A Quantile-Quantile Toolbox for Reference Intervals

Area of Science:

• Statistical methodology
• Biostatistics
• Clinical laboratory science

Background:

• Parametric statistical methods generally outperform nonparametric methods but require data to follow a known distribution, often normal.
• Applications include determining reference and detection limits, where parametric analyses provide more accurate estimates and uncertainty measures.
• Data may be normally distributed, transformable to normal, or exhibit deviations due to extreme values or data censoring from detection/quantitation limits.

Purpose of the Study:

• To present a quantile-quantile (QQ) toolbox as a versatile methodology for various statistical settings.
• To demonstrate the utility of QQ methodology in addressing challenges with data distribution and censoring.
• To provide accessible parametric methods for clinical laboratorians.

Main Methods:

• Utilized a quantile-quantile (QQ) toolbox for statistical analysis.
• Employed QQ methodology to develop methods for power transformations and normality testing.
• Applied QQ methods for estimating reference limits and constructing confidence intervals.

Main Results:

• QQ methodology facilitates the identification of optimal power transformations for data normalization.
• Normality testing before and after transformation is streamlined.
• Accurate estimation of reference limits and confidence intervals is achieved.

Conclusions:

• Parametric methods, enhanced by QQ methodology, offer statistically rigorous yet accessible solutions for clinical laboratories.
• These methods do not necessitate specialized software or advanced statistical expertise, being implementable in spreadsheets.
• Explored reference values for amyloid beta proteins in Alzheimer disease as a practical application.