Tim Cook, over at LinuxMedNews, writes:
The current landscape is that most (especially public) funding for healthcare IT projects typically come in rather small packages. Maybe a disease registry now, an MPI later and a population surveillance application to follow. In each case, the awards go to an academic group, a company, etc. and they deliver what the customer has specified. It works great for a long time and then they want to try to share some information. This is the hard part. All were written independently, on a budget without knowledge of each other. How can we even consider automated decision support across these environments? Now, someone must pay for data integration services. This is generally doable at a cost but there is usually semantic loses in the information exchange. As an additional example. What happens when a family physician needs to refer a patient to a specialist? Typically the specialist only wants an extract of the patient record and likewise the family physician only wants results in a summary format. Even if they have the same EMR, most of these systems can’t exchange information due to customization factors at each end. So, someone has to produce the needed information (usually in hardcopy) and it is sent via FAX to the other end where it may or may not be entered into another electronic system.