CLINICAL CHEMISTRY LECTURE REFERENCE INTERVALS

METHOD EVALUATION AND QUALITY MANAGEMENT A. Definition

1. Reference Interval
INTRODUCTION  Referred as reference ranges
 A pair of medical decision points that
 Quality Management span the limits of results expected for a
1. Method Evaluation given condition
 Used to verify the acceptability of new 2. Medical Decision Level
methods  Value for an analyte that represents the
2. Quality Control boundary between different therapeutic
 Insurance of a method to remain valid over approaches.
time 3. Normal Range
 Range of results between medical
Example Laboratory decision levels that corresponds to +/-
 Histopathology Laboratory 2Sd of results from a healthy patient
 Serology Laboratory 4. Therapeutic Range
 Microbiology Laboratory  Reference interval applied to a
 Hematology Laboratory therapeutic drug
 Clinical Microscopy Laboratory 5. Confidence Interval
 Blood Bank Laboratory  Range of values that include a specifies
 Clinical Chemistry Laboratory probability, usually 90% or 95%

Descriptive Statistics 6. BIas

 Used for monitoring test performance and QC  Difference between the observed mean
and the reference mean
Sources of Analytic Statistics  Negative/positive bias – test values that
1. Operator technique tend to be lower or higher than the
2. Instrument differences reference value.
3. Test accessories 7. Establishing a reference interval
4. Contamination  Done when there is no existing analyte
5. Environmental conditions (temperature, humidity) or methodology in the laboratory
6. Reagents  May require from 120 to ≈700 study
7. Power Surges individuals
8. Matrix effects (hemolysis, lipemia, serum proteins) 8. Verifying a reference interval
(transference)
A. Measure of Center  Confirm the validity of an existing
1. Mean reference interval for an analyte using
 Average (balance point of the data) the same type of analytic system
2. Median  Can require 20 study individuals
 Midpoint of the data after the data have
been rank ordered B. Categories
3. Mode 1. Diagnosis of a disease or condition
 The most frequent occurring value in the 2. Monitoring of a physiologic condition
dataset 3. Therapeutic Management

B. Measure of Spread DIAGNOSTIC EFFICIENCY

1. Range
 Largest value in the data minus the
smallest value
2. Standard Deviation (SD)
 Represent the “average” distance from
the mean
 Degree of precision of a measurement
3. Coefficient of variation (CV)
 Allows comparing SD with different units
C. Measure curve
1. Gaussian curve
 Normal distribution “bell curve”
 68.3% is between ±1SD, 95.4% is
between ±2SD, 99.7% is between ±3SD
 Definition
1. Analytic Sensitivity
o refers to the lower limit of detection for a
given analyte
2. Clinical sensitivity
o proportion of individuals with that
disease who test positively with the test
3. True positive
o patient with a condition who are
classified by a test to have the condition
4. False negative
o patient with a condition who are
classified by a test as not having the
condition
  Measures of Diagnostic Efficiency
A. Diagnostic Efficiency
1. Diagnostic sensitivity
 Ability of a test to detect a given
disease or condition
2. Diagnostic specificity
 Ability of a test to correctly identify
the absence of a given disease or
condition
B. Predictive Values
1. Positive predictive value
 Refers to the probability of an
individual having the disease if the
result is outside the reference
range
2. Negative predictive value
 Refers to the probability that a
patient does not have a disease if
the result is within the reference
range
METHOD SELECTION, EVALUATION AND MONITORING
6) LoD (Limit of Detection)
 Lowest amount of analyte accurately
detected by a method
2. Method Evaluation
a) Determination Imprecision
 Dispersion of repeated measurements
about the mean due to analytic error
 Determined by Repeated Analysis study
 Detects problems affecting
reproducibility (random error)
 Determined by Repeated Analysis study
(precision study)
 2 x 2 x 10 study
o Two controls are run twice a day
for 10 days
 E.g. glucose in
hyperglycemic stage (150
mg/L) and the
hypoglycemic range (50
mg/L)
o Estimate long terms changes
occurring over time
b) Determination of Inaccuracy

Inaccuracy
 Difference between a measured value and
its true value; due to systemic error
(constant or proportional)
 Can be determined by recovery study,
interference study and comparison of
methods (COM) study.

Measurement of Inaccuracy
a. Recovery studies – detects proportional error

Recovered = Recovered (Measured spiked sample-baseline) x

100
Concentration Added
METHOD SELECTION AND EVALUATION b. Interference studies – detects constant error
 Common interference
1. Method Selection  Hemolysis
 Reduce Cost  Lipemia
 Improve quality of results
 Bilirubin
 Increase client satisfaction
 Anticoagulant
 Improve efficiency
 Preservatives
 Journals
c. Comparison-of-methods studies
 Products specifications
 Detects systematic error
 The test method is compared with a
1) Analytical Sensitivity
standardized reference method (gold
 Ability of a method to detect small
standard)
quantities of an analyte
 40-100 specimens be run on the same day
2) Analytical Specificity
over 8-20 days
 Ability of a method to detect only the
 A plot of the test-method data (y-axis) versus
analyte it is designed to determine
the comparative method (x-axis) is
3) Specificity
generated
 Ability of a method to measure only the
analyte of interest
4) AMR (Analytical Measurement Range)
 Linear or dynamic range
 Range of analyte concentrations that can
be directly measured without dilution or
concentration
5) CRR (Clinically Reportable Range)
 Range of analyte that a method can
quantitatively report, allowing for dilution
and concentration used to extend AMR
QUALITY CONTROL
1. Introduction
 Check the reproducibility of a method by
including control specimens in the run
 Out of calibration
 Reagent may be deteriorating
 Technologist may have made an error
 Involves monitoring of analytic methods
 Done by Running QC Materials (Controls)
a. Should be the same as the specimens
actually to be tested important
b. Should span the clinically important range
of the analyte (cutoff values)
o i.e. Glucose assays
 in reference range (80 mg/dl)
 hypoglycemia (50 mg/dl)
d) R4s rule
 If the difference between the 2
controls is equal to or greater than 4s,
reject the run for probable random
error
e) 41s rule
 If 4 consecutive clues exceed the
same ẋ + 1s or ẋ -s limit, reject the
run for probable systematic error.
f) 10 ẋ rule
 If 10 consecutive control values fall
on 1 side of ẋ (either + or -), reject the
run for probable systematic error

 Types of Controls
a. Lyophilized (dehydrated powder)
 Reconstituted with its specific diluents
 Can be purchased with or without
assayed ranges
b. Stabilized frozen controls
 Needs to be evaluated for stability

2. Quality Control Charts

 Levey-Jennings Control Chart
 Used to detect errors in accuracy and
impression over time
a. Random errors
 Due to variations in techniques
b. Systemic errors
 Poorly made (contaminated)
standards and reagents
 Instrumentation problems
(function and calibration)
 Poorly written procedure
 Control values should be within statistical
limits (expresses as the mean +/- SD)
 Once a rule has been violated, reject the
analytic run, take action to remedy errors
(find the cause of error, take correction
action, reanalyze control and patient)

3. Operation of Quality Control System

1) Shift and Drift

Shift
 More than six consecutive values fall in one
side of the mean
Trend
 More than six consecutive values steadily
increase or decrease

2) Multirule Procedure
a) 12s rule
 If 1 control observation exceeds ẋ +/-
2s, a warning rule
b) 13s rule
 If 1 control observation exceeds ẋ +/-
3s, reject the run for probable random
error
c) 22s rule
 If 2 consecutive control values are
greater than ẋ + 2s or ẋ - 2s, reject
the run for probable systematic error.