This blogpost is a copy of written evidence I was asked to provide for the Commons Health and Social Care Committee inquiry into the treatment of autistic people and people with learning disabilities. It goes through some of the statistics about people in inpatient units, in my best pompous data-nerd style.
Written evidence to the Commons Health and Social Care Select Committee: Inquiry into the treatment of autistic people and people with learning disabilities
Professor Chris Hatton, Manchester Metropolitan University
This written evidence provides a brief overview of statistical information regarding autistic people and people with learning disabilities in inpatient services, focusing on:
1) The number of people in inpatient services
2) Admissions, discharges and deaths
3) Transfer planning
4) Restrictive interventions
5) Length of stay in inpatient services
6) Some observations on data
The number of people in inpatient services
There are two sources of monthly information on the number of people with learning disabilities and autistic people in inpatient services published by NHS Digital.
The first is the Assuring Transformation (AT) dataset– this information is collected retrospectively by health service commissioners in England and has been collated and reported monthly by NHS Digital from February 2015. DHSC and NHS England/Improvement communications and written policy concerning Transforming Care and Building The Right Support use statistics drawn from this dataset.
The second uses information from an extension of the Mental Health Services Data Set (the MHSDS) to include people with learning disabilities and autistic people in mental health inpatient services. The MHSDS records monthly data from service providers flagging people with learning disabilities and autistic people in mental health inpatient services (including both mainstream mental health and specialist inpatient services), reported from February 2018 onwards by NHS Digital.
There are some major differences in scope between the AT and MHSDS datasets, the most substantial being that the MHSDS reports autistic people and people with learning disabilities in mainstream mental health inpatient services, often for short durations, which the AT dataset tends to exclude (possibly as a consequence of the MHSDS being reported by providers and AT being reported by commissioners). The AT dataset also reports more detailed information on aspects of service processes, while only the MHSDS currently reports information on restrictive interventions (various types of physical, mechanical and chemical restraint, seclusion and segregation).
Chart 1 below shows the number of people with learning disabilities and autistic people in inpatient services over time, according to the AT and MHSDS datasets. There are four issues to consider when looking at this chart
1) AT numbers are retrospectively reported by commissioners, with the most recently reported numbers (2,035 people in March 2021) added to over time as commissioners update their returns. This typically adds at least 100 people to the initially reported figures. Not taking this into account will inflate the scale of any reduction in inpatient numbers.
2) Figures for 2020 and 2021 must be considered in the light of COVID-19, where admissions (particularly for children and young people) sharply decreased during the first peak of COVID-19 in spring 2020, and it is unclear what will happen once restrictions end. It is difficult to interpret recent figures as representing long-term trends in Transforming Care/Building The Right Support policy versus inpatient service responses to COVID-19.
3) The MHSDS, which includes autistic people and people with learning disabilities in generic mental health inpatient services, reports much higher numbers of people (typically around 1,000 more people) at the end of each month than the AT dataset. If ‘specialist’ inpatient units are reducing their numbers, an important part of the Building The Right Support programme should include what is happening to autistic people and people with learning disabilities in generic mental health inpatient services.
4) These are end of month figures, which under-estimate the number of autistic people and people with learning disabilities using typically generic mental health inpatient services for very short periods of time. For example, MHSDS data for January 2021 report 3,205 people in inpatient services at the end of the month, but 4,215 people in inpatient services at some point in the month. Again, the Building The Right Support programme should be interested in people using generic mental health inpatient services for very short periods of time, including for the purposes of ‘respite’ (325 people in January 2021).
Chart 1: Number of people with learning disabilities and autistic people in inpatient services at the end of the month (AT and MHSDS)
Admissions to inpatient services
Data on admissions to inpatient services are available for a longer time period for the AT dataset than the MHSDS, so these are the data reported on in this section.
Chart 2 below shows the number of people with learning disabilities and autistic people admitted to inpatient services over five years, aggregated from monthly data into annual blocks. This chart breaks down admissions into people transferred from another hospital, people re-admitted to an inpatient service within a year of leaving one, and people admitted to an inpatient service for the first time (or at least more than a year since leaving one).
From 2015-16 to 2018-19 the chart shows an increasing number of admissions to inpatient services, with the decrease in 2019-20 likely to be due to restrictions on admissions during COVID-19.
Across the five years of data, 25% of admissions have been transfers from other hospitals, 16% have been re-admissions within a year (5% of all admissions are re-admissions within 30 days of leaving an inpatient service), and 60% have been first admissions.
Chart 2: Number of admissions of autistic people and people with learning disabilities to inpatient services annually, from Oct 2015 to Sept 2020 (AT dataset)
In terms of where people were being admitted to inpatient units from, AT data available across four years (Oct 2016 – Sept 2020) shows that almost half of admissions (46%) were from the person’s usual place of residence, 29% of admissions were from acute beds in hospitals (typically acute generic mental health beds), 2% were from secure forensic locations, 14% were from other types of hospital location, 4% were from penal establishments, and 4% were from residential care.
This reinforces the importance of understanding what is happening to people with learning disabilities and autistic people in generic mental health inpatient services, particularly the possibility of people being repeatedly in and out of ‘revolving door’ inpatient services in the absence of proper support.
‘Discharges’ from inpatient services
Data on discharges from inpatient services are available for a longer time period for the AT dataset than the MHSDS, so these are the data reported on in this section.
Chart 3 below shows the number of discharges of people with learning disabilities and autistic people from inpatient services over five years, aggregated from monthly data into annual blocks. This chart breaks down discharges into the destinations that people were immediately discharged to, including transfers to other hospital locations, discharges to community locations, and ‘other’ discharges.
Deaths of autistic people and people with learning disabilities are treated as a category of ‘other’ ‘discharge’, which I discuss in the next section.
Chart 3 below shows that over the five years Oct 2015 – Sept 2020 there have been a total of 10,830 ‘discharges’. Of these discharges:
· 65% have been to various community locations, including family homes with support (22% of all discharges), supported housing (20%), residential care (17%) and independent living (4%);
· 21% have been transfers to other inpatient hospital locations, most commonly low secure hospitals (6% of all ‘discharges’) and ‘other’ types of hospital (5%), but also medium secure units, acute transfer to a learning disability unit, acute transfer to a mental health unit, forensic rehab, and complex/continuing care/rehab (each 2% of all discharges);
· A further 14% of discharges were to ‘other’ unspecified locations, a substantial number of discharges which requires further specification by NHS England/Improvement and NHS Digital.
In terms of trends over time, the number of discharges increased from 2015-16 to 2017-18, but has decreased since.
Chart 3: Number of discharges of autistic people and people with learning disabilities from inpatient services annually, from Oct 2015 to Sept 2020 (AT dataset)
Deaths of autistic people and people with learning disabilities in inpatient services
In publicly available monthly data in both the AT and MHSDS datasets, deaths of people in inpatient services are recorded as a category of ‘other discharge’. Because of standard NHS Digital rounding rules, all the data in the AT and MHSDS datasets are rounded to the nearest five, or suppressed if the number is less than five. With the exception of April 2020 during the first peak of the COVID-19 pandemic, where AT recorded 5 deaths and MHSDS recorded 10 deaths, no individual month in either dataset has recorded any figures on the number of people who have died in inpatient units in each month. If any month with suppressed data can represent 0-4 deaths, then over the course of a year this could be anything from 0-48 people’s deaths not represented in the data.
It is vital for raw data to be aggregated by NHS Digital over longer periods of time (6-monthly or annually) to enable better scrutiny of the number of people dying in inpatient services, including during the COVID-19 pandemic.
Plans for transferring people out of inpatient services
The AT dataset includes a number of indicators relating to transferring people with learning disabilities and autistic people out of inpatient services.
Chart 4 below shows the percentage of people currently within inpatient units with various types of care plan. In Sept 2020 27% of people in inpatient units did not need inpatient care according to their care plan (down from 34% in Sept 2015), including 5% of all people in inpatient units with a delayed transfer of care. 32% of people in Sept 2020 (up from 24% in Sept 2015) were recorded in their care plan as not dischargeable, and a further 42% were recorded in their care plan as needing inpatient care but with an active treatment plan.
Chart 4: Percentage of people in inpatient units with different types of care plan (AT dataset)
In Sept 2020 just under half (49%) of all people in inpatient units had a transfer planned (up from 30% in Sept 2016), and for 9% of all people in inpatient units their planned transfer was overdue (compared to 8% in Sept 2016).
In Sept 2020, among those with a planned transfer date 56% of councils were aware of this planned transfer date (down from 68% in Sept 2016).
There are also indications that fewer transfer plans are being agreed with important others. Chart 5 below shows the percentage of transfer plans that have been agreed with other people and agencies. Although there has been an improvement from Sept 2019 to Sept 2020 this has not offset sharp decreases in previous years. In Sept 2020:
· 41% of transfer plans (compared to 64% in Sept 2016) were agreed with the person
· 36% (vs 60% in Sept 2016) were agreed with the person’s family/carer
· 40% (vs 65% in Sept 2016) were agreed with the person’s advocate
· 48% (vs 82% in Sept 2016) were agreed with the provider clinical team
· 44% (vs 67% in Sept 2016) were agreed with the local community support team
· 47% (vs 80% in Sept 2016) were agreed with commissioners
Chart 5: Percentage of transfer plans agreed with other people and agencies (AT dataset)
Statistics on restrictive interventions (including various types of physical, chemical and mechanical restraint, as well as segregation and seclusion) used on people with learning disabilities and autistic people in inpatient units are provided monthly in the MHSDS dataset. How inpatient services responded to COVID-19 during this time period is an important contextual factor when interpreting these statistics.
Chart 6 below shows the total number of restrictive interventions, and the total number of people subject to restrictive interventions, reported in the MHSDS from January 2020 to January 2021. Although there are apparent large fluctuations from month to month, the figures from January 2021 are similar to those for January 2020, with 420 people during the month of January 2021 subject to at least one restrictive intervention and a total of 3,970 restrictive interventions reported.
Chart 6: Total number of restrictive interventions and number of people in inpatient units subject to restrictive interventions (Jan 2020 – Jan 2021): MHSDS
Chart 7 below takes into the account the number of people in inpatient units at the end of each month, showing that in January 2021 13.1% of all autistic people and people with learning disabilities in inpatient units were subject to some form of restrictive intervention in the month (up from 11.5% in January 2020). This chart also shows that in January 2021 each person experiencing restrictive interventions was being subjected to an average 9.5 restrictive interventions per person (up from 8.8 in January 2020), almost one every three days. Again there are large fluctuations recorded from month to month.
In terms of specific types of restrictive intervention, in January 2021 a very wide range of restrictive interventions were being reported in the MHSDS as being used in inpatient units, including:
· Chemical restraints, most commonly oral medications (used 405 times on 80 people) and rapid tranquilising injections (used 280 times on 65 people)
· Mechanical restraints (used 40 times on 20 people)
· Seclusion (used 340 times on 165 people)
· Segregation (used 15 times on 10 people)
· Eight different types of physical restraint, most commonly
o Supine restraint (used 620 times on 145 people)
o Seated restraint (used 540 times on 120 people)
o Standing restraint (used 480 times on 160 people)
o Restrictive escort (used 335 times on 85 people)
o Prone restraint (used 255 times on 85 people)
o ‘Other’ types of physical restraint (used 615 times on 120 people)
To understand the monthly fluctuations in the figures for restrictive interventions a little better, Chart 8 below shows the percentage of people in inpatient services subject to restrictive interventions broken down by NHS vs independent sector inpatient services. Chart 8 shows that figures for NHS inpatient services show some fluctuations over time but a general increasing trend over time.
Figures for independent sector services show extreme fluctuations up to August 2020, followed by consistent low levels of restrictive interventions since. This can be accounted for by variations in the completeness of reporting within the MHSDS of restrictive intervention data from independent sector services. In January 2021, only one independent sector organisation out of 17 in the MHSDS (St Andrews) recorded restrictive interventions at a level (5 or more restrictive interventions due to the data rounding rules discussed earlier) recorded at all in the MHSDS. For example Cygnet (365 people with learning disabilities and autistic people in inpatient units according to the MHSDS) and Elysium (480 people) both effectively recorded no restrictive interventions in January 2021.
Chart 8: Percentage of people in inpatient units subject to restrictive interventions by provider type (Jan 2020 – Jan 2021): MHSDS
Length of stay in inpatient services
The final chart below, Chart 9, shows AT data on the average length of stay of autistic people and people with learning disabilities in inpatient services over time. Chart 9 shows that the average length of stay for people in their current inpatient unit was 2.7 years in September 2020, barely changed from 2.9 years in September 2016. The average continuous length of stay of people in inpatient units (including transfers between inpatient units) was 5.7 years in September 2020, again little changed from 5.4 years in September 2016.
Chart 9: Average length of stay for people in inpatient units (AT dataset)
Some observations on data
This written evidence presents an overview of selected statistical indicators concerning autistic people and people with learning disabilities in inpatient services in England. I hope this will be useful to the Committee in its Inquiry. As will be obvious to the Committee, issues involved in the interpretation of the available data can be complex. I would like to offer a small number of observations on these data which are urgent if policies such as Transforming Care and Building The Right Support are to be subject to robust evaluation and scrutiny.
1) The retrospective reporting of data for the AT dataset, and how inpatient services have been operating throughout the COVID-19 pandemic, mean that assertions of continuing reductions in the number of people in inpatient services may have been overstated.
2) The continuing lack of completeness in reporting to the MHSDS, particularly by independent sector organisations recording restrictive interventions, is an urgent concern as restrictive intervention statistics will be under-estimates which make it almost impossible to evaluate policy progress.
3) The MHSDS records much larger numbers of people in inpatient services, across a much wider range of inpatient services, than the AT dataset typically used by NHS England/Improvement in reporting progress. The lack of reconciliation of these two datasets is a longstanding issue, as noted by the National Audit Office in 2017:
“NHS England does not consider the current data it uses to monitor the programme [the Assuring Transformation dataset] to be a long-term solution and is planning for it to be incorporated into a newer, data set which monitors people using mental health services. This newer data set began reporting the number of people in mental health hospitals with a learning disability in May 2016. It reports a much higher number of people compared with the programme data set (3,805 people in November 2016 compared with 2,540 people in the programme data set at the same time). NHS England considers this newer data set to be less robust, less mature and needing development and so does not use it to monitor the programme. Our 2015 report highlighted the unsatisfactory situation of having two different unreconciled data sets, where one data set reported that there were 2,577 people in mental health hospitals whereas another data set reported 3,250. We are disappointed to find this problem again.”
4) For important data, particularly the deaths of autistic people and people with learning disabilities in inpatient services, raw data from the AT and MHSDS datasets should be aggregated over longer periods of time to enable the number of people with learning disabilities who have died to become visible.
5) The data provided publicly are a series of snapshots – supplementing these with data that track people over time would be really helpful in understanding issues such as the extent of ‘revolving door’ usage of inpatient services.
6) The scale and detail of data on people in inpatient services is not matched by data on the community support recommended in Building The Right Support, such as the number and composition of community teams and the number of people using them. These data are vital if the positive ambitions of Building The Right Support are to be subject to evaluation and scrutiny.
7) Routinely recorded statistics do not currently capture anything about the experiences of people in inpatient services or of those close to them. There are challenges in collecting this type of information routinely and reliably and in sharing summary information publicly, but again it is essential to understanding the progress of Transforming Care/Building The Right Support.
 National Audit Office. Local support for people with a learning disability. London: National Audit Office. Published 3rd March 2017. Available online at https://www.nao.org.uk/wp-content/uploads/2017/03/Local-support-for-people-with-a-learning-disability.pdf