Fixing a broken system
When the Minnesota legislature established the state trauma system a decade ago, it mandated the development of a registry to which all designated trauma centers would submit data. The information would be used to track outcomes, provide feedback and create evidence-based quality improvement initiatives to improve outcomes for injured patients.
Since then, the state trauma system has grown from six designated hospitals to 123. Nine are Level I or Level II trauma centers and more than half are Level IV hospitals. That's a lot of data. But the state trauma registry, plagued with problems, hasn't worked out as planned.
David S. Morris, M.D., a trauma surgeon at Mayo Clinic's campus in Rochester, Minnesota, says some of the glitches are technical.
"Level I and Level II trauma centers already have established registries. Theoretically, we should be able to upload that information to the state registry, but we haven't been able to do that consistently and accurately. For instance, one field — length of stay in the ED — kept coming back as greater than 100 hours. We knew that wasn't correct, so there was some problem merging the data. We're working on that, and we've definitely made improvements, but there are still issues. And when you can't trust the data, it throws any data analysis into question."
Major woes for small centers
The challenges Level I and Level II centers face pale in comparison with the problems experienced by small hospitals, according to Debra L. Anderson, R.N., lead trauma registrar at Mayo Clinic's Rochester campus and architect of the hospital's trauma registry.
She explains: "Mayo has five full-time abstracters. At small hospitals, the data entry falls to the trauma coordinators or night nurse. The way the system is set up now, it's a painstaking, time-consuming and frustrating job. In order to have a valid and complete record, more than 150 fields must be filled out for each patient, many of which aren't applicable to critical access hospitals. Because so many of those fields have not been addressed, the system shows validity rates that are extremely low — around 20 percent, instead of the 80 percent we should be seeing. There needs to be some type of validation procedure, but there isn't anything like that right now."
Dr. Morris also feels the small hospitals' pain. "A lot of those hospitals are barely keeping up with the demand, and because the data isn't helpful so far, we're asking them to do something with no clear benefit; it's a burden with no upside," he says.
One solution is a stripped down, simplified data system with no more than 50 fields, so that's exactly what Dr. Morris and Anderson created. Anderson developed fields, that are specific to Level III and Level IV centers, such as patient vital signs, time from arrival to decision to transfer, and number and types of imaging tests performed.
"We've taken away a majority of the fields to get at the nitty-gritty; we want to make sure that what we have is valid before adding more data," Anderson says.
Anderson has tried the prototype and thinks it's simple and easy to use. She suspects it will take an inexperienced person about 20 minutes to enter the data for each patient — far less than the current system. The new data system was presented at the June meeting of the Southern Minnesota Regional Trauma Advisory Committee (SMRTAC).
"People are hungry for this information," Anderson says. "They want to see what they're doing right and how they can do better. This may not provide a lot of information, but it's a start."
Dr. Morris and Anderson's ingenious fix won't solve all the problems with the registry, however. Another challenge is the inability to track patients across the full spectrum of care because of privacy concerns and de-identification of data.
Dr. Morris notes that patient privacy is paramount to any health care data repository and is therefore the first concern in developing any sort of regional registry. Still, it creates difficulties.
"When a patient is injured, multiple episodes of care are generated," he explains. "The 911 call generates paperwork; EMS providers and critical access hospitals generate paperwork for their episode of care; then whatever happens in the emergency department and operating room at the Level I center generates paperwork. All these entities are operating in parallel silos, and each is doing its own performance improvement (PI) but there is no integration across the continuum of care.
"Yet for real and robust and meaningful PI on the regional level, you have to look at the whole incident from injury through posthospital care and rehabilitation. You need to access each entity's data, and right now we can't do that. The whole point of a trauma registry is to improve patient care, but without some tangible output, that's impossible."
Dr. Morris and others are trying to come up with a way around this barrier, too. Possible alternatives include data use and access agreements within the regional trauma council or the use of unique tracking numbers so data can be accessed in a de-identified way.
"There are a lot of strategies we can try," Dr. Morris says. "Some aren't favored by the state because they would affect state data acquisition, but everyone agrees the current method isn't working."