|
|
October 30, 2006
By erecting a solid data warehouse, healthcare organizations can create a foundation of useful information that can be beneficial both financially and clinically. Scattered throughout a typical hospital are islands of data that have little impact outside individual departmental borders. Yet, when bridges are built between those islands, that same data can be used to create a roadmap for best practices that can improve both clinical and business operations. To tap into the full potential of their data, a growing number of hospitals and healthcare organizations are implementing data warehouses as a way to compile the most significant information and use it facilitywide to improve care, analyze business practices, and streamline operations. “A hospital’s electronic health record systems typically contain a wealth of data around events related to patient care and business practice, both clinical and operational,” says Michael Burke, BSc, MSc, enterprise data warehouse manager for the University Health Network (UHN). “Given that healthcare workers have amassed significant anecdotal and empirical evidence on what improves patient care practices, improvements in quality of care are greatly facilitated when those empirical observations are translated into measurements supported by available data.” UHN is Canada’s largest healthcare organization, as well as a major teaching hospital for the University of Toronto. Over the past three years, Burke and his team have been working with Misys Healthcare Systems to integrate data from a number of departmental systems—including coding and abstracting, pathology, pharmacy, nursing workload, and the financial general ledger—into a data warehouse. Already, they have seen the results of their data warehouse project on operations. Among the most significant are improved resource allocation, workflow optimization, and realization of data linkages, says Burke. “Inclusion of various systems permits key performance indicators to be developed that span the silos of the departmental systems,” he says. “It is the ability to express measures of care and practice in terms of data that makes the data warehouse effective.” Measure Quality, Analyze Practices For example: • A longitudinal analysis of the reduction of urea following hemodialysis treatment across patient cohorts can identify opportunities to optimize treatment schedules. • Mining of data around medication orders can serve as a baseline for designing drug-drug interaction alerts at the point of ordering. • Analysis of lab turnaround time can lead to timelier lab results for patient care decisions. • Analysis of patient care satisfaction surveys can indicate areas for improvement; for example, improved communication of recovery instructions and information following transplant. A data warehouse provides transactional evidence of what the hospital does, says Dean Boyer, director of healthcare solutions for Sybase. By capturing information on every patient interaction—registration; number of visits; attending, consulting and referring physicians; services provided; etc—the facility gets a measure of the patient’s progress and the success of its prescribed care plan. By analyzing that level of historical patient encounter detail from a diagnosis rather than individual patient perspective, trends are revealed that can be used to improve patient care processes. “It’s that ability to look at and analyze information and trend it based on diagnoses that becomes very beneficial to the hospital as a corporate entity in developing best practices,” says Boyer. Proper utilization of a data warehouse impacts business processes by taking information from the clinical systems and formatting it into business schemas that cover a range of topics. Users can then tap into that information by running reports—whether predefined or ad hoc—designed to answer various questions, says Gary Baluta, product manager of enterprise services for Misys Healthcare Systems. The key, he says, is for the data warehouse to work closely with the clinical systems into which physicians are loading patient data. This provides clinicians with access to the data they need for multiple patients, including visit records, treatment plans, procedures performed, etc. “That information can then be mined in a series of longitudinal studies that look at trends,” says Baluta. “The warehouse is the conduit for gathering the many patient records that allow users to look across many different data dimensions, including time, physicians, illness, procedures, etc.” Building a Better Data Trap “From a technology standpoint, it is fairly simple. From an information science standpoint, it is complex,” says Boyer. Too often, he says, business owners don’t understand the source of their data, which is critical to ensuring that the data ultimately captured in the warehouse is of value. Once the source is understood, data should be audited and, eventually, aged and archived. This process allows users to follow the information flow. “One of the things I believe happens is that people begin to grab information without understanding its origin and without understanding the life cycle of that information,” says Boyer. “Because they do that, their data warehousing projects take on unusual lives of their own and they get caught up in capturing information, then trying to restructure it in a way that proves their point, as opposed to capturing the natural lifecycle of the information and letting people query against that.” According to Burke, careful management of several key areas will ensure a successful outcome. These areas include the following: • strong participation from stakeholders who can identify the project’s needs, wants, and expectations; • an effective communication plan to keep stakeholders aware of and informed about project development; • participation from technical experts and vendors on source system product specifications; and • vigorous participation of business subject matter experts on hospital area operations and the relationships to source system operations. “In our experience at the UHN, the key factors controlling project success are the project charter and the communications plan, along with careful work with stakeholders to validate data and other deliverables against expectations,” Burke says. Critical to the makeup of the project development team are individuals who possess an in-depth knowledge of the facility’s data landscape, in particular where the key data is hidden throughout the organization. Because so many of the systems implemented in healthcare organizations are vertical islands of information, it’s important to include individuals on the project team who know what data is contained in those systems, how it was entered, and how it can be used to meet the facility’s needs, says Boyer. “Where are all the hidden treasures that you’re looking for … and how will that data be provided to you?” he asks. “You need your own CSI team to look at all the evidence and see where all the data is.” A third tier of the project team should be individuals who know data warehousing and won’t get hung up on moving all the data into the warehouse, Boyer adds. Not all data can be used for the kind of analytics the warehouse is being developed to support and it’s important to distinguish what is of value. “The data warehouse is going to store enough data just because of the nature of it. You don’t need to add data that is superfluous to the kinds of queries that are going to be asked of it, and the concept that all data is required to run all queries is somewhat obsolete,” he says. Warehouse Work In UHN’s case, acceptance has not been without its challenges, key among them being perceptions around data quality and the validation of deliverables. To overcome those challenges, it has focused on user education and involvement throughout the process, says Burke. “The warehouse has been well-received at the management level as a tool to achieve strategic objectives and as a means to timelier reporting on operations,” he says. “Staff acceptance is good and improving due to the deliberate initiatives to provide meaningful descriptions of the content of the warehouse as well as usage guidelines.” Another challenge is the fact that data mining is a complex process because users “don’t know what they don’t know,” says Baluta. “They are searching for trends that they don’t know whether even exist. They need a starting point or hypothesis.” From that hypothesis, they can then use various reporting tools to evaluate the data and see whether what they think has actually occurred. If the hypothesis turns out to be correct, they move on to look at additional data and pose a new set of questions to review. “It is an iterative process that gets repeated,” says Baluta. “Users have to be patient and know quite a bit about their data before beginning the process.” The best approach to mining data within the warehouse is as a distributed activity performed by nontechnical healthcare workers. This approach, says Burke, maximizes the exposure and transparency of the data warehouse project within the organization and enlarges the analytical resource pool. “Vital to the success of this approach is the planned and vigorous development of materials to educate and direct workers on the content and use of the warehouse,” he adds. “In terms of available software, Oracle Discoverer has been the primary analysis and reporting tool and our customers have used other software to perform statistical calculations or to format reports for presentation when needed.” Achieving ROI How quickly that happens, however, “depends solely on what they are going to use the data for,” says Baluta. “If it’s used for patient quality of care or other healthcare studies, the payback is pretty quick, often less than a year. If the reports generated are only general information, the payback will be somewhat longer.” One of the most immediate areas where hospitals can realize significant ROI is through the identification of missed charge capture opportunities. By looking at the services provided and correlating them to the charges applied to the patient’s bill, a facility can see where charges are being missed. A longer-term but more beneficial ROI comes from the establishment of best practices that a hospital can use to demonstrate the level of care they are providing. Particularly for payors pushing pay for performance, hospitals can use the data mined from their warehouse to demonstrate improved outcomes and increased quality of care, says Boyer. “If the hospital is doing a good job of keeping patients out of the hospital and can prove that, then the insurance company has no recourse but to set up better financial relationships with that hospital,” he says. UHN conducted a customer satisfaction survey to identify the benefits realized during the first phase of its data warehouse implementation. The results found that: • time to produce reports to management or operations had significantly decreased; • staff was spending significantly less time routing and assembling data; • staff had the time and tools needed to focus on conducting analysis; and • the potential to analyze across data silos was realized. “Our best advice on making the decision [whether to implement a data warehouse] is to examine whether the organization can continue to improve at the same rate as peer organizations,” says Burke. “Those peers have made a strategic decision to invest in evidence-based decision making around patient care and business process improvements, the foundation of which is a solid data warehouse. In such a competitive and progressive environment, can the organization continue to ignore the value to be extracted from its data holdings?” — Elizabeth S. Roop is a Tampa, Fla.-based freelance writer specializing in healthcare and HIT. Construction Costs The answer depends on numerous factors, including the scope of the project, functionality of the technology, and the methodologies employed in the implementation process. “Data warehouses cost in the neighborhood of $300,000 to $500,000, but this usually includes the warehouse software, database licenses, and reporting tools. There is an additional cost for the IT staffing,” including database analysts and reporting analysts, says Gary Baluta, product manager of enterprise services for Misys Healthcare Systems. That price tag is why hospitals and large physicians groups have been the primary target for data warehousing. Not only are they more likely to have the financial resources necessary for the warehouse itself, but they can also afford the required electronic medical record systems and technical resources that feed the warehouses with clinical patient information. However, new products have recently been developed that provide smaller organizations with access to data warehouse capabilities. These products, which fall within the $40,000 to $50,000 range, create “virtual” warehouses that bring in data from various disparate sources but don’t need to physically store the data in tables. “They house this data in the computer’s memory while they format and display the information in reports or dashboards. They are much less expensive, providing all healthcare organizations, regardless of size and budget, the ability to mine their patient data,” says Baluta. “As you might imagine, these products have a more limited scope but can still provide the 80/20 rule in many cases—80% of the benefit at 15% to 20% of the cost. The organization would need to evaluate what it is going to use the data for and budget according to the reporting and analytic power it requires.” The number of systems within the hospital that will be feeding the data warehouse also contributes to the final construction cost, according to Dean Boyer, director of healthcare solutions for Sybase. “Each time you run into a new system, you have to develop a set of analytics, extraction, and strategies to get the data out of that system, and then you have to sort the data from that system and make it relative to all the other systems,” he says. “So one can easily see a hospital that has five systems is going to be a lot less expensive than a hospital that has 20 systems to interface with.” Finally, says Boyer, you have to factor in the service costs, which can run as much as three times the technology costs. In many cases, an incremental approach to building the warehouse is the most effective and affordable because it allows the service provider to actually use the data to create an efficient process. “If you stay more focused and grab vertical pieces of information and build on that, then you can quickly tier it,” he says. “You find that you get into a repeatable, very successful process because now the data itself is helping you build out the data warehouse.” — ESR
|
 |