[Demonstrating patient safety requires acceptance of a broader scientific palette]

It is high time the medical community recognised that patient-safety research can be assessed using other scientific methods than the traditional medical ones. There is often a fundamental mismatch between the methodology of patient-safety research and the methodology used to assess the quality of this research. One example is research into the reliability and validity of record review as a method for detecting adverse events. This type of research is based on logical positivism, while record review itself is based on social constructivism. Record review does not lead to "one truth": adverse events are not measured on the basis of the records themselves, but by weighing the probability of certain situations being classifiable as adverse events. Healthcare should welcome behavioural and social sciences to its scientific palette. Restricting ourselves to the randomised control trial paradigm is short-sighted and dangerous; it deprives patients of much-needed improvements in safety.