Laboratory Statistics Flashcards
Define reference range:
Set of values that are considered normal or typical for a specific physiological measurement based on a defined population
Used to identify abnormal results
Define repeatability:
Repeatability refers to the degree of consistency or agreement in measurements or observations when the same experiment, process, or measurement system is repeated under the same conditions
Define reproducibility:
reproducibility refers to the degree of agreement between measurements or results obtained when an experiment or test is conducted under different conditions
How do you construct a reference range when results are normal ?
When results are normal;
- range constructed from mean (𝑥̅) and standard deviation ( s ) and properties of Normal distribution
- 95% use 1.96, 90% used 1.64
- in SPSS Graphs>Legacy dialog>Histogram -> Analyse>Descriptive statistics>Descriptives
How do you construct a reference range when results are abnormal ?
Range constructed from sample percentiles
For 95% - use 2.5th percentile, 97.5th percentile
For 90% - use 5th percentile, 95th percentile
In SPSS - Analyse>Descriptive statistics>Frequencies
What is measurement error ?
Several measurements of the same sample/individual may not give the same results due to measurement error
Assuming no bias;
Value = true value + measurement error
How can measurement error be quantified ?
Can be quantified by calculating the standard deviation of repeated measurements on the same individual
The common standard deviation of repeated measurements is known as the within-individual standard deviation (sw)
How do we calculate sw ?
ANOVA to calculate average within group variance then square root
What are method comparison studies ?
Used to investigate the agreement between two methods of measuring the same property
Often used to determine if two methods can be used interchangeably
What is limits of agreement and its 2 assumptions ?
The difference between the two methods is quantified by calculating limits of agreement
d ± 1.96 s
where d is the mean of the differences, and s is the standard deviation of the differences
The method makes two assumptions;
1 - the differences are normally distributed (check using histogram)
2 - the differences are unrelated to the magnitude (check by plotting difference against average)
How would you calculate method comparison studies in SPSS ?
Data should be set up with two outcome variables
Calculate the difference using Transform>Compute
Analyse>Descriptives calculates mean and sd
Plot histogram of differences using Graphs>Legacy Dialog>Histogram
Plot scatter diagram of average against difference using Graph>Legacy Dialog>Scatter
What is quality control ?
Used to identify if laboratory process are out of control
Laboratory assays often run with a control sample (with a known target value
A Shewhart control chart represents a simple way of presenting quantitative results on each control material
What is a Shewart control chart ?
Simple way of representing run-to-run variation of the control
Uses an estimate of variability (s);
- Warning limits are ± 2s
- Action limits are ± 3s
Two results outside the warning limit (same direction) or 1 outside the action limits results in process being out of control
