Does a practicing physician need to be in the primary literature?

This post is more of a thought experiment than a suggestion for a new policy statement for medical education.  So, please understand part of why I'm putting this question up is to get you all to think a bit about this concept.

The underlying theme I've heard over and over ever since I was a resident was that getting data from a primary source was better than an aggregate source.  The idea was that everybody is much better off seeing the primary data for themselves, and then coming to their own conclusion on whether it applies to their individual patient or not.  Thus, when a student or a resident brings in some medical literature for the team to review, there are more  'points' awarded if the literature is a randomized-controlled trial or a case report, than a printout of 'UpToDate'.  I've heard one of my colleagues in a meeting essentially say that he wished 'UpToDate' didn't exist as it was making residents 'lazy'.

My argument against this idea is that we are teaching students how to behave as they get into practice.  I've talked with many friends in private practice, who have good intentions, and do read some from the main journal from their physician group.  However, not many of them have the time nor the energy to look up primary literature on every patient issue they run across.  Also, the primary literature is messy.  If you have time to look up 2-3 articles on a topic like correct choice of an antihypertensive, which primary articles should I choose to look at?  How deep of a search into the primary literature is necessary?  Should a practicing physician need to know about current animal models on hypertension research?  Where do you draw the line?  Honestly, I don't have time to go the primary literature for every patient issue I'm confronted with.

I see this as another area where it used to be OK to be able to keep up on the primary literature for a give topic.  However, as the medical literature has exploded exponentially, it is literally going to be impossible for anyone person to keep up.  That's where we need to start trusting one another more.  Services that aggregate data from primary literature are very valuable.  I honestly won't look down on a student looking things up before rounds on 'UpToDate' as that is a practice that will likely be sustainable.  Is having them look deeper into the literature sustainable or not?  What do you think?