PARTNER WITH US

Technology

Why CVS Health, Highmark, Verily and Clarify Health execs are ‘bullish’ about the future of analytics

With analytics tools advancing at a rapid pace, what do organizations at the beginning of the journey have to know to be successful? Executives at four leading health care companies discuss analytics opportunities and challenges.

Gabriel Perna | January 27, 2021

Coming from the financial industry five years ago, Richard Clarke, Chief Analytics Officer at Highmark Health, said he was in for a big surprise regarding the pace of analytics adoption in the health care industry. 

“I was naïve to the complexities of health data and thought we could go a little faster. Five years later, I see the amount of information that’s available and not being used today as a real opportunity,” said Clarke. In some regard, he said the industry is already past the tipping point. “COVID accelerated things with a massive move to digital solutions. No one knows how it will play out with how analytics are embedded into decision making and financial of care delivery, but we’re on the right trajectory.” 

Clarke’s bullishness on the potential of analytics in health care was echoed by other panelists at a recent Health Evolution industry solution webcast, The Future of Health Care Analytics. He was joined by Sree Chaguturu, MD, Chief Medical Officer, CVS Caremark, Jessica Mega, MD, Chief Medical & Scientific Officer, Verily Life Sciences and host of the discussion, Jean Drouin, MD, CEO, Clarify Health. 

“My bullishness comes from the fact we are going to ask ourselves questions we were not able to ask before. We will get answers to those questions at a speed we were not able to before. Don’t be afraid to ask those questions now because we may find we have the tools to answer them in ways we didn’t before,” Drouin said.  

Chaguturu agreed and said that more than ever before, health care organizations have the ability to aggregate and gather disparate types of patient data, develop insights and create actions based off those insights. Like the others, Mega said her bullish-ness comes from the fact there is an “exploding amount of relevant health data” and a proliferation of tech tools that can process this data into insights.  

“The ability to access much broader cloud computing capabilities, where we’re able to securely and safely share information and work together as a data science community, to make sense of this information and turn it into insights, it makes this a true inflection point. You want to avoid hype, but I truly think there is something different about this moment,” says Mega.  

Use cases 

At CVS Health, the organization is able to aggregate its pharmacy claims, demographics, benefit structure, RX claims, medical claims, and preventive screening datasets with external sources, such as consumer behavior, social determinants of health, and behavioral health risk factor. A staff of 700-plus data scientists and 100-plus PhDs have taken all that data, and to date, developed 300-plus machine learning models.  

“We’re trying to [create models that will] recognize the actions we’re trying to do, let’s say it’s an increase of flu vaccination rates. The reason that individuals actually get a flu vaccine varies by their individual desires. For one individual, it may be a value story, ‘I want to know there is no co-pay.’ For someone else it may be about convenience,” Chaguturu said. 

There are also variations in how people prefer to get messages and what time they prefer to get messages. “The end result is we’re increasing flu vaccination rates, but how you get there is about the aggregation of data we have, plus external datasets, member archetyping and then crafting consumer-based messaging that meets members where they are in their care journey,” he added. 

At Highmark Health, Clarke says the integrated delivery network is focused on using analytics to identify the right intervention for patients, as well as discovering the right site of care to engage them. He says when the tools are able to recommend an intervention at the appropriate site of care, there is a three to four times increase in patient uptake vs. just using a traditional telephonic outreach.  

Verily Life Sciences has spent the last few years working on care solutions for chronic disease management, said Mega. She says the company has used some of the lessons and principles in developing those tools to create the “Healthy at Work” and “Healthy at School” initiatives, creating modeling algorithms that help organizations operationalize their plans to return to work and campus.  

“What is most relevant is understanding what health condition you’re working on and then creating the solution. In the case with chronic disease management, what we’ve seen is when we tailor therapies and use the relevant information for an individual, we have significant reductions in Hemoglobin A1c,” Mega says. “This theme of embedding analytics within the organization, it is no longer separate vertical, it really should be the heart and soul of everything that we do.”  

Data integrity and other challenges 

Clarke said Highmark Health has seen rapid adoption of analytics tools in operational areas, but clinical settings present different challenges. The burden of proof in a clinical care setting is incredibly high and they require data scientists to be deeply embedded with those clinicians. Coming from the finance industry, Clarke noted that the burden is much higher in health care.  

“Anyone we hire, I let them know they’re going to spend a lot of time working side by side with clinicians, working through issues and questions they might have. The data are just incredibly complicated,” Clarke said. “The pairing of subject matter expertise and data technician is frequently necessary in health care and will be for a while.”  

Moreover, Mega has seen how data integrity challenges can play out if the technicians don’t work with a broad group of clinicians to create a model that will be integrated into the point of care. She said it’s important to ensure data is generalizable. “If it’s not, there are real pitfalls when it’s translated into a clinical environment,” she said.  

One thing that will help advance analytics usage is data interoperability, Mega said, as it will enable clinicians to better understand the longitudinal journey of the patient. “It will really empower people to be able to make the best choices over time and allow clinicians to offer the best care. A lot of work is going on in that area, it’s ambitious, and takes a lot of infrastructure, but that would be beneficial,” Mega said.  

The execs also said another potential challenge when integrating analytics is ensuring algorithms don’t increase implicit biases in how care is delivered. Several research studies have shown that algorithms can actually exacerbate existing disparities of care.  

“As we think about the advent of machine learning and artificial intelligence, we recognize that it can amplify a lot of the disparities that we see in the upfront data collection. So, as we’re creating these checks and balances with the data, we have to have this additional check and balance to make sure we don’t perpetuate disparities or health inequities,” said Chaguturu.  

Getting started  

As Drouin noted, many health systems, payers and life science companies don’t have the wherewithal and resources of CVS Health, Highmark Health, and Verily Life Sciences to create a substantial enterprise analytics program that’s embedded within all facets of the organization. So how should these organizations begin? 

“We have found the best engineering work was done in-house because the data are complex and there’s a lot of subject matter expertise that needs to be brought in,” advised Clarke. “Quite frankly, the best way to get started is to be use-case driven, using dedicated pods of analytics teams that sit embedded right next to the areas of the business they are helping. And then have a super clear mandate for that team where it can develop modular parts that be re-applied to other use cases. Otherwise, you’ll have 1,000 disparate things that you need to manage.”  

Mega noted that finding the appropriate technology tools is less of the challenge and the important thing is to understand what problem the organization is trying to address. “The magic starts to happen when you see a problem that really needs to be solved and you understand that problem and you understand the data you need to solve it,” she said. 

Drouin agreed with this notion and said successful transformations depend on having the knowledge, the supporting processes, and the right mindset. Chaguturu said it’s imperative to invest in people to evaluate the strategy against the tools you’re using. Upskilling and workforce development is critical, he added, because the tools are advancing faster than the ability to educate people.  

“It feels like before we were moving in one-to-two-year increments in advancements happening and now it’s happening every few months,” Chaguturu said. 

Watch “The future of health care analytics” webcast here:

About the Author

Gabriel Perna, Senior Manager, Digital Content

Gabriel Perna is the Senior Manager of Digital Content at Health Evolution. He brings 10+ years of experience in covering the intersection of health care and business. Previously, he was at Chief Executive, Physicians Practice and Healthcare Informatics. You can reach him via email or on Twitter at @GabrielSPerna