Taking our data to the next level in 2015: Q & A with Keith Chew

As senior vice president of Integrated Radiology Partners, (IRP) and also president of the Radiology Business Management Association, Keith Chew is a well-respected leader in our industry. He recently spent time with RadAnalytics to talk about his new role with IRP, the importance of applying analytics in radiology, and share his thoughts on radiology’s outlook for 2015.

RadAnalytics: Can you tell us a little about why you made the move to Integrated Radiology Partners?

Chew: I started with IRP October 1st, 2014, and one of the things that actually enticed me to make the move was IRP’s strategic vision on the impact of analytics for the future, especially within imaging, but within all of medical practice. That’s a concept to which I also ascribe. I think that imaging is right now at a bit of a crossroads, where we could truly become a commodity if we are not careful. Radiologists need to revert in some ways back to the truly consultative, valued member of the patient care team role that they were a number of years ago. To redevelop into that type of a position, it’s going to take analytics to demonstrate the value that medical imaging brings to the patient care continuum.

RadAnalytics: How can radiologists best accomplish that?

Chew: Well, I think it has got to follow more or less a step wise progression. First, you’ve got to be able to collect the data, start recognizing the patterns within the data, and then start understanding what those patterns are showing you. It’s a progression in the thought process.

To start with, radiology could begin by collecting certain types of superficial data reporting sets. We need to start reporting some very general, straightforward metrics now, such as physician productivity by work RVU, by total procedures, or by peer review activities.  

There’s a lot of information out there, but I don’t know that there is any large universe, a large n, or a large number of radiologists that are being set as the base for those studies. Instead, what we’re seeing is the study that comes out of one facility, reflecting one set workflow. Because there are workflow variances between all the different studies, you need to bring a very large number together to represent your statistical universe, the n within statistical analysis that will allow you to better develop a meaningful output. That’s your first step.

The next thing is to look at codified data within the system; whether it’s the CPT code, the ICD-9 or soon-to-be ICD-10 codes, and anything else you can actually find that puts a specific number to a piece of data. You can start looking at activities related to episodes of care, whether they are acute or chronic. How many X-rays does a normal patient in an episode of care for a hip replacement receive? We’ve got government data showing a standard based upon a very large n sampling and you can use data from your own facility to see how well you are adhering to that standard. That’s one good example there.

Then you move it into the next realm, which is the non-codified data, the data that is pulled from reports and notes within the EMRs that are really based upon an analysis of general syntax. From that data you can begin to do not only prospective but retrospective reviews to find out if the diagnostic and therapeutic imaging services you’re providing were provided at the right time. Did it have a positive outcome or did they have a positive impact on the patient’s overall outcome? 

It just continues to build from that very low level superficial-type analysis all the way up through the metadata development into the non-codified data that allows you to say we can actually now look at the whole concept of medical imaging in many terms just like imaging 3.0 is from the ACR, making certain that you get the right imaging at the right time, interpreted by the right radiologists. It allows through this analysis to get you to that point, but without the analytics, how do you get there? That to me is the big question, so that’s why I’m so strong behind analytics and that’s why I think that IRP is on the right path.

RadAnalytics: How do you feel that radiology can demonstrate value by either improving quality or decreasing cost in the particular care of a patient? 

Chew: As you’re able to collect greater data and utilize that data to generate information and demonstrate patterns, you’re going to understand better how that quality side of the formula is impacted. Part of the problem with medicine is that a lot of this is the personal medical decisions of the physician treating the patient. Physicians rely on studies to help inform clinical decisions in determining the best diagnostic or therapeutic approach.

What I don’t think there is, is a timing for the delivery of the best diagnostic or therapeutic approach.  Let’s consider a very old example when I was working with multispecialty groups in the HMO era, we had patients presenting with lower back pain. Initially, the standard approach back then from a diagnostic perspective was to do a myelogram. Well, myelograms are painful and expensive, but we were finding that orthopedic surgeons wouldn’t move forward with their diagnostic approach on the patient until an MRI was done—MRI was a new diagnostic technology at the time – definitely telling my age here. 

So we would do a myelogram, we would note that there was some type of a disc problem or what have you, send them to the orthopedist and the orthopedist wanted to do an MR. Our group rapidly decided guess what? We don’t need to do the myelogram. If we just do the MR to start with it’s a higher quality experience for the patient because they don’t have to go through the pain and discomfort of the myelogram, but it’s also a more cost effective approach because we’ve now just eliminated one imaging exam that was not going to add to the diagnostic quality for that particular patient. 

That was something we did ages ago, but that was based just upon our small amount of data and our desire to control costs within that HMO environment. In today’s medical environment we may be able to find new opportunities to use this type of analytics approach to discern whether or not the performance of a certain test is definitive or not. If it’s definitive then it should be maintained. If it’s not definitive, then why are we doing it in the first place? 

It may be because right now that’s the standard.  In my mind, the data and the analytics will help us develop more appropriate standards, and the standards may vary by region. Imaging has the potential to play a huge role in being able to bring that data to the forefront to improve the quality of the services that are offered. That’s why I think this whole analytics approach is so imperative to the value-based approach of healthcare. 

RadAnalytics: More and more, radiology practices are coming together to work cooperatively and share data. How can they use their network to their own business advantage, as well as to inform population health on a larger scale?          

Chew: By being able to pull groups together within a specific region you can start to look at the region from a population health perspective, then you can also pull regions together for a national perspective.

There are multiple tiers of information you can pull from the data. We may find that on a national basis certain patterns develop, but on a regional basis those patterns may vary slightly. If you look at the Duke study, the Duke Atlas, it’s got everything in there. If I look at the populations in different parts of the country, I would not expect healthcare in New England to be exactly the same as healthcare in Louisiana or healthcare in Washington or Oregon, the Pacific Northwest. 

Recently, I had a discussion about what happens when a patient presents in an ED with lower back pain.  If you’re in the stone belt of the United States, it may be appropriate to do either an ultrasound or an MR sooner because of the prevalence of kidney stones in the region. I’m not a radiologist, but data may show that in certain regions a diagnostic study is done much earlier for somebody presenting with certain types of low back or flank pain in the stone belt than you would in other parts of the country. 

When radiologists come together on a regional basis and share data, they can start extracting the information from that data and looking at the patterns. When the data from those regions are aggregated, you can see if that changes the patterns that you’re seeing and then you can expand it out to a national basis. You may actually find that you do have variances of the patterns that emerge based on geographic areas. What you have to be able to do is understand that the practice of medicine may vary based upon the patterns you see emerging.

To me it really just becomes a big project, but it becomes an interesting challenge because we’re trying to understand population health at a different level.

RadAnalytics: Considering the impacts of healthcare reform, specifically the shift of a lot of the financial burden of healthcare to the patients, how can radiologists use analytics to manage their businesses to mitigate the negative financial impact to their hospital relationships and revenues?

Chew: Well, as we move into some of the new payment models, such as bundled payments as a perfect example, they are based upon a standard set of utilizations. If a hip replacement typically has four X-rays from the initiation to the completion of the episode of care, and as a radiology group, you’re tracking that utilization data, you can find out that one doctor may ordinarily order three, another eight, and others 12 or 14, and on from there.

From that data, you’re able to go back to the system because remember, not only are the radiologists concerned about this, but the system will be as well. A radiologist reads a study for which they get no reimbursement and they’ve only lost their time. A system provides a service for which they get no reimbursement; they actually lose money. There’s a cost to the technical delivery of the service.

If you’ve got individuals in a bundled payment scenario who are supposed to provide four hip X-rays and are consistently requesting more than that, they’re not appropriately using the resources based upon the payment model that’s been established. Being able to bring that information to light at the hospital level will be very beneficial to the hospital being able to be more profitable. There’s an opportunity to educate those physicians and investigate why they requested additional X-rays. Did an infection develop? Was there some other type of complication that developed? If so, maybe that becomes an outlier and does not qualify for the bundled payment approach.

Imaging working cooperatively with the hospital in this way would make certain that everybody has the most information available to them and understand what the risks are in accepting contracts under some of these alternate payment mechanisms. When you understand the risks you understand what type of incentive needs to be built into the process so it allows you to actually negotiate a stronger contract that will not wind up bankrupting your facility or the payers.

RadAnalytics: Looking forward to 2015, what do you think are the biggest business challenges for radiology and the biggest opportunities for radiology for 2015?

Chew: I think the biggest challenge and opportunity, at least in my mind right now, is the development and application of analytics. I see it as a challenge because a lot of groups will not see the benefit that can be brought to bear for their particular situation by applying and utilizing analytics, but I also see that those groups that foresee the advantages that analytics can bring to them, I see a great opportunity for those organizations. 

Finding ways to team with other like practices within your specialty to be able to start building the data that is necessary to get the information and demonstrate patterns, is going to be a huge challenge, but I also see that that’s going to represent a huge opportunity as we move forward.