The work program is based on a data cleanup process to obtain normal distribution by rationing iterations.
95% Confidence Interval (2.5 и 97.5 percentile) can be established after the data “cleanup process”. To form an accurate range of RI the confidence interval must be computed; i.e.,a range of values that there is a specified possibility that the value of parameterlies within it (parametric Gaussian method). The confidence interval can be established with 90% probability.
In order to perform calculations the following stages must be exercised:
First step for using the e-source is to prepare the data, upload it from the laboratory information system (LIS) (preferably) or/and manually enter results of real measurements of one or another indicator of reference subjects for a certain time period into Excel file (extensions .xls, .xlsx, .csv).
According to the current publications and IFCC C-RIDL (the International Federation of Clinical Chemistry, Committee of Reference Intervals and Decision Limits) the following key factors for uploading must be met:
- If the purpose of the calculation is an establishing RIs of healthy reference subjects the use of outpatient laboratory databases is preferred. Thus, the use of highly specialized databases must be avoided.
- The selected analyte must be obtained by the same approach using the reagent of one manufacturer.
- The data upload must be limited to one result of the subject and include the result of the last measurement. To do so there is an option of creating a report in LIS hence, this type of data is not included in the upload or you can use the "Remove duplicates" function in Microsoft Excel. For that you need to sort results by date.
There are no specific recommendations for data quantity of an indirect approach but it is vital to gather as much information as possible so please apply a thumb rule "the more the better".
The minimum is supposed to include more than 1,000 results for one group. Small groups for example children and/or senior citizens over 75 years old can include as low as 400 results.
- Dividing into groups by age/gender must be based on known effects and literature data (such as a difference of creatinine according to age/gender). Additional verification of the division by age/gender can be done by computing a median and chosen percentiles (for example, the 10th and the 90th) of proposed groups of the general sample and comparing them with each other before establishing the RIs. To calculate the median and percentiles in Excel please use the "Median" and "Percentile" function.
- You must use a period or comma as a separator
- If the upload contains empty cells or non-numeric data; for instance no data, "1.5 (double checked)", it will not be included in the calculation.
Log in to your personal account / Sign up
Read additional information provided on the website
In your account click "Upload File".
The uploaded document must contain a column with the analyte results that you want to use for establishing RIs. The presence/absence of other columns does not affect the outcome. Computing is performed by using data of the specified earlier column. Age/gender stratification is not done automatically. Computing RIs for different groups has to be obtained separately.
- Fill in the required fields and specify the column number that is intended to be used for the calculation. It is critical to specify the number of the letter name (Latin layout) of the column correctly. Completed fields will be shown in the final report.
- Analyze the derived intervals. Compare results with published Reference Intervals as well as with intervals that were historically used in laboratories.
- For statistically accurate defining of RIs you must use the following:
- Use a minimum of 1,000 results for each group
- The sample must be heterogeneous with a minimum of 5 iteration and at least 20% of rejected results.
- For establishing RIs based on healthy reference subjects, you must use data from outpatient or non-core specific indicator databases such as laboratory of multi profile institutions. For instance establishing of RIs of tumor markers using results from cancer centers cannot be entirely correct. In such situations it is possible to establish the RI for a specific tumor marker in relation to a particular pathology. For example, computing RI for CA 19-9 (cancer antigen of pancreatic cancer and gastrointestinal cancer) whilst subject(s) have small bowel cancer.
- The data must be in a file with extensions (.xls, .xlsx, .csv.).
A program for recognizing and reading .pdf files (Adobe Reader 11 or higher) must be installed prior to using E-source.