Causal Inference Methods Based on Pseudo-Observations: A Comparative Analysis of Treatment Types for Iranian...

Area of Science:

• Oncology
• Biostatistics
• Epidemiology

Background:

• Gastrointestinal cancer poses a significant public health challenge.
• Evaluating treatment efficacy requires robust causal inference methods.
• Traditional methods for causal inference can be complex with censored or incomplete data.

Purpose of the Study:

• To investigate the impact of surgical, radiotherapeutic, and chemotherapeutic treatments on gastrointestinal cancer patient survival.
• To apply and evaluate the doubly robust estimator with pseudo-observations for causal inference in survival analysis.
• To compare the efficacy of different treatment modalities on long-term patient survival.

Main Methods:

• A historical cohort study of 602 gastrointestinal cancer patients followed for approximately 11 years.
• Utilized inverse probability weighting and the doubly robust estimator with pseudo-observations for survival analysis.
• Employed R software packages 'survival' and 'pseudo' for model fitting and analysis.

Main Results:

• Surgical treatment demonstrated a significant positive impact on patient survival at all assessed time points (1, 8, and 15 years).
• Chemotherapy also showed a significant correlation with improved survival across all three time points.
• Radiotherapy exhibited a significant effect only at the first-year time point, with no significant long-term impact.

Conclusions:

• The doubly robust estimator combined with pseudo-observations offers a more efficient and simplified approach to causal inference in survival data.
• Causal inference methodologies, particularly this enhanced approach, are valuable for evaluating the comparative effectiveness of different cancer treatments.