You are viewing documentation for the old version 2.30 of Analyse-it. If you are using version 3.00 or later we recommend you go to the Reference interval
References intervals - CLSI C28 & IFCC EPTRV
Reference intervals can be used to verify or determine a reference interval (also known as a reference range or normal range) from a sample of normal subjects from the population.
The requirements of the test are:
- A sample of normal subjects.
- A method measured on a continuous scale
Arranging the dataset
Data in existing Excel worksheets can be used and should be arranged in the List dataset layout. The dataset must contain a continuous scale variable containing the observations of the method.
When entering new data we recommend using New Dataset to create a new reference interval dataset.
Establishing a reference interval
To start the reference interval analysis:
- Excel 2007:
Select any cell in the range containing the dataset to analyse, then click Reference Interval on the Analyse-it tab/toolbar.
Excel 97, 2000, 2002 & 2003:
Select any cell in the range containing the dataset to analyse, then click Analyse on the Analyse-it toolbar, then click Reference Interval.
- Click Variable and select the variable containing the observations of the method.
- Click Method and select the method to use to calculate the reference interval.
||No reference interval will be calculated. Only an existing reference interval will be verified (see below).
||A non-parametric reference interval will be calculated (recommended minimum of 120 observations).
||A parametric reference interval will be calculated.
|Parametric (log transformed)
||A parametric reference interval will be calculated after log transforming the observations of the method. The calculated reference interval is shown back-transformed into units of the method.
- Enter the Reference interval percentage, as a value between 0 and 100, and the Confidence interval to calculate around the limits of the reference interval. Defaults are 95% and 90%, respectively, as recommended by the IFCC.
- Click Reference Limits and select Both to calculate the lower and upper limits of the reference interval, or select Lower only or Upper only to calculate just one of the limits.
- Click Normality test and select Shapiro Wilk, Anderson Darling or Kolmogorov Smirnov to show a normal plot and formally test if the sample is normally distributed. Select None to exclude the normality test and plot.
- Click Histogram and select with Reference Limits to show the reference interval on the histogram or select with Reference Limits + CI to show the reference interval with confidence intervals for each limit.
- Click OK to run the test.
The report shows the number of observations analysed and the range of observations for the method.
Descriptive statistics, including the mean, median, SD, Skewness/Kurtosis, and a normality test are shown for assessing whether the data is normally distributed. If the data is normally distributed, allowing the parametric reference interval method to be used, the mean & median should be similar and the normality test p-value should be non-significant. Otherwise the non-parametric (or the parametric log-transformed) reference interval method should be used. If the parametric (log transformed) method is used the statistics are based on the log transformed data.
The reference interval upper and lower limits are then shown, with corresponding confidence intervals. A histogram (see below) shows the frequency of observations of the method with a normal overlay (if a parametric reference interval method is used) and the reference interval. Confidence intervals for the reference limits are shown if requested.
(click to enlarge)
The normal quantile plot (see below) shows the observations of the method (X axis) against the expected normal quantile (Y axis), the number of SDs from the mean where such an observation would be expected to lie in normal distribution. When the sample is normal the points will form a straight-line, very close to the ideal normal straight-line shown in red.
(click to enlarge)
If the parametric method is used to calculate the reference interval the normality test, histogram and normal quantile plot should be used to verify the data is normally distributed. If the log transformed parametric method is used the normal quantile plot and additional histogram show the distribution of the log transformed data, which again should be assessed for normality.
Verifying a reference interval
An existing reference interval can be verified using as few as 20 observations.
To verify a reference interval:
- If the Reference Interval dialog box is not visible click Edit on the Report toolbar on the Analyse-it tab/toolbar.
- Click Compare against and select Reference interval claim.
- Enter the lower and upper limits of the Claimed interval.
- Click OK.
The goal when verifying an existing reference interval is for less than 10% of observations to lie outside the claimed reference interval. Outside claimed interval shows the actual percent of observations outside the range. The histogram (see below) shows the claimed reference interval.
(click to enlarge)
References to further reading
- How to Define and Determine Reference Intervals in the Clinical Laboratory (2nd Edition).
CLSI C28-A2; ISBN 1-56238-406-6 (2000)
- Statistical Treatment of Collected References Values. Determination of Reference Limits (Part 5).
H.E. Solberg, J. Clin. Chem. Clin. Biochem. 25 (1987); 645-656
(click to enlarge)