, and sex to produce a similar distribution to a subset of people with intellectual disabilities. The subset of people with intellectual disabilities was not preselected but was chosen on a monthby-month basis as candidates with the closest weighting criteria to the potential comparator cases.

Statistical analysis
Data not normally distributed were presented with medians and IQR and analysed with Mann-Whitney and Kruskal-Wallis tests. For categorical data, the χ² test was used, except when the expected cell count was less than fi ve, in which case the Fisher's exact test was used.

Role of the funding source
The sponsor of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had fi nal responsibility for the decision to submit for publication.

Results
The Confi dential Inquiry reviewed the deaths of 247 people with intellectual disabilities aged 4 years and older at the time of their death (table 1 25,26 ).
The total number of people with intellectual disabilities in the study area is under-estimated in offi cial sources, making it diffi cult to calculate a death rate. Using the total study area population, the crude annual death rate for people with intellectual disabilities was 7·4 deaths per 100 000 population, this fi nding equates to 16·2 deaths per 1000 of the population who have intellectual disabilites, nearly twice the rate of 8·8 deaths per 1000 of the general population. 27 The median age at death for the 247 people with intellectual disabilities was 64 years (IQR 52-75; range 4-96); 14 of the deaths were of children aged 4-18 years. This fi nding contrasts with age at death for the general population in England and Wales in 2011 (fi gure 28 ). Nearly a quarter (54, 22%) of people with intellectual disabilities were younger than 50 years when they died, compared with 9% in the general population. The median age at death of male individuals with intellectual disabilities was 65 years (IQR 54-76), 13 years younger than the median age at death (78 years) of male individuals in the general population of England and Wales. 25 The median age at death of female individuals with intellectual disabilities was 63 years (IQR 54-75), 20 years younger than the median age at death for female individuals in the general popultion (83 years Kruskal-Wallis test p=0·0005), although the median age at death of those with mild intellectual disabilities was still sub stantially younger than the general population.
When categorised according to the International Classifi cation of Diseases-10 (ICD-10), about the same proportion of people with intellectual disabilities and of the general population had the underlying cause of death as respiratory system disorders, digestive system disorders, or external causes (table 2). Signifi cantly more deaths of people with intellectual disabilities were attributable to congenital and chromosomal disorders or causes relating to the nervous system, whereas significantly more deaths in the general population were due to heart and circulatory disorders (table 2).
Avoidable deaths accounted for 24% of deaths in England and Wales in 2011. 24    intellectual disabilities for whom fi nal ICD-10 coding of cause of death was available, avoidable deaths accounted for 49% (119) of deaths. The proportion of deaths preventable by public health interventions was 21% for the population of England and Wales and 21% (52) in the cohort of people with intellectual disabilities. The diff erence between the proportion of avoidable deaths in the general population and the intellectual disabilities cohort was entirely accounted for by causes of death deemed to be amenable to good quality healthcare: 13% of deaths in England and Wales were from causes amenable to good quality health care, compared with 37% (90 of 244) of deaths in the intellectual disabilities cohort.
Within the intellectual disabilities cohort, individuals whose deaths were amenable to good quality health care were younger than those whose deaths were not amenable ( ; χ² p=0·03) than were those whose deaths were not amenable to good quality health-care.
The Confi dential Inquiry overview panel concluded that 42% (100) of the 238 deaths about which they reached agreement were premature.
For the comparator study, there were no signifi cant diff erences between the subset of 58 people with intellectual disabilities and the remaining 189 people in the intellectual disabilities cohort regarding season of death (48% [28] of the intellectual disabilities subset and 54% [102] of the rest of the intellectual disabilities cohort died in winter [October-March]; Mann-Whitney p=0·45), and although there was a slight excess of cancer-related deaths in the subset (29%, 17) compared with the rest of the cohort (18%, 34), the broad categorisation of underlying cause of death was much the same in the two groups (χ² p=0·29). There were fewer male individuals in the subset (50%, 29) than in the rest of the cohort (60%, 113), and fewer with severe or profound intellectual disabilities (19% [11] in the subset vs 32% [60] in the rest of the intellectual disabilities cohort), although neither of these diff erences reached signifi cance (χ² p=0·19 and χ² p=0·14, respectively). For this comparison, we deliberately focused on deaths in people younger than 75 years, who in population terms would have been deemed to have died prematurely according to Offi ce for National Statistics defi nitions; thus the intellectual disabilities subset was signifi cantly younger (median age at death 61 years [IQR 52-67]) than the remaining 189 people with intellectual disabilities (67 years [51-77]; Mann-Whitney p=0·006).
The 58 comparator cases were broadly weighted with the subset of 58 people with intellectual disabilities for age, sex, time of death, and broad cause of death. Median age at death was 60 years (IQR 53-63) for the comparator group and 61 years (52-66) for the intellectual disabilities subset (Mann-Whitney test p=0·36). There were more male individuals (59%, 34) in the comparator group than in the intellectual disabilities subset (50%, 29) although this diff erence was not signifi cant (χ² p=0·35). Season of death did not diff er between the groups (50% [29] of the comparator group and 48% [28] of the intellectual disabilities subset died in winter; Mann-Whitney p=0·83). Slightly more individuals in the comparator group (40%, 23) had cancer as their underlying cause of death than did the intellectual disabilities subset (29%, 17), but the broad categorisation of cause of death did not diff er between the two groups (χ² p=0·64). However, weighting the comparison by broad categorisation of death excluded a proportion of deaths largely prevalent among individuals younger at death in the general population (related to alcohol, drugs, and suicide) that were not prevalent among those who died with intellectual disabilities. Most individuals in the comparator group (95%, 55) and in the intellectual disabilities subset (98%, 57) were of white British ethnic background. In view of the paucity of work experience and reduced educational opportunities and the predominance of residential care for those with intellectual disabilities, it was not possible to use traditional markers of socioeconomic status to match the two groups.
Of the intellectual disabilities subset, 69% (40) died from underlying causes of death considered to be avoidable according to the Offi ce for National Statistics' categorisation, compared with 66% (38) among the comparators, a non-signifi cant diff erence (χ² p=0·64). However, there were signifi cant diff erences when considering whether the deaths were deemed preventable or amenable. Deaths preventable by public health interventions were more common in the comparator group  (25%, 14) than in people with intellectual disabilities (17%, ten) although not signifi cantly so (χ² p=0·33), whereas deaths from causes amenable to change by good quality health-care were signifi cantly more common in people with intellectual disabilities (38%, 22) than in the comparator group (9%, fi ve; χ² p=0·002). The proportion of premature deaths was 52% (30) in the intellectual disabilities subset, and 43% (25) in the comparator group, a non-signifi cant diff erence (χ² p=0·34). Of those deaths classed as premature, neither age nor causal classifi cation was a factor in recording the death as premature or not.
We compared the detailed circumstances leading to death in the intellectual disabilities subset and the comparator group (table 3). We identifi ed signifi cant diff erences between the intellectual disabilities subset and the comparator cases in all four domains. In particular, a signifi cantly greater proportion of deaths in the intellectual disabilities group had inadequate or inappropriate accommodation for the person's needs; family or paid carers who did not feel they were listened to; problems in advanced health care and planning; problems in recognising the person's needs and adjusting care when needs changed; and poor adherence to the Mental Capacity Act, 29 particularly in relation to assessment of a person's capacity to make a decision and to the decision making process regarding that person's health care.

Discussion
Our data suggest that, on average, male individuals with intellectual disabilities die 13 years earlier than the population of England and Wales, and female individuals die 20 years earlier. Avoidable deaths from causes amenable to change with good quality health care are more common in people with intellectual disabilities than in the general population. When compared with people without intellectual disabilities, contributory factors to premature deaths in people with intellectual disabilities include problems in advanced care planning, adherence to the Mental Capacity Act, living in inappropriate accommodation, adjustment of care as needs changed, and carers not feeling listened to.
Our fi ndings are observations of associations that might not all be causally linked to each death reviewed by the Confi dential Inquiry. However, the consistent patterns in the fi ndings suggest that meaningful changes to practice can be recommended. By reviewing the circumstances leading to the deaths of a subset of people with intellectual disabilities and a comparator group of people without intellectual disabilities, we have shown that contributory factors to the deaths of people with intellectual disabilities occur across several domains, especially in relation to care and service provision (panel 1). Although these fi ndings should be considered with caution because of the small sample sizes and the weighting criteria used, they suggest that all parties involved in the provision of care and support to people

Systematic review
We searched the Cochrane Library, Medline, and Embase, for reports in English published from Jan 1, 2000, to Oct 1, 2013, using the terms "confi dential inquiry" (or "enquiry") and "intellectual disability/disabilities" or "learning disability/disabilities". We also searched the same sources using the key words "mortality review" and "intellectual disability/disabilities" or "learning disability/disabilities".
We identifi ed only one report, that of Tyrer and colleagues, 6 which was a population-based study in one area of the UK that explored cause-specifi c mortality in adults with intellectual disabilities compared with the general population. Tyrer and colleagues concluded that strategies to reduce inequalities in people with intellectual disabilities should focus on decreasing mortality from potentially preventable causes, such as respiratory infections, circulatory system diseases, and accidental deaths.

Interpretation
The Confi dential Inquiry into premature deaths of people with intellectual disabilities was the fi rst of its kind in England. It reviewed the deaths of all known people with intellectual disabilities over a 2-year period in an area of England that had a total population of 1·7 million individuals. A comparator group of people of a similar age and cause of death as people without learning disabilities was investigated to place the fi ndings in context. Each death was reviewed in depth and involved all agencies and support services in contact with the deceased. All reports were anonymised and were then reviewed by an external multidisciplinary overview panel.
Our fi ndings show that people with intellectual disabilities were likely to die, on average, 16 years earlier than the general population. A range of potentially modifi able factors were related to care and service provision, and all aspects of care provision, planning, coordination, and documentation were signifi cantly poorer for people with intellectual disabilities than for the comparator group of people without intellectual disabilities. The fi ndings suggest that although some individual factors are of importance, factors relating to care and service provision contribute to excess mortality in people with intellectual disabilities.
with intellectual disabilities must examine problems with care and service provision as contributors to premature deaths; these factors can largely be ameliorated and are inherently unjust. With this in mind the Confi dential Inquiry proposed 18 recom mendations (panel 2) that, if implemented, would lessen the risk of premature death in people with intellectual disabilities. 28 The strength of the Confi dential Inquiry is that it has taken a population-based approach to reviewing in depth all known deaths of people with intellectual disabilities in a particular geographical area that is broadly representative of England as a whole. Without a comprehensive register of people with intellectual disabilities, it is diffi cult to be sure that we reviewed every eligible death, but the wide-ranging notifi cation system provided confi dence that few, if any, deaths were missed. The inclusion of people with mild intellectual disabilities was important to ascertain contributory factors for premature deaths across the spectrum of intellectual disabilities, but we acknowledge that there is no clear dividing line between those who do or do not have intellectual disabilities. Our reliance on past evidence of intellectual disabilities, 22 professional opinion, or descriptors to defi ne the degree of intellectual disability diff ers from the methods of other studies, although the proportions we identifi ed are in line with the predicted new entrants to social care from 2011 to 2030. 30 The number of deaths of people with intellectual disabilities was two and a half times more than the number we originally estimated on the basis of reports in the scientifi c literature. 27 Such a discrepancy draws attention to the absence of comprehensive registration and mortality data for people with intellectual disabilities in England. A need exists to link data about cause of death with appropriate registers of adults and children with intellectual disabilities, so that the age and cause of death of people with intellectual disabilities can be monitored at a national level, and can be reviewed within a health equalities framework.
Results of a study by Lavin and colleagues 5 suggested that mortality in people with intellectual disabilities in Ireland might be 10-16-times higher than in the general population, whereas Tyrer and colleagues 6 reported a three-times increase in mortality in people with moderate to profound intellectual disabilities in the UK, which is in line with our fi nding of a two-times increase in people with mild, moderate, severe, and profound intellectual dis abilities. Nearly a quarter (22%) of people with intellectual disabilities were younger than 50 years when they died, compared with about 9% of the general population. We showed that the risk of dying at an early age was greatest for people with more severe intellectual disabilities, but the median age at death of people with mild intellectual disabilities (68 years) was still substantially younger than in the general population. Therefore, our results do not support fi ndings from Finland 4 that life expectancy of people with mild intellectual disabilities is approaching that of the general population.
Twice as many deaths were deemed avoidable in the intellectual disabilities cohort as in the general population in England and Wales. Importantly, we identifi ed no diff erence in deaths preventable by public health interventions: the diff erence was wholly explained by deaths amenable to change with good quality health care. People with more severe intellectual disabilities, with congenital and chromosomal abnormalities, or without support from a partner or signifi cant friend were particularly likely to have deaths amenable to change with good quality health care, suggesting that a targeted approach to the improvement of care is needed for these groups.

Panel 2: Recommendations of the Confi dential Inquiry into deaths of people with intellectual disabilities
1 Clear identifi cation of people with learning disabilities on the National Health Service central registration system and in all health-care record systems. 2 Reasonable adjustments required by, and provided to individuals, to be audited annually and examples of best practice to be shared across agencies and organisations. 3 National Institute for Health and Care Excellence guidelines to take into account multi-morbidity. 4 A named health-care coordinator to be allocated to people with complex or multiple health needs, or two or more long-term conditions. 5 Patient-held health records to be introduced and given to all patients with learning disabilities who have multiple health conditions. 6 Standardisation of annual health checks and a clear pathway between annual health checks and health action plans. 7 People with learning disabilities to have access to the same investigations and treatments as anyone else, but acknowledging and accommodating that they may need to be delivered diff erently to achieve the same outcome. 8 Barriers in individuals' access to health care to be addressed by proactive referral to specialist learning disability services. 9 Adults with learning disabilities to be considered a high-risk group for deaths from respiratory problems. 10 Mental Capacity Act advice to be easily available 24 h a day. 11 The defi nition of serious medical treatment and what this means in practice to be clarifi ed. 12 Mental Capacity Act training and regular updates to be mandatory for staff involved in the delivery of health or social care. 13 Do not attempt cardiopulmonary resuscitation guidelines to be more clearly defi ned and standardised across England. 14 Advanced health and care planning to be prioritised. Commissioning processes to take this into account, and be fl exible and responsive to change. 15 All decisions that a person with learning disabilities is to receive palliative care only should be supported by the framework of the mental capacity act and the person referred to a specialist palliative care team. 16 Improved systems in place nationally for the collection of standardised mortality data about people with learning disabilities. 17 Systems in place to ensure that local learning disability mortality data are analysed and published on population profi les and joint strategic needs assessments. 18 Establishment of a national learning disability mortality review body.
Reproduced from the full report. 28

Contributors
PH was lead author, grant-holder, and manager of the Confi dential Inquiry, and contributed to all aspects of the study design, data collection, data analysis and interpretation, and writing of the Article. PSB was statistician for the Confi dential Inquiry, and contributed to all aspects of the study design, data analysis and interpretation, and writing of the Article. PF was chair of the overview panel of the Confi dential Inquiry, and contributed to all aspects of the study design, data collection, data analysis and interpretation, and writing of the Article. MH was lead investigator for the Confi dential Inquiry, and contributed to all aspects of the study design, data collection, data analysis and interpretation, and writing of the Article. AM was project worker for the Confi dential Inquiry with particular responsibility for management and support of the overview panel, and contributed to all aspects of the study design, data collection, data analysis and interpretation, and writing of the Article. LR was lead nurse for the Confi dential Inquiry, and contributed to all aspects of the study design, data collection, data analysis and interpretation, and writing of the Article. All authors approved the fi nal version.

Confl icts of interest
We declare that we have no confl icts of interest.