I am a consultant for a 10 year old medical records company, Records 1-2-3, Inc. which has recently introduced an electronic version (AvivaEMR) of it's template based system.
In answer to your invitation to comment:
All three of your criteria for evaluating an EMR depend on a standard definition of what it means to meet the requirements. Our EMR can (or could with moderate modifications) exchange data, perform a medical reconciliation and provide a summary of care electronically if both we and the receiving entity agreed on what information in what data format would satisfy the requirement. But there's little point in creating a capability to output data if no one will agree to accept it. Software development is expensive enough as it is. With hundreds of other commercial vendors, scores of insurers, hundreds of state and federal agencies, and countless medical institutions with whom we might some day have to interface - how do we determine what specs to build for?
In the great Chicago fire (I think it was), the city burned to the ground while firefighters responding from all over the state stood watching. There was no standard coupling on fire hydrants yet. Their hoses wouldn't fit the couplings.
Same story today. We need a standard that everyone agrees to before the functionality will appear. It's not technologically that difficult. But it's a political football. Large vendors with lots of money are (as always) dominating the legislative advisory boards so that their way of doing things become the standard. This makes some sense, but such a process often ensures that the standards we accept will comply with design decisions made 20 years ago because the large vendors have old technology that would be very expensive for them to re-write. They lobby not for the best that's available today, but for what fits their installed base.
As a nation our rush to promote (not to say "enforce") electronic medical records prior to establishing concrete, reasonable, efficient standards (HITECH, "meaningful use", common data exchange formats, etc.) means we will prod physicians into buying EMR's (with govt money) that were not built with the current (or any) standards in mind. Likewise, rewarding a vendor with high recommendation for having thousands of installations is in some ways tantamount to promoting old, non-standard technology.
So. If we want software to meet requirements, we have to define those requirements not in broad-brush politically correct terminology, but in terms that make sense to those who architect the solutions. The nation needs well written, elegant standards for it's gee-whiz national electronic medical records system.