Transcript Document
When processes go wrong; designing for 95% reliability Kevin Stewart Medical Director QIPP Safe Care workstream About me • Since September 2010 – Medical Director, QIPP Safe Care Program • 2009-2010 – Quality Improvement Fellow, Institute for Healthcare Improvement, Cambridge, MA, USA* • 2005-2009 – Medical Director, Winchester & Eastleigh NHS Trust *funded by the Health Foundation Overview • Feedback from incident review • Revision of concepts • Human Factors • Thinking about level 3 • Homework Principles • There are no “quick fixes” – This stuff is difficult • The blame culture is deeply ingrained in; – Our professions – Healthcare – Wider society • There is a difficult balance between accountability and a “just” culture Homework time! Incident investigations, RCAs • What were the common recommendations? • Are these level 1, 2 or 3? • How confident are we that these will prevent recurrence? My list • “Remind the nurses to always check the call bell” • “Emphasise the importance of good handover to on-call teams” • “re-audit use of hand gel” • “Ensure falls training is available to all staff” • ..etc… Reliability definitions Reliability levels Chaotic Level 1 Number of failures More than 2 in 10 1 failure in 10 Percentage success Les than 80% Level 2 1 failure in 100 90-99% Level 3 1 failure in 1000 90% 99-99.9% How reliable is healthcare? • Most healthcare processes are, at best, implemented in 80-90% of indicated instances • …probably reflecting reliance on Level 1 measures Are there any obvious exceptions? What about… • Blood transfusion? • Paediatric anaesthesia? • Chemotherapy administration? • Cardiac arrest calls? Designing for reliability • Level 1 – Intent, vigilance, hard work, audit, discipline • Level 2 – Design of processes informed by reliability science and knowledge of human factors • Level 3 – System-wide focus on becoming a highly reliable organisation Designing for reliability • Level 1 – Is mostly about individual action • Level 2 – Is about working on the process • Level 3 – Is about working on the whole system Human Factors • “The study of the way humans relate and react to the world around them with the aim of improving performance” – Combines psychology, physiology, behavioural science, engineering etc • Derived from design work to improves safety of WW2 aircraft Human Factors • Used widely in “safety-critical” industries – Airlines, F1, air traffic control, nuclear, defence etc • Also used widely in advertising & marketing • In healthcare has tended to be used in “high tech” areas (ITU etc) PS • We use HF all the time in real life, e.g – Child care – Caring for patients with dementia – Our own daily routines – etc Basic HF principles • Human performance follows predictable patterns • Individual performance capacity is limited • Individual performance can be severely impaired by external factors • Systems should be designed around knowledge of these principles Have you ever….? • • • • • • Put the wrong fuel in the car? Sent an e mail to the wrong person by mistake? Deleted the wrong document? Locked yourself/your keys in/out? Forgotten to lock the door? Given a patient the wrong medicine? • And if so, when?? Factors affecting human attention • • • • Fatigue Stress, anxiety, fear Competing demands Environmental conditions – Clutter, motion, poor lighting, noise • Too many handoffs • Shift work Designing to overcome Human Factors • Avoid reliance on memory, vigilance and hard work • Simplify and standardise • Use cues • Make the desired action the default • Take advantage of habits and patterns • Use constraining and forcing functions • Build in redundancy Using what we know about memory Visual and auditory cues Standardise Constraining and forcing functions Redundancy Example • Patients diagnoses with sepsis in A&E had delays up to 7 hours before the first dose antibiotics. • ..resulting in; – Increased ITU admission – Increased mortality – Prolonged length of stay Example • Audit showed; – Delays in getting blood cultures – Delays in cannulation – Delays in getting the right antibiotics prescribed – Delays in administration of antibiotics – Confusion between A&E and MAU staff about antibiotics What should we do? The Deterioration Recognition (DR) Box With thanks to Dr Matt Inada-Kim Winchester With thanks to Dr Matt Inada-Kim Winchester Human failure taxonomy Human failures Intended actions Violation - Intended consequences When the person decided to act without complying with a known rule or procedure Unintended actions Errors - Unintended consequences Mistakes When the person does what they meant to, but should have done something else Adapted from HSE, after J Reason Lapses When the person forgets to do something Slips When the person does something, but not what they meant to do J Reason 1990 Why do we fail? • Current systems in healthcare are highly dependent on level 1 measures; intent, vigilance, hard work • We focus on outcomes, so only measure the process where there is catastrophic failure • We miss process defects where the patient does well despite the system (rather than because of it) • We don’t really analyse failures and learn from them systematically Level 3 reliability • Usually only achieved by system-wide approaches • Organisations in some industries (air, nuclear etc) have achieved this • Difficult to identify healthcare examples in the UK – ? Virginia Mason, Mayo Clinic in the US Level 3 and beyond • To improve reliability beyond level 2 usually requires fundament system redesign • Based on Failure Modes and Effects analysis which analyses process failures in detail and changes the systems accordingly IHI 3 step design for reliability 1. Prevent initial failure -Use intent and standardisation 2. Identify failures and mitigate -Use redundancy function 3. Critical failure analysis -Identify critical failures then redesign system IHI design for reliability Prevent initial failure Identify failures and mitigate Measurement Redesign Critical failure analysis For next week… Review your incident report or RCA again. Think how you might re-write it to include at least one Human Factors Level 2 action. Summary • At best, most healthcare processes deliver level 1 reliability (i.e. around 90% success) • Hard work, vigilence, training & audit is unlikely to make this any better • Level 2 (human factors) changes, can deliver up to 90-99% success • To get more reliable than this requires organisation-wide change Acknowledgements • Frank Federico & Carol Haraden, IHI • IHI white paper “Improving the reliability of healthcare” at www .ihi.org • Nolan T “System change to improve patient safety” BMJ 2000;320. 771-3