I recently presented at the International MUSE conference on the topic of HIEs. The main message was about how to control your data when connecting to an HIE. Of course, when talking about HIEs the conversation can turn to the subject of EMPIs. Several of the attendees mentioned having an EMPI, and, probing further, it was clear there is still a lot of confusion around how to make their EMPI solution work well for their specific situation.There are a lot of factors that determine if your EMPI solution is working for you. For me to explain, I will need to get a little more technical (but not too much), so bear with me. To start, ask yourself a few clarifying questions:
- How does your EMPI work?
- EMPIs are based on methodologies that make them either probabilistic, deterministic, or a hybrid of both. Which methodology does your system utilize?
- Are based on a sophisticated matching method that utilizes various attributes (name, dob, gender, etc.)
- Assign a weight to those values based on uniqueness/randomness
- Add the weights of the attributes together to create a “score.” This score can then have levels of confidence in which either the system can automatically link records together above a certain score, create a task based on a range of score, or do nothing as the score is too low to link across records.
A misconception of EMPIs is that a “merge” occurs. This is not entirely true as a “merge” is designated by a single source system with the ability to “remove” a record (chart) and place that into another record (chart).
This “process” was normally a physical activity that occurred at a single site/facility and involved physically pulling multiple charts and combining those charts into a single chart and identifier. This would also involve combining the records electronically by making a record obsolete and keeping a record as permanent in some form or fashion, depending on the system involved. Once this occurs the system will normally create a “merge” transaction that is sent outbound to let other systems know what has happened (HL7).
When there are multiple participants in an HIE, and none of those systems are within the same jurisdiction, then a physical moving of a record (chart) to another record (chart) will never ever happen. That is the concept of a link, and is “soft” in nature, and could potentially change based on scoring, weights, etc.
Deterministic EMPIs: are:
- Based on an approach that evaluates different pieces of demographic information across multiple systems to look for exact matches. Think rules. An example:
- “If first name = first name” then do something. So if ROB matches ROB then it passes the rule
- If ROBERT = ROB then you did not pass the rule and as such you would not get the benefit of any matching.
- Very binary in nature and exactness is almost always needed in order to link records together.
The best of both worlds would be to use a combination of the two EMPI approaches in which you get a weighed score above a certain threshold, and then you can apply specific logic in the form of rules. I’d be happy to go into more detail on this with you so feel free to give me a call or drop a line.
Does it work for your HIE, and how well?
This is the most important question no matter which approach you choose, probabilistic, deterministic, or a hybrid of both. To determine if it is working for your organization and your HIE, you should be asking these additional questions:
- Can you configure auto-linking thresholds?
- Do you leave it and let it do its job or do you have a task list of items your staff needs to make decisions?
- Does your HIE/Organization have a data governance document/policy?
You must also understand your duplicate rate — your rate of tasks created along with false positives and false negatives.
For an HIE, its all about providing an aggregated or complete view of the patient record across multiple facilities. If you can’t successfully match them across facilities, then you aren’t providing the maximum benefit for your participants.
I tell my clients that the work is not done when the EMPI is implemented. It requires on-going feeding and care to make sure patients are being matched accurately across all systems and putting the right processes in place can help you achieve success in your organization and when connecting to an HIE.
If you have questions about your EMPI solution, you’ll find answers by listening to a recorded webcast version of the presentation I mentioned earlier: Best Practices in Joining an HIE: How to Make Your Connection Work for You.