Quantification of Difference in Nonselectivity Between In Vitro Diagnostic Medical Devices

Area of Science:

• Medical Diagnostics
• Biostatistics
• Analytical Chemistry

Background:

• Accurate results from in vitro diagnostic (IVD) medical devices (MDs) are essential for patient care.
• Method comparison studies are used to assess IVD-MD performance but can be affected by various factors, including nonselectivity.
• Quantifying nonselectivity differences between IVD-MDs is crucial but historically lacking.

Purpose of the Study:

• To introduce a novel measure for quantifying differences in nonselectivity (DINS) between pairs of IVD-MDs.
• To develop a plug-in estimator for DINS and a criterion for identifying excessive DINS.
• To evaluate the performance of the DINS estimator using simulations and clinical data.

Main Methods:

• Developed a novel measure for quantifying differences in nonselectivity (DINS) using elements of ordinary least squares regression and IVD-MD repeatability variances.
• Introduced a plug-in estimator for DINS, linked to prediction interval widths.
• Employed Monte Carlo simulations and analyzed clinical data to assess the estimator's behavior.

Main Results:

• The proposed DINS measure effectively quantifies nonselectivity differences between IVD-MDs.
• The DINS estimator is significantly influenced by DINS and, to a lesser extent, by heteroskedasticity, with study design size modulating these effects.
• Simulations and clinical data confirmed that DINS impacts the estimator, validating its utility.

Conclusions:

• The novel DINS measure and its estimator provide a quantitative approach to assess nonselectivity in IVD-MDs.
• This metric aids in evaluating the quality of method comparison studies, ensuring reliable diagnostic results.
• The findings support improved quality control and performance assessment of IVD medical devices.