When choosing infrastructure to support clinical analytics, many healthcare organizations use a monolithic and siloed approach, which can hinder implementation and limit opportunities to scale up and gain efficiency, according to John Reeves, solutions development manager of the Data Center for Analytics and Security at Lenovo Health. He believes organizations must view — and implement — clinical analytics systems as strategic assets allowing for greater economies of scale.
This content is sponsored by Lenovo Health
Mr. Reeves joined David Reis, PhD, CIO of Burlington, Mass.-based Lahey Health, and Jennifer Esposito, worldwide general manager of health and life sciences at Intel Corp., to share best practices for implementing infrastructure to support a real-time clinical analytics program.
Experts discussed the benefits clinical analytics systems can offer to both patients and healthcare organizations and shared advice for health systems looking to implement the technology.
Note: Responses have been lightly edited for length and clarity.
1. What key technology infrastructure components does a system need to support a real-time clinical analytics program?
Dr. David Reis: For organizations with a full-suite EHR, a majority of the relevant data needed for real-time clinical analytics is centrally located in the system. Along with a documented and well-understood data model, a real-time clinical analytics computing environment also needs very fast processing speeds, which are supported by large amounts of system memory and disk storage that is as fast as possible. These components enable the real-time part of clinical analytics, which knows about the patient’s current status and takes that into account during the analysis and prediction modeling.
John Reeves: The main component is a general purpose, high-performing computing platform that accounts for the depth and types of data being processed. In a high-performing computing environment where you’re bringing together different elements required for patient care, you may have traditional column and row data, along with imaging analytics data, data from an external project or mixed data from other modalities. So you have to look at infrastructure as a fabric. You want to be able to use and reuse, so accounting for scheduling and cloud deployment models is important. There should also be multiple use cases embedded in its architecture. Development of the analytics program cannot be solely driven by cost. For it to become part of the health system’s culture, it must be driven by the clinical perspective.
Jennifer Esposito: We often see health systems discuss very narrow or individual use cases under the guise of a technology infrastructure or clinical analytics program. These programs are often initiated by specific departments or lines of business. The discussions lack broader planning for scale up or how they can leverage the same data for multiple uses throughout the overall health system. These types of pilot activities are more likely to remain pilots because they do not align to bigger technology infrastructure planning and do not leverage an overarching strategic discussion. If health systems are seriously considering implementing a clinical analytics program on a broader scale, they have to pull together a multidisciplinary team, understand what individual stakeholders are looking for and identify how they can intersect with the health system’s overarching strategic imperatives.
2. How important is it for healthcare organizations to consider mobility, security and cloud when choosing the best technology infrastructure to support a clinical decision support system?
DR: It’s vital to the success of the clinical decision support system. Healthcare organizations must consider several factors, including technology and funding, when deciding between on-site or cloud infrastructure. Cloud technologies can often provide greater scalability, accessibility and security than on-site infrastructure, but the trade-off is funding. Typically, on-site infrastructure is owned by the organization, and they can capitalize the acquisition and setup cost. Conversely, the acquisition and setup of cloud services is typically funded out of operating expenses, which can be difficult to afford.
JR: These things can’t be an afterthought. We’re hearing more about cloud-first design, where organizations think of security in front of the conversation rather than in the back. There are many ways of addressing provisioning and cloud computing, which need to be better understood in a security framework. It’s important to understand where the data is needed; in what environments of care or settings outside of the hospital. This consideration will inform internal processes that should be set around addressing data access in concert with primary stakeholders — it’s not just a technology conversation.
In terms of mobility, consumer expectations are already a big impetus for driving point-of-care analytics adoption in the clinical care setting. Decisions on tablet use, phone use, etc., are changing the landscape in terms of security. If organizations look at mobility properly and implement technologies agnostic to the end device, they will be well on their way to creating a device strategy that will accommodate how clinicians and caregivers feel about interacting with the organization.
JE: It really depends on how the healthcare provider wants to implement the clinical decision support system and how the clinicians are going to use real-time analytical information and the insights gained. Health systems should ask questions — such as “Will the analytics require data, including protected health information, to reside on the clinician’s end-point device?” — to gauge how critical device and data protection will be up front. Is the philosophy of application development within the health system to be “mobile first” or is mobile nice to have because most of the users of this data will be using laptops or notebooks? Will the healthcare organization need to support many clinicians and require a multi-tenant architecture that could have high utilization or bursts of utilization at different times? If so, health systems should consider tapping into cloud resources so user experience doesn’t suffer. Will older data sets desired to help infer decisions that the hospital doesn’t want to store on premise anymore be needed? If so, integrating with the cloud storage provider who is managing the hospitals cold storage tier will be needed.
Once health systems tackle these questions, they can factor in their answers to choose the right technology infrastructure. At the end of the day, the infrastructure has to integrate into the clinical workflow and fit with how care is delivered at the health system.
3. What expectations do clinicians have for the end-point devices used to access real-time analytical information?
DR: We have passed the tipping point of only using Windows PCs to access clinical information. Our clinicians rightly expect device-agnostic support for accessing any type of clinical information, including real-time analytical information. The IT division had to significantly adjust how it thinks about end-user devices — and also data or information visualization — since factors like screen size and interaction methods vary widely when an end-user device agnostic approach is implemented in a clinical setting.
JR: New clinicians are entering the workforce from academic medical centers and other environments of care that are more electronic and paperless. Their default capabilities and skills are skewed toward a heightened awareness of technology and what it can do for them. Healthcare organizations need to consider what impedes and benefits clinicians’ interactions with the data and care team. They should establish a feedback loop to maintain conversations around continuous expectations and support with clinicians. These types of decisions can’t be made on behalf of clinicians. Instead, they must constantly be involved in the decision-making process.
JE: I think clinicians bring their own devices to work because they’re not happy with the device they’re getting, and they are very comfortable with the experience they have at home with their personal devices. It’s really important to think about the end device when implementing a clinical analytics system. If clinicians have difficulty using a specific device — or have to use multiple devices for different purposes — it can hinder the overall benefits or adoption of any new capability. Providing clinicians with a device that they enjoy using, facilitates workflow and allows them to collaborate with peers will help drive the overall culture change associated with integrating the new analytics platform into clinical workflows.
4. Besides improved patient care, what other benefits do hospital or health systems stand to gain from clinical analytics?
DR: While clinical analytics offer numerous benefits for patient care, it also offers various qualitative and quantitative benefits for the health system. Qualitative benefits include higher patient experience scores and improved reputation. Colleagues feel good about where they work when they know their organization is a leader in patient care quality. From a quantitative point of view, if we can reduce 30-day readmissions or length of stay, we generate more income that we can then use to drive our mission for enhancing patient care.
JR: Healthcare providers need to see not only the value of what’s right in front of them, but also the value of what’s to come for clinical analytics. Once the data is quantified and codified in an open fabric, health systems can start to look at their processes via more in-depth meta data analysis. A focus on mining data for different needs leads to greater efficiencies and greater overall outcomes. By taking a measured and prescriptive approach in these areas, hospitals can capitalize on efficiencies outside of their focus of a vertically integrated delivery network. They’ll also be able to see adjacencies, grow with the market and better absorb mergers or acquisitions.
JE: Using clinical analytics to improve patient care often goes hand in hand with improving the financial management of health systems, too. With all the payment reforms underway, including value-based care, it’s really important to stratify the riskiest of patients and apply scarce resources to that population. When healthcare providers can predict which patients are more likely to be readmitted, be transferred to the intensive care unit, acquire infections, etc., they can focus on and monitor these patients more carefully to avoid penalties from CMS, improve metrics and improve their bottom line.
Other key areas include population health management where analytics can help case managers proactively intervene with patients who are at risk for missing appointments, need specialized services, or have early signs of disease indication, prevalence or deterioration. Operational or financial analytics can use claims and billing data to determine which patients may be more likely to default. Precision medicine can use analytics to determine a more accurate diagnosis based on the molecular profile of a cancer patient, and then determine a more targeted therapy that aligns to their unique disease. A lot of the same data you use for clinical analytics can also be used from an operational and financial perspective. Leveraging the same data and technology infrastructure for operational and financial analytics reporting capabilities is an interesting way to kill two birds with one stone using a very similar set of data.
5. What advice do you have for healthcare providers looking to implement a clinical decision support system?
DR: Before launching a decision support system, the healthcare organization must ensure clinicians agree with the philosophy of the system and the evidence embedded in the system, and understand how it works, why it functions the way it does and what the intended benefits are. Fully engaging the end user population is a crucial, yet often overlooked, step during the ideation, design, build and implementation phases. Healthcare organizations should also try to integrate the clinical decision support system into the EMR as tightly and closely as possible. The analytics system’s logic and capability should be built inside of the organization’s EMR, so clinicians don’t have to go outside of the EMR to reference the system.
JR: You need to be rooted in your core mission when it comes to making decisions on what system to use. You need to challenge — or be challenged by — the reasons why these decisions are made. Is the change initiative driven by an external factor that isn’t necessarily rooted in the core business challenge? This will affect longevity or how well the organization will receive the project. The decision must be made from motivating factors that are clinical in nature and anchored in the organization’s beliefs.
JE: It’s one thing to assemble a small team to create a proof of concept that involves getting various historical data sets into one place and building a predictive model with high accuracy that the IT department is excited about. Getting this into production and implemented for real clinical practice is a whole different thing. To do this, health systems must establish a cross-functional team with executive sponsorship who can collectively look at the end game and determine the policies or processes that need to be addressed from a governance perspective. They must consider how the key performance indicators will need to change from a performance management perspective, and how care management and coordination will need to change to accommodate changes in workflows. Health systems should also have a clear initial use case in mind that ties back to measurable return on investment with a business case behind it. They can measure this benchmark using a milestone-based implementation plan and point to small wins along the way to garner continued support for the project and inspire continued investment in the project over the long haul.
Like many other industries, data is now a differentiator and competitive advantage in healthcare. If healthcare providers are not getting new insights out of this data and improving, they will fall behind their competition using the resource. Finally, to ensure a successful culture change with clinicians, your clinical analytics goals must align with clinicians’ missions. You don’t want clinicians to feel like this initiative could potentially jeopardize patient care. The clinical decision support system should be the core of what is now considered an emerging standard of care.
More articles on health IT:
Epic, InterSystems boast most new hospitals in 2016, KLAS finds
Cedars-Sinai researchers identify patient medication concerns via social media
Dignity Health subsidiary taps American Well for urgent care
