LMP299Y1 Lecture Notes - Lecture 3: Statistical Process Control, Analyte, Viscosity

68 views8 pages
LMP301 Lecture 3
Analytical Factors
- methods (or assays) for chemical analysis may affect the final result
- the appropriate choice is determined by the Laboratory Professionals
- example – sodium can be measured by flame photometry, electrodes, or colorimetrically
- some methods measure sodium concentration and others sodium ion activity, which depends on solution
and therefore water contend
Ideal Methods
- need little or no sample
- the sample must be easily obtained
- give results instantly
- cost nothing to do
- accurate
- precise
- free from interference
- appropriate range for measurement
- sensitive for low concentrations
Analytical Errors
- methods in the clinical laboratory have some uncertainty in them
- not the “best” methods but are cheap, give results quickly and on small samples
-random errors (or imprecision) – small variations in ambient temperature, viscosity of fluids, electrical
surges, operator technique, etc.
- analytical errors have a bit of error as well – random
-systematic errors (or bias) – due to differences in standardization and calibration of the methods
- when making measurements – potential for errors
- bias – even when checking the time (systematic)
Precision vs. Accuracy
-precision – how well repeated measurements on the sample agree with one another (how reproducible this
is; example of “can you keep telling the same lies”)
-accuracy – how close a measurement is to the true value (example of “can you tell the truth”
- measurements can be imprecise, precise but inaccurate (bias, off the mark), or precise and accurate (ideal)
Statistical Quality Control
- how well a method is performed – QC
- laboratory staff use Quality Control (QC) to define and monitor error
- significant portion of tests in North American laboratory are to do with Quality Assurance – large part of
testing is in fact quality control
- whether or not a method is working properly – use certain samples that are known to produce a particular
result
- in many clinical situations it is more important for a test to be precise than accurate
- precision despite inaccuracy – to be able to notice changes (even if “normal” is not accurate, you can still
see a change that can indicate disease)
- the best informed physicians are cognizant of the analytical performance of the laboratories that they use
- the typical way to look at quality control is a QC chary – y axis is the concentration, x axis is the dates of
testing
- expect the values to be stable (should stay at target value, “close to the mean”)
- small amounts of variation is to be expected
Unlock document

This preview shows pages 1-3 of the document.
Unlock all 8 pages and 3 million more documents.

Already have an account? Log in
- also, the mean may stay the same but the variation is further (not precise – also bad)
- when analyzing standard deviation – Gaussian curve – 95% should stay within -/+ 2 S.D., whereas 65%
should be between -/+ 1 S.D.
Interferences
- example of using method that measures the change in light/colour, or cross-reactivity (other molecule may
bind as well and interfere with the assay)
- interferences are constituents in the sample that alter the measurement of the desired analyte and lead to
an erroneous result
- may be due to cross-reactivity, light interference, consumption of reagent, non-specific effects, etc.
- common interferences that can lead to false results (mainly in chemistry)
hemolysis – RBC break open and contents from inside the cell can go into the serum or plasma
(like the potassium levels)
lipemia – lipid particles in the sample (lipemic sample is usually milky/creamy)
bilirubinemia (icteria) – bilirubin (breakdown product of Hb) causes a greenish tinge to the
sample in diseases with higher levels, or can cause certain reactions
drugs – mimic certain particles (example of steroids similar in structure to cholesterol)
- effect depends on the nature of the assay
Post-Analytic Factors
- want to be able to determine whether or not a disease may be present – reference interval
- a patient is either healthy or sick – test should be able to help identify these populations (tell them apart)
- laboratory test used to assess if the disease is getting worse or better – management of the disease and in
prognostication
Reference Intervals
-reference interval (reference range) – range of values in health or disease
- usually a healthy (or diseased) population is recruited
- usually test the people who do not have the disease – sometimes disease reference intervals are done but
not nearly as often
- laboratory test measurements are made
- interval defined by the central 95% of the values derived from the population (Gaussian distribution)
- usually the bottom and top 2.5% are excluded
- there are challenges because it is not always easy to get a population for this type of study
- example of needing to test children, or people not willing to undergo testing when it does not benefit them
(because they do not have the disease themselves, etc.)
- does this “normal” range represent everyone (usually, no)
Is the Normal Really Healthy?
- not all populations are the same
- one “normal range” may work for the GTA but not another region
- range of results may not have a Gaussian distribution
- no apparent disease does not equate to health – example of serum cholesterol or body weight
- statistical requirement for a minimum of 120 people tested for a “normal” range to be determined
- age, time also a factor
- if you were to measure for “normal” cholesterol it would be much higher than in the past, and due to the
“Western diet” and bad habits, it is likely not good or healthy, despite having a “normal” 95%
- the best reference range in health is the patient’s own when healthy
- tough to get reliable health intervals for a population, and difficult to reach 120 people tested – therefore the
best results are one’s own “baseline value” when healthy
Unlock document

This preview shows pages 1-3 of the document.
Unlock all 8 pages and 3 million more documents.

Already have an account? Log in
Results Outside Normal Range
- this could mean
the patient is sick
patient is well but a statistical outlier (people whose “normal” is an extreme)
patient is well but not of the age, sex, race group, etc. of the reference range population
patient is well but is carrying out a proscribed activity (jogging, eating, etc.) before the sample was
made – therefore abnormal result explained
- note that the population’s reference range may not be the healthy reference range – as with cholesterol
example
Change or Variation?
- need to distinguish between heath and disease
- healthy vs. disease state – ideally they would be distinct – becomes tough when there is a bit of an overlap
- normal and disease result values usually overlap slightly
- how much imprecision in the analysis
- also challenging when there is variation (example of cholesterol level variation even throughout a day –
biological variation)
- what is important is to have a lab test that has analytical variation that is smaller than biological variation –
to be able to see if level is beyond its usual limits
- need a tight precision so that you can see if there is a major change (different from what can be considered
biological variation) or if there is a change in the biological variation
- the total variation is the sum of biological variation and analytical variation (number of measurements)
What is a Real Change?
- when two tests are done – how much difference between the results can there be before the change is
significant
- significance determines whether or not it is a “real change”
(total varrance)2 = (analytical variance)2 + (biological variance)2
95% probability limit = 2.80 x total variance
- certain t values for confidence – for 95% it is 2.80
- the total variance times 2.80 gives you number which is the 95% probability limit (19 times out of 20) of the
change in the result which can be expected because of analytical imprecision and the individual’s biological
variation over time
Case 2
- a man on a hunger strike is noted to have his serum albumin change over 3 months from 40 g/L to 30 g/L
Unlock document

This preview shows pages 1-3 of the document.
Unlock all 8 pages and 3 million more documents.

Already have an account? Log in

Get OneClass Notes+

Unlimited access to class notes and textbook notes.

YearlyBest Value
75% OFF
$8 USD/m
Monthly
$30 USD/m
You will be charged $96 USD upfront and auto renewed at the end of each cycle. You may cancel anytime under Payment Settings. For more information, see our Terms and Privacy.
Payments are encrypted using 256-bit SSL. Powered by Stripe.