As many of you will know, laboratory results are primarily exchanged electronically using the electronic data interchange standard called HL7, or HealthLevel 7. Unfortunately, there is considerable variation in HL7 implementations which ultimately means for EMR users that each laboratory has to have an interface coded specifically for them; if you want codified results which can be compared between laboratories. For example, one laboratory might call a complete blood count a CBC, whereas another might term it a FBC or full blood count. You will know that they are equivalent, but if you wish to search on patients who have had a recent CBC, you will miss those who have had a FBC instead!
This places a rather heavy burden upon the EMR writer who wishes to provide a comprehensive laboratory inteface, as clearly it is not possible to account for all the variations that have been implemented. In most cases, the interface is undertaken as a contract programming task which may possibly be paid for by the laboratory in question, the physician requesting the interface, and/or the EMR vendor.
As an alternative to these possibilities, at Synapse EMR, we hope to create a generic interface to receive blood results in HL7 format, and have a process whereby the user
trains the interface so that if they do use different laboratories, discrepancies such as the one quoted above needs to be resolved only once by the user. Thereafter Synapse EMR will remember the equivalence mapping.
Currently there is a HL7 parser included in the current version of Synapse EMR. If you navigate to the
Inbox tab, you will see a
Import HL7 button. This will parse a directory of HL7 messages and report back whether it has managed to read each file correctly, and whether or not it has found a match for a patient inside the database. For the current implementation, it is required that each HL7 file only refers to a single patient. That is, you can not have an HL7 file that contains more than one patient's results.
Work is now in progress to create the generic trainable interface.