Frailsafe collaborative

This is the Frailsafe members' area where you will be able to discuss research and statistics from the programme. If you believe you should have access but don't, please get in touch via the contact page.

Login to the Members' Area

email: Password:

Forgotten password?

Welcome Guest

Stroke 90:10

The below information is taken from the Health Foundation's review of the Stroke 90:10 Collaborative. A link to the original article can be found at the bottom of the page.

What was it?

Stroke 90:10 was a two year improvement collaborative, involving 26 hospitals in the  North West of England which completed in December 2010. Each hospital had a participating multidisciplinary  team led by a Stroke Physician. Data on patients  admitted to participant hospitals with a stroke  diagnosis (ICD 10 codes: 61, 63 and 64) between July  2008 and July 2010 were analysed for this submission.

Why did they do it?

Stroke affects up to 110,000 people per year in the  United Kingdom. Mortality remains unacceptably high  with up to 30% of patients dying within one month.  In the North West stroke outcomes are amongst the  worst in Europe. The National Sentinel Audit of Stroke (England) [] carried out by 100%  of acute trusts biannually has repeatedly shown a gap between the performance of NHS North West and  other regions on key tenants of care.  

The National Sentinel Audit of Stroke collects  data on nine key process indicators. We were  able to determine that over six years the health  economy had improved by only 18% (from 54% to  72% on these key indicators). Despite significant  documented improvements at one hospital (Salford  Royal NHS Foundation Trust) regional spread had not been achieved. Analysis of the regional  challenges included isolated pockets of excellence, an absence of a focussed aim/vision, an absence  of regular data collection and an absence of a community for sharing improvement ideas. 

Study design

Stroke 90:10 is a cluster randomised control trial in which 26 hospitals were randomised either to 
intervention (n=13) or control (n=13). Hospitals in  the intervention group participated from January 
2009 to October 2010 (22 months). Hospitals in the  control joined the intervention teams on month 13 
and participated for 10 months. We used The Breakthrough Series Collaborative  (BTS) model, from the Institute for Healthcare  Improvement. This model structures the change  process around three face to face learning events  (90 days apart) and focuses improvements around  a number of evidence based clinical processes.  This cycle of learning sessions and actions periods  repeated twice. 


We worked on nine processes of care grouped into two bundles. Bundle 1 (early hours care) comprised CT scan, swallow screen, aspirin and weight within 24 hours of admission. Bundle 2 (rehabilitation) comprised physiotherapy assessment, occupational therapy assessment, multidisciplinary goals, mood assessment and 50% stay on a stroke unit.

Non-identifiable patient level data were collected on the nine key process indicators through retrospective case note review. The data were reviewed as ‘all or none measures’ in two bundles i.e. credit was only given if patients received all elements of the bundle. Twenty cases per month per  site were randomly selected from coded discharges from July 2008 until December 2010 (30 months) to include a ‘run in’ period of six months (July 2008 – December 2008) for all hospitals.

Effects of change

For Bundle 1, the intervention sites improved from 20% to 65% (up 45%) and the control sites from 
27% to 62% (up 35%) during the same period (graph 1). For Bundle 2, the intervention sites improved from 29% to 71% (up 42%) and the control sites from 18% to 74% (up 56%).A breakthrough series collaborative model can be used as a framework to deliver improvements in stroke care. Improvement-naive teams can learn quickly in an established collaborative. These findings have significant implications for models of diffusion and spread. 

Stroke 90:10

Implementation Science Article 1

Implementation Science Article 2