Walsall CCG Mortality Review Group
Exploring hospital standardised mortality ratios at
Walsall Healthcare NHS Trust between 2010 and 2013
Lead authors
Dr David Pitches, Locum Consultant in Public Health, NHS Walsall
Dr Paulette Myers, Consultant in Public Health, Walsall Council
Item 8.3 Appendix 1
2
Item 8.3 Appendix 1
3
Contents
1. Executive Summary
1.1 What was already known?
1.2 What was our approach?
1.3 Were there any other driving factors?
1.4 What were the principal activities of the group?
1.5 What were the main findings of the group?
1.6 What lessons can be learnt from this group?
4
2. Introduction to hospital mortality statistics
2.1 Definitions
2.2 Causes of high mortality statistics
6
3. Historical perspective and relevance to Walsall
3.1 Early attempts to publish hospital mortality statistics
3.2 Publication by the Dr Foster unit at Imperial College London
3.3 Alternative explanations for high mortality rates
3.4 Changes to palliative care coding
3.5 Monitoring of changes to coding and effects on mortality by Walsall PCT
3.6 Extending patient choice and availability of preferred place of death
3.7 Redevelopment of Goscote Lane and the opening of the Palliative Care Centre 3.8 Additional consultants in palliative medicine recruited to Walsall Healthcare NHS
Trust
3.9 Hospital review of mortality
9
4. Walsall CCG Mortality Review Group 14
5. External Drivers 16
6. Summary of Phase 1 findings of the Mortality Review group
6.1 Performance monitoring and assurance
6.2 Hospital activities
6.3 Mortality and care homes
6.4 Data and intelligence
6.5 Addressing and strengthening the role of General Practices
6.6 End of life
17
7. Impact of mortality review group
23
8. Independent management consultants’ report
25
9. Lessons learnt and conclusions
26
10. Next steps
27
11. Acknowledgements
28
Item 8.3 Appendix 1
4
1. Executive summary
As a local leader and commissioner of NHS services, Walsall CCG (formerly PCT) has a
responsibility to ensure services are safe and of high quality. In 2011-2 Walsall (then) PCT and
Walsall Healthcare NHS Trust (WHNT) identified concerns at the apparently high standardised
hospital mortality ratio (HSMR) observed at Walsall Manor hospital.
1.1 What was already known?
Walsall Manor Hospital has had a history of having a high HSMR which is investigated, falls, and
rises again. Importantly therefore, the mortality review group therefore was not approaching the
question in isolation, but coming aware of contributing factors and interventions (including
coding reviews) that had been undertaken previously.
1.2 What was our approach?
WHNT launched its own review and agreed an action plan to tackle the issues within its direct
control but as a system also agreed that there was a need to take a wider approach to reducing
mortality.
The PCT convened a mortality review group to examine the contributory causes of the high
HSMR during 2011/12 and exploit opportunities to improve choice and quality of care throughout
the local health system. This group included partners from across the health economy – public
health, primary care, hospital and palliative care services and social care, with a lay Chair at arms’
length from the CCG (since for most of the duration of the mortality review group the CCG was
operating as a sub-committee of the PCT cluster). The group was aligned both to PCT corporate
core business (for example clinical quality review and safety, quality and performance groups)
and to regional (SHA) work.
The collective focus of the group was to understand the current situation and make
improvements where opportunities presented themselves, not only in the hospital but in the
community too, since it was well recognised that a high HSMR may have its origins in factors
outside the control of the hospital alone. The group had two main objectives:
(i) to ensure that the actions taken by the hospital trust were having the required effect
(ii) to identify and address the wider health economy issues contributing to high
mortality rates (e.g. performance of nursing homes or availability of palliative care).
System wide leadership underpinned both the group as a whole and the separate workstreams
that reported into the group. Strong leadership, high levels of trust, transparency and openness,
and being prepared to challenge one another in each of the workstreams were vital to the overall
success of the group.
Item 8.3 Appendix 1
5
1.3 Were there any other driving factors?
Separately to the PCT group, the council Health Scrutiny and Performance Panel commissioned
an independent review of hospital mortality. Meanwhile the Mid Staffordshire NHS Foundation
Trust public inquiry was expected to report imminently, with the potential to attract further
attention to other hospitals with high HSMRs. Discussions about hospital mortality cannot be
separated from palliative care, and the Liverpool Care Pathway is an important framework of
best practice in palliative care. However it has recently come under the spotlight itself in the
wake of criticism that it might not always be being used appropriately.
1.4 What were the principal activities of the group?
Several workstreams were set up under the mortality group to analyse specific areas, including:
(i) Assurance
(ii) Hospital care
(iii) Nursing home care
(iv) Data collection and hypothesis testing
(v) Primary care
(vi) End of life care
1.5 What were the main findings of the group?
• Historically Walsall has been under-provisioned for end-of-life care meaning a higher
proportion of terminally ill patients die in hospital than the national average and this
effect can be reflected in hospital mortality statistics
• Opportunity and choice for hospice and other end of life care has been expanded
recently
• Many cancer patients are dying in their preferred place of death
• Admissions in nursing home pilots have reduced
• Clinical pathways have been strengthened
• HSMR has been steadily declining since April 2012 and SHMI is within the expected range
• Assurance has been strengthened
1.6 What lessons can be learnt from this group?
• Secure and trusted ownership of the exploration of HSMR has to come from all partners
• The group must have independence – it was totally separate from a hospital Trust
mortality review group and had an arm’s length lay Chair
• There needs to be good evidence and the ability to interrogate data
• Findings need to support actions
• This was no small undertaking for the members of the group
Item 8.3 Appendix 1
6
2. Introduction to hospital mortality statistics
Walsall CCG has a responsibility, as local leader and commissioner of NHS services, to provide
assurance that health services, including hospital care, are delivering high quality and safe
healthcare to the people of Walsall.
Hospital mortality statistics, of which there are a number of different measures, are a commonly
used tool that can be used to scrutinise safety and quality of hospitals. However there are many
caveats and limitations on the use of hospital mortality statistics and their interpretation requires
great care to ensure a complete understanding of the contributory factors to an unacceptably
high mortality statistic.
2.1 Definitions
Hospital mortality statistics include the Hospital Standardised Mortality Ratio (HSMR), the
Summary Hospital Level Mortality Indicator (SHMI) and the crude mortality rate. These are
published at different intervals (monthly, quarterly or annually) by a range of organisations and
each is calculated using different formulae and definitions and each has its limitations. These
currently include:
(i) HSMR (produced by the Dr Foster company) includes the deaths of patients who die
in hospital from the most common conditions causing 80% of deaths (thus some
patients who died may be excluded from the calculations). It is defined as the ratio of
the number of actual (observed) deaths in a time period e.g. a month, divided by the
number of “expected” deaths, where Dr Foster calculates a probability of death for
each patient who is admitted to hospital and estimates how many deaths a hospital
might reasonably be expected to have had during that time. There is substantial
disagreement of expert opinion however about what factors to take into account
when estimating how many patients might have been expected to die over a given
time period. In addition for statistical reasons, any one hospital’s HSMR cannot
meaningfully be compared against any other hospital, because no two hospitals will
admit exactly the same patients. Despite this, HSMRs are routinely published and
compared against one another.
(ii) SHMI (NHS health and social care information centre) includes patients who died in
hospital and patients who died up to 30 days after coming out of hospital and thus
may include patients whose death might have been caused by the hospital but which
would not be detected by the Dr Foster company. Conversely, deaths occurring in the
community as a result of poor out-of-hospital care may appear to adversely affect the
hospital’s mortality rate. The SHMI also fails to account for the effect that deprivation
can have on a patient’s likelihood of dying, and the likelihood that if they are receiving
palliative care they are probably being admitted with a terminal illness and their
death is unlikely to be preventable.
Item 8.3 Appendix 1
7
(iii) Crude mortality rate: this is typically the number of deaths divided by the number of
admissions, but takes no account of the variation in severity of illness that patients in
different hospitals present to hospital with so cannot be used to compare hospitals.
In addition, mortality statistics are usually given in the context of a range of values within which a
hospital would be expected to be by chance. Outside of (ie outlying) this range, chance alone is
unlikely to explain why the rate is higher (or lower) than could reasonably be expected by
chance, implying other factors may be at work. This is important because one factor that
seemingly self-evidently might explain why the rate is higher is that there is poor quality of care.
2.2 Causes of high mortality statistics
The reality is however far more complicated and it is invariably premature to jump to the
conclusion that a high mortality hospital has a poor standard of care. It is essential to stress that
hospital mortality statistics are a consequence of a complex mix of factors including
• case mix (how severely ill patients are who are admitted to hospital)
• lifestyle choices such as smoking and died
• disease coding (how accurately the patient’s presenting conditions are categorised to
one or more standardised disease categories, to allow comparison between patients with
different conditions)
• quality of care in primary care and community settings e.g. care homes
• quality of care in hospital
• availability of end of life care options e.g. local hospices
• patient choice of where to die
• chance variations
Therefore a high mortality rate or ratio in itself does not necessarily imply that there is any reason
for concern about the quality of clinical care at the hospital; rather it is described as a “smoke
alarm” that should trigger examination of pertinent factors including clinical care to see whether
they can explain the outlying mortality statistic. In fact often the majority of factors that cause a
hospital to have a high mortality statistic are outside the direct control or influence of the
hospital. Furthermore, within hospitals there may be areas of exceptional care and other less
strong areas and in theory if a department had a high mortality rate as a result of poor care, this
could be masked by stronger performance in other areas.
A recent very detailed review of a thousand deaths in a range of NHS hospitals in England found
that on average around 5% had factors in their clinical care that might have contributed to the
patient’s death, which as a headline figure is actually similar to research studies of avoidable
hospital death in other industrialised countries1. Mathematical modelling suggests that at this
level of “preventable death”, avoidable mortality can only explain 8% of the variation in hospital
mortality rates and only one in every eleven hospitals identified as having a significantly elevated
1 Hogan et al. Preventable deaths due to problems in care in English acute hospitals: a retrospective case record review study. BMJ Quality and Safety 2012
Item 8.3 Appendix 1
8
mortality statistic may have problems with quality of care2. The majority of hospitals with high
HSMRs or SHMIs are unlikely to have more quality of care problems than hospitals with normal or
low mortality statistics.
Two research papers illustrate the difficulty of linking poor quality care to high mortality rates.
Firstly, a systematic review was published in 2007 of all studies that attempted to measure quality
of care and mortality rates (after adjusted for patient case-mix) across several hospitals3. It was
evident that there was a very tenuous link if any between quality of care and hospital mortality;
for every study that showed that higher mortality rate hospitals delivered poorer care, another
study found either no association or even a negative association (that is, the better quality
hospitals had higher mortality rates).
Secondly a unique experiment in Massachusetts in 2008 cast doubt on the ability of
commercially-calculated hospital mortality statistics to discriminate between “good” and “bad”
hospitals4. They provided a complete dataset of all admissions over a period of several years for
83 hospitals and invited four companies, including the British hospital mortality statistics
company Dr Foster, to try to identify the “highest” and “lowest” mortality outlying hospitals. 12
of the 28 identified by one company as “significantly worse than expected” were simultaneously,
and using the same dataset, identified as performing “significantly better than expected” by the
other companies. The researchers suggested that it was not safe to assume that hospitals could
be identified as good or bad on the basic of mortality rates alone.
There are three key mistakes that can be made when reviewing hospital mortality statistics:
(i) attempting to explain a high HSMR or SHMI solely through use of data on the
grounds that they are imperfect indicators of quality of care – for example improving
coding can bring down mortality without changing clinical quality of care
(ii) being falsely reassured by a low or “within expected limits” mortality and not
noticing clinical or organisational failings that are causing harm to patients
(iii) only focussing attention on the hospital when attempting to understand hospital
mortality statistics, and failing to consider the impact that events in the wider health
economy are contributing to the mortality rate.
Any approach undertaken in Walsall therefore had to be sufficiently robust to ensure that careful
attention was paid to understand where the data and intelligence was pointing and what clinical
and wider health economy areas should be considered, rather than simply trying to “improve”
the data. Conversely. whilst improvements in quality of care do not automatically lead to changes
in hospital mortality statistics, potential for improved care could not be ignored where it was
identified.
2 Girling et al. Case-mix adjusted hospital mortality is a poor proxy for preventable mortality: a modelling study BMJ Quality and Safety 2012 0:1–5. 3 Pitches et al. What is the empirical evidence that hospitals with higher risk adjusted mortality rates provide poorer quality care? Biomed Central Health Services Research, 7. p.91. 4 Shahian et al. Variability in the Measurement of Hospital-wide Mortality Rates. New England Journal of Medicine 2010;363:2530-9.
Item 8.3 Appendix 1
9
3. Historical perspective and previous actions in Walsall
3.1 Early attempts to publish hospital mortality statistics
The first attempt to calculate and compare mortality rates in different hospitals was undertaken
by Florence Nightingale in 1853. Her attempts were highly criticised and generally considered
unsuccessful for a number of reasons, some of which remain valid today and are worth studying:
(i) Firstly, there were shortcomings in the formula she chose to use to calculate
mortality rates: she divided the number of deaths by the number of beds in the
hospital. In some hospitals this meant they had apparent mortality rates of nearly
100%, which does not convey meaningful information.
(ii) Some hospitals with high mortality rates claimed this was something to be proud of,
since they were clearly taking care of the sickest patients.
(iii) Others tried to conceal deaths by discharging very sick patients before they died.
(iv) Finally, getting hospitals to provide her with comparative statistics proved more
difficult than she had anticipated.
3.2 Publication by the Dr Foster unit at Imperial College London
Hospital mortality statistics first began to be routinely published in England in 2001 in the Sunday
Times as the Dr Foster “Good Hospital Guide”. Three quarters of the hospitals with the lowest
HSMRs were in London and the South East (University College London Hospital’s HSMR at 68
was the lowest) whilst several hospitals in the Midlands were identified as having the highest
HSMRs in the country. These included the hospitals of Walsall (119, where 100 was the “England
average”), Sandwell (117) and Nuneaton (114) and prompted much analysis and investigation to
understand why the mortality ratios varied so much across the country.
Following the publication of the Good Hospital Guide a number of changes were introduced in
Walsall Hospitals NHS Trust. These included the formation of specialist clinic governance groups
to implement changes in a wide range of clinical disease areas such as respiratory, renal and
elderly care, and changes to many different management areas such as audit, education,
information services and bed management. Over the period 2000 to 2004 the HSMR dropped
from a high of 130 to 92.8. This suite of improvements coinciding with a falling HSMR was written
up in the British Medical Journal by Dr Foster as an exemplar of good practice5, though the Dr
Foster researchers reserved judgement about the extent the changes made to the HSMR
compared to other potential factors (better disease coding (classification), different admission or
discharge policies or simply the hospital emerging from a chance period of unusually high
mortality). Additional improvements included audits of unexpected deaths and (from 2008) the
adoption of the Global Trigger Tool.
5 Jarman et al. Monitoring changes in hospital standardised mortality ratios. BMJ 2005;330:329
Item 8.3 Appendix 1
10
3.3 Alternative explanations for high mortality rates
At the same time a separate line of inquiry was focussing on a different possible cause for
variations in hospital mortality measures – the relatively high proportion of people in Walsall who
died in hospital compared to the national average for place of death. After taking into account
the high proportion of hospital deaths in Walsall it was concluded that if people in Walsall had the
same access to other places of death (home, hospice) as the national average, the HSMR at the
time (126) would actually be only 1106.
3.4 Changes to palliative care coding
The Department of Health issued new guidance in March 2007 about coding hospital patients
with terminal disease and palliative care. The significance of this is that the Dr Foster organisation
calculates the number of expected deaths based, amongst other things, on the clinical disease
codes that they are classified as having been admitted to hospital with in order to estimate the
probability of dying. According to these calculations, patients whose medical conditions include a
palliative care code are likely to have a terminal condition and the effect of coding more patients
than normal with a palliative classification code increases the number of patients who are
expected to die in hospital, which potentially can reduce the HSMR. By the summer of 2007 the
Walsall Hospitals NHS Trust HSMR was identified as being high again by Dr Foster. An internal
review was undertaken of how patients with terminal disease were being coded and it was
observed that Walsall hospital was a relatively low user of palliative care codes and patients for
whom they might well have been appropriate were not being coded with palliative care.
Fig 1 Changes in palliative care coding and effect on HSMR, 2007-2010 Walsall Hospitals NHS Trust
6 Seagroatt V. Hospital mortality league tables: influence of place of death. BMJ 2004; 328
Item 8.3 Appendix 1
11
Along with a number of other hospitals at the same time, the Manor Hospital coders
misinterpreted the guidance as recommending an increase in the use of palliative care codes and
hence during 2008 and 2009 more patients who died were recorded as having a palliative care
code then previously. The increase in palliative care coding was accompanied by a commensurate
decrease in HSMR (Fig 1).
Upon the appointment of a new coding manager in September 2009 the palliative care coding
approach was reviewed again and the amount palliative coding used was changed downwards
again. The Stafford Inquiry noted the misinterpretation of Department of Health guidance in a
number of hospitals at the time, including Walsall, but did not consider this to have been
deliberate.
3.5 Monitoring of changes to coding and effects on mortality by Walsall PCT
The PCT and now CCG has been actively monitoring the change in coding practice and its effect
on HSMRs. Mortality ratios as calculated by Dr Foster are presented on a monthly basis to the
safety, quality and performance (SQP) and clinical quality review (CQR) committees. Assurance
was sought through a WHNT board paper on palliative care coding, and detailed analysis of
coding was undertaken by the public health intelligence team in early 2011.
A number of tools have been routinely used by both Walsall PCT (now CCG) and WHNT to
monitor the mortality rates, including the Healthcare Evaluation Data (HED) tool produced by
University Hospitals Birmingham and used by the former Strategic Health Authority (SHA); the
SHMI, the HSMR, crude mortality rates and actual numbers of deaths.
3.6 Extending patient choice and availability of preferred place of death
Historically the proportion of patients dying in hospital in Walsall has been very high in
comparison to other parts of the country (fig 2), and it would be appropriate to expect a
relatively high proportion of deaths in Walsall hospital to be coded as palliative care. In other
areas, where fewer patients died in hospital and more died in a hospice, all the hospice patients
by definition would be terminally ill and would receive palliative care. So logically if those patients
were instead to die in hospital it would be anticipated that they should be coded as palliative
care. However, by 2011 the proportion of patients coded as palliative care had been reduced to
pre-2008 levels. Meanwhile the HSMR increased, and the six month period to February 2012 it
was 118 making Walsall a high outlier and thus unlikely to be that high just by chance.
There has long been the option for Walsall patients to die under hospice care, but prior to the
opening of the St Giles Walsall hospice, this was limited to a small number of places at out-of-
borough units including Compton Hospice (Wolverhampton) and St Giles Hospice Lichfield.
Hence end of life profiles prior to the opening of the St Giles unit in 2011 still show a small
proportion of hospice deaths (in 2008-2010 the proportion was 1.8% of all deaths) but since the
opening of the St Giles Walsall hospice in mid-2011 the proportion has increased to over 5% of
deaths, meaning that Walsall is on a par with the national average for the proportion of people
Item 8.3 Appendix 1
12
dying in a hospice. Note that this statistic was calculated locally; the national data calculates a
three year average and so will continue to suggest that Walsall has a lower access to hospice care
than the national average for the next two years until the pre-Walsall hospice opening year
figures are completely out of the equation.
Fig 2 Proportion of Walsall residents who die in hospital, home, care home or hospice, 2008-10
(National end of life profiles)
Since the opening of the Walsall hospice the proportion of patients dying in hospital before and
after has changed from 61.9 to 58.1% - a fall of 3.8%. The effect of opening the hospice has been
primarily to reduce the numbers of people dying in hospital and balance this with an
approximately equal number of people dying in the hospice, with the proportions of people
dying at home or in care homes remaining largely unchanged.
Nationally the majority of people say they do not want to die in hospital, though nationally
around half of us will eventually die in hospital. Choices of preferred place of death can change at
different stages of a patient’s illness, and also may depend on the condition. A recent survey of
cancer patients referred to the Walsall community palliative care team found that under 1% of
patients preferred to die in hospital, and of those asked where they wanted to die, the vast
majority ended up dying in their preferred place of death (home, care home or hospice – not
hospital).
3.7 Redevelopment of Goscote Lane and the opening of the Palliative Care Centre
Well before any concerns were raised in late 2011 about the way the HSMR had crept back up,
significant work had been undertaken by Walsall PCT to improve patient choice in end of life care.
The land occupied by the former Goscote Hospital in Goscote Lane was redeveloped into a
hospice and end of life unit, the £5 million Palliative Care Centre following up to two decades of
planning and preparation. Construction began in 2009 and the unit officially opened in April 2011,
though throughput did not get up to full capacity for the first few months.
3.8 Additional consultants in palliative medicine recruited to Walsall Healthcare NHS Trust
At the time the hospice opened there was only one consultant in palliative medicine working half-
time in the Manor hospital. With the opening of the Goscote hospice (requiring four consultant
sessions, or two full time days per week input) in early 2011 this presented an opportunity to
completely rethink the provision of palliative medicine across the health economy. Funding was
Item 8.3 Appendix 1
13
agreed from the PCT to create two new posts, employed by the WHNT and arranged so that one
worked primarily in the Manor hospital and the other worked primarily in the community, but
with significant overlap and cross-cover, in addition to the sessions provided in the hospice (and
for completeness, there are additional sessions funded in Wolverhampton hospital for Walsall
patients who are admitted there). Two new consultants were appointed in April 2012 and
provision of consultant palliative medicine sessions has greatly expanded as a result.
3.9 Hospital review of mortality
WHNT developed an action plan to address HSMRs which was presented at the January 2012
Health Scrutiny and Performance Panel. This had four key areas:
(i) Improving the care of patients with respiratory conditions
(ii) Enhanced senior medical review at evenings and introducing six day ward reviews
(iii) Improvements in palliative and end of life care
(iv) Reviewing all deaths and seeking best practice from other organisations
Item 8.3 Appendix 1
14
4. Walsall CCG Mortality Review Group
The recent concern about mortality rates at Walsall is therefore not new but is a continuation of
publication of mortality statistics over a period of over a decade flagging the hospital as being
higher than the “expected-by-chance alone” range, followed by internal quality and process
reviews and subsequent reductions in mortality rates.
Independently of this and following discussions during 2011 at Health Scrutiny and Performance
Panel (HSPP), the CCG in collaboration with the Black Country PCT Cluster has been undertaking a
range of actions culminating in a project plan for a mortality review group consisting of a number
of partner agencies across Walsall to effectively address the high HSMR, looking at the wider
health economy as well as the hospital. The purpose of this project plan was
• To understand the factors that contribute to HSMR and SHMI
• To devise and prioritise effective strategies that impact on these measures
• To monitor the impact of the actions put in place
In recognition of the contribution of the wider health economy to HSMRs, the project working
group was designated the CCG mortality review group, rather than HSMR review group.
The project has consisted of several workstreams and functions, including
• Performance Assurance – formal critique and performance management of the WHNT
action plan for reducing HSMR by the SQP Committee of the CCG.
• Quality of care provided by WHT in the acute and community settings
• Quality of care provided by nursing homes, including reviewing and implementing
effective options for improving care and improving referral options for care homes with
deteriorating patients
• Review of data, estimations of mortality rates, coding and reporting of figures
• Addressing and strengthening the role of General Practices
• Review of local end of life services, in particular, the performance of hospices and
community end of life services
• Quality and accessibility of community services provided by the Local Authority
Each project workstream has reported into the mortality review group which meets monthly and
is a subcommittee of the Walsall CCG Improving Outcomes Programs Board, accountable to
Walsall CCG governing body. A wide membership of health economy stakeholders includes CCG,
WHNT medical director, other clinicians and senior managers, Walsall Council and CCG Board. A
lay Chair was appointed who was at arms’ length from the CCG but had past experience of the
PCT to chair the mortality review group meetings. A formal project plan was agreed by the CCG
Board in June 2012. Progress updates were provided to the National Commissioning Board Local
Area Team. The mortality review work is aligned to regional work examining mortality rates and
is aligned to the corporate business (SQP and CQR committees) of the CCG.
A key purpose of each workstream was to understand the issues that it was responsible for
which could influence hospital mortality, and ensure that opportunities for improving quality
Item 8.3 Appendix 1
15
could be identified and acted upon. Underpinning this was both agreement that robust data and
evidence would be required in order to gain the fullest understanding of the contributory factors
to hospital mortality, and recognition that the groups were not intended to try to improve the
data for its own sake, but to improve quality, regardless of the impact this might have on the
data.
Item 8.3 Appendix 1
16
5. External Drivers
A number of external factors also directed the work of the mortality review group. These
included the decision of the HSPP to commission a review of hospital mortality rates at WHNT
that would be entirely independent of the Walsall healthcare economy and provide additional
assurance during the period in 2010 to 2012 when the HSMR was a significantly higher than
expected outlier. The management consultancy Mott Macdonald was commissioned to
undertake the review, the findings of which were presented to HSPP on 23 April 2013.
Meanwhile in late 2010 the Government had set up a formal public inquiry into standards of care
at Stafford Hospital, the second “Francis Inquiry”. This reported in early 2013 and was highly
critical of patient care at that hospital. However, during throughout the time that the Walsall
mortality review group has been meeting, it was felt inevitable that there would be wider
implications and recommendations from the Stafford inquiry for the entire health service. Any
hospital known to have high mortality rates, such as Walsall, could potentially expect to be put
under intense scrutiny.
Following the publication of the Francis Inquiry in February 2013 it was announced that fourteen
hospitals would be investigated for their quality of care by the Medical Director of NHS England
during the summer of 2013 as a result of persistently high SHMIs or HSMRs. WHNT is not included
in these reviews.
At the same time concerns were raised at the suggestion that hospitals might attempt to
“massage” their data to lower their mortality statistics, prompting the Health Secretary to
announce that this would become a criminal offence. Bolton NHS Foundation Trust was initially
highlighted as a possible example of a trust “gaming” data to improve mortality statistics
because coding rates for sepsis were much higher than expected, increasing the apparent
number of “expected” deaths. It was later completely exonerated, since it transpired the Trust
was following the UK Sepsis guidelines on early identification of sepsis patients and patients
were being coded with sepsis who elsewhere would not have been identified. Ironically the
combination of following a model of best clinical practice and an ambiguous diagnosis code
without national definitions had led to a situation that at first sight appeared to be suspicious,
and is a helpful lesson in the limitations coding puts on hospital mortality statistic calculations.
Finally, the Liverpool Care Pathway, though offering a framework of best practice in palliative
care when applied properly, has recently come under intense scrutiny and generated significant
public anxiety as a result primarily of media reporting of its potentially inappropriate use. This
therefore features prominently in contemporary discussions of improving end of life care.
Item 8.3 Appendix 1
17
6. Summary of Phase 1 findings of the Mortality Review group
6.1 Performance monitoring and assurance
Routine and ongoing monitoring of HSMR provided by WHNT happens on a monthly basis during
the PCT/CCG’s regular SQP and CQR committees. These committees monitor the quality of care
and critique and performance manage the WHNT action plan for reducing HSMR.
For additional assurance, the GP chair of the SQP committee attends the hospital mortality
review group.
6.2 Hospital activities
Actions undertaken in the hospital to improve quality and reduce mortality have included six-day
consultant ward rounds, standardisation of ward rounds, implementation of care bundles
modelled on the Nightingale Foundation approach, in-depth reviews of every patient who dies
and improvements in clinical coding. There are particularly detailed reporting processes for
patients dying from causes that have historically had high mortality rates, especially pneumonia
and sepsis.
A review of consultant job plans enabled extension to six day specialty ward rounds. This is in
addition to existing emergency admission areas which receive seven day 24 hour consultant
cover. Additional investment enabled the appointment of six acute medical physicians and four
further Accident and Emergency consultants plus an orthogeriatrican and renal specialist.
A “consultant dashboard” has been developed based on each consultant’s activity, mortality,
length of stay, and readmission rates to identify outliers and is used in the consultant appraisal
process.
The model chosen for the internal mortality review group was developed with assistance from Dr
Martin Farrier of Wrightington, Wigan and Leigh NHS Foundation Trust, who is a national leader
in this field. Membership comprises a senior multi-professional team including medical director
(chair), GP (CCG) representative, retired consultant, Nurse Director and Chief Executive.
Published evidence and experience from other NHS organisations has shown that reviewing all
deaths in hospital is a key to identifying issues in care delivery, and two senior consultants are
specialist reviewers, along with a senior coding manager and the patient’s consultant in all cases.
Feedback and learning points are disseminated widely. This model of mortality review is
considered best practice.
A monthly mortality report has been delivered during the delivery of the action plan and
continues, supplemented with in depth analysis from data analysts within the organisation, the
Dr Foster organisation, University Hospital Birmingham’s HED team and regional experts.
Analysis of data provided both new areas for attention and tracking of progress.
There have been significant improvements in palliative care coding and improvements in coding
of comorbidities. The HSMR has been recalculated by University Hospitals Birmingham excluding
Item 8.3 Appendix 1
18
palliative care coded patients and this has fallen, suggesting improvements irrespective of
palliative care coding practices.
Care Bundles have been launched for pneumonia and sepsis (“surviving sepsis” campaign). These
evidenced based checklists promote prompt care and have been adopted well in clinical areas.
Review of existing clinical guidelines commenced in 2012 including fluid & electrolyte guideline
which is being prepared with teaching provision.
The respiratory & heart failure service provide guidelines and assertive in-reach to acute medical
admission areas. In addition the orthogeriatric service is supported via in reach from older adult
physician to orthopaedic wards to ensure medical management is optimised.
WHNT recognises that communication skills are essential and is funding two-day communication
skills courses for consultants, following an accredited model. There is an ongoing programme of
education for nurses and allied health professionals.
6.3 Mortality and care homes
A six month pilot project sought to improve quality of clinical care and resident care in care
homes. Audit of reasons for why patients are admitted from nursing homes to hospital has
demonstrated opportunities for synergy between care homes and the Manor hospital for
reducing avoidable admissions and potential mortality for conditions that could have been
managed in the community.
A clinical wrap around team to support frail elderly patients at the end of life was introduced in
four nursing homes with support from two GP practices and secondary care domiciliary elderly
care physician. This resulted in fewer admissions to hospitals and those patients who were
admitted were more appropriately admitted. It was notable that patients with cancer tended to
receive a more collaborative approach than patients with non-cancer long term conditions. Of 15
unplanned admissions during the pilot 9 were due to respiratory conditions and audit suggested
that all these admissions were appropriate. 89% of the deaths reported during the project were in
the patient’s preferred place of death. GPs reported fewer emergency call-outs to the homes,
whilst the homes reported greater confidence in applying end of life care pathways. The majority
of patient reviews were for respiratory or urinary tract infections.
All nursing homes in Walsall have been supplied with new syringe drivers in line with NICE
guidance and a comprehensive education package was provided for the use of them in end of life
care in nursing homes.
Medicines management reviews in nursing homes have identified where long-term prescribed
medicines might be substituted by more effective or better tolerated medicines, enhancing
compliance and avoiding waste. This has made savings of over £76,000 whilst ensuring that
patients with repeat prescriptions were on the most appropriate medicines.
A locally developed and delivered five-day leadership program has been undertaken for care
home managers, and the majority of managers have now attended this program. A self-
Item 8.3 Appendix 1
19
assessment contract monitoring framework has been developed to assess clinical care quality,
support improvement and benchmark performance.
6.4 Data and intelligence
A desktop review of current literature on hospital mortality statistics was undertaken to identify
current thinking and emerging trends. The 2010 “Dying to know” document7 is particularly
helpful in explaining what factors outside the control of hospitals – particularly palliative care
services – can influence mortality rates, and recommends how to investigate high mortality rates.
The NHS Confederation has also published guidance on investigating HSMRs8.
A meeting of the Health Statistics Users Group in London was attended in January 2013 to
observe a detailed discussion of hospital mortality statistics, including limitations regarding social
and demographic details and the problems with palliative care coding. For instance the HSMR
methodology includes palliative care coding and attempts to adjust for deprivation but the SHMI
does not. The NHS health and social care information centre (which calculates SHMI) illustrated
the difficulty of defining when an episode of care could be classed as palliative care – for
instance, should it be applied to any patient admitted to hospital who is cared for by a
community palliative team, should it only include patients who are seen by a palliative medicine
consultant (or maybe nurse) during their admission, or should it include patients who are not
seen by a palliative care specialist but whose admitting team discusses their care with a palliative
care specialist during admission. The significance of this is that these definitions can affect the
“expected” mortality of a patient and hence can influence the hospital’s HSMR.
One reasons for a high HSMR could be that a comparatively high proportion of hospital deaths
occurred soon after admission in patients referred in from care homes. Transferring patients to
hospital from care homes at the end of life is often not considered the most appropriate care. In
Walsall around 6% of patients aged over 65 who died in hospital died on the day of admission and
31% died within 3 days of admission. Discussion with national experts suggested that deaths so
early in admission were unlikely to result from poor care and most likely reflected admitting
behaviour. However national comparative data was unavailable to benchmark Walsall’s position.
On average around four Walsall GP-registered patients a month are admitted to Walsall Manor
hospital from care homes and die within three days of admission (using postcodes of care homes
as a proxy for care homes). A further six patients die in a care home within 30 days of discharge.
It is reasonable to ask whether for some of these patients the hospital admission was
appropriate or whether the patients could have been better supported to die at home. A limited
number of care homes contributed many of the admissions and could be targeted for extra end-
of-life support. Around one patient a month was referred from a care home to the Accident and
Emergency department but died before they were even admitted.
7 Dying to Know - How to interpret and investigate hospital mortality measures. Flowers et al. Association of Public Health Observatories 2010 8 Hospital standardised mortality ratios: their uses and abuses. NHS Confederation 2010
Item 8.3 Appendix 1
20
A relatively low proportion of patients (4%) admitted from care homes died from cancer within 3
days, whereas 30% of deaths within 30 days of discharge were from cancer, suggesting
considerable success in averting hospital admissions of patients with cancer at the end of life and
success in discharging patients with cancer away from hospital to their preferred place of death.
Fig 3A & B Proportion and actual numbers of patients dying per month in hospital, hospice, home
and care homes in Walsall 2008-2013, showing changes when the palliative care centre opened.
Review of local data and comparison with national benchmarks confirmed that historically
Walsall has had a very high proportion of people dying in hospital compared to other parts of the
country (almost 62% in the period 2008-10) and comparatively few deaths happened in care
homes or in hospices (due to lack of availability – prior to the opening of St Giles only 1.8% of
people died in hospices, and Walsall has comparatively low availability of care home beds too).
Since the opening of the St Giles Walsall hospice in April 2011, a steady increase in the proportion
of people dying in the hospice has been matched by an equal reduction in people dying in
hospital, so that the proportion currently dying in the hospice is similar to the national average
(over 5%) and the proportion dying in hospital has fallen to 58%, though this is still some way
above the national average (Fig 4A & B).
Item 8.3 Appendix 1
21
6.5 Addressing and strengthening the role of General Practices
Heart failure pathways have been developed jointly between lead GPs and cardiologists and
presented to CCG locality groups. Care plans are issued by secondary care to patients with severe
heart failure who are approaching the end of life, and the care plans are then used by GP
practices to manage patients. The palliative care team now has a joint heart failure clinic at the
Goscote palliative care centre.
GPs highlighted concerns that some patients in care homes were experiencing repeated
admissions, sometimes of patients who had not been brought to the GPs’ attention even when
they were visiting the home earlier the same day. This led to the development of the pilot twice
weekly joint proactive “ward rounds” with a community elderly physician of all nursing home
patients in four selected nursing homes, rather than only those that the nursing homes were
requesting the GPs to visit.
A workshop was convened in early 2013 jointly with WHT to look at the contribution primary care
can make to end of life pathways. A communications skills training package for GPs has been
developed to build GPs’ confidence in initiating conversations about “do not resuscitate” orders
in primary care. West Midlands Palliative Care Guidelines have been circulated to every GP.
Other conditions have been identified that can be supported by formal end of life services,
allowing local care pathways to be developed, beginning with respiratory disease.
6.6 End of life
The primary aims of end of life care in Walsall are to ensure that patients, their families and carers
are treated with dignity and respect, being supported to die in their preferred place of death
where possible, whilst patients will be listened to and able to contribute to a personalised
advance care plan wherever possible.
This depends on recognising and acknowledging the palliative and/or end of life phase, offering
patients and carers the opportunity to discuss their priorities for care and place of death, treating
on assessment of need, reviewing and amending the plan, identifying the most appropriate
location for care to be provided avoiding admission to acute trust services unless clinically
appropriate, and to achieve a seamless and timely transition across care environments.
The palliative care and end of life service aspires to achieve the following:
• Increase in the number of patients being identified as entering the palliative or end phase
of life and benefitting from the development of an advanced care plan
• Increase in number of patients having their preferred place of care and death identified
and recorded within their records
• Increase in number of patients achieving their preferred place of care and death
• Year on year increase in number of non-cancer patients accessing palliative and end of life
care services
• Increase in number of patients receiving end of life care closer to home
Item 8.3 Appendix 1
22
• Reduction in number of patients receiving end of life care in the acute trust environment
where this is safe and viable to do so
• Increase in number of patients receiving co-ordinated care
• Improved patient and carer experience
In 2011 the St Giles Hospice Walsall opened in Goscote, and during 2012 two consultants in
palliative medicine were employed by WHNT, one based primarily in the hospital, the second
working in the community. Following the publication of the NICE Quality Standards and advice
for commissioners for end of life care, and more recently from the National End of Life
Programme, “The route to success in end of life care - achieving quality in acute hospitals”, the
palliative care team are exploring new models of provision, focusing on the following “enablers”
to allow acute hospitals to deliver excellent end of life care:
1. Advance Care Planning (ongoing within NHS Walsall). As of March 2013, the new Advance
Care Plan for the Walsall borough is available for health and social care providers to
adopt;
2. Electronic Register of Palliative Care Patients. WHNT are currently using the Gold
Standard Framework and are planning some Patient and Public Engagement before
extending this further into the hospital on a rolling basis throughout all specialties;
3. AMBER Care Bundle – this is a new opportunity for the Trust and is being implemented on
a ward-by-ward basis. It will allow staff to identify patients whose recovery is uncertain,
and to make plans with the patient and family in a timely manner;
4. Use of Liverpool Care Pathway – this is already embedded in the hospital and existing
community nursing teams however the working group continue to seek assurance on the
quality and impact on the team, the patients and members of their families.
5. Rapid discharge home to die. This enables people who are in hospital with terminal illness
to go home to die, and the Trust is working with the CCG to implement this.
There has been considerable success ensuring cancer patients have been able to die in their
preferred place of death (which is rarely a hospital setting). Palliative care capacity in the Trust
has been reflected in the increase in patients coded as palliative, which is now above the national
average again (and appropriately so, since even after opening the hospice a higher proportion of
patients still die in hospital compared to the national average).
Whilst end of life services have got most experience in cancer pathways, it is recognised
nationally that end-of-life services for patients with non-cancer terminal conditions are in general
less well developed, even though more patients have non-cancer terminal conditions than
patients with cancer. A heart failure outpatient clinic has been set up at Goscote and there is a
joint heart failure and palliative care multi-disciplinary meeting, as part of the ongoing work to
improve access to palliative care services for non-cancer patients with life-limiting conditions.
Item 8.3 Appendix 1
23
7. Impact of mortality review group
Nursing home pilot studies have observed fewer admissions and more appropriate admissions.
Patients are increasingly dying in their preferred place of death and there has been a shift of
around ten patients each month who are now able to choose to die in the St Giles Walsall hospice
who prior to 2011 would have had no choice but to die in the Manor Hospital. The palliative care
coding rate between April 2012 and January 2013 was 3.7%, against a national average of 2.5%
which is entirely to be expected given the additional resources of two new consultant positions
in a hospital in which historically a high proportion of the population died. Clinical pathways that
can prevent admission to hospital and improve care in the community, and improve the safety of
patients in hospital have been prioritised. Assurance of safe, high quality clinical processes and
pathways has been strengthened.
Between April 2012 and March 2013 the HSMR has been below 100 for twelve consecutive
months. The latest available month at the time of writing (June 2013) is March 2013 with a ratio
of 94.6 and the year to date HSMR is 89.3. The latest SHMI is 1.11 (October 2011 – September
2012), and this is within the expected range. The SHMI data does lag behind the HSMR data and
currently includes a time period during which the HSMR was higher, though the SHMI and HSMR
are measuring slightly different things – the SHMI includes patients who died within 30 days of
discharge whereas HSMR only counts those who die in hospital. So certain patients who are no
longer included in the HSMR may still be counted by the SHMI if they die within 30 days of
discharge.
In addition to changes in HSMR and SHMI, it is important to consider absolute numbers
(observed numbers of deaths) and crude (unadjusted) death rates. Over the past two years it is
notable that the absolute number of deaths, and the crude death rate, have fallen (Fig 4), whilst
the expected number of deaths has increased. The fall in absolute numbers is probably directly
attributable to the opening of the St Giles Walsall Hospice. Crude mortality tends to increase in
months when there are higher numbers of deaths, particularly during the winter season
(coloured green in Fig 4), but it is less clear why HSMR, which is supposed to take account of
casemix, should also increase during these months. Most likely the HSMR is failing to completely
adjust for casemix, especially if coding has improved since early 2011. A fall in crude mortality can
signal the effect of improvements in quality of care.
Any mortality statistic that is presented as a ratio of observed to expected deaths (for example
the HSMR and the SHMI) can be affected by one or both of observed and expected deaths. For
example if observed deaths falls (e.g. because improvements in choice for end of life mean some
patients can now die in the hospice who previously would have died in hospital, or because
clinical pathways have been improved) HSMR will fall. If expected deaths rises (e.g. because of
improved coding or a sicker patient population), HSMR will also fall. In the case of Walsall, the
expected mortality over the past two years has been rising whilst the observed mortality has
fallen, which helps to explain why the HSMR has fallen as sharply as it has in Walsall.
Item 8.3 Appendix 1
24
Fig 4 Changes in HSMR, crude mortality, observed and expected deaths 2011-12, including the winter
peak (green shading).
Item 8.3 Appendix 1
25
8. Independent management consultants’ review
The independent Mott Macdonald report was presented to Council HSPP in April 2013. It included
assessments of the wider health economy, quantitative analysis of data and qualitative appraisal
of quality of care. The remit and scope of the study precluded detailed ward-level observations
however.
A number of hospital-led initiatives were highlighted as potentially contributing to the reduction
in HSMR, including the introduction of six-day consultant ward rounds, increased staffing levels
in the Acute Medical Unit, introduction of care bundles, mortality reviews of all patients who died
in hospital with discussion across many participants and dissemination of lessons learnt, and
standardisation of ward rounds
The report drew attention to the improvements in end of life care, including the appointment of
the two new palliative care consultants, improved palliative care coding and the contribution of
the St Giles Walsall hospice to removing observed deaths from the calculations underpinning
hospital mortality and thus reducing both HSMR and SHMI. The contribution of the PCT/CCG,
both to improved quality of care and confidence managing frail elderly patients in care homes,
and to establishing the CCG-led mortality review group was acknowledged.
The report identified additional areas that further work should investigate, including a perception
that discharge planning arrangements were not working optimally, leading to longer than
necessary length of stay, and the need to ensure palliative and end of life services that have been
so successful to address the needs of cancer patients also can ensure they meet the needs of
non-cancer patients. Certain specific areas of improvement, such as relieving pressures in
Accident and Emergency, and improving access to non-invasive ventilation support were also
recommended.
Item 8.3 Appendix 1
26
9. Lessons learnt and conclusions
A number of important lessons were learnt in the process of undertaking this review that could
be considered steps for other organisations considering setting up a review group. Firstly,
establishing a core purpose that required system-wide support and board level signup.
Commitment of senior management attendance, openness and transparency and establishing
high levels of trust are essential to a robust and wide-ranging discussion. Strong leadership of
each individual organisation and their separate workstreams complemented the collective
leadership of the CCG mortality review group, creating a synergy in the group. Willingness to
consider from the outset the whole health economy rather than a narrow focus on just the
hospital was essential since it was clear from early examination of local data that many of the
factors that contributed to the HSMR were beyond the immediate control of the hospital in
isolation.
The importance of secure ownership of the problem and trust by all parties to find solutions
cannot be overemphasised. The subject matter of hospital mortality statistics is technically
complex, easily misunderstood, professionally controversial, emotionally distressing and
politically challenging. Another key lesson was the importance of independence and
interdependence, with both CCG and hospital leading their own review groups. It is essential to
use all the available evidence and interrogate and challenge interpretations and explanations,
and use the findings to support and review actions.
Use of data was important to help focus attention on what were likely to be the key areas
contributing to hospital mortality – particularly end of life care and community factors – but data
was not studied for its own sake nor was it analysed with the intention of assuring the group that
there was no scope for further improvement. Conversely, even if some of the improvements in
quality of care identified by the project have no measurable impact on HSMR they are still
important.
Finally it is no small undertaking to commit to a review of hospital mortality statistics, especially
when it includes wider health economy and community stakeholders.
Item 8.3 Appendix 1
27
10. Next Steps
Care homes
Phase two of the CCG project workplan will focus particularly on the contribution of community
services and of primary care to reducing HSMR. The Interim Director of Social Care & Inclusion is
developing a project plan for the community services workstream.
Primary care
The CCG Board member lead on quality is developing a project plan for the primary care
workstream, including work on end-of-life respiratory pathways, communication skills training for
GPs to initiate end of life discussions, supporting community nursing teams in end of life care and
linking the primary care workstream to the work in care homes. The nursing homes pilot is
expected to be rolled out further.
Palliative and end of life care
A working group is looking at rapid discharge home to die, with the intention of minimising visits
to Accident and Emergency Departments and supporting patients to die in their preferred place,
i.e. at home or their care home with the appropriate support package in place. Consideration
needs to take into account how out of area patients may be managed. A specification for a
Hospice at Home Service to include supporting those patients who may live alone or lack
support, is being considered.
Work is ongoing to improve access to palliative care and end-of-life services for patients with
non-cancer life-limiting conditions, especially heart failure, respiratory failure and dementia.
The End of Life Quality Assessment Tool (ELQUA), a national quality assurance and measurement
tool is being adopted by Walsall. WHNT intend to use the tool to benchmark themselves against
providers and track progress and share good practice.
Palliative care coding within WHNT is expected to be independently audited and benchmarked
for further assurance.
Item 8.3 Appendix 1
28
11. Acknowledgements
This was a highly complex and detailed piece of work that required the support of many people.
Particular recognition and thanks must be recorded to Ms Salma Ali, Geoff Archenhold, Mr James
Avery, Mr John Bolton, Ms Reena Bhardwaj, , Prof. Michael Campbell, Dr Steve Cartwright, Ms.
Simone Chung, Ms Sue Crabtree, Mr Paul Deeley-Brewer, Dr Isabel Gillis, Mrs Wendy Godwin,
Miss Ruth Hall, Mr. Andy Hood, Mr. David Hughes, Ms Suzanne Joyner, Mr Amir Khan, Mr Richard
Kirby, Dr Radka Klezlova, Mr Roman Kuciaba, Dr Raj Kumar, Dr. Sue Laverty, Dr Raj Mohan, Dr
Paulette Myers, Dr Su Nambisan, Mr. Bharat Patel, Dr David Pitches, Dr Najam Rashid, Mrs Sally
Roberts, and Dr Esther Waterhouse for their invaluable contributions.
Item 8.3 Appendix 1