evaluate if differences in outcome among ITU can be related to different managements in term of sedation, fluid filling, inotrops ( global severity seems to be similar in the 8 centers, they are all general ITU with surgical+medical+/- trauma pts)
Good. See to above - have included this, but in essence we will examine the importance of the 'hospital/centre' effect after adjusting for important practice parameters. Practically, we will build a model to predict 24 hour noradrenaline dose (?log transformed with an adjustment for 'zero' doses) at 24h controlling for patient characteristics at baseline and assess the intra-class correlation (for patients nested within hospital). Then repeat this after including the 'practice parameters'.
Good. See to above - have included this, but in essence we will examine the importance of the 'hospital/centre' effect after adjusting for important practice parameters. Practically, we will build a model to predict 24 hour noradrenaline dose (?log transformed with an adjustment for 'zero' doses) at 24h controlling for patient characteristics at baseline and assess the intra-class correlation (for patients nested within hospital). Then repeat this after including the 'practice parameters'.