RESULT AUTO-VERIFICATION

IMPLEMENTATION AND VALIDATION

MOHD JAMSANI BIN MAT SALLEH

MD, MPATH (USM), B.C.N
CHEMICAL PATHOLOGIST
HOSPITAL SEBERANG JAYA
INTERNATIONAL GUIDELINE AND STANDARD

“If the laboratory implements a system for automated selection and

reporting of results, it shall establish a documented procedure to ensure
that a) the criteria for automated selection and reporting are defined,
approved, readily available and understood by the staff”
(ISO 15189 – 5.9.2)
IMPLEMENTATION OF AV
1. Choose a process improvement framework (process flow chart)
2. Select an AV software (instrument/MW/LIS)
3. Assemble the team (include SME/IT/frontline staff)
4. Select metrics and gather data on the current state
5. Develop AV algorithms
6. Convert to computer code and complete AV programming
7. Validate programmed algorithms
8. Perform prospective risk assessment
9. Compile and maintain record documenting AV preparation and validation
10. Laboratory director sign-off
11. Train staff, “Go live” with AV and intensively audit
12. Regularly monitor AV performance and improve
IMPLEMENTATION OF AUTOVERIFICATION

• Planning
• Develop
• Validate
• Go live
• Monitor
PROCESS FLOW CHART

• Process of designing should be deliberate and systematic

• Setting up individuals and processes help to ensure a successful auto
verification project
GENERAL CONSIDERATION

• Instrument, LIS, Middleware or any combination

• Organize rule sets to be easily maintained
• Unlimited number of variables
• Obtain spesific information from vendors the precise transmittal
of result
“SOURCE OF TRUTH”

Sources of information for building AV rules

u Critical risk result
u Interfering substances; HIL, package insert, user experience, clinical literature
u Pre-examination algorithms (e.g NH3)
u Patient demographics
u Chartable and unchartable comments
u QC results (value range, Westgard, moving average)
u Delta checks (published literature or customized user experience)
QUALITY CONTROL

• Stop autoverification if QC fail

• Time limit (e.g every four hours)
• Control per patient sample (e.g every 50th sample)
• Analyzer off line – until QC issue is resolved to prevent AV
DESIGNING THE RULES

• Middleware vs. LIS

• Interferences
• Filter for auto-verification
MIDDLEWARE VS. LIS

• Auto-verification – middleware
• First level technical result verification
FILTER FOR AUTO-VERIFICATION

1. HIL checks / interference

2. Delta checks and Delta Day
3. Delta range / Review range
4. Consistency check
OUTLINE

uDesigning the rules

uEvaluating the rules
uChallenges
uConclusion
 Start

1ST LEVEL TECHNICAL VERIFICATION.

1ST LEVEL VALIDATION!
Middleware ANY PROBLEMATIC RESULT STRANDED.
RESULT MANUAL TRANSMISSION TO LIS

NO
Verify
Demo
FILTER FOR AUTO-VERIFICATION

1. HIL checks
 Verify
Demo

HIL
HIL rules Check

Previous
result
FILTER FOR AUTO-VERIFICATION

1. HIL checks
2. Delta checks and Delta day
 YES FAIL
Previous Delta
result Checks

NO

PASS
FAIL
Manual Review Manual
verification range Verification

PASS

Auto-verification
DELTA CHECKS

• specimen misidentification
• specimen related issues
• Analytical issues
• important change in patient status.
LIMITATION OF DELTA CHECKS

• No access to prior results (transferred patient, first time to the

hospital etc)
• Infrequently performed lab – Outpatient
• Dramatic change – patient disease state or possible due to the
treatment of the disease state (dialysis, transfusion etc)
DELTA CHECKS INTERVAL (DAY)

• Insufficient information in literature to make recommendations

• The trial and adjustment technique may prove adequate
FILTER FOR AUTO-VERIFICATION

1. HIL checks
2. Delta checks and Delta day
3. Delta range (Review range)
ALGORITHMS

• Auto-reflex
• Auto-comment
• Consistency Check
• Alert system
ALERT SYSTEM

• Hemolysis Alert
• Lipaemic Alert
• Paraprotein Alert
• Icteric Alert
• AKI Alert
AUTO-REFLEX
• If TSH abnormal add FT4
• If Chemistry Urinalysis abnormal add Microscopic Urinalysis
• If Lipemic, auto-reflex Lipemic verification
• If Grossly Hemolysis, auto-reflex In-vivo Hemolysis verification
• If Total Protein > 90 g/L, auto-reflex paraprotein verification
• If Total bilirubin high, add Direct bilirubin
• If Icterus index high, add Total bilirubin
• BHCG fail delta checks, auto-reflex Chemical pathologist verification
• Creatinine Delta Checks – AKI Alert
• If Urine creatinine low, auto reflex urine SG, pH
AUTO-COMMENT

• If TSH normal, add auto-comment

• If both TSH and FT4 normal, add auto-comment
• If Ferritin low, add auto-comment
• If Ferritin High, add auto-comment
• If normal Ferritin with low TSAT, add comment
• If Random cortisol requested, add comment
• If Chemical Urinalysis and Microscopic Urinalysis discordant
VERIFICATION TEST (NON REPORTABLE)

• Sodium Verification Direct ISE

• Sodium Verification indirect ISE
• Calcium Verification
• ALP Verification
• Magnesium Verification
• Chloride Verification
CONSISTENCY CHECK
• Albumin > Total protein
• Direct bilirubin > Total bilirubin
• Creatinine/Urea ratio
• 2H Glucose < Fasting
• TDM Pre vs Post
• Discordant TSH and FT4
• Low Thyroglobulin with positive Thyroglobulin antibody.
• Discordant Chemical Urinalysis findings and Microscopic urinalysis
findings.
VALIDATION OF AUTO-VERIFICATION
• “dry testing” – electronically simulated cases (different scenario)

• Verified patient historical data

• Real samples to challenge spesific rules (“wet testing”)

- unusual and abnormal patient data
- PT samples
- manipulated samples
- intentionally misidentified patient results
AUTO-VERIFICATION

• General biochemistry
• Endocrine
• Tumor Markers
• Therapeutic Drug Monitoring (TDM)
• Arterial Blood Gases
• Urinalysis
• HbA1c
EVALUATING THE RULES
1. Develop algorithms for automatic release
2. Verify the data set in the LIS
3. Validate the performance of algorithm using stimulated result and actual test result
4. Introduce system for validating patient results
5. Perform periodic revalidation of the algorithm and criteria
6. Review the performance of Auto-verification
7. Perform revalidation of new algorithm or criteria whenever there are any changes
made to the rules.
CHALLENGES

• Manual hospital
• References
• Staff
• Change management
• Programmer
WHY AUTO-VERIFICATION?

• Consistency
• Improve LTAT
• Patient safety
• Proactive measure
• Predictive medicine
• Added value
AUTO-VERIFICATION

Enhancer Labor saving

Awesome
Revolution Advancement
Excellent
Efficiency
THANK YOU