RCTs without important limitations or overwhelming evidence from observational studies