Top Actions for Healthcare Delivery Organization CIOs, 2014: Avoid 25 Years of Mistakes in Enterprise Data Warehousing
Healthcare has a compelling need to use more information, better. With EHRs often in place, and a gusher of new data streams on the way, HDO CIOs cannot afford the cost, time or agony of repeating classic blunders, just to get to an integrated warehouse with clinical data.
- Integrating business/financial and clinical data into an effective EDW is the top new IT initiative for HDO CIOs once a Generation 3 EHR system is deployed.
- Health systems that must become population health managers have urgent requirements that stretch well beyond the internal data sources driving the post-EHR EDW.
- HDO CIOs often severely underestimate just how tough and time-consuming it is to build strong information governance, and to bake accountability for data quality into the health system.
- The scope of an EDW project must find its level between trying to "boil the ocean" and simply "data marting to death" just to demonstrate a fast time to value.
- Use all the education and influence you can muster with top executives to make the case for a strong top-executive commitment to high-value use, strong information governance and data quality.
- Do not overestimate the value of a commercial vendor's data model, avoid underscoping the total effort and personnel needs, and never treat an EDW as just another module from the EHR vendor.
- Create an IT EDW or EIM vice president or director job that recognizes the importance of persistent marketing and communication about information management and analytics architecture use cases, as well as about current value achieved, as a primary focal point.
- Break out of the HDO CIO's traditional forums to learn more about EDW technologies and best practices, as well as about healthcare quality improvement efforts.
Healthcare IT's potential value is changing in profound ways. While the big value boost started with electronic health records (EHRs), efforts must be heavily directed to advancing the state of retrospective and real-time analytics. Superior use of analytics will be a dominant factor in health system success for the rest of this decade, and it is a growing component of the CIO work. Indeed, in 2013, Gartner changed the name of one of our Hype Cycles from "Hype Cycle for Healthcare Provider Applications and Systems" to "Hype Cycle for Healthcare Provider Applications, Analytics and Systems" to emphasize this point. A quick look at that Hype Cycle reveals analytics' emerging dominance (see Figure 1). Over half of the initiatives we chose to feature on the left-hand (early-stage) side of the cycle either are purely analytics (highlighted in green) or have a significant analytics component to them (in yellow).
Enterprise data warehousing (EDW) is a key and difficult element of that value, and creating an EDW in healthcare is a particularly weighty endeavor due to the complexity of data and uses. At the same time, EDW practices are evolving in significant ways — even to the point that "data warehouse" may be better referred to as "analytic data management." This is why we have specifically highlighted EDW among our Top Actions for 2014. We have already seen too many healthcare delivery organization (HDO) CIOs repeat the EDW mistakes and misestimates of the past. Use the Fatal Flaws for HDO CIOs and Recommendations in this research as triggers for dialogue within IT, with executives and with vendors.
Source: Gartner (February 2014)
While the term "data warehouse" has been around since the 1970s, it is generally in the last 25 years that more technologies were introduced, and broad enterprise data warehousing efforts started to become common. Gartner has been evaluating the technologies, tracking project successes and failures, and discerning best practices and common mistakes for many years. This work has resulted in our current list of "nine fatal flaws" in business operations improvement (BOI). Here, we emphasize the five of those flawed beliefs most germane for HDO CIOs.
An EDW is an enterprise asset and not an IT project, though an IT project nests within it. Many Gartner clients report that their IT organization sponsored, funded and led their business intelligence (BI) initiatives from a technical, data-centric perspective, the assumption being that an EDW would assemble and model crucial data, and that its value would become obvious eventually. The danger is that its value is not obvious (and is perhaps even entirely obscure) to the business, so all the hard work does not result either in big impact or in heavy use. This is why we so heavily emphasize information governance, and make a point of distinguishing it from also-important data governance. Historically, the "Build it and they will come" model was used in almost every vertical during the late 1990s and early 2000s. More than 80% of those IT-directed warehouses took three years to fail, and were replaced by data marts built to purpose by individual business units. In contrast, data warehouses sponsored by business executives exhibited a success rate of greater than 70%.
Some managers resist EDW initiatives out of a fear of loss of control of their work. The process they have in place seems to be working OK, the reports they generate are required on a mandated deadline, and they're afraid IT will take it over and muck it up.
But others resist for political, control, performance accountability and transparency issues. HDO CIOs — thinking logically and systemically — see the EDW as a rational way to meet multiple needs, and to leverage common data management and technologies. Resistance to both data access and shared data quality responsibility can catch IT off guard in any enterprise, but because the healthcare EDW portends enterprise scrutiny of medical decisions and care processes, it can really strike a nerve. Gartner has heard directly from some clients that warehouses that expose errors in calculations and data have actually been directed to recreate the errors as portrayed in those other systems. Don't do this.
Industries where the enterprise resource planning (ERP) provider is the dominant strategic vendor can be tempted to assume that the best EDW option must be to extend that relationship. This isn't necessarily either better or cheaper. In fact, Gartner strongly recommends against just blindly taking this path. Dimensional data is the key: If more than two-thirds of all the warehouse dimensions come from one ERP system, that only puts the packaged warehouse on the "shortlist." Note how readily this syndrome translates to the same flaw with healthcare megasuite/EHR vendors.
Pressure on the IT organization from discrete stakeholders for quick dashboards in narrow domains with small budgets is to be expected — an urgent need can arise from payers or regulators, especially. But a constant stream of these shows a lack of executive cohesion around enterprise information management (EIM), and is a failure of IT governance to adhere to sound principles for enterprise leverage of IT resources. Additionally, that magical dashboard data has to come from somewhere, and when it is aggregated by users, there is a strong tendency to follow their own agenda in what data will be presented and how it will be put together.
The biggest flaw of all is the lack of a documented BI strategy, or the use of a poorly developed or socialized one. Many other issues can be avoided completely, or mitigated significantly, if the HDO CIO can orchestrate a collaboration with top business, financial and clinical leadership.
There are other analytics systems that the HDO will already have in place or may need to pursue, including financial reporting, mandated quality metrics reporting and cost accounting/key performance indicator (KPI) systems. The new requirements in EDW for the CIO focus on one or two overlapping initiatives. If the health system is not assuming population health management responsibilities or risks, only the first applies. For both needs, the decision and development effort include:
- Several build/buy decisions
- An architecture
- A data intake/quality (or extraction, transformation and loading [ETL]) process
- A data model
- A database
- One or more BI platforms
An HDO may decide to begin with separate efforts, but we believe they eventually coincide for the most aggressive, larger HDOs.
The integrated clinical/business EDW is incorporating at least:
- Claims/revenue cycle
- Cost accounting
- Patient satisfaction data
The biggest trigger of a new EDW or a major overhaul is the EHR, and it is clinical data and users that add the most complexity. Most vendors working in healthcare and other industries observe that healthcare has the most-complex data of any industry (possibly excluding government intelligence efforts). The EDW enables large communities of users to develop/receive reports and dashboards/queries, and to perform data discovery around core patient care, and business processes and outcomes.
This must support both large-group and small-cohort analysis to develop a more-personalized and precision approach for individual patients who are higher risk/higher cost. Strong population health analytics is a very iterative effort. It might start with claims, but it pretty quickly evolves to meld EHR and other patient data. It uses clinical epidemiology-type techniques to discern patterns and correlations among patient risk factors, community and environmental influences, disease characteristics, engagement and treatment approaches, resources applied and outcomes. Its purposes are:
- Risk assessment and prediction
- Flexible patient cohorting and registries
- Identification of care gaps
- Prioritization of patient engagement
- Effectiveness benchmarking
- Innovation research
The shift in payment models fueling this need includes both incentive-based and risk-reward payment models, and is illustrated by the U.S. Centers for Medicare & Medicaid Services' Shared Savings and Pioneer Accountable Care Organization (ACO) programs.
Use all the education and influence you can muster with top executives to make the case for a strong top-executive commitment to high-value use, strong information governance and data quality
Don't underestimate the significant resistance to accountability, and the surprising amount of data quality issues that will surface — nor the time it will take to resolve them.
Ignoring or underestimating information governance is the most common mistake we've seen HDO CIOs make in recent years. And, it is often made by CIOs who are of long tenure, and are respected in their organizations. They just don't believe they will have the resistance and data accountability problems — until they do, and this has delayed a number of EDW projects.
CEOs live in strategy; CIOs live day-to-day in infrastructure. Information governance is where they meet (see Figure 2).
CIOs too often track success by volume instead of by impact. The value of the integrated EDW is high for organizations whose leaders grab hold of it with both hands. That starts with a determination around high-impact uses of information and accountability among business and clinical executives for data quality.
Information governance means focusing the executive body on setting a clear and effective process for making decisions about information to drive important enterprise change, to establish principles, and to put appropriate subcommittees and responsibilities in place.
We find that many CIOs and consultants say "information governance" when they are just acting on data governance. The executive-level CIO's work starts with seed planting here. This requires executive education. It also requires the CIO to translate technical lingo into business language, and to relate use cases to business priorities and "return on information."
The CIO must also seek the right alignment with the CFO and senior quality officer, two veterans in information management. Instead, IT folks tend to plunge quickly into their project mindset, which fixates on the first building and stumbling block of the EDW — the data. Governance goals include endorsement, funding and commitment to using the EDW, an equal commitment to principles and actions around enterprise data standardization, and the assignment of data stewardship responsibilities to business and clinical leaders (the chief medical informatics officer [CMIO] plays a role here). Equally important is a commitment to culture change to ensure acceptance of transparency and accountability. Note: Active use of executive-level dashboards that move to lead indicators of change, not just KPIs (which are lagging indicators), is a good sign of the engaged CEO.
Source: Gartner (February 2014)
For additional advice on building strong information governance, which usually will take eight to 15 months, see "EIM 1.0: Setting Up Enterprise Information Management and Governance" and "Information Governance Best Practice: Adopt a Use Case Approach."
Do not overestimate the value of a commercial vendor's data model, don't underscope the project, and never treat an EDW as just another module from your EHR vendor — an EDW effort for any large, varied enterprise is complex
In healthcare, it is made more so by the variety and complexity of the clinical data involved, and by creating a data model. Therefore, a number of BI stack vendors, healthcare-specific entrepreneurs and consultancies have developed such models. Some have added in starter sets of use cases, content, navigation screens and reports. Some of the healthcare megasuite vendors (for example, Cerner and Epic) have been very focused on offering new capabilities in this realm.
The notion of vendor "acceleration" help is appealing — healthcare is very accustomed to leveraging the similar IT needs of HDOs through vendor relationships. But this can also create some specific risks for the CIO, chief among them that the vendor overstates the simplicity of the project, and underestimates the time and effort that internal staff have to invest. This approach can easily end up with the EDW being treated as an IT project, or with executives attributing all the value to the vendor, and seeing IT's work as a commodity expense.
Therefore, don't focus on the vendor "who" before focusing on the "what" and "why." A successful initiative combines relevance and time-to-value considerations with strong architecture. Use vendor presentations and success stories as valuable tools in educating executives. Take care in calculating the time/resource savings, and in negotiating the price and specific deliverables for a commercially provided healthcare data model. Data modeling is a very significant percent, about 45%, of the total effort to build an EDW, but a commercial model only addresses part of that work. Any model will have to be localized to the HDO's circumstances, physically instantiated and evolved over time. It is very difficult to hold strictly to a vendor's model over the long term, so its value is mostly in the initial effort. However, EDW data-modeling expertise in healthcare is scarce, and this makes vendor assistance more attractive.
Investigate how vendors have assisted others with governance and requirements processes, not just the technical build. Be sure the vendor doesn't gloss over this need. Experienced suppliers recognize that neglect of these is more likely than technical challenges to waylay or delay the EDW.
Just blindly buying an EDW from your EHR vendor is as flawed as a manufacturer or retailer just buying from the ERP vendor. The EHR vendor has a close and strategic relationship with executives. Most of them know that HDOs need an EDW, and more of them each year want their clients' EDW business. Make build/buy and procurement decisions formally and objectively. Be clear that governance, planning, deployment and management are not just another application go-live.
Create an IT EDW or EIM vice president or director job that recognizes the importance of persistent marketing and communication about information management and analytics architecture use cases, as well as about current value achieved, as a primary focal point
As seen in the "flaws" above, many EDW efforts have been done in by an "If you build it, they will come" approach. Every successful EDW vice president//director we have encountered has embraced a continuing EDW marketing role. Not all started that way, or were originally comfortable with it, but each has recognized how critical this is. Initial expectation management and communication are key in these initiatives, and the CIO helps set the tone. But the direct relationship between the EDW lead, BI core competency center management and key stakeholders is continuous. Identifying and nurturing the champions who raise their hands early to drive specific impact is invaluable. HDOs have found that leaders in nursing, in the cardiovascular and emergency departments and in intensive care units are among those who most commonly engage early. Of course, if population health management is a goal, those investigators are another key user community.
Break out of the HDO CIO's traditional forums to learn more about EDW technologies and best practices, and about healthcare quality improvement approaches
Data warehousing is not an area of core experience for many HDO CIOs, and many of the best practices and technology advances have been developed outside of healthcare. As with efforts like adapting Lean to IT, CIOs need to expand their information sources and familiarity with two domains:
- Increase your familiarity with EDW and advanced analytics technologies/best practices.
- Gain a deeper understanding of how the best healthcare quality improvement and population health management get done (see Gartner Recommended Reading and Note 1 for a number of helpful resources).
The Intermountain Healthcare Institute for Healthcare Improvement (www.intermountainhealthcare.org/qualityandresearch) is an excellent source of education programs on healthcare quality management for executives, quality leaders and physicians. CIOs and CMIOs who have attended these courses cite their high value.
(Please note: While their names are similar, these are two separate organizations.)