CLINICAL CHEMISTRY BSMDT 2A

QUALITY CONTROL & QUALITY ASSESSMENT

QUALITY ASSURANCE • The median is the middle of the data after the data have
Process that guarantees that lab results are as accurate as been rank ordered.
possible and includes 3 phases of testing: • The median is often used with skewed data, so its
1. Pre analytic calculation is not significantly affected by outliers.
2. Analytic • The median is the value that divides the data in half.
3. Post analytic • To determine the median, values are rank ordered from
least to greatest and the middle value is selected.
QUALITY CONTROL
• Specific process of monitoring & accesing the analytical Mode
phase of testing. • The mode is the most frequently occurring value in a data
• Before implementing a new test, it is important to set.
determine if the test can perform acceptably by meeting or • The mode is rarely used as a measure of the data’s center
exceeding defined quality criteria but is more often used to describe data that seem to have
two centers (i.e., bimodal).
QUALITY ASEESMENT
Ensures that the results generated conform to predetermined MEASURE OF DISPERSION/SPREAD
standards and citeria. There are four commonly used descriptions of spread:
1. range,
2. standard deviation (SD),
MEASURES OF CENTER 3. coefficient of variation (CV), and
The three most frequently used descriptions of the center of a
4. standard deviation index (SDI).
data set are the mean, the median, and the mode.
Range
Mean
The range is simply the largest value in the data minus the
• The mean is most commonly used and often called the smallest value, which represents the extremes of data one
average. might encounter.
• The mean is calculated by summing the observations and
dividing by the number of the critically evaluated Standard deviation (SD)
observations
• the most frequently used measure of variation.
• SD describes the distribution of all data points around the
Median
mean.
CLINICAL CHEMISTRY BSMDT 2A
• Although calculating SD can seem somewhat intimidating, negative errors occur on both sides of the assigned target
the concept is straightforward; in fact, all of the descriptive value.
statistics and even the inferential statistics have a • Random error can be a result of many factors including
combination of mathematical operations that are by instrument, operator, reagent, and environmental variation.
themselves no more complex than a square root. • Random error is calculated as the SD of the points about the
• The SD and, more specifically, the variance represent the regression line (Sy/x). Sy/x essentially refers to average
“average” distance from the center of the data (the mean) distance of the data from the regression line.
and every value in the data set. • The higher the Sy/x, the wider is the scatter and the higher
is the amount of random error.
Coefficient of Variation (CV)
• Allows a laboratorian to compare SDs with different units Systematic error
and reflects the SDs in percentages. • influences observations consistently in one direction
(higher or lower).
Standard Deviation Index (SDI) • The measures of slope and y-intercept provide estimates of
• refers to the difference between the measured value and the the systematic error.
mean expressed as a number of SDs. • Systematic error can be further broken down into constant
• The SDI may be positive or negative. Similar to CV, it is a way error and proportional error.
to reflect ranges in a relative manner regardless of how low
or high the values are. QUALITY CONTROL AND QUALITY IMPROVEMENT
• QC in the laboratory involves the systematic monitoring of
To calculate the SD (or “s”), simply take the square root of the analytic processes to detect analytic errors that occur
variance: during analysis and to ultimately prevent the reporting of
incorrect patient test results.
ERROR
• The difference between test and reference method results Control Limits
• There are two kinds of error measured in method • The expected values are represented by intervals of
comparison experiments: acceptable values with upper and lower limits
❖ Random error • When the expected values are within the control limits, the
❖ Systematic error operator can be reasonably assured that the analytic
method is properly reporting values as approved during
Random error the method validation.
• Present in all measurements and can be either positive or • However, when observed values fall outside the control
negative; typically, a combination of both positive and limits, the operator must be aware of possible problems
CLINICAL CHEMISTRY BSMDT 2A
and the need for further investigation before reporting Control limits are expressed as the mean ± SD using formulas
potentially erroneously patient results. previously described in this chapter.
Control charts can detect errors in accuracy and imprecision
QC Materials over time.
• Specimens analyzed for QC purposes and commonly
referred to as controls. Analytic errors
• These materials must be available in sufficient quantity to that can occur can be separated into random and systematic
last for extended periods determined by QC material errors. The underlying rationale for running repeated assays
stability and manufacturer-determined expiration dates. is to detect random errors that affect precision.
• QC materials should be the same matrix as the patient
specimens. Random errors
• Control material concentrations should span the clinically may be caused by variations in technique.
important range of the analyte at appropriate decision
levels Systematic errors
• arise from factors that contribute to constant differences
QC Charts between measurements; these errors may be either
• A common method to assess the determination of control positive or negative.
materials over time is by the use of a Levey-Jennings • Systematic errors may be due to several factors, including
control chart ❖ poorly made standards,
❖ reagents,
Levey-Jennings control chart ❖ instrumentation problems,
• Results are plotted over time to identify acceptable quality ❖ poorly written procedures,
control and failures due to inaccuracy and imprecision. ❖ inadequate staff training.

• Control charts graphically represent the observed values of The distribution of error is assumed to be symmetrical and
a control material over time in the context of the upper and bell shaped, examples of:
lower control limits in relation to the target value.
• When the observed value falls with the control limits, it Shift
can be interpreted that the method is performed an abrupt change in the analytic process. Drift of values from
adequately. one level to another
• Points falling outside the control limits suggest that Trend
problems may be developing. a gradual change in the analytic process, can be seen. Drift of
values continuously in one direction..
CLINICAL CHEMISTRY BSMDT 2A
CATEGORY OF DIAGNOSTIC TESTS
WAIVED
• This includes simple qualitative tests
• Method validation is not required.
• The CLIA recommendation for using this is to follow the
manufacturers instruction. It does not require quality
control and proficiency testing.

NON-WAIVED
• These tests are required to perform quality control and to
participate in proficiency testing.
• Method validation is mandatory
QUALITY ASSESSMENT
• Quality assessment (QA) is the continuous process of NON-WAIVED: Moderate Complexity
monitoring the effectiveness of the Quality Control Plan • This refers to all automated methods
(QCP).
• The process of QA involves ongoing monitoring and NON-WAIVED: High Complexity
assessing of performance in a defined way • This refers to manual methods that require more specific
and detailed interpretation by a medical technologist.
METHOD EVALUATION • This has high patient impact.
is the process of assessing a testing
methods performance in order to Clinical Laboratory Improvement Amendment requires
determine its utility or acceptability. waived test to be used in accordance with the manufacturer's
instructions, whereas non-waived tests must undergo method
Steps In Method Evaluation validation.
1. Determine the imprecision and accuracy of a test method.
2. Assess the type, magnitude, and the clinical significance of Importance of Method Evaluation
the degree of analytical error. • Diagnostic tests have inherent sources of analytical
3. Compare the error of the results of both manual methods variability that should be minimized, such as:
and automation against the allowable error or analytical ❖ operator technique
goal. ❖ instrument differences
❖ test accessory contamination
❖ changes in environmental conditions
❖ reagent storage
CLINICAL CHEMISTRY BSMDT 2A
❖ shelf life fluctuations not compensated by forby instrument
❖ power surges circuit, temperature fluctuations, operator and
❖ matrix effects environmental conditions, bubbles in the reagent,
• Compliance with these steps ensures lower risk of bubbles in the lines, incomplete mixing of reagents,
occurrence of analytical errors. inter-tech variation, fluctuating temperature, and
electrical supply
VALIDATION
• This is a confirmation that requirements for specifically Systematic Error / Inaccuracy
intended use or specific applications were met through • It is an error that is observed at one side of the mean and
objective evidence causes the test to be low or high
• It confirms that the level of measurement is sufficient and • Westgard Rules: 2v2s, 4v1s, 1v0x
correct, and the calibration was done properly. • It influences the observations to be consistently at one side
of the mean, either higher or lower.
TYPES OF VALIDATION • It is represented by changes in the slope and y-intercept
Analytical Validation • Causes:
It is the ability to accurately and reliably measure the analyte of ❖ Inadequately manufactured standards, reagents and
interest in the clinical laboratory and in specimen's instrumentation calibration errors, deterioration of
representative of the population of interest. reagents (control and standard), sample instability,
unstable and insufficient reagent blanks, instrument
Clinical Validation drift or material changes, contaminated solutions, and a
It is the ability to diagnose or predict risk for a particular health poorly written standard operating procedure. Lot of
condition, measured by clinical or diagnostic sensitivity and reagents, improper reagent/calibrator preparation,
specificity, as well as predictive values. mechanical pipette issue, temperature changes, and
gradual deterioration of the light source
Random Error/Imprecision
• It results from variation of technique. Constant Error
• Imprecision is usually quantitated by calculating the • It is an error that exists regardless of the concentration
standard deviation (SD) from the results of replicate between the test method and the comparative method.
measurements • The magnitude of change is constant and independent of the
• Westgard rules: 1, 1g, R45 (Two levels of controls are amount of analyte
needed.)
• Causes: Proportional slope / Percent
❖ Misabeling a sample/specimen, pipetting errors, The magnitude of error increases with increasing sample
improper mixing of sample and reagent; voltage concentration
CLINICAL CHEMISTRY BSMDT 2A
Total Error 3. Recognizes an increase in the frequency of both excessively
• It is the net or combined effect of random and systematic high and excessively low minimally acceptable values
errors. (dispersion)
• Mathematically, it is the sum of random error and 4. Detects any progressive drift of values away from the
systematic error. average value over a period of at least three days (trends)
• GOAL: Total Error < Allowable Error 5. Exhibits an abrupt shift or change in value from the
established average value for three consecutive days (shift)
Clerical Error
Highest frequency occurs with the use of handwritten labels What is a Quality Control Chart?
and request forms. • Control charts represent the observed values of a control
material over time graphically in relation to the upper and
Error Assessment lower control limits. When the observed value is within the
A method of estimating how much error might be present in a control limits, it can be concluded that the method was
test result produced by a method. This is to ensure that this sufficiently performed. Excessive deviations from the
amount of error will not affect the interpretation of the test control limits may show the development of problems. The
result and compromise quality patient care. term "control limits" refers to the point at which a value
becomes statistically improbable or to the expected values
QUALITY CONTROL represented by intervals of acceptable values with upper
• It is a process that ensures the reproducibility and accuracy and lower bounds.
of results by utilizing control specimens or by ensuring that
a method remains valid over time. Types of Controls Used in the Clinical Laboratory
• It is a system for ensuring precision and accuracy in the • Assayed Control: The value of the analyte is
laboratory by utilizing quality control reagents in each predetermined. Most commercially available controls have
series of measurements. predetermined values of various analytes. The target value
• The purpose of quality control is to check the stability of the must be verified before use.
machine and quality of the reagent and check for technical • Unassayed Control: The target value is not predetermined.
error committed by the operator. This control must be fully validated by running a test for at
least 20 times in a single batch run and then perform testing
Functions once a day for 20 consecutive days to establish a target
1. Serves as a reference for the proper operation of equipment, value.
reagents, and individual techniques (machine stability, • Homemade Control: If the assayed control material is not
reagent quality, and error detection) easily commercially available (e.g., for an esoteric test), the
2. Confirms the testing's accuracy when compared to control material may be prepared by a trained laboratory
reference values
CLINICAL CHEMISTRY BSMDT 2A
staff by dissolving correctly weighed pure material in an Standard Deviation (SD)
aqueous-based solvent or in serum or whole blood. • For samples, the square root of the sum of the squared
deviations of each data point from the mean divided by n-
Types of Quality Control Plot • The standard deviation of an individual value from the
1. Tonks-Youden Plot - best plot for comparison of accuracy mean
and precision among clinical laboratories
2. Shewhart or Levey-Jennings Chart - best plot for Standard Deviation Index
detecting all types of quality control errors • Most appropriate test when comparing laboratory's
3. Cumulative Sum Graph - provides the earliest indication monthly mean to its peer group to determine if bias is
of shift or trend present
Formula:
LABORATORY STATISTICS
DESCRIPTIVE STATISTICS: Measure of Center
1. Mean
• Sum of all results divided by the number of results or Coefficient of Variation (CV)
average of all values in the set if the numbers are • Enables the medical technologist to compare standard
distributed equally on both sides of the curve deviations for various units and concentrations expressed
(symmetrical curve) - normal distribution of data as percentages
2. Median • Indicates the overall precision of a test or two methods
• Data in the middle after data has been ranked from least Formula:
to greatest, divides the distribution of data points into
upper and lower halves, or the 50th percentile, where
50% of values are greater and 5% are lower
3. Mode Variance
• The most frequently occurring or most prevalent value • Quantity under the square root sign
in the data set / the point at which the greatest number • It is the average of the squared distances of all values from
of values occur the mean.

DESCRIPTIVE STATISTICS: DIAGNOSTIC EFFICIENCY

Range It is important to recognize that there is both diagnostic
• Largest value in the data minus the smallest value; it (clinical) and analytic sensitivity.
represents the extreme data one might encounter
Diagnostic sensitivity
• also referred to simply as sensitivity
CLINICAL CHEMISTRY BSMDT 2A
• the ability of a test to detect a given disease or condition. • SPECIFICITY has few false positives
• SENSITIVITY has few false negatives
i. Diagnostic sensitivity refers to a test's ability to detect a
Analytic sensitivity particular disease or condition.
• refers to the ability to detect small quantities or changes ii. Diagnostic specificity is a test's capacity to accurately detect
in an analyte. the absence of a particular disease or condition.
• Sensitivity can be calculated from simple ratios. iii. Positive Predictive Value is referred to as the probability that
• Patients with a condition who are correctly classified by an individual will develop a particular disease or condition if
a test to have the condition are called true positives the test results are abnormal.
(TPs). iv. Negative Predictive Value is a probability that an individual
• Patients with the condition who are classified by the test does not have a particular disease or condition if the test
as not having the condition are called false negatives results are within the reference interval.
(FNs).

Another measure of clinical performance is diagnostic

specificity

Diagnostic specificity (also known as specificity)

• defined as the proportion of individuals without a
condition who have a negative test for that condition
• Patients who does not have condition and have a negative
test are called true negatives (TNs),
• Those who are incorrectly classified has condition by the
test are called false positives (FPs)