Flashcards in Medicine - Measurement and Data Reliability Deck (37):
Name 5 steps in drug analysis
1) Take representative sample
2) Extract analyte(s) from interfering matrix
3) Separate analytes
4) Identify, detect and quantify analytes
5) Determine reliability/validity of results
5 Steps of analysing a sample
1) Select representative sample
2) Extraction -e.g. a lot of matrix is present
3) Purification - e.g. is the test highly susceptible to impurities, excipients, degradation products
4) Characterisation - the target analyte should be identified, especially if the test is not very specific
5) Data validity - us stats, comparison with standards and/or another method. If you are testing by a newly developed method or if the method has been changed significantly then it is critical to validate the new method first
Ways to present and handle analytical data
-Can you group it? - frequency table - histogram if numerical or bar chart if numbers and text
-Did you change only one parameter? Line graph suitable where x axis having independent variable (what you altered) and y axis being dependent variable (what you measured)
-If you changed more than one variable then a scattergram may be suitable
How to get the line of best fit?
Linear regression analysis - better than manual determination
equation of the line of best fit
y = a + bx
Use equation where:
a and b are parameters of the line
b is the slope, a is the y intercept
N= number of pairs of values
Analytical measurements are not exact. What do we do in reaction to this?
Assume the last digit quoted in a measured value is uncertain
An analyst weighs 25 tablets and finds the average weight per tablet.
Then he calculates that 38 tablets should weigh 44.2239672g. The accuracy of the balance is +/-0.1mg
How should he write the result?
As Weight= 44.2240g
When calculating drug content, how many significant figures is favoured for the answer?
3 significant figures
Rules for sigfigs when adding or subtracting?
Round off to the same number of decimal places as the value having the fewest decimal places
10 + 28.621 = 129
Rules for sigfigs when multiplying or dividing?
Round off to have the same number of sig figs as the value with the least significant figures
2.1 x 11.31 = 24
How should one take measurement on a scaled instrument?
Go for one more figure than you can read with certainty
Manufacturers often give accuracy value to their apparatus. A graduated 50ml burette may be accurate to only +/-0.02mL
Therefore when measuring 30ml of solvent you should report to this level of accuracy only as 30.00ml
What 3 types of errors affect validity of analytical measurement?
1) indeterminate (random) error
2) determinate (systematic) error
3) Gross error
What is indeterminate error and what does it affect?
Evaluate with statistics
affects precision more than accuracy
what is determinate error? how to evaluate it?
evaluate with standards
affects accuracy more than precision
Gross error is what?
A big mistake, like spilling everything on the floor, sample evaporated etc. the value stands out as an outlier.
What do random errors follow statistics wise?
The normal distributions or Gaussian pattern
With the normal distribution, use standard deviations to explain how sure one can be where the values are
68% values within one sd (or 68% chance that random error will not exceed +/-1sd)
95% are within +/- 2sd
99.7% are within +/- 3sd
of the mean value
What is precision?
How close (repeatable) the measurements are to each other
How is precision expressed?
Range (quickest) - difference between highest and lowest results
Standard deviation or the relative standard deviation - preferred. Calculate by statistical analysis
What is relative standard deviation?
The spread of data expressed as % of the mean value, so units become unimportant
+/- RSD also called the coefficient of variation (CV)
use RSD to remove units
divide by mean
What is accuracy?
how close the measurement is to the accepted value
How can we evaluate accuracy?
by calculating the absolute error (measured value minus a true value)
and relative error
Equation for absolute error
where xi= observed value and xt= true value
Equation for relative error
E=(xi-xt)/xt * 100%
Name and describe 3 origins of systematic error
Calibration - apparatus (cylinders, dispensers etc) or equipment (balances, pH meter etc)
Reagents (unstable), reactions (incomplete)
3) Personal error
e.g. eye sight (colour), height (can not read burette at eye level) etc
What can a null hypothesis that assumes to bias verify?
If a difference is due to a random or systematic error (bias) in the method or not
When do you reject the null hypothesis?
where xbar = mean
xt = true value
SRN = square root of N
ts = standard deviation
When do you accept the null hypothesis?
What is t-Test?
One of the most powerful and widely used statistical tests
It is used to calculate the confidence intervals of a measurement when the population standard deviation is not known, which is the case with limited number of data points
What can the t test be used to do?
What does the t test correct for?
Used to compare two averages
Corrects for the uncertainty of the sample standard deviation caused by taking a small number of samples
How to detect gross error?
use a Q test
You can use a Q distribution table to compare the Q experimental value at a given confidence level with Q critical value in the table
Using the Q test, when is a value an outlier?
where xq=value in doubt
xn=nearest value to doubted
What is method validation?
The process of demonstrating that analytical procedures are suitable for their intended use and that they support the identity, strength, quality, purity and potency of the drug substances and drug products
In method validation, in addition to the details of how to carry out an assay (SOP) what else must the method state?
How the sample is extracted/prepared or handled before it is analysed. Quality of reagents including reference or standard materials used. Procedures for instrument calibration, reagent preparation, method of sample preperation and handling
When do methods require complete validation?
When they are new methods
When are pharmacopoeial methods not valid?
If they are significantly changed. Any changes must be validated (partial validation)