Modelling the reproducibility of acoustic rhinometry.

Area of Science:

• Medical instrumentation
• Biomedical engineering
• Clinical measurement science

Background:

• Acoustic rhinometry (AR) is a non-invasive technique to assess nasal cavity geometry.
• Evaluating the precision and reproducibility of AR measurements is crucial for its reliable application in clinical medicine.
• Understanding sources of variability in AR data is essential for accurate interpretation.

Purpose of the Study:

• To apply a linear model with correlated errors for assessing the precision of acoustic rhinometry.
• To quantify the reproducibility of nasal volume and minimal cross-sectional area measurements using AR.
• To provide guidance on selecting appropriate reproducibility measures based on study characteristics.

Main Methods:

• A reproducibility study involving 20 healthy subjects, 5 investigators, and 3 replicate measurements for 4 nasal variates.
• Application of a parametric linear model accounting for inter-subject variation, intra-subject serial correlation, and measurement error.
• Estimation of the model-based reproducibility correlation coefficient and within-subject coefficient of variation.

Main Results:

• The model-based reproducibility correlation coefficient was estimated to be fair, averaging 0.65 for both nasal volume and minimal cross-sectional area.
• The within-subject coefficient of variation averaged 15 percent.
• Reproducibility measures are influenced by population heterogeneity, time trends, and measurement scales.

Conclusions:

• Acoustic rhinometry demonstrates fair reproducibility for nasal volume and minimal cross-sectional area measurements.
• The choice of reproducibility metric depends on specific study parameters and measurement scales.
• Repeated measurements are essential in clinical practice to ensure reliable acoustic rhinometry results.