Q&A: It takes a map-test-tweak cycle to integrate device data
 |
Where are you in the integration process now?
We’re currently integrating ventilator and hemodynamic monitors. We plan to add telemetry and IV pumps; we’re in preliminary stages of that as well, and anesthesia. Our EMR has an application built for anesthesia integration that we are considering. But for now, just the hemodynamic monitors and pumps. We’re integrating 165 beds for monitors and 77 for vents, and [we’ll have a] total of 305 beds, come April, downloading vent data, monitor data or both.
Currently we allow [the critical care unit, emergency room, labor and delivery, neonatal intensive care unit, endoscopy, observations unit, cardiac cath, pain management clinic and post-anesthesia care unit] to download monitor data. Only the critical care unit and neonatal intensive care unit are downloading vent data right now. On Dec. 2, 2010, we upgraded to an Epic EMR system and added quite a few departments.
Can you talk about the challenges of integrating data from those devices?
There was a variety we had to deal with and overcome. One of the things is synchronizing time stamps among these devices. We have a Philips network of servers which is an independent network to support our hemodynamic monitoring system. [Each] Philips monitor time-syncs with the server it’s connected to. When we went live, we found that two of those servers were no longer time-synching correctly, so we had monitors that were displaying different times. We have the Capsule Technologies DataCaptor server receiving all of this [data], which also has to be time-synched. Also, we have our Epic EMR server, so making sure they’re all speaking the same time is very important when you’re dealing with device data.
If the devices sending the data are not properly time synched, it can be a problem particularly when administering medications in a clinical environment [such as the NICU], where you’re tracking very precise data.
We also had some things that were challenges based on Epic EMR, the way it handles device data and displays it in flow sheets—there’s a look-back and look-forward feature that will only display times in a column for given intervals, and we had to do some fine tuning with this.
How did you handle those issues?
You really need to do a thorough job of unit-testing all of your bed locations, and all of your device variables. Each vent has its own flavor of variable and you need to be very careful and organized in how you map them.
We build each [data integration] variable in the EMR and each bed location in the EMR, we call them device records. You also have to build a mapping based on the node IDs for each bed that has a particular Philips monitor and data capture. Therefore, using a really good naming convention is extraordinarily helpful. You want to have the whole team agree on a convention so there’s a logical labeling in each device because you’re going to be naming things in each environment. In the EMR, as you’re adding variables, you want to use the labels that your device uses.
After our initial go-live [in 2009], there were three or four variables we had to go back and test and figure out why they weren’t crossing, and get them tweaked. Having actual messages is key. If you can get messages flowing using a simulator of some sort, then you can actually see what the device is sending out and it makes your work a lot easier.
Has the hospital seen benefits from this integration project?
Nurses are able to document many patients in one centralized station without disturbing them in their rooms, and [the system is] a lot less error-prone because the data is validated manually. There’s a pending data area where the nurse goes in and selects the data she wants and validates to make sure it’s correct. It’s far easier because they’re not keying in values.
Are you planning other device/data integration?
Integrating our telemetry data would really help nurses and tele-techs. Right now they’re spending a lot of time pulling out strips and pasting them in. Our health information management team is spending a lot of time scanning these and filing them in the EMRs. It’s a lot of manual work, it’s error-prone and it could be handled a lot more elegantly.