Presentation on theme: "University of California, San Diego"— Presentation transcript:
1University of California, San Diego Overview of Quality Improvement Focus on Designing Reliable InterventionsGreg Maynard MD, MSProfessor of Clinical Medicine and Chief, Division of Hospital MedicineUniversity of California, San Diego
2Quality Improvement: Bridging the Implementation Gap Scientific understandingProgressTimeImplementation GapPatient careSo we start out by noticing an area of focus. This focus for quality improvement projects should develops from recognition of a gap between the level of care that is optimal and best supported by the evidence contrasted with the care that is actually being delivered to our patients.We are all here today because we know that management of hyperglycemia in the hospital is one of these areas. What is actually being done, is far from what is supported in the literature.How do we ultimately bridge that gap? It can be accomplished by using qi principles and following steps that have been determined to be essential by those who have gone before you.
4The Evolving Culture of Medicine 20th Century CharacteristicsAutonomySolo PracticeContinuous learningInfallibilityIndividual Knowledge21st Century CharacteristicsTeamwork & systemsGroup practiceContinuous improvementMultidisciplinary problem solvingChangeShine, KI. Acad.Med. 2002;77:91-99
5How Do We Close the Gap? Essential Elements Institutional support and multidisciplinary teamsStandardized order setsInfusionSubcutaneous which promote basal / bolus regimensAlgorithms / protocols / policiesAddress dosingNutritional intakeSpecial situations: TPN, enteral tube feedings, perioperative insulin, steroidsSafety issuesTransitions in care and discharge planningMetrics: How will you know you’ve made a difference?Comprehensive educational program
6Traditional Quality Assurance Focus on the statistical outliers in the ‘tail’Faulty assumption: Care not in the ‘tail’ must somehow be acceptable and needs no improvementFocus on intervention for a few individualsLittle focus on process improvementNo potential for breakthrough improvement even if complete success in getting rid of outliersEvent related - reactiveoutliers
7Quality Improvement Before After Quality Quality better better worse Focus on processes of care, not just outliersImprovement in processes reduces variation and shifts entire curve toward better carePotential for radical change through changing the design of careReally bad apples now isolated for picking!Goal related - proactivebetterQualityworsebetterQualityworse
8Quality Improvement is… Focus on processes of care Reduced variation by shifting entire practiceA change in the design of careQuality Improvement is NOT…Forcing people to work harder / faster / saferTraditional QA or peer reviewCreating order sets or protocols without monitoring use or effect…not just the outliers…shifting entire practice toward better care by process change…potential for radical change through changing the design of care…Yelling at people to work harder / faster / safer
10Features of a Good Team Safe Inclusive Open Consensus seeking no ad hominem attacksInclusiveopen to all potential contributorsvalues diverse views; not a cliqueOpenconsiders all ideas fairlyConsensus seekingfinds a solution all members can support
11Models for Improvement In use around the globe for decadesSuccess in many fields of endeavorHealthcare late to the game!Alternative to the usual:Predictable breakdowns in reliability leading to common problemsIgnoring improvement concepts & trying the first thing that comes to mindNot measuring effectiveness of implementation outcomes or process until bad events happen…..again
12A Model for Improvement Setting Aims Improvement requires setting aims. The aim should be time-specific and measurable, with a defined population.Establishing Measures Teams use quantitative measures to determine if a specific change actually leads to an improvement.Selecting Changes All improvement requires making changes, but not all changes result in improvement. Organizations therefore must identify the changes that are most likely to result in improvement.Testing Changes The Plan-Do-Study-Act (PDSA) cycle is shorthand for testing a change in the real work setting — by planning it, trying it, observing the results, and acting on what is learned. This is the scientific method used for action-oriented learning.
13Features of Good Aim Statements SpecificMeasurableAggressive yet AchievableRelevantTime-bound
14Sample Aim Statements: Glycemic Control on the WardsWithin 6 months the use of sliding scale only regimens will be reduced by half.Within 12 months the % of patients with POC glucose testing achieving a mean glucose of < 200 mg/dL will improve from 65% to 85%.Within 12 months the % of our patients suffering from hypoglycemic events will be reduced from 11% to 6%.
16Measurement Principles Seek usefulness, not perfectionIntegrate measurement into daily routineUse qualitative and quantitative dataUse samplingPlot data over timeUse a balanced set of measures for all improvement efforts
17A Blend of Measures Structure Process Outcomes Do you have a multidisciplinary steering committee?Do your SQIO sets include a prompt for A1c?Process% of SQIO written using your order form% with basal insulinOutcomesLOS, Mortality: Glycemic control, Hypoglycemia
18Picabo Street and Communication Olympic Gold Medal Winner….AND a Critical Care Nurse!
20Hierarchy of Reliability PredictedSuccess rateLevel1No protocol* (“State of Nature”)Decision support exists but not linked to order writing, or prompts within orders but no decision supportProtocol well-integrated (into orders at point-of-care)Protocol enhanced (by other QI and high reliability strategies)Oversights identified and addressed in real time40%50%2365-85%490%595+%
22High Reliability Design Solutions (as applied to Insulin Protocol) Standardize insulin choices for common situationsMD must “opt out” of default choices (not opt in)Prompts for basal insulin if over glycemic target, prompts for HgA1c, etc.Scheduled assessments of glycemic control / insulin regimenRedundant responsibility to maintain glycemic targetSTANDARDIZE:Choosing one preferred option for these situations is advantageous because:You can communicate preferred regimens more simply and succinctly to all staff.You eliminate all inappropriate choices for insulin regimens for that situation, as well as some other less preferred, but acceptable choices.You can encourage regimens that are most economical (by promoting the insulin regimens that reflect your hospital formulary choices).Staff members can become very familiar with a few regimens, instead of being confused by a multitude of them. They can identify variations from your preferred choices and target these patients for extra scrutiny and actions should they fail to meet glycemic targets.It lends itself to building protocols more than the inherent variability in usual practice.
23CAUTION!!!! Be Sure to Insert a Brain Between Protocol and Patient! Education for broad range of providersConsider special team of focused providers
24Engineering Change: Hints for Success Empower nursingExpedite passage through medical staff committeesBetter to implement an imperfect, compromise change than no change at allProvide hot line or support for difficult situationsFollow metrics continuously as you implement
25Engineering Change: Hints for Success Measure, learn, and over time eliminate variation arising from professionals; retain variation arising from patientsKeep big picture in mindNegotiate ‘speed bumps’Time delays in getting dataIncomplete buy-inGo around obstacles instead of through them (can always go back to them later)Some who disagree with you may be correctMake changes painless as possible: make it easy to do the right thing
26PDSA: Plan-Do-Study-Act The use of PDSA has been referred to as the “democratization of the scientific method.” (Paul Miles, MD)Do small scale tests of change.Everyone can do it!ActPlanStudyDoPlan itDo itCheck itAct on the results found - quickly
27Benefits of rapid cycle change: Increases belief that change will result in improvementAllows opportunities for “failures” without impacting performanceProvides documentation of improvementAdapts to meet changing environmentEvaluates costs and side-effects of the changeMinimizes resistance upon implementation
28Examples: integration of best practice A1c level within last 30 days.Specify hyperglycemic diagnosisEach patient should have a glycemic target.
29A1c LevelIncorporate prompt for A1c level in insulin order sets and protocols.Ordering can be accomplished with checkboxMonitor performance, feedback to providersGlycemic control team obtains it
30Proper diagnosis Diagnosis: Uncontrolled –or– Controlled Diabetes type: 1 2 Gestational –or–Secondary to another cause;Specify–or– Stress/situational hyperglycemiaImproves reimbursement: define “uncontrolled DM” and monitor coding accuracyOrder set docmentation translates into ICD-9
31Identify non-critical care glycemic target Preprandial target 90–130 mg/dL; maximum random glucose < 180 mg/dL (ADA/AACE consensus target)80–150 mg/dLPreprandial target 90–130 mg/dL for most patients, 90–150 mg/dL if hypoglycemia risk factors
32“Actionable” Glycemic Target The “what” is common to all institutions: push for changes in regimens when glycemic target not being met.Variable by institution:Glycemic target definitionHow to generate reportWho acts on reportPutting this in place moves you up hierarchy of reliability.Opportunity to Learn from variation!What happens if your institution’s glycemic targets are not being met? Glucose values consistently out of your target range without attempts by physicians to tighten control warrants institutional action. Consider creating parameters for calling the physician if glycemic targets aren’t met as a standing order, or integrate a few limited choices for glycemic targets into your insulin order sets, which trigger calls from nursing should glucose values fall persistently outside this range. Some institutions print out all glucose values for their monitored patients on each ward on a day-to-day basis, along with their glucose control regimen. Pharmacists or nurses get involved in asking for physician action and presenting recommended choices (such as adding scheduled basal insulin). Others refer similar information to a special team (a “glucose control team” or “hyperglycemia hit squad”) who intervenes via consultation or direct ordering strategies. Certain electronic medical records may have the capability of prompting physician action when glucose values are out of the target range. The what to do is common across all institutions: push for order changes when glycemic targets aren’t met. The specifics of the glycemic target and the who and the how the spur to action occurs is institution specific.
33Hierarchy of Reliability PredictedSuccess rateLevel1No protocol* (“State of Nature”)Decision support exists but not linked to order writing, or prompts within orders but no decision supportProtocol well-integrated (into orders at point-of-care)Protocol enhanced (by other QI and high reliability strategies)Oversights identified and addressed in real time40%50%2365-85%490%595+%eliminate variation arising from professionals; retain variation arising from patients
34Setting Academic teaching medical centers with over 400 beds Adult inpatients on non-critical care wards with POC glucose testing.Nov 2002 – Dec 2005Excluded:Critical care, OB, Psych, Senior Behavioral Health
35Questions What is current state? Baseline Nov ’02-Oct ’03. Insulin Use PatternsGlycemic ControlHypoglycemiaOtherWhat is effect of implementing a standardized SQIO set?Main Intervention #1 Nov ’03-May ‘05What is the incremental effect of an insulin management protocol?Main Intervention #2 May ’05-Dec ‘05
36Intervention #1 (Nov 2003): A Basic Subcutaneous Insulin Order Set Basal / Nutritional / Correction dose terminology introducedMultiple correction dose scales available, based on total insulin dose required.Sliding scale only regimens discouragedCheck box simplicitySome guidance for dosing and adjustmentHypoglycemia protocol incorporatedPaper, then CPOE versions
37Intervention #2 (May 2005) Insulin Management Protocol One page algorithmGlycemic TargetPrompt for A1CDC Oral Hypoglycemic AgentsGuidance on dosingSuggested regimens for eating patient, NPO patient, patient on enteral nutritionGuidance on dosing adjustmentIntroduced with case based teaching
39The Use of Basal Insulin Increases (sliding scale only regimens decline) UCSD clinicians have done a good job of switching to regimens that have some scheduled basal insulin in them, but we still see a lot of variability. The first generation order set was introduced 10/03, and the computerized order entry version was initiated at UCSD Thornton 1/04, then at Hillcrest in the summer of ’04.30-90 patients sampled per month, no formal analysis done, results sustained
40Glycemic Control Days 1 – 14 of admission Exclude patients with < 8 POC tests5,800 patients37,516 patient days111,473 POC testsBy patient stay% of patients with mean glucose < 180 mg /dLBy patient day% of patient days when all glucose values were between 60 – 180 mg / dLPearson chi-square statistic to compare:TP 1 (Baseline) Nov ’02 – Oct ‘03TP2 (Order Set) Nov ‘03 – Apr ’05TP3 (Algorithm) May ’05 – Dec ’05
4173%69%62 %5800 patients w/ > 8 POC glucose values, day 1-14 values: p value < .02 (Pearson chi-square statistic)
46Hypoglycemia All non critical care patients with POC values 11,057 patients / 53,466 days / 148,466 POC testsHypoglycemia: ≤ 60 mg/dLExtreme Hypoglycemia: ≤ 40 mg/dLBy patient day% of patient days with one or more hypoglycemic eventsPearson chi-square statistic to compare:TP 1 (Baseline) Nov ’02 – Oct ‘03TP2 (Order Set) Nov ‘03 – Apr ’05TP3 (Algorithm) May ’05 – Dec ’05
47Percent of Patient Days with Hypoglycemia / Extreme Hypoglycemia decreased by 30% and 31%, respectively. (Pearson chi square p < .02)> 53,000 patient days > 148,000 POC glu tests
48Approximately 100 fewer patients with Hypoglycemia per year Month
49Summary Large opportunities for improvement A safety and quality issue Systems approach is neededSHM and others now provide resources to assist implementation teams with all “essential elements”Use “Talking Points”, local anecdote, and small sample data to gain institutional supportReduced hypoglycemia can be compatible with improved glycemic control on the wardsControversy exists, but time for action is now
50The first time subcutaneous insulin is ordered, the prescriber is asked for an actionable glycemic target. A prompt to order HbA1C is also presented.
51The weight and markers of insulin sensitivity are elicited, as well as the form of the patient’s nutritional intake. (in this case, the patient is an obese 80 kg woman eating regular meals)
52The Total Daily Dose (TDD) is calculated for the clinician, based on the information provided on the patient’s obesity and weight. The TDD can be adjusted by the physician. Alternate methods of calculating the TDD are also presented.