On Modelling Minimal Disease Activity.

Area of Science:

• Rheumatology
• Biostatistics
• Clinical Research

Background:

• Assessing minimal disease activity (MDA) in psoriatic arthritis is crucial for treatment monitoring.
• Intermittent clinic visits present challenges for accurate MDA status determination.
• Existing statistical methods may not fully utilize available patient data.

Purpose of the Study:

• To develop and evaluate statistical modeling methods for minimal disease activity (MDA) using data from infrequent clinic visits.
• To compare the performance of complete case analysis versus hidden model methodology in estimating MDA parameters.
• To assess the impact of incorporating data from visits with incomplete MDA information.

Main Methods:

• Employed a 2-state statistical model for analyzing MDA.
• Compared "complete case" analysis (using only visits with known MDA status) with hidden model methodology.
• Utilized data from an observational psoriatic arthritis cohort, including 856 patients and 7,024 clinic visits.

Main Results:

• Analysis incorporated data from nearly all clinic visits, though only 62.6% had complete MDA information.
• Hidden 2-state model analysis yielded smaller estimates for mean MDA episode duration (3.10 years) compared to complete case analysis (4.18 years).
• The hidden model also estimated longer expected times spent in MDA (>1 year) and higher probabilities (0.91) over 10 years.

Conclusions:

• A 2-state model is effective for analyzing minimal disease activity (MDA).
• Incorporating data from visits with undetermined MDA status enhances analytical precision.
• Complete case analyses risk bias and loss of precision; hidden models offer a more robust approach.