J. Bryan Bennett, MBA
Executive Director, Healthcare Center of Excellence, LLC, Buffalo Grove, IL
Since 2014, the Healthcare Center of Excellence (HCOE) has been tracking the healthcare industry’s progress towards population health analytics using their proprietary Healthcare Transformation Change Model. This model was developed in 2013 to help healthcare organizations understand the requirements to become analytics-focused. The concepts were based on a customer relationship marketing (CRM) implementation model, co-developed by this author. The CRM model was considered by the technology research firm, Gartner, Inc. to be one of the top three CRM implementation visions at the time and has become the basis for most successful CRM implementations today.
The 2016 State of Population Health Analytics (SOPHA) study revealed that the healthcare industry is making progress, but at a slower pace than expected. This is due to the challenge of integrating multiple sources and types of data and a focus on the data and technology to the detriment of people, processes and leadership. Larger healthcare organizations are making more progress than smaller ones due to their greater access to resources.
Each year, the SOPHA study is conducted using a combination of primary and secondary research. This research includes:
- Interviews with healthcare executives
- Conversations with healthcare conference attendees
- Presentations from healthcare conference presenters
- News and journal articles
In 2016, the study was enhanced by utilizing a 20-question survey to add more empirical insights. Once analyzed, the results were validated through conversations with several healthcare executives.
The SOPHA study is based on the proprietary Healthcare Analytics Transformation Model developed by the HCOE, which is comprised of three continuums: Organizational/People, Data/Technology and Process/Workflows. To be successful, an organization must move along all three continuums at the same pace to progress from non-analytics focused to analytics focused, by completing the various steps at each stage of the transformation.
By applying the different types of analytics to the model, an organization can determine what steps must be taken and at what stage they will be able to perform specific types of analytics. (See Figure 1.)
- Descriptive analytics is just past the EHR Implementation stage on the data/technology continuum. Most organizations can perform some type of descriptive analytics from their EHR systems. Answers to questions such as, “How many patients were diagnosed with X last year?” can easily be determined.
- Diagnostic analytics could be performed once the organization is beginning the data integration process. Questions such as, “Why did these patients develop X?” can be addressed at this stage.
- Predictive analytics requires the organization’s data to be fully integrated in order to obtain a complete view of the patient’s health management. Questions such as, “What can be done to prevent X?” can be answered at this stage.
- Prescriptive analytics will require fully integrated data plus a robust analysis and modeling function. Additionally, for prescriptive analytics to be most effective in a healthcare organization, the results need to be delivered to the providers in real-time to help support their diagnosis and treatment plan. Questions such as, “What is the relationship between X and Y and how can X be minimized?” can be addressed at this stage. (See Figure 2.)
Based on the model, the HCOE determined that in 2014, the industry was just past the “EHR Implementation” stage on the data/technology continuum. In 2015, the dial moved a bit farther as many organizations began achieving success with integrating data sources. (See Figure 3.)
Impact of the 2015 Healthcare Analytics Challenges Study
In a separate study conducted in 2015 by the HCOE, participants were asked to identify five challenges faced in implementing healthcare analytics. The participants included people from all levels of healthcare organizations from locations across the U.S. The challenges were classified into 10 categories for further examination. Those categories were:
- analytics tools
- change management
- data management
The top three categories chosen were leadership (29 percent), data management (18 percent) and talent (14 percent). (See Figure 4.)
Leadership included common terms/phrases such as “lack of priority,” “lack of vision,” “need for buy-in from staff” and “lack of direction,” but also included “disparate EHR systems,” “siloed systems” and “teams not working together.” These issues can all be resolved through proper leadership.
Data management included “lack of data standardization,” “lack of data stewardship,” “lack of definition of key variables to study,” “where does the data reside” and “quality of data.”
Talent included “lack of adequate analytics talent,” “hiring the right people,” “lack of skills,” “not enough qualified staff” and “retaining talent.”
Overall, the distribution of responses was decidedly pointed at leadership.
Even more surprising was the source of the comments. As part of the study, the participants were asked to include their title on the survey form they submitted. The titles were summarized into general categories which indicated the responses included all levels of the healthcare organization, from C-level to the analyst. The participants were divided into two groups: wide span of control, which included participants at the director level and above; and narrow span of control, which included all others.
The only challenges where the disparities were significant were integration and processes. This is likely because of perspective. The wide span executives are anxious to see results of their investments through data integration versus the narrow span personnel desire for more and better processes to make their jobs easier. (See Figure 5.)
Based on this additional research, leadership was incorporated into the transformation model (shaded in yellow in the model). Leadership is not a continuum, but it must be present in each stage of each continuum and applied consistently throughout the transformation. It is represented in the model like a thermometer—going from the bottom, low-level of leadership to the top, high-level of leadership. It is the most important critical success factor in any change management situation and is also known as Executive Sponsorship in Lean Six Sigma or project management training. (See Figure 6.)
The results of the 2016 State of Population Health Analytics study reveals that the needle has only progressed slightly. (See Figure 7.)
Integration of data is still a challenge for many organizations. This not only includes integrating clinical data, but also other internal and external data, such as operating costs, billing and reimbursement data, to achieve a complete picture of the patient population.
Many organizations are focused on the data and technology and not leveraging the people and processes necessary to build a successful population health analytics program. If the data is not collected and recorded in a timely manner, then the risk of incorrect or out of date information increases.
Lack of leadership is still an issue identified by participants. Analysis indicates that leadership is about one-third of where it needs to be for the organization to be successful. This is due to a variety of reasons, including:
- The complexity and timing of the transformation required by the industry. Healthcare organizations have had to integrate electronic health records solutions, ICD-10 and value-based care all within a short amount of time. It’s like asking a major electronics retailer to also become a fine French restaurant in 24 months. It can be done, but the transformation is painful.
- Many organizations are led by clinically-trained professionals who may not have the broad business and change management skills necessary to navigate this transformation. This has led to an autocratic management style versus the inclusive style needed for success with such a complex transformation. (See Figure 8.)
Increasingly, the industry is facing a situation of the “haves” versus the “have-nots” due to an imbalance in resources enjoyed by the larger institutions versus those of the smaller institutions. If the transformation model was segmented by size, it would reveal many of the top institutions well into, and some past, the predictive analysis stage. These organizations can afford the personnel and technological resources required to be successful in the later stages and typically have exceptional leadership. At this rate, the smaller institutions will have little choice but to lag behind or be acquired by their larger brethren.
The HCOE has several recommendations that could help an organization be more successful with their population health analytics implementation, including:
Understand and document where your organization is on population health analytics implementation.
Before an organization begins implementing a population health analytics initiative, they must truthfully assess their current capabilities and identify gaps that need to be overcome to become an analytics focused organization. The HCOE has created tools that help healthcare organizations with this step. These tools include the Analytics Readiness Assessment tool for beginning to intermediate analytics organizations and the Competing on Healthcare Analytics Discovery tool for intermediate to advanced organizations. Although either of these tools can be utilized internally, the HCOE highly recommends having an unbiased third-party manage the process to ensure accurate and candid responses.
Invest in Leadership Development
Strong leadership is required in such a challenging transformation to motivate people to go over and above the basics of their job requirements. Healthcare is different from other industries and has unique needs as well as consequences for improper or incorrect actions. This requires a new approach to leadership. Many don’t understand that leadership is not a position or a skill, but a process that can be learned and must be practiced daily like a professional athlete. The HCOE calls these professional leaders. Most leadership training falls short because of the focus on “skills” development, when the focus should be on the process.
Build an Analytics Function Using Alternative Methods
Once the other continuums have been addressed, a key component of a population health analytics initiative is the data scientist. Most healthcare organizations are finding that hiring a qualified data scientist is a real challenge. Experienced data scientists are expensive and are usually employed elsewhere. An alternative to hiring a data scientist could be building your data scientist out of a team of people currently on staff or readily available in the marketplace. This is called the “DataScienceStein” approach modeled after Mary Shelley’s Frankenstein monster built from several human parts. Instead of an individual data scientist, the organization would deploy a data science team.
By following these recommendations, an organization can be assured that its population health analytics program is built on a strong foundation of people, processes and technology powered by effective leadership.