Modeling smoking in systemic sclerosis: a comparison of different statistical approaches

Area of Science:

• Rheumatology and Vascular Health
• Epidemiology and Biostatistics

Background:

• Vascular complications are significant in rheumatic diseases.
• Accurate assessment of smoking's impact on vascular health is challenging.
• Previous studies may have underestimated smoking's effects due to inadequate modeling.

Purpose of the Study:

• To evaluate diverse smoking modeling techniques.
• To determine the optimal method for assessing smoking's effect on vascular outcomes in rheumatic disease patients.
• To identify the influence of smoking on Raynaud's phenomenon and digital ulcers.

Main Methods:

• Utilized data from the Canadian Scleroderma Research Group Registry (n=606).
• Compared five smoking models: ever/never, current/past/never, polynomial contrasts, separate intensity/duration/cessation, and the Comprehensive Smoking Index (CSI).
• Vascular outcomes included severity of Raynaud's phenomenon and digital ulcers.

Main Results:

• Simple models (ever vs. never smokers) showed no significant effect.
• Analysis revealed a 'healthy smoker bias' when isolating past smokers.
• The Comprehensive Smoking Index (CSI) demonstrated a significant negative association between smoking and vascular outcomes.

Conclusions:

• The method of smoking modeling critically influences the observed association with vascular outcomes in rheumatic diseases.
• The Comprehensive Smoking Index (CSI) provides a more accurate representation of smoking's detrimental vascular effects.
• Implementing robust smoking assessment is essential for future research in this field.