Targeted test evaluation: a framework for designing diagnostic accuracy studies with clear study hypotheses.

Area of Science:

• Medical research methodology
• Diagnostic test evaluation
• Biostatistics

Background:

• Randomized controlled trials (RCTs) typically pre-specify hypotheses and perform statistical testing.
• Diagnostic accuracy studies often lack pre-defined hypotheses and sample size calculations, potentially compromising rigor and leading to biased interpretations ('spin').

Purpose of the Study:

• To propose a strategy for defining meaningful hypotheses in diagnostic accuracy studies.
• To enhance the scientific rigor of diagnostic accuracy research.
• To provide a framework for hypothesis formulation and sample size calculations.

Main Methods:

• Developing a strategy based on the index test's role in the clinical pathway.
• Weighing the consequences of test misclassifications (false positives and false negatives).
• Establishing minimally acceptable criteria for test performance (sensitivity and specificity) that justify the test's intended use.

Main Results:

• The proposed strategy allows for the definition of specific, meaningful hypotheses.
• Minimally acceptable performance criteria can be determined based on clinical utility.
• This approach provides a basis for robust sample size calculations.

Conclusions:

• Pre-defined hypotheses and sample size calculations are crucial for diagnostic accuracy studies.
• A strategy considering clinical consequences can establish meaningful performance criteria.
• Implementing this approach will improve the reliability and reduce bias in diagnostic accuracy research.