Transcript Dr A Willson (NHS Wales)

The NHS in Wales
New challenges needing New methods
Prof Alan Willson, NLIAH
Challenges
• Rising need
• Rising expectation
• Less money
Handicaps – being human
• Discount the future and accept the status
quo
• Greater attention to the harm of action
than inaction
(Predictable surprises, Bazerman and Watkins, HBS, 2004)
• We need an extra dimension
(Flatland, Abbot, Dover Thrift, 1992)
Solutions
• Shift to primary care
• Patient focus and involvement
• Improve our reliability
WASTE
HARM
VARIATION
An example from another
setting
Acute MI Care in US (Mayo clinic)
• Aspirin at discharge
• ACEI for LVSD
• Beta-blocker at arrival
• Beta-blocker at discharge
• Door to lytic
• Door to PCI
• Smoking cessation advice
• Composite and all-or-none scores
• Survival rate/index
• Aspirin at arrival
NHS Wales Examples
• 1000 Lives Campaign
• Reliable care of Central Lines in ITU
(↑ “bundle compliance)
• Improving acute stroke care
Welsh Central Line
Maintenance
• Review necessity of central line every day - and
remove promptly if it is not needed.
• TPN should be given via a separate line or a
dedicated lumen.
• Access to line must be made using an aseptic
technique.
• Entry site to be checked every day for signs of
leakage or inflammation.
Welsh Central Line
Insertion
• Wash hands before and after procedure: soap
and water or alcohol-based agents.
• Use barrier precautions: gown and gloves must
be worn; as much as possible of the patient
should be covered with sterile drapes.
• Sterilise skin with chlorhexidine in alcohol and
wait until the skin is dry.
• Avoid the femoral site unless it is the last resort.
National Aggregate CVC
Bundle Compliance
1
80.00%
1
Jun-06
Aug-06
Oct-06
Dec-06
Feb-07
Apr-07
Jun-07
Aug-07
Oct-07
Dec-07
Feb-07
I Chart of CVC bundle
6
90.00%
3
60.00%
6
3
1
6
5
1
6
5
30.00%
1
Jun-06
1
Aug-06
Oct-06
Dec-06
Feb-07
Apr-07
Jun-07
I Chart of CVCI bundle
Aug-07
Oct-07
Dec-07
Feb-07
_
X=97.11%
Helics defined CVC Infection rate by month of line insertion,
Wales: 09/2007 - 12/2009*
2.5
Source : W HAIP
2.0
1.5
1.0
0.5
Sep-07
Oct-07
Nov-07
Dec-07
Jan-08
Feb-08
Mar-08
Apr-08
May-08
Jun-08
Jul-08
Aug-08
Sep-08
Oct-08
Nov-08
Dec-08
Jan-09
Feb-09
Mar-09
Apr-09
May-09
Jun-09
Jul-09
Aug-09
Sep-09
Oct-09
Nov-09
Dec-09
0.0
Infection Rate (per 1000 catheter days)***
Mean Average
UCL
LCL
Acute stroke care
Patient
arrives in
A&E
Patient
departs
A&E
Patient
arrives in
ACU
Patient has
scan
Patient
transfer to
stroke unit
Patient
discharge
from
stroke unit
What if we change
the process?
Patient
arrives in
A&E
Patient
departs
A&E
Patient
arrives in
ACU
Patient has
scan
Patient
transfer to
stroke unit
Patient
discharge
from
stroke unit
Modelling
• Provide a template to match resource with
process change
• What if
• Increase the “importance of the future”
Medicines
• what is the system, how should we
improve it?
Non-Adherence
Patient
Harm from
medicines
Health
Professionals
Molecule
Pharmacoepidemiology
Errors
© N Barber www.pharmacy.ac.uk
Errors and Harm
• Harm much rarer than errors, however the
enormous volume of prescribing – 2.5m Rx per day
in the UK - means that even small percentages
cause harm to a lot of people
• ~1 in 20 hospital admissions (=cancer admission
rate) result from avoidable adverse effects of drug
therapy – ie from errors/ nonadherence/ avoidable
ADRs
We are scraping burnt
toast
How do we improve?
• Polish the cog
• Redesign the system
Medication errors in
primary care
BMC Medicine 2009 7:50 Garfield et al
© Garfield, Barber, Walley, Willson
100
Intention
80
Prescribing
Adherence
Quality (%)
40
20
Dispensing
60
Presentation
Effect
Intention
Prescribing
Presentation
Time
Dispensing
Adherence
Effect
Intention
Prescribing
Presentation
Dispensing
Adherence
Effect
Error rate 7.46% items
(Shah et al 2001):
•No directions 25%
•Prescribing something not needed 18%
•Directions incomplete 11%
•Over supply 11%
•Strength missing 9%
•Quantity missing 8%
•No Signature 5%
•Other 13%
100
Intention
Adherence
40
Quality (%)
Dispensing
20
Presentation
60
80
Prescribing
Effect
Intention
Prescribing
Presentation
Time
Dispensing
Adherence
Effect
Intention
Prescribing
Presentation
Dispensing
Adherence
Effect
Error rate 2.9% prescriptions
(Jones & Britten 1998)
Error Rate 5.2% items
(Beardon et al 1993)
100
Intention
80
Prescribing
Adherence
Quality (%)
40
20
Dispensing
60
Presentation
Effect
Intention
Prescribing
Presentation
Time
Dispensing
Adherence
Effect
Intention
Prescribing
Presentation
Dispensing
Adherence
Effect
Error rate 3.3% items
(Dean Franklin et al 2007):
•1.6% labelling
•1.7% content
Clinical Severity:
•67%: minor
•32% moderate
•1% severe
100
Intention
80
Prescribing
Adherence
Quality (%)
40
20
Dispensing
60
Presentation
Effect
Intention
Prescribing
Presentation
Time
Dispensing
Adherence
Effect
Intention
Prescribing
Presentation
Dispensing
Adherence
Effect
Error Rate 30-50% patients
(Cochrane 2008, Nice 2009)
100
Intention
80
Prescribing
Adherence
Quality (%)
40
20
Dispensing
60
Presentation
Effect
Intention
Prescribing
Presentation
Time
Dispensing
Adherence
Effect
Intention
Prescribing
Presentation
Dispensing
Adherence
Effect
Medication ineffective = 50%-90%
(NNT medication 2-10)
Drug related admissions = 6.5%7.5% admissions
(Pirohamed 2004, Howard 2003,
Green 2000)
•69% of these are preventable
100
Intention
80
Prescribing
Adherence
Quality (%)
40
20
Dispensing
60
Presentation
Effect
Intention
Prescribing
Presentation
Time
Dispensing
Adherence
Effect
A Tale of Two Tanks
German Panzer III
Russian T34
Thanks to the Tank Museum, Bovingdon
Where should we improve
the system?
• Some suggestions:
– Start with the largest error
– Start closest to the patient
– Focus on harm
• We decided on two targets:
– To reduce nonadherence
• causes harm through too little drug
– To reduce the number of hospital admissions
caused by avoidable harm from medicines
• causes harm through poor management
Can we improve
adherence?
 Proof of concept RCT of 500 patients in primary care
 Developed an intervention that was based on theory,
pragmatic (delivered by phone) and delivered from
the patient’s perspective
 Pharmacist phones at 2/52 “How are you getting on
with your medicines?”
 A chance to talk about adherence, wants, ADRs, errors
 Main evaluation by phone at 4/52 + q’iare at 6/12

Clifford et al Pharmacy World & Science, 2006; 28:165-170


Clifford et al J. Psychosom Res, 2008; 64: 41-6
Elliott et al Pharmacy World & Science, 2008; 30:17–23
Results
• Mean duration of phone call 12 minutes
• Non adherence fell 16% to 9% (p=0.032)
• Reported problems fell 34% to 23%
(p=0.021)
• Beliefs more positive (p=0.02)
• Modelling of cost effectiveness (by
bootstrapping) shows .....
Cost-effectiveness
of the intervention
Elliott et al PWS 2008
Developing a measure for
non-adherence
• No ‘gold standard’ method exists
• We need a measure that:
–
–
–
–
–
Can be scale or dichotomous measure
Includes diagnosis of cause of nonadherence
Can be self administered or by pharmacist
Is quick and easy to use
Can be used for any medicine or combination of
medicines
– Is based on theory
– Is sufficiently valid and reliable
Reducing avoidable
admissions to hospital
• There is potential.....
• Reliable prescribing and management
• We can estimate diuretic usage figures in
the community
• Count admissions
Measuring avoidable
admissions to hospital
• We would like to coin a new phrase - NNA (number
needed to admit). How many patients do we need on a
group of drugs before we should expect an
admission?
• 1 in 12 of all people in Wales is taking a diuretic using primary care prescribing data and DDDs. That's
250,000. The Howard and Pirmohammed studies
would give us about 4500 admissions a year from
diuretics for our population. If each patient is taking it
for 5 years, each has about a 1 in 10 chance of being
hospitalised as a result of their diuretic care (sic).
• The 5 year NNA for diuretics is10.
• 5 year NNTs for thiazides are around 50-100 for stroke
and MI.
• Willson et al The Pharmaceutical Journal 2009 283 651
New challenges need new
methods
•
•
•
•
System focus
Reliability
Map and shift resource
Moving from predictable surprises