Dave Goehrke:
Hi, everyone. It's Dave Goehrke here with you for another edition of our electronic health records podcast. During our inaugural discussion this year, we spoke with two executives about electronic health records, Dave Dorans from Clareto, a Munich Re Company, and Tim Morant, Chief Risk Assessment Officer at Munich Re. During our last podcast earlier this summer, we spoke with two underwriters about the actual usage of electronic health records in the market today.

Today we're going to talk about how Munich Re Life US is actually evaluating the electronic health records data. We'll be sharing the results of a very large pilot study, and also a cost benefit analysis that we performed based on that pilot study. With me today are June Quah and Murali Niverthi, both actuaries at Munich Re. June is joining us from Toronto and Murali is with us from Atlanta. Welcome both.

June Quah:
Hi, Dave. Nice to be here.

Murali Niverthi:
Hi, Dave. Excited to join today's chat.

Dave Goehrke:
Great. Thanks for coming. First, Murali, for the benefit of the listeners, tell us about your role at Munich Re.

Murali Niverthi:
So Dave, I'm an actuary and my background is primarily in experience studies, developing mortality assumptions in the traditional and accelerated underwriting space. Another area of interest for me is researching emerging trends, both medical and nonmedical, that impact life insurance underwriting.

Dave Goehrke:
And June, you're an actuary as well, and also a data scientist. What is your role at Munich?

June Quah:
That's right. I transitioned into a data science role about 15 years ago, and today I lead an analytic team. So my team is focused on developing risk selection models using both traditional and emerging data sources to improve insurance underwriting and pricing.

Dave Goehrke:
So June, we have this very large pilot study that you led. It's not what I think of when I think of a typical underwriting pilot. Can you tell us about the pilot? Tell us how it was structured.

June Quah:
Of course. The study involved partnering with several carriers to conduct a retrospective study on post-issue policies. So what our team did was we integrated with Clareto via API which then enabled us to submit case information and then to retrieve EHR documents from Clareto. So what makes our study different from past studies is that past studies tended to be small scale because it required a lot of manual effort to run those studies. Carriers would need to resource for the effort required such as typing in an applicant's information, waiting and checking for results, downloading the files, and then finally manually reviewing the traditional evidence and the EHR file side by side.

So what we did instead was we used technology to automate the EHR order and retrieve process, and then we worked very closely with our underwriting team on digital data extraction and validation. So what this meant is that we could conduct a much larger and more credible study, and then at the same time focus our underwriting team on reviews on cases where the digital data indicated review was needed. All in all, our study included over 1,000 cases with EHRs. These cases were a mix of fully underwritten and accelerated policies. Using digital data, our underwriters then narrowed down cases for extensive review.

Dave Goehrke:
It was a big pilot study. It was a lot of contributors to it as well. What were the goals of this pilot, June?

June Quah:
Yeah. We had two main goals for the study. The first was to measure metrics such as hit rates and turnaround time. Everyone is interested in those metrics because it allows them to evaluate how they would use EHRs in their program. And then the second was to assess the quality and usability of data to understand the value of EHRs in underwriting. So for this portion of the study, we ended up using a smaller subset of 200 cases that were recently underwritten from an AUW program.

Dave Goehrke:
Great. So in the end, what were the key takeaways from the pilot?

June Quah:
So hit rates were excellent. 65% of applicants were found and 46% had documents returned. So now most of this difference in applicants found versus those that had no documents were due to individuals that have a file at a healthcare facility, but they did not have a recent medical encounter. We were also very interested in special authorizations because this was an early challenge for EHR access. So in these situations, an additional contact point was needed to obtain the authorizations from the applicant. In our pilot, less than 5% of cases required these authorizations. Now this is clearly a testament to the strides that Clareto and others have made to improve EHR access for insurance.

In terms of turnaround time, we found that most documents were returned within one day. HIEs or health information exchanges were the fastest where documents were returned in minutes. And this is because they use an automated release workflow. Some facilities are still on manual workflows, and that can take somewhere between three and five days, but this was a minority in our pilot.

Dave Goehrke:
Okay, so those were some really good, high level results of the actual data, acquisition. We talked about the quality of the data as well. How did that look?

June Quah:
You beat me to the punchline. So in terms of data quality, our team was able to successfully extract underwriting relevant information. So our underwriters specified the type of medical information they were interested in. So this is the typical list of height, weight, BMI, blood pressure, diagnosis, procedures, lab results, medications, and smoking history, to name a few. And we were very happy to see that generally speaking, for any one type of data, about 80% of EHRs had this type of information and the data was very readily extractable.

Now the EHRs also contained other rich information such as clinical encounter notes and family history. All of these are also available digitally and can be processed for the underwriters to use. We did have one very interesting finding to note. While looking with the underwriting team, we drilled down further and looked at EHR data compared to applicant self-disclosed information. For example, using BMI alone, nearly 20% of cases had EHR BMI that would have landed them one class or worse compared to self-disclosed BMI.

Dave Goehrke:
Great. I'm an underwriter. I didn't say that earlier. But as you know, I'm an underwriter, so this is super exciting for me. I'm going to keep grilling you, June. What does this mean, in your mind, for the future of EHRs and life underwriting?

June Quah:
I think that's a lot. So in the last podcast we heard our underwriters talk about the value of EHRs and how they go beyond the traditional APS replacement use case. So now seeing the data from our pilot shows how this can be put in practice. There's a clear use case in detecting applicant misrepresentation in accelerated programs, especially with the availability of build, BMI and blood pressure in EHRs. So these measured values are not found in any of the other data sources that are used today, be it prescription records, medical billing, or clinical labs. So it's new data that we are seeing in EHRs.

And the fact that the data is digital means that they can be ingested and used directly in rules, engines and algorithms. This information is very rich, and in the pilot we've only scratched the surface of what is possible. The longitudinal data and clinical notes can help underwriters get a better picture of an applicant's health trajectory and how they are managing their health. So instead of one cholesterol reading, for example, there's now multiple cholesterol values over time, and they're all available digitally.

Dave Goehrke:
This is great. I love the concept of accelerated programs and net new data. Even if you had all of the typical accelerated tools at your fingertips, this is still going to be net new data. So it's a lot. What do you actually do with all of this data?

June Quah:
We're doing a lot with all this data. It is requiring a lot of resources and commitment, but at Munich Re, we are committed to building out solutions that will make this data easier to use in underwriting. As a start, we've grown our team of data science, engineering and underwriting experts who are actively working together. And we're working together to build tools and models to achieve this. As a re-insurer, our focus is underwriting and risk assessment for the insurance industry. So our solution is designed by underwriters for underwriters.

Dave Goehrke:
And that's the key. We talked about this during the first discussion with Dave Dorans, and Tim Morant why this all makes sense. And the last point you made June, really brings it all together.

Like I mentioned, Murali, these data sources really do kind of add up. There's a lot out there. We're committed at Munich Re to building out solutions. But one question our listeners might have is about the cost, the actual cost of using electronic health records. You recently compared, Murali, the cost of using EHRs versus mortality savings realized based on the EHRs data. Can you tell us a little bit about this?

Murali Niverthi:
Yeah. Thanks, Dave. So, that's the key question isn't it? So for the past few years, EHRs have got a lot of press. But what we tried to do is take it a step further. What happens when the rubber meets the road? When you have actual carrier data, how do EHRs perform? So Munich Re did cost-benefit analysis to really answer the question, do EHRs provide value for the money? And what we found was, yes, they do provide value and they're cost-effective for using accelerated underwriting, and they provide substantial value in mitigating the adverse impact of risk classification. So in a nutshell, the dollar savings in mortality or premium differences from risk class changes, however you look at it, that you get by adding EHRs to the accelerated underwriting process more than covers the procurement costs. To be a little more specific, if I'm talking from a numbers standpoint, for our sample of 200 cases, the mortality savings was about three times the procurement cost of $11,000. And what this translates to is per case it's a net benefit of $100.

Dave Goehrke:
Great. So we've talked about this 200 case study. You and June both, and I'm very familiar with it as well. And these were actual cases from one particular carrier. And we have these results that you just mentioned, Murali. But how do we ensure that if you looked at another set of cases or looked at the same data from multiple different perspectives, that there would still be this mortality savings impact?

Murali Niverthi:
So Dave, any time you evaluate a relatively new tool, especially something like the EHRs which has generated so much buzz, we need to be mindful of a couple of things. First of all, we need to be careful that we do not overstate the benefit. And secondly, do the results hold over a broad range of scenarios or other target populations? We need to keep both those things in mind.

So for the first one regarding calculating the mortality benefits, for instance, we use a tinier window for the calculation as opposed to the more typical 20-year term period. At the same time, if we're getting into the demographics, ours was a younger population. So 95% were below the age of 50. It was healthier in general. There were no declines in our sample compared to the industry average of about 2%. And the maximum face amount was $1 million. Why do I say all this? Because when you put these factors together, they have the impact of putting a lid on the mortality savings and not letting it get unduly inflated.

Moving onto the second point regarding generalizing these results over a wide range of scenarios, for this population of 200 cases, we generated a few scenarios. At one extreme which was an overly restrictive and conservative case, we assumed overlap between the information in the EHRs and other tools. This generated $25,000 of mortality savings. The other end of the spectrum was where we assumed an industry declined rate of 2% resulting in $125,000 of mortality savings. So clearly a pretty broad range there. But what it tells us is each case, the EHR was cost-effective.

And then we ran a study on an entirely different carrier sample. Less than 100 cases but with double-digit decline percents, and this resulted in $800 of net benefit per case. Why do I state all these numbers? Because when you put all of them together, it gives us comfort in the results, and gives us confidence that EHRs hold their value across a broad range of scenarios.

Dave Goehrke:
Great. So those numbers really speak for themselves, and that's great work looking at it from many different perspectives. Just from a pure, I guess, qualitative perspective what kind of conclusions would you make from this study?

Murali Niverthi:
So there was clear benefit to using EHRs in AUW - in accelerated underwriting programs, and they proved to be cost-effective even accounting for the presence of other tools. And just to add to the point which June made earlier, where EHRs stand out as they're often times the primary source of verifiable medical information about the applicant. So that would be the first conclusion. In addition, I would say that over time, the prevalence of fluidless and accelerated underwriting programs is increasing. With that, the ceiling for the value of EHRs is only going to get higher. As we speak, some carriers are raising their maximum face amount limits, some going beyond five million. And others are also relaxing the age limits. And the reason I'm emphasizing this is, in those situations, all it takes is a handful of cases for EHRs to be cost-effective.

And, finally I would end with, we saw that EHR, depending on the target population, we can get a range of a mortality savings estimate. So what does it mean for carriers? When carriers are thinking of using EHRs in their accelerated underwriting programs, they need to take into account the unique cost structure, the target market, and the applicant demographics so that it can help them scope out the sweet spot where EHRs are going to give them the biggest bang for the buck.

Dave Goehrke:
Great. And I should mention, by the way, that you published all of this and it's available on our electronic health records landing page. I think it was around mid-September that that was published. And, our listeners are welcome to read that and explore in detail everything that you just outlined for us.

Murali Niverthi:
Yes, I did. And would welcome any feedback there.

Dave Goehrke:
Great. So June, back to electronic health records. We talked about hit rates. Not 100%. That’s probably not realistic for any data source. But we hear that there is some perception that electronic health record hit rates aren't high enough. They need to be, they need to be really high to get the benefit. Definitely improving. But how can we reassure carriers that electronic health records today really are worth it?

June Quah:
Yeah. There's a misconception there that carriers would be paying just for making an order. With Clareto, a carrier's account can be set up such that there is no charge at all if no documents are returned. So that means carriers are only paying when they have received EHR documents. So what that means, or said another way, is you only need to see value on cases with EHRs to make it worth it. And as Murali has discussed earlier, this is very clearly the case.

Now the pilot has also showed that EHRs were returned on almost 50% of cases. This is an impressive increase from just one to two years ago. And as new health providers are continually added, the value will also continue to increase.

Dave Goehrke:
Great, great. And June, we talked briefly about solutions being built today by Munich Re, the work your team is doing right now. I'm really excited because this is really going to be a compilation of data science, actuarial work, the technology, and considerable input from our underwriting teams, as well. So beyond the mortality savings, there are considerable efficiencies that can be gained by maximizing the use of EHRs, and that's actually a topic that Murali will be developing in another cost-benefit analysis very, very soon, a research paper looking at the perspective of how can this data actually be used more efficiently than it's, in most cases, being used today?

June, Murali, thank you very much for joining us today.

June Quah:
Thanks for having us.

Murali Niverthi:
Thanks, Dave.

Dave Goehrke:
And thanks to all the listeners. If you'd like to learn more about our pilot studies and the solutions we're building, please reach out to any one of us. I'm really looking forward to another podcast soon that will feature the perspectives of direct writers. Thanks, everybody, for listening.