Sramana Mitra: If the interoperability is within the hospital data and diverse proprietary systems, it is easier. When it comes to outside the bounds of that particular hospital system, what kind of issues do you face?
We are part of the Stanford hospital system. For certain procedures, our family also goes to UCSF. UCSF and Stanford have some level of interoperability, but there is also competitiveness, mistrust, and frictions that we noticed.
If you could speak on what is happening and how that tension surfaces in your work, that would be great.
Drew Ivan: You identified one of the biggest differences between communication within a hospital and communication across hospitals. Since they are separate organizations, they may have competitive reasons for are not being motivated to share the information.
That is changing now because there are federal rules that prevent information blocking. That is finally being addressed, but that does not change the fact that there are still reasons why organizations prefer not to cooperate.
To take care of patients, they often do need to transfer that information. You then run into a few different issues that you typically don’t have within a single hospital system. For example, the level of security may be different when you are communicating over the open network from one hospital to another compared to when you are communicating within a private network within a single hospital.
One of the other big problems that we have when we start communication across the organization is that each organization has a different master file record for that patient. They have different identifying numbers. They are not going to have the same medical record number at both hospitals. They might not even have the same name for the patient.
The patient may have registered under a slightly different name with one institution versus another, so you cannot necessarily use the name for matching. Not to mention the fact that there could be problems with data entry. You could get the address wrong because they could have moved and yet still get the same person.
Patient matching across the organization can be difficult. When you need to transmit one record from one hospital to another and make sure it lands in the right patient chart, even something as basic as that becomes a challenge as you try to cooperate across different organizations.
Sramana Mitra: How do you resolve these kinds of issues of identifiers? How do you reconcile that it is the same patient? Is it with social security numbers?
Drew Ivan: There are different approaches. The one that is commonly used is that the system that is receiving the record will look at the data that has been supplied and try to match it to a patient that they already know. They will use a combination of name, birthdate, address, and telephone number.
They shouldn’t use the social security number because in a lot of cases they shouldn’t even be recording that. It is not meant to be used as a unique identifier. What we found is that the telephone number is one of the better identifiers.
What they will do is they would have a system called a master patient index that uses an algorithm to check all these different fields. If it’s the perfect match, then they are pretty sure that it is the same person.
Usually, it is not a perfect match, so the algorithm determines how close the match is. If it is over a certain threshold, then it matches the patient. If it is under a certain threshold, then it’s a different patient.