clinical care algorithms: the good, the bad, and the ugly r. matthew sailors, phd uth medical school...

39
Clinical Care Clinical Care Algorithms: Algorithms: The Good, The Bad, The Good, The Bad, and The Ugly and The Ugly R. Matthew Sailors, PhD R. Matthew Sailors, PhD UTH Medical School UTH Medical School Department of Surgery Department of Surgery

Upload: michael-schwartz

Post on 27-Mar-2015

220 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Clinical Care Algorithms:Clinical Care Algorithms:The Good, The Bad, The Good, The Bad,

and The Uglyand The Ugly

R. Matthew Sailors, PhDR. Matthew Sailors, PhD

UTH Medical SchoolUTH Medical SchoolDepartment of SurgeryDepartment of Surgery

Page 2: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

OverviewOverview

• Modern World / Why Use AlgorithmsModern World / Why Use Algorithms• Types of / Uses for AlgorithmsTypes of / Uses for Algorithms• Clinical Care AlgorithmsClinical Care Algorithms

– Types, Use, AutomationTypes, Use, Automation

• Good, Bad, and Ugly AlgorithmsGood, Bad, and Ugly Algorithms• Algorithm Classification & ExamplesAlgorithm Classification & Examples• Evaluating AlgorithmsEvaluating Algorithms• Writing Good AlgorithmsWriting Good Algorithms

Page 3: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Modern WorldModern World

• Society is making ever greater demands on our Society is making ever greater demands on our healthcare delivery system and, in turn, on the healthcare delivery system and, in turn, on the healthcare providers. healthcare providers.

• It is imperative that the workflow of healthcare It is imperative that the workflow of healthcare delivery be altered if quality of care and access delivery be altered if quality of care and access to healthcare are to be maintained or to healthcare are to be maintained or improved.improved.

• One of the many ways of accomplishing this One of the many ways of accomplishing this alteration is the automation of clinical alteration is the automation of clinical algorithmsalgorithms

Page 4: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Why Do We Use Algorithms?Why Do We Use Algorithms?

• Share or extend expertiseShare or extend expertise– TrainingTraining– Disseminate processes / proceduresDisseminate processes / procedures

• Reduce variabilityReduce variability

• Standardize careStandardize care– Improve overall quality of serviceImprove overall quality of service– Serve as baseline for new strategiesServe as baseline for new strategies

• Medico-Legal reasonsMedico-Legal reasons

Page 5: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Types of AlgorithmsTypes of Algorithms

• ClinicalClinical

• AdministrativeAdministrative

• FinancialFinancial

• Time-basedTime-based• Data-basedData-based• State-basedState-based

• Evidence-basedEvidence-based• HeuristicsHeuristics• Model-basedModel-based• WAGWAG

Page 6: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Clinical Care AlgorithmClinical Care Algorithm

• Specifically clinical (patient care)Specifically clinical (patient care)

• NOTNOT– FinancialFinancial– AdministrativeAdministrative– Resource allocationResource allocation

• Neutral, high-level termNeutral, high-level term– No biases or preconceptionsNo biases or preconceptions

Page 7: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Clinical Care AlgorithmClinical Care Algorithm

• Description of a process intended to Description of a process intended to guide sequential clinical (therapeutic or guide sequential clinical (therapeutic or palliative) interventions. palliative) interventions.

• Usually single patient-centricUsually single patient-centric

• Time or data-drivenTime or data-driven

• Evidence-based, models, heuristic, WAGEvidence-based, models, heuristic, WAG

Page 8: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Clinical Care AlgorithmClinical Care Algorithm

Protocol (I)Protocol (I)

GuidelineGuideline

Care Path(way)Care Path(way)

Protocol (II)Protocol (II)

ProcedureProcedure

PracticePractice

Care PlanCare Plan

Knowledge BaseKnowledge Base

This is not a hierarchy diagram, just a terminologyThis is not a hierarchy diagram, just a terminology

Page 9: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Use of Clinical AlgorithmsUse of Clinical Algorithms

• Serve only as guidesServe only as guides

• Only good inside the design envelopeOnly good inside the design envelope

• Professional clinical judgment overrideProfessional clinical judgment override

• Handle “routine” situationsHandle “routine” situations

• Allows experts to concentrate on difficult Allows experts to concentrate on difficult cases cases

Page 10: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Automation of Clinical AlgorithmsAutomation of Clinical Algorithms

• guide (but not directly provide) therapiesguide (but not directly provide) therapies

• manage information flowmanage information flow

• assist in diagnosis and treatment assist in diagnosis and treatment planningplanning

• provide a safety net for the patient for provide a safety net for the patient for the times when the inevitable human / the times when the inevitable human / technical / system errors occur. technical / system errors occur.

Page 11: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Automation of Clinical AlgorithmsAutomation of Clinical Algorithms

• Computers have no native intelligenceComputers have no native intelligence

• Algorithms must be as detailed as Algorithms must be as detailed as possible possible – streamline the implementation process streamline the implementation process – computerized algorithm must representcomputerized algorithm must represent

• what we want to dowhat we want to do

• not just want we told the computer to do.not just want we told the computer to do.

Page 12: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Good Algorithms -- RequiredGood Algorithms -- Required

1.1. Concise description Concise description • Content and intent of the algorithm Content and intent of the algorithm

• Patient groups to which it can and cannot Patient groups to which it can and cannot be safely applied be safely applied

2.2. Structured, repeatable algorithm Structured, repeatable algorithm • textual or graphical formtextual or graphical form

3.3. Fully specified concepts Fully specified concepts • (e.g., “high nasogastric tube output is defined (e.g., “high nasogastric tube output is defined

as nasogastric tube output > 1200 cc/12 hr”)as nasogastric tube output > 1200 cc/12 hr”)

Page 13: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Good Algorithms -- RequiredGood Algorithms -- Required

4.4. Fully specified decision points Fully specified decision points • E.g., PaOE.g., PaO22 >= 60 and PaO >= 60 and PaO22 <= 80 <= 80

5.5. Fully specified action steps, Fully specified action steps, • Therapeutic interventions suggested by the Therapeutic interventions suggested by the

algorithm algorithm

• Calculations to be performedCalculations to be performed

• Patient-specific recommendationsPatient-specific recommendations

Page 14: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Good Algorithms -- DesiredGood Algorithms -- Desired

1.1. Formal expression language Formal expression language • Describe the decision and action steps Describe the decision and action steps • Delineated scope and purpose Delineated scope and purpose • Define entry and exclusion criteriaDefine entry and exclusion criteria

2.2. Formalism to describe the flow of the Formalism to describe the flow of the algorithm from one state to the nextalgorithm from one state to the next

3.3. Encoded linksEncoded links• DidacticsDidactics• Reference materialsReference materials• On-line resourcesOn-line resources

Page 15: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Bad AlgorithmsBad Algorithms

• Full of vagaries (“weasel words”)Full of vagaries (“weasel words”)– ““optimize patient’s respiratory status”optimize patient’s respiratory status”

• Fail to adequately describe the decisions Fail to adequately describe the decisions and actions that are required to care for and actions that are required to care for the patientthe patient

• Important entry or exclusion criteria and Important entry or exclusion criteria and conditional values missingconditional values missing

• Concepts poorly definedConcepts poorly defined– ““high NG output”high NG output”

Page 16: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Ugly AlgorithmsUgly Algorithms

• Unstructured / poorly structured algorithmUnstructured / poorly structured algorithm

• Algorithm follows no sequential orderAlgorithm follows no sequential order

• Important entry or exclusion criteria appear Important entry or exclusion criteria appear at the end of the algorithm or in footnotesat the end of the algorithm or in footnotes

• No standard formalism used to describe No standard formalism used to describe algorithmalgorithm

Page 17: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Algorithm ClassificationsAlgorithm Classifications

• Proposal to HL7 Clinical Decision Proposal to HL7 Clinical Decision Support Technical CommitteeSupport Technical Committee

• 5 levels5 levels– 0 – 40 – 4– Increasing detail with higher classification #Increasing detail with higher classification #

Page 18: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Class 0Class 0• Often encoded only in textual form. Often encoded only in textual form. • Full of vagariesFull of vagaries• Fail to adequately describe the decisions and Fail to adequately describe the decisions and

actions that are required to care for the patientactions that are required to care for the patient• Actual algorithmActual algorithm

– often unstructured or poorly structured often unstructured or poorly structured – may follow no sequential ordermay follow no sequential order

• Important entry or exclusion criteria and Important entry or exclusion criteria and conditional values often appear at the end of conditional values often appear at the end of the algorithm or in footnotes, if at all.the algorithm or in footnotes, if at all.

Page 19: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Class 1Class 1

• Improve upon Class 0 algorithmsImprove upon Class 0 algorithms• All of the entry and exclusion criteria specified All of the entry and exclusion criteria specified

at the beginning of the description. at the beginning of the description. • Algorithms steps are coarsely structured and Algorithms steps are coarsely structured and

are arranged in a temporal or logical are arranged in a temporal or logical progression. progression.

• Algorithms are usually still represented in Algorithms are usually still represented in textual form, but may also be represented in textual form, but may also be represented in other forms.other forms.

Page 20: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Class 2Class 2

• Improve upon Class 1 algorithms Improve upon Class 1 algorithms

• Explicitly defining all thresholds and Explicitly defining all thresholds and decisions within the algorithms. decisions within the algorithms.

• Some action steps are also defined.Some action steps are also defined.

Page 21: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Class 3Class 3

• Distinguished from Class 2 algorithms byDistinguished from Class 2 algorithms by– Representation format Representation format – Presence of definitions for all stepsPresence of definitions for all steps

• Represented using structured formalismRepresented using structured formalism– flow diagramsflow diagrams– formal, structured text (pseudo-code)formal, structured text (pseudo-code)

Page 22: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Class 4Class 4

• Include all of the details necessary for a non-Include all of the details necessary for a non-expert or computer to negotiate the algorithm expert or computer to negotiate the algorithm in a reliable and repeatable manner. in a reliable and repeatable manner.

• All logical and clinical concepts are explicitly All logical and clinical concepts are explicitly spelled out and are described in terms of spelled out and are described in terms of patient-specific values. patient-specific values.

• Most often disseminated as either flow Most often disseminated as either flow diagrams or encoded using a knowledge base diagrams or encoded using a knowledge base formalism.formalism.

Page 23: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Intermediate ClassificationsIntermediate Classifications

• A given clinical algorithm may fulfill all of the A given clinical algorithm may fulfill all of the requirements for a given classification and part requirements for a given classification and part of the requirements for a higher classificationof the requirements for a higher classification

• May be necessary to classify the algorithm as a May be necessary to classify the algorithm as a intermediate value. intermediate value.

• Separate the two levels with a forward slash (/), Separate the two levels with a forward slash (/), such as, “Class 3 / 4”. such as, “Class 3 / 4”.

• This notation, while less precise than a decimal This notation, while less precise than a decimal or true fractional notation, has the advantage or true fractional notation, has the advantage of being simple and efficient.of being simple and efficient.

Page 24: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Classification OverviewClassification OverviewAlgorithm Class

Elements of a “Good” Algorithm 0 1 2 3 4

concise description of content and intent of algorithm +/- + + + + description of inclusion and exclusion patient groups +/- + + + + structured repeatable algorithm +/- +/- + + + fully specified concepts - +/- + + + fully specified decision points - - + + + fully specified action steps - - +/- + + formal expression language - - - +/- + formalism to describe the flow of the algorithm - - - - + encoded links to didactics, references, on-line resources - - - - +/- + := always present - := always absent +/- := may be present

Page 25: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Class 0 AED AlgorithmClass 0 AED Algorithm

1.1. ABC’s, start CPR, apply AEDABC’s, start CPR, apply AED

2.2. Push “analyze”, if indicated defibrillate at 200 JPush “analyze”, if indicated defibrillate at 200 J

3.3. If no conversion, defibrillate at 300 JIf no conversion, defibrillate at 300 J

4.4. If no conversion, defibrillate at 360 JIf no conversion, defibrillate at 360 J

5.5. Check pulse, if present, support airwayCheck pulse, if present, support airway

6.6. If no pulse, CPR for one minuteIf no pulse, CPR for one minute

7.7. Check pulse, if absent press “analyze”Check pulse, if absent press “analyze”

8.8. If advised, defibrillate up to three times at 360 JIf advised, defibrillate up to three times at 360 J

9.9. Repeat steps 2 thru 8 until arrival at medical facilityRepeat steps 2 thru 8 until arrival at medical facility

Page 26: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Class 0 AED Algorithm (cont.)Class 0 AED Algorithm (cont.)

Notes:Notes:A.A. Single rescuer with AED should verify Single rescuer with AED should verify

unresponsiveness, open airway give two breaths, and unresponsiveness, open airway give two breaths, and check pulse. If full arrest, AED should be attached and check pulse. If full arrest, AED should be attached and proceed with algorithm.proceed with algorithm.

B.B. Pulse checks are not required after shocks 1, 2, 4, and 5 Pulse checks are not required after shocks 1, 2, 4, and 5 unless “no shock indicated” is displayedunless “no shock indicated” is displayed

C.C. Only to be used on pulse-less, non-pediatric patientsOnly to be used on pulse-less, non-pediatric patientsD.D. If advanced personnel are present, they can use the If advanced personnel are present, they can use the

manual modemanual modeE.E. Advanced personnel can enter the above algorithm at Advanced personnel can enter the above algorithm at

any point and continue with appropriate advanced any point and continue with appropriate advanced protocolprotocol

Page 27: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Class 1 AED AlgorithmClass 1 AED Algorithm

Notes:Notes:

A.A. If advanced personnel can use the If advanced personnel can use the manual modemanual mode

B.B. Advanced personnel can enter the Advanced personnel can enter the algorithm at any point and continue algorithm at any point and continue with appropriate advanced protocolwith appropriate advanced protocol

Page 28: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Class 1 AED Algorithm (cont.)Class 1 AED Algorithm (cont.)1.1. If patient has pulse or is a pediatric patient then do not continue with If patient has pulse or is a pediatric patient then do not continue with

algorithm. Instead use alternate algorithms for VFalgorithm. Instead use alternate algorithms for VF2.2. Single rescuer with AED should verify unresponsiveness, open airway Single rescuer with AED should verify unresponsiveness, open airway

give two breaths, and check pulse. If full arrest, AED should be give two breaths, and check pulse. If full arrest, AED should be attached and proceed with algorithm. If multiple rescuers then ABC’s, attached and proceed with algorithm. If multiple rescuers then ABC’s, start CPR, apply AEDstart CPR, apply AED

3.3. Push “analyze”, if indicated defibrillate at 200 JPush “analyze”, if indicated defibrillate at 200 J4.4. If “no shock indicated” then check pulseIf “no shock indicated” then check pulse5.5. If no conversion, defibrillate at 300 JIf no conversion, defibrillate at 300 J6.6. If “no shock indicated” then check pulseIf “no shock indicated” then check pulse7.7. If no conversion, defibrillate at 360 JIf no conversion, defibrillate at 360 J8.8. Check pulse, if present, support airwayCheck pulse, if present, support airway9.9. If no pulse, CPR for one minuteIf no pulse, CPR for one minute10.10. Check pulse, if absent press analyzeCheck pulse, if absent press analyze11.11. If advised, defibrillate up to three times at 360 JIf advised, defibrillate up to three times at 360 J12.12. Repeat steps 3 thru 11 until arrival at medical facilityRepeat steps 3 thru 11 until arrival at medical facility

Page 29: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Class 2 AED AlgorithmClass 2 AED Algorithm

Notes:Notes:

A.A. If advanced personnel can use the If advanced personnel can use the manual modemanual mode

B.B. Advanced personnel can enter the Advanced personnel can enter the algorithm at any point and continue algorithm at any point and continue with appropriate advanced protocolwith appropriate advanced protocol

Page 30: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Class 2 AED Algorithm (cont.)Class 2 AED Algorithm (cont.)1.1. If patient has pulse or If patient has pulse or patient age <= 8 yearspatient age <= 8 years then do not continue with then do not continue with

algorithm. Instead use alternate algorithms for VFalgorithm. Instead use alternate algorithms for VF2.2. Single rescuer with AED should verify unresponsiveness, open airway give Single rescuer with AED should verify unresponsiveness, open airway give

two breaths, and check pulse. If full arrest, AED should be attached and two breaths, and check pulse. If full arrest, AED should be attached and proceed with algorithm. If multiple rescuers then ABC’s, start CPR, proceed with algorithm. If multiple rescuers then ABC’s, start CPR, apply AEDapply AED

3.3. Push “analyze”, Push “analyze”, if AED displays “shock indicated”if AED displays “shock indicated”, defibrillate at 200 J, defibrillate at 200 J4.4. If “no shock indicated” then check pulseIf “no shock indicated” then check pulse5.5. If If AED displays “shock indicated”AED displays “shock indicated” (no conversion), defibrillate at 300 J (no conversion), defibrillate at 300 J6.6. If “no shock indicated” then check pulseIf “no shock indicated” then check pulse7.7. If If AED displays “shock indicated”AED displays “shock indicated” (no conversion), defibrillate at 360 J (no conversion), defibrillate at 360 J8.8. Check pulse, if present, support airwayCheck pulse, if present, support airway9.9. If no pulse, CPR for one minuteIf no pulse, CPR for one minute10.10. Check pulse, if absent press analyzeCheck pulse, if absent press analyze11.11. If If AED displays “shock indicated”AED displays “shock indicated”, defibrillate up to three times at 360 J, defibrillate up to three times at 360 J12.12. Repeat steps 3 thru 11 until arrival at medical facilityRepeat steps 3 thru 11 until arrival at medical facility

Page 31: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Class 3 AED AlgorithmClass 3 AED Algorithmstart

pt. age > 8 yrs pulse present?

use alternatealgorithm

end

# rescuers = 1

open airway

check pulse

give 2 breaths

performs ABCs

initiate CPR

apply AED andturn unit on

pulse ?

wait 1-3 minutes

perform CPR for 1 min.

support airway

defib. at 200 J

press "analyze""shock advised" ?

defib. at 300 J

defib. at 360 J

conversion ?

pulse ?

conversion ?

check pulse

pulse ?

at hospital?

defib. at 360 J

defib. at 360 J

defib. at 360 J

conversion ?

conversion ?

breathing

responsiveyes

no

yes

no

yes

no

yes

no

yes

no

yes

no

yes

no

yes

no

yes

no

yes

no

yes

noyes

no

yes

no

yes

no

Page 32: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Class 4 AED Algorithm (Part 1)Class 4 AED Algorithm (Part 1)start

pt. age > 8 yrs pulse present?

use alternatealgorithm

end

# rescuers = 1

open airway

check pulse

give 2 breaths

performs ABCs

initiate CPR

apply AED andturn unit on

pulse ?

wait 1-3 minutes

perform CPR for 1 min.

support airway

defib. at 200 J

press "analyze""shock advised" ?

defib. at 300 J

defib. at 360 J

conversion ?

pulse ?

conversion ?

check pulse

pulse ?

at hospital?

defib. at 360 J

defib. at 360 J

defib. at 360 J

conversion ?

conversion ?

breathing

responsiveyes

no

yes

no

yes

no

yes

no

yes

no

yes

no

yes

no

yes

no

yes

no

yes

no

yes

noyes

no

yes

no

yes

no

check pulse

Page 33: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Class 4 AED Algorithm (Part 2)Class 4 AED Algorithm (Part 2)

place AED close to supinepatient's right ear

rescuer position on patient'sright side

clear area for AED andrescurer on patient's left side?

place AED close to supinepatient's left ear

rescuer position on patient'sleft side

turn AED on by lifting monitorscreen to "up" postition

AED brand = X ?

turn AED on by pressingpower switch

remove patient clothingthat may interfere with

AED use

open first defibrillator pad

attach first pad in the anterior position --right (patient's right) of the upper sternum (breast

bone) and below the clavicle (collar bone)

open second defibrillator pad

attach second pad in the apex position --left (patient's left) of the nipple with the

center of the electrode in the midaxillary line(where the chest and armpit meet)

Page 34: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Critically Evaluating AlgorithmsCritically Evaluating Algorithms

• Identify target audienceIdentify target audience– ExpertsExperts– NovicesNovices– Related fieldsRelated fields

• Identify intended useIdentify intended use– Author’sAuthor’s– YoursYours

• Look for well-defined decision and action Look for well-defined decision and action targets (no “weasel words”)targets (no “weasel words”)

• Look for individual-based outputsLook for individual-based outputs

Page 35: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Critically Evaluating AlgorithmsCritically Evaluating Algorithms• Look for well-defined decision and action Look for well-defined decision and action

targets (no “weasel words”)targets (no “weasel words”)• Look for individual-based outputsLook for individual-based outputs• Use the table to help classify algorithmsUse the table to help classify algorithms

Algorithm Class Elements of a “Good” Algorithm

0 1 2 3 4 concise description of content and intent of algorithm +/- + + + + description of inclusion and exclusion patient groups +/- + + + + structured repeatable algorithm +/- +/- + + + fully specified concepts - +/- + + + fully specified decision points - - + + + fully specified action steps - - +/- + + formal expression language - - - +/- + formalism to describe the flow of the algorithm - - - - + encoded links to didactics, references, on-line resources - - - - +/- + := always present - := always absent +/- := may be present

Page 36: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

Writing Good AlgorithmsWriting Good Algorithms

• Start with general and work to specificStart with general and work to specific– Iterative processIterative process

• Avoid Avoid Gotchas -- Gotchas -- later slidelater slide

• Think like a child (or engineer)Think like a child (or engineer)– Simple, discrete, decisionsSimple, discrete, decisions

• Keep it simple at firstKeep it simple at first

• Add complexity as neededAdd complexity as needed

Page 37: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

TipsTips

• Simple binary (yes / no) decisions Simple binary (yes / no) decisions involving 1 or 2 data pointsinvolving 1 or 2 data points– X < 25X < 25– X > 36 or Y <= 18X > 36 or Y <= 18

• String together lots of small steps rather String together lots of small steps rather than having one or two big onesthan having one or two big ones

• Nest complexities awayNest complexities away

Page 38: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

GotchasGotchas

• Over generalizationsOver generalizations

• ““Weasel Words”Weasel Words”

• Being Too AmbitiousBeing Too Ambitious

• Not Understanding Problem DomainNot Understanding Problem Domain

• Trying to Solve Wrong ProblemTrying to Solve Wrong Problem

• Trying to Use Wrong TechniquesTrying to Use Wrong Techniques

Page 39: Clinical Care Algorithms: The Good, The Bad, and The Ugly R. Matthew Sailors, PhD UTH Medical School Department of Surgery

SummarySummary

• Algorithms – many uses: for good, for badAlgorithms – many uses: for good, for bad

• Good, bad, and ugly algorithms Good, bad, and ugly algorithms

• Good algorithms share expertiseGood algorithms share expertise

• Algorithm classifications: 0 (low) – 4 (high)Algorithm classifications: 0 (low) – 4 (high)

• Critically evaluate algorithmsCritically evaluate algorithms

• Writing good algorithms is about attention Writing good algorithms is about attention to detailsto details