Memorandum by Brian Capstick (PS 17)
1. There is a need to focus the National
Reporting and Learning System (NRLS) on the more serious untoward
incidents, with a view to identifying the correctable causes of
them and then making changes in the healthcare system that will
deliver a more consistent standard of care.
2. The correctable causes of serious incidents
comprise shortcomings in a relatively small number of processes,
such as supervision, that have already been identified and should
be hard coded into the classification systems of the NRLS and
local incident reporting systems. Serious incidents could then
be screened for the presence of these process failures so that
the evidence accumulating in the NRLS and local systems could
then point the way to feasible changes in the healthcare system.
The relevant processes could then be continuously validated, monitored
and, if necessary, adjusted at local and national level.
3. A major strength of the recommended approach
is that the existing framework of standards could provide a mechanism
for ensuring that appropriate process changes take place. Periodic
inspections against the standards would then provide a means of
audit and feedback. The NHS would then have a joined-up system
of patient safety management that would yield many times the benefit
of the existing arrangements at little extra cost.
4. Recommendations for action by the government
(a) policy should aim to achieve a consistent
standard of care that can be relied upon not to cause serious
harm to patients
(b) determined action should be taken to
improve maternity care in such a way as to reduce the number of
children born with birth-related cerebral palsy.
(c) messages should move away from the emphasis
on clinical error and towards the improvement of processes that
would assist clinicians in their work
(d) public and staff should be reassured
that most NHS care is good care
(e) the NRLS should focus on the more serious
(f) there should be a list of, say, a dozen
or so sentinel events that all Trusts should record, including
maternal death and intrapartum-related hypoxic injury
(g) serious incidents should be screened
for the more common correctable causes of these events, which
should be hard-coded into the NRLS.
(h) an improved taxonomy is required for
the NRLS and policymakers should ensure that it is developed in
conjunction with local incident reporting systems, in particular
the Datix Common Classification System.
(i) the NHS should learn from private providers
in the measurement and analysis of indicators and outcomes to
improve patient safety.
5. My experience of patient safety began
in the early 1980s when I founded a law firm which defended NHS
hospitals against about 10% of the clinical negligence claims
brought in England at that time. I became concerned about the
number of incidents that gave rise to serious injury, especially
in maternity care, and began to investigate the causes of them.
In 1986, I founded a software company, Datix, that now supplies
incident reporting software to healthcare providers serving a
population of about sixty million people. This includes most of
the NHS Trusts in the United Kingdom, large parts of Canada and,
soon, the Military Health System in the USA. I have researched
several aspects of patient safety and published the results in
the peer-reviewed, academic press.
6. I declare interests as a non-executive
director of Datix Ltd and of the NHS Litigation Authority.
1. What are the risks to patient safety?
7. The risk to patient safety is of lapses
in the standard of care that are liable to cause injury. A high
proportion of the risks that cause serious injury come from a
fairly small area that can be defined and improved. In the majority
of cases, the injury is not caused directly by the doctor in the
form of wrong site surgery, retained foreign objects and so on,
but is caused by the disease after an avoidable failure to recognize
or act on the severity of the patient's condition. Some of the
process failures that are apt to cause serious injury are identified
in Paras 15-16.
To what extent are the risks to patient safety
8. Risks to patient safety can be reduced
but probably not eliminated. Lapses in the standard of care occur
as the result of human error and weak systems for the delivery
of care. "Pure" human error that is not compounded by
any discernible system error is difficult to reduce other than
by training, except in the case of serial offenders who should
be removed from the front line. However, there are a number of
processes whose shortcomings create enlarged spaces for human
error. Supervision is an example of such a process which is familiar
to most people. A proportion of the risks to patient safety may
be reduced by the improvement of these processes.
9. The potentially correctable causes of
patient safety incidents have been narrowed down to a list of
usual suspects and patient safety policy should move quickly towards
validating, assessing and rectifying them. They include the processes
referred to in Paras 15-16 below and in the article that accompanies
The definition of avoidable risk
10. A risk is avoidable if it can be mitigated
or removed by delivering a standard of care that is acceptable
to the community at large. The aim should be to provide a consistent
standard of care that seldom or never fails in such a way as to
cause serious injury.
11. The extent of the patient safety problem
is that the NPSA receives reports of about 25,000 serious patient
safety incidents a year. The NHS Litigation Authority receives
about 5,000 clinical negligence claims a year, of which about
half cannot be successfully defended. Most claims in excess of
one million pounds are brought by or on behalf of children with
birth-related cerebral palsy and prompt action should be taken
to end this state of affairs, each of which represents a tragedy
for the family concerned.
12. While these are sizeable numbers, they
do not amount to a pandemic of poor treatment except, possibly,
in the case of maternity care and the public, including those
who work in the Service, should be reassured that the NHS continues
to provide a preponderance of good care and conscientious clinical
practice. Scare stories which over-emphasize the scale of the
problem undermine morale and are not helpful in winning the support
13. The task of eliminating process failures
that are apt to have serious consequences should be driven by
NHS leaders with the support of the public and NHS staff.
What is the role of human error?
14. In 2006 I carried out a study to assess
the extent to which human error was responsible for the incidents
serious enough to give rise to litigation claims and found that
human error was the proximate cause of 97% of them. This is not
surprising when one considers that, on the whole, healthcare is
delivered by human beings and it is their errors which compromise
the safety of patients. However, there are numerous systems whose
shortcomings create enlarged spaces for human error and there
was evidence of such system shortcomings in 50% of the cases in
15. The focus of patient safety should be
on improving weaknesses in the processes which are apt to cause
serious harm to patients when they fail. In an article published
in 2004 which accompanies this submission,
I identified about two dozen processes which are apt to cause
serious injury to patients when they fail. Other commentators
such as Professor Vincent and Professor Arulkumaran have since
produced comparable lists and there seems to be a broad consensus
about where the problems are thought to lie.
16. For the purpose of illustration, the
relevant process failures include:-
delays in making a primary diagnosis
of a relatively small number of conditions or in recognizing the
severity of the patient's condition generally
the omission to act on adverse results
the lack of an adequate management
plan for high risk cases that have been identified
the omission to allocate known high-risk
cases to a reliable pathway or person
allowing or encouraging clinicians
to work outside or beyond their skill set;
the unavailability of skilled supervision,
guidelines or protocols;
shortcomings in clinical leadership
shortcomings in the recruitment or
induction processes and
deficiencies in educating the patient
to deal with his or her own condition.
17. It is possible incorporate such a list
in an incident reporting system and so produce a powerful analytical
tool. Datix has largely completed this work (the "Datix matrix")
and recommends using it in the manner described below.
2. Incident reporting
18. The NPSA deserves recognition for the
success in IT terms of the National Reporting and Learning System,
the NRLS. It has succeeded where so many other healthcare-related
IT projects have disappointed. The NRLS provides a reliable means
of collecting data locally and relaying it to a central system
that spans the entire NHS. However, the NRLS has historically
achieved more in the direction of Reporting than it has in Learning
and it is now time to redress the balance.
Whether adequate assessment is undertaken and
19. The NRLS would benefit from a more comprehensive
classification system. In particular, the taxonomy needs to identify
and codify the causes of serious adverse clinical events in such
a way that the data accumulating in the NRLS point the way to
feasible improvements in the healthcare system. These "correctable
causes" are likely to be the process failures referred to
in Paras 15-17 and should not just be broad descriptions of general
factors such as "communication" or "training".
The lack of a classification system that functions in this way
has made it difficult for greater learning to be gleaned from
20. Amongst the possible sources for a revised
classification system are the WHO's International Classification
for Patient Safety (ICPS), the "Common Formats" being
produced by a federal agency of the United States government,
the Agency for Healthcare Research and Quality or AHRQ, the Datix
Common Classification System or CCS and the NRLS itself.
21. The Datix CCS is probably the most widely
used of these alternatives at this time and is probably also the
most comprehensive, although it is mapped back to the NRLS. The
CCS is currently being revised under the aegis of Project Linnaeus
(http://forums.datix.co.uk/linnaeus/index.php?board=1.0 ) to take
into account some elements of the WHO's and the AHRQ's taxonomies.
Linnaeus combines the existing CCS with contributions from many
eminent people and sources. Apart from Datix itself, the main
participants in Linnaeus are from Scandinavia, the USA and Canada.
The NPSA has been invited to participate but has not yet responded
at the time of writing in mid September 2008.
22. Linnaeus is aimed at producing a classification
system designed to be embedded in a computer software program
and is expected to produce the world's most comprehensive and
usable classification system for adverse events in healthcare.
Policymakers should ensure that there is a process to ensure that
the NRLS and the Datix taxonomies develop in conjunction with
each other. Otherwise, there will be a wasteful duplication of
effort in creating two systems and then of mapping the NRLS and
the CCS to each other.
A focus on the more serious events
23. In addition to the development of its
classification system, the NRLS should begin to focus on the 25,000
or so more serious untoward incidents or SUIs that occur each
year. In the past, the NRLS has functioned as a system to notify
the NPSA of large numbers of patient safety incidents with a relatively
light amount of data about each one. The next step might be to
retain a notification system for the less serious events while
collecting a greater amount of data on the more serious incidents.
The additional data that needs to be collected in relation to
the more serious events includes a consistent set of data about
their causes coded in a way that could be aggregated for large
numbers of cases. The causes that need to be codified include
the correctable causes referred to in Paras 15 and 16.
24. Another element of the taxonomy which
would benefit from development and, following a process of consultation,
endorsement by the NPSA would be a defined list of SUIs or "sentinel"
events. A list of sentinel events is to be preferred to a formula
that will be interpreted differently from one region to another
and should include deaths resulting from the wrong route for the
administration of medication, wrong-site surgery, maternal death
and moderate or severe intrapartum-related hypoxic injury.
25. The intrapartum-related hypoxic injury
group is important because about half the surviving infants will
show signs of neurological impairment, including cerebral palsy
in about one sixth of cases. Besides being an ongoing tragedy
for the family, the cerebral palsy group is the source of most
clinical negligence claims above one million pounds against the
NHS. Intrapartum-related hypoxic injury is evidenced by stillbirths,
early neonatal deaths and the syndrome of moderate or severe HIE
(hypoxic ischaemic encephalopathy) in newborns. The incidence
of stillbirth and early neonatal death is about 8 per thousand
births and of moderate or severe HIE about 1.5 per 1,000.
26. When sentinel events or SUIs occur,
they should be the subject of a consistent analysis aimed at screening
for a defined list of contributory factors. A consistent list
of pre-defined causal factors is already embedded in the CCS.
If causal factors were consistently recorded, the data could then
be aggregated for large numbers of cases in such a way as to reveal
the frequency of the various factors which, in turn, would point
the way to the possible nature and impact of the measures required
to mitigate them. Conversely, if little effort is made to standardize
the classification of causes, it is more difficult and perhaps
impossible to discern common themes in large numbers of cases
and correspondingly difficult to learn lessons from a database
of adverse events.
27. In common with several other experts
in this field, I would suggest a review of the benefits of root
cause analysis (RCA) as it is currently applied in the NHS. One
fundamental problem is that RCA does not provide sufficient consistency
in the identification of causes for them to be automatically processed
in a database. Another problem is that it is too time-consuming
an approach to apply consistently well to large numbers of cases.
Policymakers should consider a move away from large-scale RCA
and towards the application of a simple screening test as described
Impact of the changing public-private mix in provision
28. Choose and Book and the Extended Choice
Network have increased the importance of the role of private providers
in the NHS. To qualify for inclusion in the Extended Choice Network,
private providers must meet certain standards in risk management
and patient safety. To assist with compliance, a number of large
private providers have used Datix software to evidence improvements
in care brought about by learning from patient safety incidents.
29. Some of the larger private providers
use the data they collect in Datix on clinical and patient safety
indicators and variances from care pathways to measure the quality
of outcomes by hospital, procedure and clinician. Lessons learned
from the analysis of this data have a tangible impact on improving
clinical outcomes, one of the areas targeted by Lord Darzi in
High Quality Care for All. The measurement and analysis of indicators
and outcomes to improve patient safety are processes where the
NHS could learn from the work done by private providers.
Whether past spending on patient safety has been
30. Past spending by definition has been
enough to get us to the point where we are. Future spending should
focus on enhancing the benefits of the existing infrastructure
and should try to avoid the proliferation of too many initiatives.
There is a need for improved leadership in patient safety and
a greater sense of direction, but these may not be supplied by
more money. Once policy has been settled, it is essential that
Trusts are given adequate resources to ensure that patient safety
is properly implemented.
Effectiveness of national bodies
31. The inquiry invites comments on the
current effectiveness of national bodies. It seems to me that
the NPSA under its current management team is likely to be effective
in delivering the benefits expected of the NRLS, provided government
policy is aimed in the right direction.
32. The NHS Litigation Authority seems to
me to have done a good job since its inception. It defends claims
against the NHS professionally and expeditiously and has developed
useful and well-regarded risk management standards. More recently,
is has begun to initiate various programmes to learn lessons about
patient safety from the claims experience.
33. Other health service bodies are aware
of the importance of patient safety but would benefit from guidance
about what an effective patient safety programme entails.
34. Education for health professionals should
include modules in patient safety management, emphasizing the
process failures that have been identified as most frequently
implicated in serious incidents. The RCOG already does this.
3. Next steps
35. The emphasis should be on delivering
a consistent standard of care that seldom, if ever, fails in such
a way as to cause serious injury to patients. Evidence of benefit
for this or that method would have to be a matter of comparing
Trusts with good records with those with bad records. However,
the process improvements advocated here are those that any organization
of a comparable size should carry out as a matter of course.
36. Best practice is for clinicians to determine.
The only comment that might be appropriate to make here is to
point out that best practice is not necessarily in the domain
of patient safety. Patient safety is about ensuring the delivery
of a consistent quality of care that can be relied upon not to
fall below an acceptable standard. It is not hard to see how the
aim of delivering a consistent standard of care could become a
popular national policy that the public has a right to expect.
37. The inquiry invites comments about how
to ensure that learning is implemented. A major strength of the
recommended approach is that there is an existing framework of
standards that could be used to ensure that effective processes
are devised, implemented and audited.
38. As a minimum, serious untoward incidents
or sentinel events should be measured and assessed for the presence
of correctable causes, as set out above. Data relating to the
number, type and causes of events on the sentinel list should
39. The main incentive for improving patient
safety should be to create a consistent standard of care that
does not let the patient down. NHS staff, patients and the taxpayer
have a right to expect policymakers, management and clinicians
to pursue this aim without further incentive. This is what most
of them are motivated to do and, if given the right leadership,
will do without the imposition of targets or intrusive supervision.
In an increasingly competitive world, there are also commercial
benefits to be gained by a provider that can supply a reliable
standard of care. In our experience, private providers of healthcare
are very conscious of this.
131 http://cr.rsmjournals.com/content/vol10/issue6/£RISK_MANAGEMENT_IN_PRACTICE Back