Unwarranted clinical variation in ischaemic stroke.

Actions from the NSW Stroke Network.

NSW Health Symposium Breakout. Sydney 2014.

SESSION 1: Responding to Unwarranted Clinical Variation: A Case Study

Conjoint Associate Professor John Worthington Clinical Co-Chair, ACI Stroke Network

Translational Cancer Research Unit

John M Worthington,1,2,3, 4

South Western Sydney Clinical School, Faculty of Medicine, UNSW

1Ingham Institute for Applied Medical Research, Liverpool, NSW, Australia.2Associate Professor (Conjoint) South Western Sydney Clinical School, The University of New

South Wales. 3 Senior Staff Specialist Department of Neurology, Liverpool Health Service, Sydney, Australia.4 Medical Co-chair Stroke Services New South Wales, Agency of Clinical Innovation.

Unwarranted clinical variation in ischaemic stroke.

Actions from the NSW Stroke Network. NSW Health Symposium Breakout. Sydney 2014.

Evidence based practice in ischaemic stroke

There is substantial evidence around what constitutes good ischaemic stroke care.

The major elements of good stroke care include:

• Stroke units. With co-localised stroke beds served by a multidisciplinary stroke team that uses evidenced-based pathways improve stroke outcomes by approximately 30%, at all ages. What reduces death reduces disability.1

• Clot-busting. IV rt-PA within three hours, reduces death and disability by 44% (Cochrane), with more modest benefits at 3-4.5 hours (favourable Odds Ratio 1.34). There is an all-hours cost-of-readiness and no DRG.

1Gattellari et al Stroke 2009; 40: 10-7. 2 Wardlaw et al, Cochrane Database of Systematic Reviews. 2003 (3).

Ischaemic stroke care in NSW

• 20 Acute Thrombolysis Centres (ATCs) are now nested in 30 acute stroke units across NSW. Nine other hospitals have stroke services.

• The 30 and 365 day ischaemic stroke mortality in NSW is 17 and 27%, respectively (Gattellari et al, Cerebrovascular Diseases, 2011).

• 35% of stroke patients are discharged to in-patient public rehabilitation, 5% are discharged directly to nursing home and 20-25% return directly to independent living.

• NSW outcomes for stroke compare favourably with OECD countries (BHI).

• However, many patients do not reach a stroke unit hospital or stroke unit bed (NSF) and there is unwarranted clinical variation between hospitals.

In 2012 BHI published de-indentified hospital level data on 30 day stroke mortality in NSW (Health Care in Focus). This analysis, by hospital of death, started a collaborative process to examine

unwarranted clinical variation and the 2012 analysis itself.

Unwarranted clinical variation in Stroke. Process 2013.

Data

• Identification of clinical variation in ischaemic stroke.• Health Care in Focus (BHI). Published December 2012.• Ongoing process to refine ascertainment, analysis, reporting.• Regular Work Group meetings of ACI and BHI members.

LHD Feedback

• ACI letters to LHD CEs and clinician leaders. 7th March, 2013.• Clinical Variation Workshop. ACI, BHI, UCV Taskforce. 3rd April.• Workshop feedback on required reporting and QI processes.

Audit and analysis

• Expert Reference Group meetings to discuss site audit tools and processes. BHI, ACI and SSNSW and the Florey and Ingham Institutes. First meeting 22nd April.

• Finalising audit tool and access to other site data. 28th June.

QI

• Implementation Team meets to plan 6 pilot site visits May 10.• Invitation letter to LHDs for participation in audit, feedback and QI process.• Site audits begin 7th July and site feedback visits start 7th August.• Audit data and further consultation refines the later 2013 BHI data analyses.

5

Six rural and metropolitan pilot site visits.

Site and date Type NCrude30 day

Mortality%

Standardised(Adjusted)Mortality %

CI 95%

Hospital 114th August

Principal referral ATC/SU

353 17.1 20.7 15.3-27.2

Hospital 27th August

Principal referral ATC/SU

289 8.0 8.2 5.0-12.6

Hospital 314th August

Non-principal, Metro. SU

138 11.6 9.2 5.2-15.1

Hospital 415th August

Rural. SS/No SU

197 20.8 19.1 13.6-26.15

Hospital 529th August

RuralNo SU

83 22.9 30.6 7.6-63.0

Hospital 6 30th August

Rural ATC/SU

213 8.9 9.6 5.6-15.3

6

The NSW Stroke Network accepts that stroke care varies and there is a clinical variation in stroke outcomes.

The ACI team selected 6 rural and metropolitan sites

with above or below average mortality on the 2012 BHI

analysis and different service characteristics

Example: Hospital 6 Audit and Feedback

• Rural site. Favourable 30 day mortality.

• Stroke Unit and Acute Thrombolysis Centre.

• All ischaemic strokes were admitted to the stroke unit!

• 75% were on a clinical pathway during the admission.

• 65% had a CT within 2 hours and 100% in 24 hours.

• Stroke investigation rates shown in the figure.

• 100% received neurological observations in the first 24 hours.

• 72% received aspirin in the first 24 hours.

• Documented swallow assessment within 4 hours of 40% (45% in speech impaired patients).

7

No hospital unit performed consistently well across all clinical care processes that are likely to

influence patient outcomes. Where outcomes appeared worse the gaps

in evidence-based care were generally greater

There was local surprise at rates of pathway use and swallow assessment with an immediate QI response

ACI Stroke Audit Tool

Extracted details from the three metropolitan site audits.

Site and date TypeAdjusted

mortality % Selection of audit characteristics

Hospital 114th August

Principal referral ATC

N=353

20.7 July-Aug 20113 transfers in. Nil reported palliative. Rapid CT brain; rate 100%. 100% reached stroke unit or HDU. 100% Neuro obs in 1st 24 hours. Low rate of cardiac ultrasound 30%. No use of a clinical pathway. Only 78% on antithrombotics at discharge. 44% on aspirin in 24 hours. Documentation of swallowing at 4 hours 25%. Low statins.

Hospital 27th August

Principal referral ATC

N=289

8.2 Aug 2011-Nov 2011.1 transfer in. 1 documented for palliative care and 2 t/f to a Pal care facility. Rapid CT brain; rate 100%. 100% reached stroke unit or HDU. 95% Neuro obs. Cardiac ultrasound TOE + TTE 76%. Clinical pathway 45%. 84% on antithrombotic at discharge. 58% on aspirin in 24 hours. Swallowing documentation at 4 hrs 70%.

Hospital 314th August

Non-principal Metro. SU

N=138.

9.2 July 2011-Jan 2012. Note: major service changes. No transfers in. Two documented as palliative care. 63% reached the stroke unit. TOE + TTE 97%. 63% Neuro obs. 85% on a clinical pathway. 93% on antithrombotics at discharge. 60% on aspirin at 24 hours. Swallowing documentation < 4 hrs 20%.

8

The face-to-face feedback to managers and clinicians was almost universally well met and has impacted on care

A comparison of some processes that may influence or reflect outcomes.

Hospital Adjusted

30 day

Mortality (%)

SU/HDU

Bed

(%)

24 hr

Neuro

Ob's (%)

Clinical

P’way (%)

Swallow test< 4 hrs

(%)

%Discharged

on A’thrombotics

Aspirin at 24 hours

(%)

Pall’

Care (N)

% D/C on

Statin

1 20.7 100 100 0 25 78 44 0 28

2 8.2 100 95 45 70 84 58 3 63

3 9.2 63 63 85 20 93 60 2 60

4 19.1 0 55 80 10 71 47 0 43

5 30.6 0 9 0 0 80 20 3 20

6 9.6 100 100 75 40 100 72 0 67

No hospital performed consistently well across all the clinical care processes likely to influence patient outcomes. Where outcomes appear worse the gaps in evidence-based care are generally greater.

A catalyst for further work: The BHI publication of 30 day ischaemic stroke mortality with identification of hospitals in 2013.*

*The Insights Series: 30-day mortality following hospitalisation, five clinical conditions, NSW, July 2009 – 2012

After the pilot audits the BHI analysis was modified

to measure outcomes according to hospital of first

presentation.

There is a strong argument to benchmark against the better

hospitals rather than an arithmetic mean

Conclusions: Where do we go next?

A process is needed and underway to reduce unwarranted variation in stroke care.• Expanding access to stroke unit care with new units and consider further bypass of sites.• Improving access to stroke unit beds in stroke unit hospitals.• Improving existing organised stroke care through the ACI Statewide Stroke Clinical Variation Strategy

(SSCVS) which offers expansion of the piloted process of audit and feed-back to 30-40 sites.

Evaluate and improve Stroke Thrombolysis and the Early Stroke Reperfusion Programme.• Reinforce local Thrombolysis Committee governance, deploy a qualitative audit tool (SRAT) to assess

barriers to rapid door-to-needle times and collect an already agreed minimum dataset needed for QI. • With the ABF Taskforce develop a DRG for thrombolysis and re-cost existing stroke care DRGs.• Improve access to stroke thrombolysis through deployment of Tele-Stroke, with HealthShare.

Improve data quality and analysis and provide site reporting to management, clinicians and consumers.• Use audit and feedback and consultation to drive improved data collection, analysis and reporting.

11

Unwarranted clinical variation in stroke is explicable variation. At present stroke patients do not always receive evidenced-based care. This may be the result of being admitted to a smaller hospital with no organised stroke care and little prospect of providing it, admission to a hospital where stroke unit care could reasonably be provided but

no unit has been established, because patients fail to reach stoke unit beds in a hospital with a stroke unit or because of a variations in the quality of care in existing stroke units.

Thank you

12

Special thanks to Mark Longworth (ACI), Melina Gattellari (Ingham Institute), Kim Sutherland and Doug Lincoln (BHI), Dominique Cadhillac (Florey Institute), site auditors and the clinicians and managers of the 6

pilot sites and the ACI Unwarranted Clinical Variation Taskforce.