Latent dry eye disease state variable.

Area of Science:

• Ophthalmology
• Ocular Surface Disease
• Biostatistics

Background:

• Dry eye disease (DED) is a common condition with diverse underlying pathologies affecting the ocular surface.
• Clinical assessments of DED severity may reflect a single underlying disease state variable, denoted as Θ.
• Existing theories propose a method for estimating Θ from clinical observations.

Purpose of the Study:

• To develop a theory for estimating the dry eye disease state variable (Θ) from scaled clinical observations.
• To test the hypothesis that a single latent variable (Θ) exists and underlies DED severity.
• To evaluate the relationship between various clinical indicators and the proposed disease state variable.

Main Methods:

• Developed a theoretical model based on assumptions of monotonic indicator-to-Θ mapping, random between-person mapping function differences, and normally distributed random perturbations.
• Digitized data from published scatter plots of seven different DED indicators against a composite severity score.
• Analyzed data using a model assuming independence of between-person variance in mapping functions and indicator values.

Main Results:

• Tear osmolarity emerged as the most sensitive indicator of DED severity, while tear breakup time was the least sensitive.
• Model-predicted indicator values generally aligned with observed values, with residuals matching expectations.
• Tear osmolarity exhibited larger-than-expected residuals, and the composite severity score showed smaller-than-expected residuals.

Conclusions:

• The findings support the existence of a single latent dry eye disease state variable (Θ).
• Tear osmolarity appears to be an exception, not consistently mapping monotonically or unidimensionally onto the latent variable.
• This research provides a framework for understanding and quantifying DED severity through a unified disease state variable.