Comment by finnh

exactly. that one clause casts doubt on all the other reasoning; randomization controls for patient selection bias but not diurnal clinic performance