Insert name of presentation on Master Slide Developing the measurement of harm 25 th November 2010
Global Trigger Tool Standardised measurement of harm in acute hospitals. Wales is the only country to use it as a national measure. Purpose –Track harm –Characterise harm
Developments Research Primary Care Further developments Support
Measuring harm and informing quality improvement longitudinally in NHS Wales Four year programme funded by the National Institute for Health Research Led by Cardiff University in collaboration with Imperial College Aims to build on the progress NHS Wales has made in the global adoption of the Global Trigger Tool Phase 1: (Years 1 and 2) will assess the validity of the GTT for widespread use across the NHS Phase 2 aims to concentrate NHS GTT resources and the research teams resources around a single tool
Automated Universal Trigger Tool Graham Copeland Consultant Surgeon
UK global Trigger Tool Adult care –32 variables –5 modules General care Surgical care Intensive care Medication care Laboratory tests
Usual technique Review random case notes –50 initially –Then 20 per month Identify trigger events Assess causation
Problems Very labour intensive Only a tiny sample of the whole –Credibility with clinicians Random selection –Can be few triggers –Difficult to show improvements over time Not all triggers result in harm
Avoiding these problems Assess ALL patients admitted –Increase credibility with clinicians –Allows comparison over time Prioritise selection of cases for review –Maximises potential improvements –Focuses review on harm
Automated data sources Hospital episode statistic (HES) –Demographic data –Admission Source –Diagnostic codes –Procedure codes –ITU data –Outcome data Mortality Re-admission
Automated data sources Pathology system Radiology system Bed management system Theatre management system
Universal Trigger Tool HES Surrogates Manual comparison of 10,000 random cases over a 2 year period –Coding vs manual review Manual comparison of all mortalities over a 3 year period (120 cases/month) –Coding vs manual review Correlation between trigger and best fit to code combinations 95% accuracy of trigger detection for the more complex trigger events Only harm events produce automated trigger
Universal Trigger Tool HES Surrogates Some direct mapping codes Some triggers require multiple codes Complications – 147 codes Some triggers require complex code combination –Failure to respond to deterioration in EWS- combination of 3-5 codes Some have no code but are still part of the HES dataset
Universal Trigger Tool HES Surrogates Some are easy (Direct single code maps) –HyponatraemiaE87.1 –HypernatraemiaE87.0 –HypokalaemiaE87.6 –HyperkalaemiaE87.5 –Wound infection T81.4
Universal Trigger Tool HES Surrogates Some require multiple single codes –Complications 147 –Hypotension and cardiac arrest 8 –Sudden cessation of treatment 7 Some require combinations –Nosocomial pneumonia 5 Code sets + Y95X
Universal Trigger Tool HES Surrogates Complex combinations –Elective admission + procedure code + myocardial infarction code = Post-op MI –Admission source + episode identifiers and critical care flow dates gives HDU/ITU information
Universal Trigger Tool HES Surrogates Potential problems with HES –iLAB suggested 40% inaccuracy –Takes first man standing However accuracy in surgery 95% Automated system takes –‘Last man standing approach’ –Radically decreases inaccuracy
PS If you think I’m intending to reveal all the CRAB code THINK AGAIN !!! THINK AGAIN !!!
Universal Trigger Tool HES Surrogates For single relationship codes 99% For multiple codes relationship 99% For more complex relationships 95% Some areas may potentially over-assess –Orthopaedic prosthetic complications –Orthopaedic prosthetic infections –Pressure ulcers/MRSA/C difficleCausation Some areas may potentially under-assess –Falls Harm
SOFTWARE CRAB trigger tool module CRAB Clinical Informatics –c-ci.definingdesign.co.uk –c-ci.definingdesign.co.uk Internalised system –Real time monitoring once coded –Produces automated exception report for non-coded –28 day delay to allow readmission –Automated reports
Triggers Triggers recorded –10% of admissions –Water muddied by day cases Large volume in surgery, endoscopy Very few trigger events (<0.2%) Unless converted to overnight or readmitted Excluding day cases –Trigger events in 31%
Triggers Consultant specific data –Patients with triggers31.2% –Patients with 3 or more3.7% –Average number of triggers per admission0.5% –Range Excluding the 1 patient 100%’s 4.4 to 51.1% 2 lowest orthopaedic consultants
Triggers Commonest –General medicine37% Elderly care26-52% Respiratory25-42% Gastrointestinal30-44% Other high groups –Obstetrics/gynaecology
Module types General care 71.3% Lab test20.3% Surgical care6.9% Intensive care0.3% Medication 1.3%
General care module % of total in group Readmission54% Complication20% EWS/Cardiac arrest/shock14.2% Transfer to higher level4% Decubiti3.5% DVT/PE2.3% Fall 1.4%
Lab test module (% of total in group) Haematological25.7% Urea/electrolyte34.6% Hypoglycaemia6.4% Troponin16.1% C.difficile/MRSA2.2% Wound infection4.2% Nosocomial pneumonia3.6% Blood culture7.1%
Variation of numbers of patients with trigger events > Trigger events
Variation of numbers of patients with trigger events (Mortality) > Triggers events
Effect of mortality on trigger events 85% patients who die have at least 1 trigger event 34% have 3 or more 5% have 6 or more 5 patients had 10!
Trends (myths and facts) No sessional variation –General care module –Lab test module Medication –Highest Jan to May –Lowest in the summer
Trends (myths and facts) Intensive care –Small numbers –Large variations –Related to occupancy levels Surgical care –Peak July/ August –Lowest Winter
Change technique Automated e mail to all consultants –Monthly trigger report –Performance in speciality (%) Automated management report –Performance overall and in all modules –Variation over time –Highlights for Cquin LIPS
Change technique Mortality review group –Review all deaths >4 triggers Divisional governance groups –Review all patients >3 triggers –Review variations in individuals –Automated data sheets Overseen by Benchmarking Group Report to Clinical Governance Committee and Board
CONCLUSION Automated Trigger tool analysis possible Accurate and reproducible Can allow focused review Can monitor changes over time Perhaps the time has come to revisit the methodology Could increase clinician engagement
Developments Research Primary Care Further developments Support Automate?
Any Questions?