Quality Control & Quality Assurance

Good laboratory practice requires that controls be run with each calibration curve.

A statically significant number of controls should be assayed to establish mean values and acceptable ranges to assure proper performance.

It is recommended to use control samples according to state and federal regulations.

Use controls of control samples is advised to assure the day to day validity of results. Use controls at both normal and pathological levels.

IMPORTANT POINTS ABOUT QUALITY CONTROL

Employ appropriate statistical methods for analyzing control values and trends, if the results of assay do not fit to the established acceptable ranges of control materials patients results should be considered invalid.

In this case, please check the following technical areas: pipetting and timing devices, photometer, expiration, dates of reagents, storage and incubation conditions, aspiration and washing methods.

The test results are valid only if all controls are within the specified ranges and if all other test parameters are also within the given assay specifications.

It is recommended to do the test sample duplicate.

The controls should be treated as unknowns and values determined in every test procedure performed.

Significant deviation from established performance can indicate unnoticed change in experimental conditions or degradation of kit reagents. Fresh reagent should be used to determine the reason for the variations.

Microbial contamination of reagents or specimens may give false results.

• Quality Assurance (QA): the practice which encompasses all activities, procedures, formats of activities directed towards ensuring that a specified quality or product is achieved or maintained.

• QA involves in every step in the analysis process from the initial ordering of a test and the collection of the patient sample (pre analytic), analysis of the sample (analytic), and finally the distribution of result to the proper destination.

• QA program involves every person in the lab. From the director to the lab helpers, also includes everyone who has contributed to the enterprise such as the phlebotomy team and data processors.

QUALITY ASSURANCE

1. Select the most accurate and precise analytical methods, in a time period that is most helpful to the physician.

2. Adequately train and supervise the activities of the lab personnel and conduct them with continuing education sessions.

3. Good instruments and institute a regular maintenance program.

4. Good quality control program, which is concerned with the analytical phase of QA.

5. Make available printed procedures for each method, with explicit directions, an explanation of the chemical principles and a listing of the reference values.

Roles of QA:

• Statistical QC is concerned with the analytical phase of quality assurance.

• It monitors the overall reliability of lab results in terms of accuracy and precision according to the criteria specified for each measurement.

Quality Control (QC)

WHY IS QUALITY CONTROL NECESSARY?• Proper validation and standardization of the

assay.• A reliable endocrine monitoring program.

Subsequent assessment of assay quality and consistency is absolutely necessary to assure the biological relevance of results.

• For every assay system there is an inherent level of error which must be accepted. A quality control program indicates when that level of error becomes unacceptable.

• Regular testing of quality control products along with patient samples.

• Comparison of quality control results to specific statistical limits (ranges).

• The result may be quantitative (a number) or qualitative, (positive or negative) or semi-quantitative (limited to a few different values).

The statistical process requires:

1. How is quality control monitored? Assay consistency is monitored by analyzing ‘internal control samples’ which are treated as unknowns, but are run in every assay.– How many to run? Usually 2 - 3 controls,

assayed in duplicate or triplicate. Much of this depends upon the variability of the assay.

– What concentrations to run? Controls should provide an estimate of variability over the working range of the standard curve.

Monitoring Assay Quality

– What samples can be used as controls?

Almost any biological material containing the relevant antigen, including serum pools, fecal extracts, purified steroids diluted in buffer or other fluid (provided it does not interfere with the assay), pituitary extracts, urine pools, purchased standards, etc.o prepare a large pool projected to last for

several years.o divide the material into small aliquots to

avoid repeated freeze-thawing.o keep the aliquots in a safe freezer

(preferably one with a temperature alarm).o store all materials, including samples, in a

no frost-free freezer.o do not allow controls to run out before

making up new controls.

• Assay coefficients of variation:(%CV = standard deviation * 100/mean times)intra-assay CV• determines the within assay error (the error

associated with running the same sample in one assay). Most accurately determined by calculating the variation in assaying multiple replicates of one sample throughout the assay (e.g., n = 10 replicates).

• More typically, the average intra-assay CV is calculated from the internal controls assayed at the beginning of the assay.

2 .coefficients of variation(CV)

• A third method is to calculate the average CV of all unknowns run in an assay. If more than one assay has been run for a particular study, then the mean intra-assay CVs for those assays should be averaged.

inter-assay CV:• determines the between assay error (the error

observed when the same sample is run in different assays).

• Determined by calculating the variation in values for samples run in every assay.

• Within a study can calculate the individual inter-assay CVs for each internal control and then average those numbers.

• intra-assay CV:– generally is the result of the presence of

unequal amounts of sample, tracer or antibody (i.e., poor pipetting, or incomplete mixing of sample or reagents), but also can be due to inconsistencies in counting or decanting.

• inter-assay CV – caused by system-related problems like

reagent instability, procedural variation or changes in standards (do not allow old standards to run out before making up new standards!).

3 .Causes of variation:

4. What is an acceptable level of error?– Sample rejection: a level of error needs

to be established that objectively determines when a sample needs to be re-analyzed. However, the amount of error tolerated is subjective and determined, in part, by how critically the data need to be interpreted. For example, the error rate would be lower in a human endocrinology laboratory where health status and potential treatments hinge on accurate measurements. Conversely, more error might be tolerated in assays conducted to define hormonal trends (i.e., the absolute values are not as important as the profile).

• rules of thumb for defining acceptable levels of error– fixed percentage of the CV- often

designated as 10% such that all samples with a CV >10% are re-analyzed.

– assay rejection criteria- often subjective, but can involve exclusions based on control values falling within 2 SD of the mean of previous values, or a set proportion of sample CVs being below 10%.

• Analytical, determined or systematic error: Which is usually due to an analytical factor (instrumental, operational or errors of method) this bias (inaccuracy) can be determined and corrected.

• Random error: Lab also subjected to imprecision or random variability

TYPES OF ERRORS:

Accuracy: is defined as the extent to which the mean measurement is close to the true value. The accuracy of a method is generally reflected by it is ability to produce the values of reference samples of known concentration.

Accuracy and Precision

Precision: the variation of results when numerous tests are performed multiple time, may be graphically depicted by a frequency curve, which gives symmetric distribution.

• Mean: the average values (the central point).

• Median: the middle value within the range.• Mode: the most frequency occurring value.

X is the mean Xi is an individual measurement∑ is the operation of summation

• Note: the degree of precision of method is best expressed in terms of SD, the greater SD the less precise method, due to the larger deviation from the mean.

The coefficient of variation (CV): Is the SD expressed as a percentage of the mean (average) and is more reliable means for comparing the precision at different concentration level of units.

• The precision of method varies inversely with the CV; the lower the CV the greater the Precision.

• Range of CV < 5-10% to consider the method is precise.

• Variance: less used to express precision, equal squared of SD= (SD)2

• Fluctuations in CV can be caused by:Inaccurate pippeting (ensure pipette

tips are sealed to the pipette before use so they draw up to correct volume of liquid)

Splashing of reagents between wellsBacterial of fungal contamination of either

screen samples or reagentsCross contamination between reagentsTemperature variations across the plate.

Ensure the plates are incubated in a stable temperature environment away from drafts.

Some of the wells dry out. Ensure the plates are always covered at incubation steps

ReproducibilityReproducibility of results can be assessed by calculating

CV values to compare mean concentrations. Intra-assay variationThe variation within assays determined by repeated

analysis of samples, typically low, medium and high, within a single

Inter-assay variationThe variation between assays determined by repeated

analysis of the same sample in several assays. These samples are called controls and should bind at 30% and 70%.

• Example (use manual method and computer program – Excel)

• The following results was obtained for a sample when assayed several times,

17, 18, 19, 20, 20, 20, 21, 21, 22, 22

Find the mean, SD, CV, Variance??

• Solution:1. Calculate mean= Σ x\n= Σ (18+20+21+17+22+19+20+20+21+22)/10=20.2. Calculate the difference of each result from

the mean.3. Square the difference.4. Σ Of squares .

(Xi-X)2 Xi-X Xi

4 2 180 0 201 1 219 3 174 2 221 1 190 0 200 0 201 1 214 2 22

Σ(Xi-X)2 =24SD = 1.6 unitCV = 8.0%

The simplest, most straight forward way to check the reproducibility of a method is by including control specimens in the run.

If control serum or urine are included with patient's samples, and observed results are the expected results, we can feel confident about the assay and can probably safely assume that the results on the patient's samples are also correct.

Control specimens are often consisting of lyophilized pool sera or urine, which may be in the normal rang or abnormal when reconstituted.

Control specimens

QC specimens may be commercial prepared or by the laboratory By pooling plasma, freezing, then to be used when needed.

A quality control product usually contains many different analytes. For example a general endocrine control can contain any number of hormone analytes including TSH, FSH, LH, Testosterone and others.

characteristics of good control:1. The composition should be as similar to the patient

sample. 2. The concentration should be stable under storage

for long period of time.3. Material should be low vial-to- vial variability.4. After vial has been opened and material prepared,

it should remain stable for the period of use.5. The material should be reasonably priced (Not

expensive).6. The material should be available in large

quantities.

Standard specimenA substance that can be accurately weighed or measured to produce a solution of an exactly known concentration.

• QC can be divided into two major types:

1. Internal QC (intralaboratory QC): This primarily monitors the day to-day performance of laboratory results, precision.

2. External QC (inter laboratory QC): which primarily monitors the accuracy of the results

• Shift: Is defined as a drift of values from one level of the control chart to another, which may be sudden or gradual and may be due to failure to recalibrate when changing lot numbers of reagent during an analytical process.

Some Quality Control Techniques

• Trend: Is the continuous movements in one direction over six or more consecutive values. Trends are difficult to detect without continual charting, and may start on one side of the mean and move across it or it can occur entirely on one side of the mean this problem is usually due to deterioration of reagents or light source of the instrumentations.

• Dispersion: Values may be within the acceptable range (2s and 3s) but are unevenly distributed outside the ±1s limits. This is indicating a loss of precision and due to random error.

• Always run ELISA samples in duplicate or triplicate. This will provide enough data for statistical validation of the results. Many computer programs are now available to help process ELISA results in this way:1. Calculate the average absorbance values for

each set of duplicate standards and duplicate samples. Duplicates should be within 20% of the mean.

2. Standard curve : Create a standard curve by plotting the mean absorbance for each standard concentration (x axis) against the target protein concentration (y axis). Draw a best fit curve through the points in the graph (we suggest that a suitable computer program be used for this).

Calculating and evaluating ELISA data

Example with human HIF1 alpha Simple Step ELISA™ kit

A representative standard curve from Human HIF1 alpha SimpleStep ELISA™ kit (ab171577) was diluted in a serial two fold steps in assay buffer.

Each point on the graph represents the mean of the three parallel titrations. We recommend having a sample of known concentration to use as a positive control.

A separate protein/peptide sample of known concentration can be used to create the standard curve.

The concentration of the positive control sample should be within the linear section of the standard curve in order to obtain valid and accurate results

3. Concentration of target protein in the sampleTo determine the concentration of target protein

concentration in each sample:– First find the mean absorbance value of the

sample. From the Y axis of the standard curve graph, extend a horizontal line from this absorbance value to the standard curve. E.g. if the absorbance reading is 1, extend the line from this absorbance point on the Y axis (a)

• At the point of intersection, extend a vertical line to the X axis and read the corresponding concentration (b).

4. Samples that have an absorbance value falling out of the range of the standard curve

For example, samples in the image above where the standard curve has a plateau, are too concentrated and will provide an inaccurate quantification.

To obtain an accurate result, these samples should be diluted before proceeding with the ELISA staining. For these samples, the concentration obtained from the standard curve when analyzing the results must be multiplied by the dilution factor.

Using Microsoft Excel to plot calibration curve :

• There are many different versions of MS Excel, and these instructions may vary slightly in different versions! 1. The first step in creating a graph using

Microsoft Excel is entering the data. The data should be in two adjacent columns with the x data in the left column. In the figure below, I have labeled the columns but this is not necessary to create the graph. Note that your absorbance data should include the origin {concentration = zero, absorbance = zero}:

2. Position the cursor on the first X value (i.e., at the top of the column containing the x values, or "Concentration” values), hold down the left mouse button and drag the mouse cursor to the bottom Y value (i.e., at the bottom of the column containing the y values, or "Absorbance" values). All of the X-Y values should now be highlighted (do NOT highlight the labels!):

3. Click on the “Insert” tab at the top of the toolbar.

4. Under the “Chart” section, choose “Line”. At the bottom of this menu, choose “All Chart Types”.

5. Under “Chart Type” choose “XY (Scatter)”, and pick the first option, (without lines). (On older versions of Excel, you will have to choose the “chart subtype” – choose the option that plots the points without lines).

6. After you click “OK”, a reduced version of your graph will appear; it should be a set of points that makes a straight line with a positive slope.

7. If you wish, you may label your graph (ie, put a title at the top of the graph). You may wish to label the y-axis “absorbance” and the x-axis “concentration”. You may also modify the style of line, the legend, the data markers, and other options if you wish, although this is not necessary.

8. Next you want to determine the slope and intercept of your line. Move the mouse cursor to any data point and press the left mouse button. All of the data points should now be highlighted. Now, while the mouse cursor is still on any one of the highlighted data points, press the right mouse button, and click on “Add Trendline” from the menu that appears.

9. From within the "Trendline" window, select the type of “Trend / Regression Type” you want – for a Beer’s Law plot the function should be “Linear”.

10.Click “OK”. A line and an equation should appear on the graph, as shown below. Notice that this equation is in the format {y = mx + b}, and numerical values are provided for the slope and intercept. The value of the intercept should be close to zero, a small number, but it may not be exactly zero. In my example, the slope is 21033 and the intercept is 0.0039.

The data points may be difficult to see on the standard axes that Graph provides. To zoom in on the data, select the menu “Zoom” and click on the choice “Fit”; this will re-scale your graph so that all the data points fit nicely on the screen and can be easily read. Now you want to determine the slope and intercept of your line. Select the “Function” menu and click on the choice “Insert Trendline”.

• Several choices will appear (linear, logarithmic, polynomial, exponential, etc.); choose the “linear” fit and click “OK”. When your graph reappears, you will see a line that fits your data, and the function appears in the upper left corner of the screen, like this: f(x) = mx + b where, of course, m is the slope and b is the intercept. The program will provide numerical values for m and b; the value of the intercept should be close to zero, a small number, but it may not be exactly zero. You may print the graph, or save it as a bitmap or jpg image, by selecting the appropriate choice from the “File” menu.

• You may also save the data set, from the “File” menu, so you can recall it later without entering the numbers again. Using a graphing calculator to plot calibration curve: Almost all scientific calculators (e.g. TI-83, TI-84, TI-89) will calculate slopes and y-intercepts for linear data sets. Please note that you don’t necessarily need a graphing calculator; almost any scientific calculator will calculate slope and intercept for a data set of x,y data. Refer to the documentation for your calculator if you want to determine your slope and intercept directly, using your calculator. (This will require you to plot and print the calibration curve separately).

Definition of Terms• Accuracy: The degree to which the measured

concentration corresponds to the true concentration of a substance. This is related in part to the specificity of the assay.

• Precision: Refers to the repeatability of a measured value or the consistency of results. It is a measure of random error defined as the variation among replicate measurements of a defined sample. It is expressed as the Coefficient of Variation (%CV) which is the(Standard Deviation/Mean)*100.

• Sensitivity: method sensitivity refers simply to the lowest level of analyte that can be detected by a given method with low CVs.

• Specificity: refers to how specific a test is for a certain substance without interferences.

• Bias: the difference between the true value and value obtained.

• Concentration: a measure of the amount of dissolved substance per unit of volume.

• Lyophilized: freeze-dried.

• Analyte: the constituent or characteristic of the sample to be measured.

• Assay: to analyze a sample of a specimen to determine the amount, activity, or potency of a specific analyte or substance.

• Out of control: indicates that the analysis of patient samples is unreliable.

• Run: a period to time or series of measurements within which accuracy and precision of the measuring system are expected to be stable.

• Range: the difference between the largest and smallest observed value of a quantitative characteristic or statistical limits.

• Notes:When evaluating laboratory findings, the

pathologist and clinical chemist must use the information on the patient that is available to them to decide whether values are to be regarded as ‘normal’, ‘pathological’, ‘critical’ or ‘ambiguous.

In deciding, they must also consider the reference intervals and corresponding cut-off limits distinguishing pathological from normal values.

In addition to this information, the Pathologist or clinical chemist will often also provide the physician with recommendations for further tests, and in some cases even for an appropriate therapy.

In matters of quality control, the physician must be able to trust in the reliability of the laboratory.

Quality management systems are used in every diagnostic laboratory today. Nevertheless, he should not trust laboratory results blindly, since errors can also go undetected in the laboratory.

In cases of doubt, improbable measuring results must be checked, or it may even be necessary to generate a new sample.

Thank You