I'm the Informatics Manager for a research support center within an academic healthcare institution in the Southeastern United States. IMO, PHRs are a novelty which will be shortlived. This is because they trust that the patients:

1) will document everything relevant

and

2) understand enough about his/her own healthcare to get it right

Storing data is one facet of Healthcare Informatics. Honestly, if it's done right, there won't be a problem as to where the data is stored. However, the "if it's done right" is the key. You're main complaint seems to be that there isn't an effective way to collect all the information on a patient. While this is true, collecting information is only a small piece of the puzzle. It's collecting the information in a *sensicle and meaningful way* that truly matters. If you can't do that, it doesn't matter where the information is housed.

Right now, there are ways to pool information from different datasources together (HL7 being the most prominant). However, little is being done to ensure that DataSource1.Table1.Column1 is really the same thing as DataSource2.Table3.Column25. In addition, there is vast amounts of research needed to convert:

"This patient has Condition1, Condition2, Condition3. In addition he/she is allergic to Drug1, Food 1. However, they have no history of Problem1."

into a representation that can quickly be retrieved from the storage mechanism.

While PHRs do allow for patients to conglomerate their data, I am worried what they will be conglomerating. If all their going to be doing is sweeping all the crap from many different places into one location...well, I think you can guess what they'll end up with.