Multivariate statistical modeling: alternative approach to test evaluation, applied to counting reticulocytes by flow...

Area of Science:

• Hematology
• Statistical Modeling
• Biomarker Validation

Background:

• Evaluating diagnostic tests often relies on a gold standard method.
• The absence of a gold standard presents challenges in comparing novel assays to existing ones.
• Reticulocyte counting is crucial for assessing red blood cell production.

Purpose of the Study:

• To develop and apply a statistical path analysis model for comparing two diagnostic tests without a gold standard.
• To evaluate the performance of flow-cytometric reticulocyte counting against visual counting.
• To assess the utility of hemoglobin concentration (HGB), mean cellular volume (MCV), and erythrocyte density width (EDW) as comparison parameters.

Main Methods:

• Statistical path analysis model was developed.
• Flow-cytometric and visual reticulocyte counts were compared.
• Comparison utilized HGB, MCV, and EDW measurements, assuming inverse correlation with HGB and direct correlation with MCV and EDW for higher reticulocyte counts.
• The model was applied to data from a study involving these parameters.

Main Results:

• The statistical model successfully evaluated two reticulocyte counting methods without a gold standard.
• Flow cytometry demonstrated superior performance compared to visual reticulocyte counting, particularly in the low-value range.
• The assumed relationships between reticulocyte counts and HGB, MCV, and EDW were utilized within the model.

Conclusions:

• The developed statistical path analysis model is effective for comparing tests in the absence of a gold standard.
• Flow cytometry is a more accurate method for reticulocyte counting than visual methods, especially for low counts.
• This modeling approach has broad applicability for evaluating comparative diagnostic tests across various fields.