Q&A with Best Practices Awards Winners—Healthcare
Last year, the TDWI Best Practices Awards saw an unprecedented number of winners in the healthcare industry. The Awards recognize implementations that demonstrate significant business impact, maturity, relevance, and innovation, so in creating What Works in Healthcare, we thought nothing would be more appropriate than to check back with the business and IT sponsors of these winning solutions to see what’s transpired over the past year, what they’ve learned, and what benefits their implementations continue to deliver.
Enterprise Data Warehousing
Co-winner: Blue Cross and Blue Shield of Kansas City
Solution Sponsor: HP
Blue Cross and Blue Shield of Kansas City (Blue KC) is the largest health benefits provider in the greater Kansas City region, serving nearly one million members in 32 counties. Founded in 1938, the not-for-profit health insurer employs approximately 1,000 Kansas City area residents and offers a variety of health and related benefit plans for individuals, families, and employers. Blue KC is an independent licensee of the Blue Cross and Blue Shield Association.
To maintain its market leadership position in a changing environment, Blue KC brought in Hewlett-Packard BI Solutions to help design and deploy a fully integrated data management architecture. The objective was to turn data into an asset that created a competitive advantage. The resulting solution incorporates data from 45+ sources and provides a single view of vast amounts of information. In addition, a center of excellence was formed that brought technology and business staff together into a single organization accountable for data governance, data stewardship, and program execution.
The solution has enabled efforts that improve member health, enhance operations, and reduce medical and administrative costs. The quantified benefits made possible by this initiative fall into four primary categories—improved data management efficiency (cost avoidance), reduced medical costs, elimination of outsourced health management programs, and increased revenue through customer retention. A conservative estimate quantifies the results at a 332 percent return on investment over a six-year period, with the breakeven point at 20 months into the effort.
What Works in Healthcare: Are you still seeing significant benefits from your solution? What areas would you say have been most successful?
Blue Cross and Blue Shield of Kansas City: Blue KC continues to reap significant benefits from our enterprise data warehouse (EDW). The solution not only supports traditional health plan efforts, such as reporting and transaction system data feeds, but also serves as a vital foundational element for our health and wellness solutions. Blue KC is committed to providing our customers affordable healthcare and building programs that enable improved health outcomes. We leverage our valuable data asset to tailor the most effective programs for each individual we serve. In addition, the integrated data has allowed Blue KC to create an effective 360-degree view of each member, which makes possible an enhanced member experience when interacting with our plan representatives.
WWiH: What are some of the unexpected challenges that have come up in the past year? What lessons did you learn in solving them?
Blue KC: It’s no secret that one of the biggest industry challenges over the past year has come through the introduction of the Health Care Reform Bill. The introduction of this bill has caused Blue KC and all other health plans to make adjustments to current and future operating procedures and to evaluate their business models moving forward. Having a solid data platform that can be used as a basis to create complex, proactive forecasts is more important than ever. Since Blue KC already has a dependable data management program, the focus and effort can be placed on advanced analytics and predictive modeling instead of tedious data collection efforts.
WWiH: Have you expanded the program to other groups or users, found new uses or expanded functionality, or made other changes since it was first implemented?
Blue KC: Blue KC continues to grow its EDW by adding new types of data on a routine basis to outpace future requirements. The newest area of focus includes further leveraging our strong data by taking advanced analytics to new heights. Late in 2010 we created a centralized division that manages the EDW, has accountability for the company’s portfolio of analytic technology solutions, and delivers analytics at an enterprise level. As part of this new focus, we are implementing a new database platform for performance gains and a new business intelligence application for better analytic and visualization capabilities, and we are further enhancing capabilities by creating advanced analytic algorithms. Examples of such algorithms include models to help patients avoid inpatient readmissions, models to help determine a customer’s propensity to purchase other products, and models to determine the likelihood of a person having a negative health event before it actually occurs. This new focus on analytics comes at a dynamic time in the healthcare industry and creates exciting opportunities at and for Blue KC.
Customer Intelligence
Winner: Blue Cross Blue Shield of Massachusetts
Solution Sponsor: Netezza
Blue Cross Blue Shield of Massachusetts (BCBSMA) provides quality healthcare coverage to approximately 3 million members. As a non-profit, it is committed to delivering world-class service to its members and to being a good corporate citizen and community partner.
Previously, BCBSMA’s Web interactions had been limited to members logging in to check their health insurance plan benefits and claims history—a passive, static system that failed to engage members in their healthcare. BCBSMA also educated its members about health screenings and wellness programs through generic newsletters, which were rarely read. The challenge was to find a cost-effective mechanism to deliver secure, personalized messages to members. BCBSMA also had several new initiatives to reduce health costs, which depended heavily on an engaged, educated consumer making informed decisions about preventive care, provider networks, and appropriate care sites.
The purpose of the project was to (1) leverage the wealth of information and tools available on the data warehouse to deliver personalization; (2) build a system that was system- and delivery-channel- agnostic; and (3) build a mechanism to measure the efficacy of personalization.
BCBSMA built and deployed an innovative solution that successfully integrated its data warehouse and member Web portal to produce near-real-time personalized messages for its members. The framework is also scalable to handle new data sources, alerts, reminders, and campaigns with minimal coding changes.
WWiH: Are you still seeing significant benefits from your solution? What areas would you say have been most successful?
Blue Cross Blue Shield of Massachusetts: The Member Health Intelligence (MHI) application was launched in January/February 2010 with five alerts. Since then, we have expanded the scope of the application to include health-condition-related reminders (e.g., overdue screenings) and are currently generating 13 alerts. The primary delivery channel for the personalized alerts/reminders continues to be the Web (Member Portal). There has been an expansion in the realm of static alerts wherein a list of members is first created based on specific criteria and an alert is sent to this same set of members until a predefined expiration date. The design of the application allows for an exceptionally quick time to market for static alerts (three or four days from concept to implementation).
With regard to the efficacy of personalization, we have noticed an impact on certain alerts more than others. For example, the alert to remind members to use their $150 annual fitness benefit has generated the most impact.
WWiH: What are some of the unexpected challenges that have come up in the past year? What lessons did you learn in solving them?
BCBSMA: The application has only a single delivery channel and the impact of personalization is constrained by the use of the channel by our members. While there has been a steady growth of registered users on the Member Portal, the usage patterns are highly variable. There are efforts under way, outside of the MHI application, to increase the usage of the portal, which should result in increased responses to the alerts/reminders. Expansion of the application to use the e-mail channel is also under way.
Another challenge has been achieving a consensus on rules used to arrive at a list of members for a specific alert (e.g., different business areas define “high-risk diabetics” using different metrics). The short-term solution has been to assign business ownership of every new alert. Initial discussions are under way to implement a centralized business rules engine at the enterprise level to get around this problem.
WWiH: Have you expanded the program to other groups or users, found new uses or expanded functionality, or made other changes since it was first implemented?
BCBSMA: The MHI application has gone through a set of enhancements over the last year, with the primary focus being expansion—both in terms of the number and complexity of alerts being generated. The dashboard to measure efficacy of personalization (Track IT) is being redesigned to be the central location for reporting numbers from all campaigns (not just Web alerts).
On the business aspects of the application, a governance body will be in place before the end of 2011 to oversee business rule standardization. This team will be responsible for certifying the rules used to select a member segment for an alert, and for ensuring that proper metrics have been defined to measure outcomes.
Government and Non-Profit
Co-winner: Centerstone Research Institute
The Centerstone Research Institute (CRI) is a private, not-for-profit company dedicated to improving healthcare delivery through the marriage of research and information technology. Based in Tennessee and Indiana, CRI works with the community mental health centers of Centerstone to conduct clinically relevant research and provide high-quality services to more than 70,000 individuals with mental illness. For decades, CRI researchers and affiliated community mental health centers have conducted hundreds of service and clinical studies to secure more than $50 million in federal and private funding. The purpose of CRI’s BI/DW solution was to empower end users, who previously had no access to data, with actionable information to inform their business management and clinical practices.
Like many other community behavioral health providers, Centerstone faces increasing pressure to keep costs down and increase output. In addition, Centerstone has limited financial resources to invest in an enterprise BI solution. The bulk of the data warehouse infrastructure relies on open source technologies (Postgres, Jaspersoft, Pentaho, and KNIME) combined with vendor solutions that add value in targeted areas such as dashboards (QlikView) and predictive modeling (IBM/SPSS Modeler & Statistics, SAS).
End users can now see and explore vital information related to service quality, financial viability, and client outcomes. In addition to operational and management tools, predictive models have been developed to better inform clinical care decisions. Finally, CRI’s low-cost model may help other behavioral healthcare organizations to affordably deploy similar solutions, or CRI’s data warehouse might be extended to provide similar functionality for other healthcare providers.
WWiH: Are you still seeing significant benefits from your solution? What areas would you say have been most successful?
Centerstone Research Institute: We are still seeing significant benefits from our solution. We have continued to expand the BI infrastructure to support organizational needs by further embedding actionable information in clinical and business processes, enhancing transparency of data and information throughout all layers of the organization, and developing improved tools to model what-if scenarios and predict performance. For example, our Indiana sites recently went through a significant change in about 20 complex payer rules, and the analytics team developed what-if models that provided real-time insight into the impact of each rule change. Management used those modeling tools to identify which rule changes were most likely to impact operations, which allowed targeted development of management processes and monitoring tools that helped the organization navigate changes with minimal disruption when similar organizations struggled. This enabled continuity of clinical services and stable revenue through a very turbulent external environment.
WWiH: What are some of the unexpected challenges that have come up in the past year? What lessons did you learn in solving them?
CRI: One unexpected challenge was that we saw a plateau in user adoption of our primary dashboard tool, QlikView, despite the ongoing addition of new content. Upon talking to management, we learned that we needed to strengthen our training efforts with staff who were not early adopters or power users. We also discovered that this user group needed extended training over a period of months, including (1) initial overview and orientation, (2) follow-up after some period of use and experimentation, and (3) advanced skills training. Over the last six months, we significantly increased our training efforts with key layers of management, and have seen a 53 percent increase in monthly unduplicated users during that period.
WWiH: Have you expanded the program to other groups or users, found new uses or expanded functionality, or made other changes since it was first implemented?
CRI: The solid foundation of widely distributed BI tools and corresponding information transparency has significantly improved data quality, which has enabled opportunities for more advanced predictive modeling and data mining. Most notably, we are in the midst of implementing a learning algorithm that will sort through clinical data when patients present for treatment and make targeted recommendations regarding the treatments that are most likely to be effective for a given individual. Unlike most one-size-fits-all treatment algorithms based on comparisons of group averages derived from randomized controlled trials, these recommendations will be personalized to the presenting problems of a specific individual. This will provide clinicians with prompts regarding optimal treatment choices and their corresponding probability of success, while still leaving ultimate treatment decisions in the hands of clinicians. By increasing the likelihood of effective treatment, we have an opportunity to significantly improve treatment outcomes while decreasing healthcare costs associated with duplicative or ineffective treatments.
This article originally appeared in the issue of .