Q&A: NHIN will have role in disease management
 |
CMIO: Will the NHIN framework enable disease management through better quality data?
We’ve been managing various diseases for 15-20 years now using healthcare claims data to do patient identification and perform case management interventions. In the past, providers focused on disease management have been dependent on using claims data to identify patients for inclusion in disease management programs.
There are a lot of known limitations on data quality from claims and because of the paucity of the data, there ends up being a lot of false-positives and missed cases.
For example, if you’re looking for patients who have multiple sclerosis (MS), you’re look for patients who have filled prescriptions for Avonex supporting an ICD-9 code for MS, from claims data. A lot of times, the people you’re doing this for are payors--which have access to these data. I hope we’re starting to see more of the specialty associations trying to get involved, because they have a lot of clinical expertise to apply. In the future, as the NHIN matures, we will see clinical experts developing patient identification protocols that can work with the clinical data along with the provider and independent of the financial data processes.
With the new framework, the overall architecture will enable the provess because everyone involved will benefit.
CMIO: Are there other drawbacks to using claims data?
When we were using claims data as our only source—and it was the only available source—it had a lag time. Before the data were aviable, it had to be captured, it had to be transcribed and submitted to the payor, and the payor had to get it to the point where it was clean enough for them to extract from their system and send it to us.
I’m looking for this patient who needs some case management, and there’s a 90-day lag already. I’ve got my own workflow once I’ve written this algorithm that finds them, then my nursing staff has to call these doctors on false positives. It could take 150 days go by before we identify someone who could really benefit from case management.
That’s where we were [in the 1990s], with this sort of data that was difficult to interpret – we’d end up with false-positives, or had to have some complicated algorithms so the cost of identifying patients was pretty high, and then the actual costs of getting to the point where we were intervening with somebody was pretty high.
CMIO: How will the NHIN framework improve this process?
When we have the NHIN framework in place [in 2015 and beyond], we’ll be able to rapidly do not just disease management and patient registries, but start getting to near-real-time outcomes management; get closer to the point where we’re able to say, for example, white males age 40-45 who have congestive heart failure respond better to a particular treatment.
There are two differences that this evolution in health IT (in relation to disease management) will enable: One is to get away from claims data, and get closer to standardized content directly from the EHR, which is more clinical in nature, so we’re not only identifying the right patients for certain treatments, but also those patients that might be at higher risk.
We can do population science [with] unsupervised algorithms if the data is properly de-identified, and see what unusual things are happening. We can get beyond where we are today, with a lot of controlled trials on a very specific population, and be able to explode it out to a whole population, to where it really matters.
If we get to 40 to 50 percent EHR adoption, where we’re getting data in near-real time, we’re going to be able to do a lot more signal detection at the population level and at the particular intervention level than we’ve ever been able to do before.
CMIO: What privacy measures are necessary for this to go forward on a nationwide scale?
The consensus that I’ve found from a lot of my peers and a lot of patients and so forth, [is] that an informed consent [approach is necessary]. ‘I am going to use your data, this is how it will be be used. Are you comfortable with that?’ It’s a layered approach. Hopefully, we’ll get to the point where patients [expect a provider to be sharing their data and say], ‘What do you mean you’re not doing that?’
We’re not just trying to solve new problems, we have to change embedded behaviors. There’s a certain way we see our doctors, and certain expectations that we have in how our data is treated today. And we have to go through the cultural and educational process with the appropriate privacy measures in place. We need to have smarter tools that allow people to configure their preferences.