Transcript Slide 1
What can we learn from serious
clinical complaints
Graham Neale
Imperial College
Errors in hospital care.
Hospitals are not High Reliability Organisations
The accepted mantra for reducing adverse events:
Active failures are committed by those in direct contact with the
patient – very difficult to modify – sanctions and exhortations have
little effect.
So use root cause analysis to define the underlying causes of error
Identify latent failures in organisational and managerial spheres
and concentrate on avoiding these.
.
But how does the public see medical mistakes?
Much influenced by the legal process
The legal process – Tort negligence
(from L. torquere – wrong, twisted)
The tort of negligence has developed this century largely as a
result of the judgement: Donoghue v Stevenson (1928).
May Donoghue, a single mother of modest means but of much
determination, was enjoying a ‘Scotsman ice-cream float’. The café owner
poured on the ice-cream ginger beer from a brown opaque bottle labelled
Stevenson’s of Paisley. May took a drink and then saw the partly
decomposed remains of a snail. She was distressed and shocked.
The judge articulated the rules that we live by today
The legal process – Tort negligence
• The defendant owed a duty of care to the claimant
• That duty was breached
• The breach caused harm
In medical practice we need to ask
in what way, how and why
was that ‘duty breached’
Specialty
im
ar
y
e
164
rs
in
g
P)
(G
th
al
422
Nu
Ca
r
He
449
Pr
e
nc
752
Am
bu
la
vic
es
tS
er
ol
og
y
774
po
r
Pa
th
gy
sia
1,121
Pu
bl
ic
lS
up
ed
ica
st
he
al
th
He
cy
e
1,210
Ra
di
ol
o
An
ae
ta
l
ge
n
Em
er
2,000
Pa
ra
m
&
ici
n
gy
lo
y
6,000
/M
en
sy
ch
ol
en
t
ec
o
M
ed
G
yn
a
8,000
/P
hi
at
ry
&
Su
rg
er
10,000
Ac
cid
tri
cs
18,000
Ps
yc
bs
te
O
Number of Claims
Claims recorded by NHSLA (to March 2008)
20,000
17,817
16,000
14,000
12,000
9,477
7,920
5,056
4,000
0
149
93
Tr
ea
tm
en
t
Su
rg
er
y
D
ia
gn
os
is
M
ed
ic
at
io
Ad
n
m
in
is
tra
ti v
e
M
on
i to
In
rin
tra
g
ve
no
us
flu
id
An
ae
st
he
sia
Ps
yc
hi
at
ry
Eq
ui
R
pm
is
en
k
m
t
an
ag
em
en
t
Tort Claims VHA 2000
1800
1600
1400
1200
1000
800
600
400
200
0
Data from Weeks WB et al J Law Med Ethics 2001; 29: 335-45
Some background data
Diagnostic errors are a significant problem (Harvard study)
Diagnostic error (14%) > Medication error (9%)
Misdiagnosis - carries a worse outcome (serious 47%)
than medication error (serious 14%)
Diagnostic errors more likely to be unrecognized or unreported
Example cerebrovascular accidents – 9% missed initially – and precursor
event events missed in 40% (Ann Neurol 2008; 64(suppl 12) S17 – S18)
Causation of misdiagnosis has been little studied
In To Err is Human diagnostic error mentioned twice, medication error 70 times
In 93 AHRQ-funded studies only one addressed misdiagnosis
Breakdowns of process in diagnostic error or delay
(Gandhi TK et al Misdiagnosis in the ambulatory setting: A study of closed malpractice claims)
(Ann Intern Med 2006; 145: 488-96)
120
100
60
40
20
0
Po
or
ju
La
ck d gm
of
en
La
t
v
ck
ig
ila
of
nc
kn
ow e
le
Po
dg
or
e
co
m
m
un
ica
tio
n
Po
or
su
pe
rv
is
O
io
ve
rw n
or
In
k
te
rru ed
pt
io
ns
Fa
t ig
ue
Pa
tie
nt
-r
el
at
ed
%
80
Diagnosis is one aspect of human error where the final hole
in the ‘Swiss cheese’ model cannot be easily protected
Diagnostic errors – the next frontier for patient safety
Newman-Toker DE and Pronovost PJ
(Johns Hopkins Medical School, Baltimore)
JAMA 2009; 301: 1060-2.
Can we explore the diagnostic process more
profoundly and make potentially useful
recommendations for improvement
• Relatively little evidence regarding diagnostic error
• Doctors reluctant to discuss
• Reporting is limited
But an important area for psychologists to consider
– can their analyses be applied to diagnostic fault?
Diagnosis is underpinned by data collection
and decision-making
Misdiagnosis: “No fault” errors
Presentation
Silent
Atypical
Non-compliant patient
Related data
Example
None available
Pulmonary emboli
(in the elderly or very sick)
Hidden
Early meningitis
Data not revealed
Vomiting and pregnancy
Misdiagnosis: System errors
System
Interpretation of data
Defect
Interpretation radiographs;
Example
Missed fractures
ECGs
Training
(supervision)
Quality of senior input
(ward rounds and OP clinics)
Epilepsy overdiagnosed by 20%
Follow-up/handover
Abnormal test missed
Anti-coagulation
But most diagnostic errors appear to be
cognitive or behavioural
(Croskerry et al Acad. Med 2003; 78: 775-80)
The 2 systems of decision-making
Characteristic
Type 1
(intuitive)
Type 2
(analytic)
Cognitive
Heuristic
Systematic
Awareness
Low
High
Processing
Parallel
Serial
Automaticity
High
Low
Rate /Effort
Fast/ Low
Slow/High
Errors
Normative distribution
Few but significant
Scientific rigor
Low
High
How might the problem of diagnostic errors be addressed?
Attempts made to analyse diagnostic errors are anecdotal
Data very hard to come by – hidden clinically
and by the medico-legal process
Here describe a pilot study based on cases sent for advice examining
what are described as “cognitive dispositions to respond” (CDRs)
(Croskerry describes > 30 CDRs in article in Acad Med 2003; 78: 775-80)
Breakdown of 35 claims
submitted for advice
Diagnosis
Management
Medication
Monitoring
Surgery
A case of severe headache
31 years’ old woman (39 weeks’ pregnant) woke at midnight with severe headache
Seen in A&E at 6am. Started on antibiotics for UTI
No cause for headache associated with pregnancy and delivered within 24 hours.
Medical registrar diagnosed ‘tension headache’ but queried for CT
Over next week recurrent headache especially at night (in the early hours) –
requiring opiate analgesia. Occasional vomiting. No neurological signs.
Subsequent medical assessment – low grade pyrexia ‘probable viral infection’
Consultant note “Improved. Complete course of antibiotics. CT not indicated”
Patient discharged despite persistent symptoms over next 4 days
(without mention of headache in discharge note)
Headaches and vomiting persisted – then on day 6 grand mal fit
CT scan: brain abscess with ventriculitis
Neurosurgical intervention – recovered but left with severe mental impairment,
The dangers of hospital organisational structure
Hierarchical leadership – the apprentice-bias
Premature closure
A case of abdominal pain
Female aged 27 with RUQ pain and mildly disturbed liver function. To physician
Palpable liver; normal ultrasound. Diagnosis – hepatic steatosis. Advised re weight
To private surgeon. Minimal assessment – need laparoscopy.
Gallbladder removed – cholesterolosis
Post-operative – persistent severe pain and vomiting. Weight loss 3Kg
Surgeon states possible “Median arcuate ligament syndrome (MALS)”
Laparoscopy ligament divided with difficulty
Symptoms persisted
Surgeon states “probably incomplete division of ligament” – so open operation
Symptoms worse – now requiring home care and morphine-dependent
Surgeon states “probably phrenic nerve involvement” referred to Pain Clinic
Pain doctor – “I think that there is a huge psychological element here”
Look at past medical history
Aged 3 Constipation and severe abdo pain………………………Psychological
Aged 9 Headaches and vomiting………………………………….Stress-related
Aged 11 Severe pains in legs – cant walk ………………................Not organic
Aged 12 Atypical abdo pain – private care…………Normal appendix removed
Aged 13 Severe temper tantrums…………………………Child Guidance Clinic
Aged 14 Severe leg pain after minor injury…. All sorts of tests for non-existent
anterior compartment syndrome
Aged 16 Back pain – private care
Aged 21 Boy friend killed
Subsequently started a small successful business – cheated by partner
Aged 25 Depression with suicidal ideation
Aged 27 onset of abdominal pains
The patient has ‘Briquet’s syndrome somatisation
The danger of doctors going it alone
• “God” syndrome
• Over-confidence bias
• Diagnostic momentum
Note the importance of past medical history –
whose responsibility to maintain the list
Atypical chest pain
Private patient (female aged 57 years) on treatment with
capecitabine for Ca ovary develops odd chest pain.
Seen as an emergency by junior doctor from
Department of Oncology
Transcript from case record (after assessment in A&E)
Review
No further episodes chest pain
History suggestive of GORD
But Troponin I = 0.06 and might be PE
d/w cardiology who suggest admission
Will need repeat troponin I and D-dimer mane
Relatives upset regarding length of stay in A&E…………….
Admit
Have written blood forms for Monday
***********************
Progress of patient
Patient admitted to Oncology ward
Further attack of chest pain at midnight
Then very breathless
Chaotic management by newly appointed junior doctor
and inexperienced nurses
Patient died
At examination post-mortem – pulmonary oedema
– cause uncertain (normal heart)
Cognitive dispositions in action
• Multiple alternatives – none really satisfied
• Sutton’s slip – considered only the obvious
• Yin-yang out – nothing more to do
Never forget the effect of drugs
‘Google’ Capecitabine and chest pain
Clinical case Capecitabine can induce acute coronary syndrome ...
took three doses of capecitabine (7500 mg/m. 2 total dose)
and12 h after the last ingestion of capecitabine before chest pain
developed. ...annonc.oxfordjournals.org/cgi/reprint/13/5/797.pdf –
Similar pages by N Frickhofen - 2002 –
Cited by 56 - Related articles - All 7 versions
An 86 years’ old female with constipation
and urinary frequency
Mrs EC aged 86y
SHO notes at 1230h
PC
HPC
PMH
1. Constipation
2. Urinary problems
Treated for UTI one week ago with Trimethoprim
Has dull ache lower abdo
No burning on PU but frequency – small amounts
Bowels last open 4 days ago
No vomiting, no nausea. Appetite reduced.
Staghorn calculus
Hypertension
Hypothyroidism
NIDDM
AF
(NKDA) No drug allergy
Meds Trimethoprim
Na docusate
Senna
Fe sulphate
Digoxin
Frusemide
Spironolactone
Amlodipine
l-Thyroxine
Lorazepam
Lansoprazole
Glipizide
Paracetamol
Dipyridamole
SHO
examination
Investigations and next steps
Blood tests
Na137 K4.6 U14.0 Cr 224 LFTs normal Alb 43 Glob 36 Amylase 24 CRP 251
Hb 13.2 MCV 93 WCC 20.4 (N 19.1) Plts 344 INR 1.0
X-rays
Chest – ectatic aorta
Abdo – n.a.d. (see next screen)
Reviewed with senior resident:
Soft abdo passing wind
Urine noted (Bld tr Prot+ Leuk – Nitrites -)
Cognitive dispositions to respond (CDRs)
that affected assessment
CDRs that affected surgeon’s assessment
• Value bias – didn’t put worst scenario at top of list
• Satisficing – stopped searching
• Probably person bias – 86-years’ old – what the hell!
And that terrible end-piece – impression – might there be
a better method of ending a clinical assessment
Instead of impression suggest
Analysis:
Diagnostic issues
Plan
Sepsis
Hunt – clinical examination;
scanning; blood culture
Urinary tract infection
Check MSU past and present
Renal failure
Follow progress (hydrate)
Medications
Discontinue digoxin
Cardiac disease
ECG. cardiologist
Diagnostic issues
Plan
Action by
Sepsis
Urgent hunt – clinical
examination; scanning;
blood culture
SHO/registrar
(resident)
Urinary tract infection
Check MSU past and
present
Ward doctor
Renal failure
Follow progress to assess
cause - hydrate
Ward doctor
Nurse
Medications
Check digoxin levels
? Diuretics
Stop ferrous sulphate
SHO/registrar
(resident)
Co-morbidities
CVS: ECG. Discuss cardiologist
NIDDM: keep an eye on blood sugar
SHO to arrange
Nurses to monitor – TPR BP 6-hourly – contact ward doctor if concerned
Registrar to review progress in next half-day
CDRs that appeared to underlie misdiagnoses
(unvalidated data from a pilot study)
• Being too easily satisfied
• Minimising the serious
• Playing the odds
14
11
8
(take the easier option)
• “Silo thinking”
8
(economy, utility, efficiency)
• Failing to get help
8
(knowing when one does not know)
• Making an inadequate investment
7
(pride/duty)
• Allowing diagnostic momentum
• Failing to think ‘outside the box’
• Failing to elicit all the information
6
6
5
Misdiagnoses in hospital care: how might they be reduced?
•
Improve recording and analysing
•
Hunt for explanations – discuss and look (‘Google’)
•
Improve supervision– at ward rounds
for assessment of unsolved or complex cases
of quality of case records
•
Promote teamwork – regular multi-disciplinary meetings
with ‘transformational leadership’
•
Teach: Cognitive errors and how to minimise
•
Promote national learning
– Scottish Surgical mortality Study
– Work of NCEPOD
– Look at the Nordic countries – move away from Tort law
Over to
medical
team
CDRs that affected medical assessment
Diagnostic momentum
Confirmation bias
Day 2
CDRs that affected ‘post-take’ assessment
diagnostic momentum
‘silo-thinking’ – economy, utility, efficiency
no time for ‘complex unpacking’
CDRs that affected assessment by FY1
Unable to think outside the box
Failure to get help
Lack of commitment/supervision by senior staff
Day 3
CDRs on day 3
Framing effect – ARF (acute renal failure)
(if the clinicians had checked blood urea would have seen
improvement – and what does “looks dry” mean!)
Omission bias – clearly thought problem in abdomen (/pelvis) –
therefore ultrasound – actually I found that CT was requested and
refused because of renal failure (radiologist wouldn’t use contrast!)
Investigation bias – easier than examining the patient
Day 4
CDRs on day 4
Order effect – listed symptoms not problems
Unpacking error – failed to unpack the problems that were
shown initially
Base rate neglect and omission bias - ignored significance
of low BP
Sunk cost – invested his pride in spotting that patient still
taking digoxin
Day 4 pm
Presented as a difficult case
But was it?
What did the GP think?
Walk-in clinic referral
Transcript from case record (1)
18.30h 67F – Known ovarian Ca
PC Chest pain, vomiting
HPC This morning onset of a burning retro-sternal pain, similar
to indigestion pains that have troubled her before.
Went to private clinic for a quick check that was normal.
Went home and lay down flat. Awoke with discomfort in chest again
- burning in nature and went away after vomiting once.
Gaviscon helped. No sweating. No radiation to back or arms
Now OK
Transcript from case record (2)
OE: Well, undistressed
T36.4, BP126/66, SaO2 97% R.A.
Head and neck normal
CVS P reg JVPNE HS I +II +0 No murmurs
RS S>N>T. Good BS
Abdo No masses. No organomegaly
ECG Sinus R at 80/min Flat Ts laterally
?1/2 square ST depression V3
Plan: Bloods, CXR nad, Repeat ECG, Review
Transcript from case record (3)
Systems:
Chest No sputum, SOB, haemoptysis
Abdo: No pain, change bowels
Neuro: No headache, collapse, LOC
PMH: Ovarian Ca. Laparotomy and resection last year
2 cycles chemo.
Nil else. No DM/CVA/MI/HTN
DH: NKDA. Chemo: Capecitabine
SH: Non-smoker. <4 units EtOH per week
22.30 Admitted Oncology ward
00.30
Further attack of pain
20 minutes’ delay in getting medical help – chaos on the ward
Became very breathless and died
PM: Pulmonary oedema; no PE; patent coronary arteries
Treated as a Serious Untoward Incident (SUI) but still no diagnosis
Never forget the importance of medications
nor the importance of teamwork
- all sensitive experienced medical clinicians
will remember the times
when they were rescued from error
by nurses, pharmacists and other care workers
Classifying error in surgery
(Fabri and Zayas-Castro 2008)
Defined a template from literature search and faculty input - validated by expert
opinion, paired assessment and national survey
Residents reported complications electronically (weekly reminders) –
specifically requested to classify by
Presence of error
Type of error
Severity of outcome
The data were stripped of personal health information – analyzed over 12
months
Classifying error in surgery
(Fabri and Zayas-Castro 2008)
Results from 9,800 surgical procedures
Major complications in 332 (3.4%) – underlying error in 78%
Individuals doing the wrong thing
20%
Individuals doing right thing incorrectly
60%
Problems with organisation, systems, communication etc
were relatively rare
Misdiagnosis in hospital care: how to reduce?
•
Make clinicians aware of the biases to which we are all exposed in making
decisions
•
Consider how best to analyse clinical data and record conclusions with the
action to be taken
•
Consider the place of using ‘Google’ in diagnosis e.g. to examine the effect of
unusual drugs
•
Re-structure ward rounds so that the difficult case gets detailed attention –
preferably from an experienced clinician who takes day-to-day responsibility
•
Full review of cases that are slow to resolve– consultants to examine quality of
case records (cf Sydney) and possibly re-invent post-mortem assessments
•
Regular unit meetings and Grand Rounds to face issues of Quality and Safety
•
National learning e.g. Scottish Surgical Mortality Study; the work of NCEPOD;
and build on the way clinical complaints are handled in the Nordic countries.
•
Value of teamwork – transformational leadership
A new starting point
Human error,
not communication and systems,
underlies surgical complications
Fabri PJ and Zayas-Castro JL
(School of Graduate Medicine, University of South Florida)
Surgery 2008; 144: 557-65.