Analytical Methods for Observational Data to Generate Hypotheses and Inform Clinical Decisions

Abstract

Randomized controlled trials have been considered the gold standard in informing clinical decision-making while observational studies have generally been utilized to generate hypotheses for future studies. The rising cost of randomized studies along with increased difficulty in accrual has led the clinical community to consider utilizing observational studies to inform clinical decisions. Various statistical methods exist to analyze observational data. Researchers must consider each method carefully, paying specific attention to its ability to answer the hypotheses, while ensuring the underlying assumptions are met. While each has its own strengths and weaknesses, research has shown that each method may yield similar estimates of treatment effect when conducted appropriately. We describe several commonly used analytical methods including their: strengths, weaknesses, and common missteps in order to inform and serve as a reference to the broader oncology community.