Minimal important change (MIC) based on a predictive modeling approach was more precise than MIC based on ROC...

Area of Science:

• Health Outcomes Research
• Psychometrics
• Statistical Modeling

Background:

• Health-related quality of life (HRQOL) is crucial for patient-centered care.
• Estimating minimal important change (MIC) is vital for interpreting HRQOL score changes.
• Existing methods like ROC analysis have limitations in precision and subgroup analysis.

Purpose of the Study:

• Introduce a novel predictive modeling approach to estimate MIC for HRQOL scales.
• Compare the performance of the predictive MIC method against the traditional ROC-based MIC.
• Demonstrate the utility of the predictive method in handling MIC modifiers across subgroups.

Main Methods:

• Developed a new MIC estimation method using logistic regression and likelihood ratios.
• Conducted simulation studies to evaluate concordance, accuracy, and precision of both methods.
• Investigated the impact of distributional assumptions on MIC estimation.

Main Results:

• The predictive MIC and ROC-based MIC were congruent under specific distributional conditions (equal variances, normal or oppositely skewed distributions).
• The predictive MIC demonstrated superior precision compared to the ROC-based MIC.
• The predictive method effectively identified and estimated modifying factors, such as baseline severity.

Conclusions:

• The predictive modeling approach for MIC provides a more precise estimation than ROC analysis in many scenarios.
• This increased precision enhances statistical power for studies utilizing MIC.
• The new method facilitates a more nuanced understanding of MIC across different patient subgroups.