Bayesian methods in health technology assessment: a review.

Area of Science:

• Biostatistics
• Health Technology Assessment
• Decision Analysis

Background:

• Bayesian methods quantitatively incorporate prior evidence into health technology assessments (HTAs).
• They combine prior distributions with study likelihoods via Bayes's theorem to form posterior distributions.
• These methods address how new evidence modifies existing beliefs and extend to prediction and evidence synthesis.

Purpose of the Study:

• To review Bayesian analysis concepts and applications in health technology assessment.
• To critically compare Bayesian and conventional statistical approaches.
• To provide guidance on reporting, case studies, and assimilation of Bayesian methods in HTAs.

Main Methods:

• Searched BIDS ISI, EMBASE, and MEDLINE databases (1990-1998) using 'Bayes' or 'Bayesian'.
• Reviewed approximately 500 relevant references for health technology assessment applications.
• Focused on studies using 'proper' Bayesian methods that update informative prior distributions.

Main Results:

• Bayesian methods offer flexible statistical approaches adaptable to different HTA contexts.
• Prior distributions are crucial, and sensitivity analyses using various options are recommended.
• While incorporating judgment, Bayesian methods acknowledge potential biases and offer tools for exploring prior opinions.

Conclusions:

• Bayesian methods provide a framework for updating beliefs and making decisions in health technology assessment.
• The choice and assessment of prior distributions are critical components of Bayesian analysis.
• Proper application of Bayesian methods requires careful consideration of judgment and potential subjectivity.