When it comes to creating consistently rich healthcare experiences, integrating the elements of modern data science across an organization may not be easy, but it is most certainly the best way to quickly adapt data-driven innovations that will help improve care, cut costs, and improve healthcare overall.
Yet, while many other industries are rapidly adapting their business model to adhere to that of a data-driven one, healthcare seems to be stuck in the past.
That’s why last week, we held a webinar with Craig Kartchner, AVP of Marketing at HonorHealth and Angelique Russell, Healthcare Data Scientist formerly of City of Hope, to unpack some of the hurdles, frustrations, and solutions that come with integrating data science into the industry.
If you missed the live session or would like to give it another listen, access the full recording here. This post summarizes some of the key learnings from the session.
Customer-centricity is at the forefront of healthcare business strategies — or at least it should be…
“Customer” isn’t a dirty word in healthcare, says Kartchner. Thinking of patients as customers helps transition the traditional mindset to true patient centered care, according to the Institute for Healthcare Excellence.
Fully adopting this mentality not only asks the healthcare business model to shift, but requires the doctor to entirely rethink their relationship with his or her patients, which is not necessarily a bad thing. Metrics such as patient loyalty, satisfaction, and trust will benefit from a new approach.
To some, calling a patient a customer is “disgusting” or “not appropriate” according to Kartchner.
Yet, a traditional relationship with a more cut and dry industry such as retail, for example, offers frequently used customer feedback channels and other market signals help keep it flexible to adjust to fit the customer’s needs over time. This is a major opportunity healthcare is missing out on.
This craving for choice relates primarily to clinician factors, such as inconvenient office hours or locations, long waiting times, personal characteristics or qualities of the provider, and/or insufficient communication time between the patient and clinician, according to the National Institutes of Health (NIH).
“A retailer is trying to target as effectively as possible, target their customers with the best message, at the right time, and in the best manner to inform the customer, and get them to take action,” says Kartchner.
With the vastly complex infrastructure in healthcare, the volume of players further complicates what is otherwise considered table stakes in other industries.
Seeing eye-to-eye across departments — from marketing and clinical practitioners to billing, insurance companies, care coordinators, front office staff, and a myriad of other functions — getting those essential market signals and customer feedback is tough.
“One of the basic tenets of marketing is to differentiate yourself,” says Kartchner. “Companies want to differentiate themselves from their competition to allure clients and to allure business. In healthcare, you can’t very well do that on costs or quality so most health systems are trying to differentiate themselves on experience.”
This too, has resulted in a saturated playing field with health systems positioning themselves and marketing their core differentiating factor as providers of personalized care. The gold rush toward offering personalized care is causing the words to become devalued despite the best of intentions to follow through.
“You look at other companies like American Express, Nordstrom, Marriott, Amazon, these are good examples because they have good reputations and generally provide good customer experience,” says Kartchner. “They know exactly what your customer lifetime value is, and they will spend up to exactly that amount to either get you as customer or retain you as a customer.”
Kartchner outlined some basic steps the healthcare industry can adopt in order to get the most out of their customer’s (or patient’s) intel:
Step one: Find out what matters to your customers. Look at social channels, take data inputs from elsewhere, surveys and studies (both secondary and primary) to understand what your customers want.
Step two: Change operations. Think about revising some things like online scheduling, telehealth operations, convenient hours and locations, easy wayfinding, and appropriate staffing.
Step three: Decide what matters to you. Make it simple and approachable. Are you focusing on revenue? Patient volume? Do you want to improve HCAHPS scores? Raise your star ratings so you have higher bonuses? Increase the customer experience and satisfaction?
Step four: Get buy-in from senior management, but also the rest of the health system from the front line employees. This has to be centrally driven throughout the system.
Step five: Inform everyone what the next best action is for each customer (patient). You’ve got to get down to the front office staff to the call center employee to the marketing person who is sending out an email to the nurse or physician assistant who’s interacting with patients to the billing department — everyone needs to know what the next best action is. This requires integrations with a customer relationship management systems (CRM).
“That customer relationship management machine has to receive the data from all these different entities, the marketing departments, and the doctor’s office,” says Kartchner. “It doesn’t matter the channel or medium you’re using, whether it’s text message, or mail, or phone — it all needs to route through that central CRM database.”
From collection to action: how data is being leveraged by healthcare data scientists
Health system’s data often operates in database silos, according to Russell.
“Just on our own campus, we have many different databases that contain clinical data,” says Russell. “But we’re also just an ancillary provider to a patient’s primary medical system and their primary care provider.”
From patient portals, pharmacy databases, and lab portals, data lives in many different parts of the health system. Due to a lack of data centralization, the associated challenges are interesting, says Russell.
One of the most pressing challenges is knowing which database or which silo to depend on and therefore make decisions from.
Exchanging the necessary vital patient information to make sound decisions from depends solely on interoperability, which is yet another top challenge.
“One of the limitations is we still lack very consistent standards,” says Russell. “One of the reasons why we need data standards is because different vendors and different organizations are using different terminology to refer to the same medical condition or the same test, and making sure that we can share that information in a way that can be received by both organizations is really complicated.”
In light of these universally felt exchange-based challenges, clinicians are finding creative ways to leverage data to improve care by reading the patient’s progress notes — data that is otherwise overlooked in data transmissions and not included in daily labs or patient orders, according to Russell.
“Thanks to natural language processing( NLP), we have begun to work on how to identify patients who have sepsis, and we’ve found that the most reliable approach was leveraging the information and the physician notes,” says Russell. “If a physician is diagnosing a patient with sepsis, there’s going to be mention of that. If you can then extract that information from where the physician is documenting it, that’s most effective.”
If you want to leverage data science at your healthcare organization, Russell says it’s worth taking a pause and figuring out what you can do with the data that you currently have.
“As people understand that more and more, the focus is going to shift a little bit for data-driven innovation,” says Russell. “To that end, healthcare organizations really need to up their IT staff to optimize for the next phase.”
In our live session, Russell discussed another way organizations tapping into data by adopting early warning systems while other smaller systems such as community health facilities lack the resources needed to implement clinical decision support solutions.
Flagging and documenting early warning signals is one part of the very complicated and layered process, but breaking down how to then communicate what the data is saying to the clinician remains a tall hurdle that data scientists like Russell wrangle with day in and day out.
This is where accurate data visualization and the dire need for experts with such skillsets come into the conversation.
Current teams inside health systems should be driving analytics and data science, a skillset that requires a very different proficiency than maintaining clinical applications, report writing, and dashboard building.
“Healthcare really does have a long way to go not just in the standards arena, not just in interoperability, but in getting organizations ready to do data science,” says Russell.
The challenges associated with switching toward a data science business model in healthcare and throughout the national health system is no easy feat. To learn more about the challenges and solutions discussed during the webinar session, download the full recording.
“Other organizations in other industries are way further than we are in healthcare,” says Russell. “It’s a little silly I can log in to Amazon and it knows what I’m likely shopping for. But if I go to my hospital, they’re going to struggle with what could possibly be my medical issue. We really need to get this technology in place so that we can take care of our patients.”