Criteria for Distinguishing Between Healthy and Ill

Laboratory results are assessed using reference limits. The term reference limit encompasses reference intervals, decision limits, and action limits. By definition, reference ranges contain 95% of results from non-diseased individuals. Decision limits allow for efficient separation of diseased and non-diseased states. Action limits are empirically established thresholds linked to specific actions when exceeded. Statistical methods play a significant role in determining these intervals and limits and in assessing their validity.

In a long-term collaboration, Prof. Dr. Rainer Haeckel, along with Mr. Dipl. Math. Werner Wosniok and Dr. Farhad Arzideh from the Institute of Statistics at the University of Bremen, have worked on this complex of questions. In 2007, the previously loose collaboration was incorporated into the DGKL’s Reference Limits Working Group, which transitioned into a section in early 2020.

The section typically meets twice a year, offers courses on the use of indirect methods for reference range estimation, and publishes on algorithms and collected data.

Section Objectives

The section addresses questions related to the evaluation of laboratory results.

These include:

• Methods for determining reference intervals, decision limits, and action limits
• Procedures for estimating the validity of these limits and ranges
• Methods for evaluating laboratory findings
• Procedures for evaluating method comparisons

The goal is to develop solutions that can be integrated into patient care. Results are published and presented in small conferences and workshops during annual congresses.

Reference Limit Estimator (RLE)

In clinical chemistry laboratories, a large number of measurement values are determined, evaluated, and documented daily. A database exists with measurement results and other relevant information such as the date of sample collection, age, and gender.

Generally, for each measurand, the proportion of pathological values is small compared to the total number of values. Various “indirect methods” exist to estimate upper and lower reference limits from these sometimes very large datasets. Additionally, it is possible to determine age- and gender-specific reference limits. From version 50 onwards, the ‘Reference Limit Estimator’ supports the TML method by Farhad Arzideh et al. and the TMC method by Werner Wosniok.

The advantage over the conventional or direct method, which examines a pre-selected reference population, is enormous: The data is already available in laboratory databases, eliminating the need for time-consuming examination and selection of subjects for a reference population. Moreover, for organizational, cost, and time reasons, it is hardly possible to find such a large number of subjects that further stratifications (e.g., by age) can be made in addition to gender. Additionally, doubts about the representativeness of a selected population are often warranted. If the measurement method is poorly standardized, reference limits determined in one laboratory using subjects cannot be adopted by other laboratories. However, when using measurement data from the laboratory’s own database, laboratory-specific reference limits can be determined. Universally applicable reference limits are obtained with these methods when multiple laboratories participate in the determination and the measurement procedure has a high degree of standardization.

The user operates the program through a Microsoft Excel interface, which was developed with Excel 2003/2010 32-bit but requires 64-bit versions since version 50 (Excel 2021). The statistical calculations are very extensive and complex and are performed using the statistical program “R” with several additional modules (packages) (last tested version is 4.4.1-64Bit). The statistical results and graphical representations are then partially transferred back to the Excel interface. The operating language of the Excel interface can be adjusted in the tool.

DOWNLOADS

Reference Limit Estimator (Program as ZIP-File, version RLE50 20231002)
Reference Limit Estimator (PDF Manual, version 20180426)

Permissible Measurement Uncertainty

The Reference Ranges section of the DGKL has developed a concept to derive the permissible measurement uncertainty (permissible imprecision and permissible bias) from the reference interval. This concept and the underlying algorithms were described in Clin Chem Lab Med 2015 (Permissible limits of uncertainty in laboratory medicine). The Excel program automatically calculates the permissible measurement uncertainty (permissible coefficient of variation, pCVA) and the permissible limits of deviation for individual values according to RiliBÄK 2008 (∆max). The table contains almost all measurands from RiliBÄK 2008.

Excel File Permissible Measurement Uncertainty

TMC – A new indirect estimation of reference intervals: truncated minimum chi-square approach. (Download as R-script with instructions)

ReferenceRangeR – Making reference range estimation easy

The platform offers a reliable and user-friendly way to estimate reference ranges from real-world data. Several methods are available that are at the current state of science (TML, TMC, RefineR, kosmic, reflimR). An intuitive interface makes it possible to import data quickly and easily through copy-and-paste in the browser. No additional software is needed. The data can be selectively pre-filtered or stratified by gender and age. The platform is provided by the University Medical Center Oldenburg.

Link: https://kc.uol.de/referenceranger/