Although the strategy has taken hold in healthcare, using data to drive performance comes with a host of challenges. For starters, from electronic medical record to practice management and financial systems, physician practices employ a number of information systems, each with its own set of data. This proliferation often causes confusion because the data is disorganized and unstructured, rendering analysis virtually impossible.
Breaking through this paralysis requires turning the disparate bits and pieces of data into useful information through the use of tools that provide business intelligence. Unlike customary reporting, business intelligence doesn’t just produce static reports that require the user to search through to dig out useful information. Business intelligence actually analyzes data to generate dashboard-type reports that identify problems in clinical, operational and financial areas and automatically signal the need to act. These alerts, in turn, foster faster, more effective decision-making.
The challenge of BI
But simply buying and installing a BI tool is not enough to improve performance. Organizations can invest significant amounts in sophisticated BI systems only to generate reports that receive either a cursory glance or are ignored entirely.
To be effective, business intelligence must be accepted and understood by the physicians, clinical directors, financial managers and other users who are charged with delivering better clinical and financial outcomes. To gain this buy-in, organizations must identify the kind of metrics that can drive performance, employ the appropriate processes for gathering the data that produces the metrics and prepare physicians and others on their roles and responsibility in using such information most effectively for their organization.
Incorporating volume and value metrics
The segue from a volume- to value-based payment approach has necessitated the use of metrics from both the financial and clinical halves of the healthcare world — metrics that not only originate from different information systems but are also used differently.
Financial metrics are part of the curve 1 economics of healthcare.
These metrics gauge productivity and may include such measures as evaluation and management volumes, work relative value units or claims charges. Clinical metrics are part of the curve 2 economy and gauge outcomes that represent quality performance, such as with patient experience (e.g., patient access to care) and process of care protocols (e.g., the percentage of patients in a diabetics panel who receive monitoring on five evidence-based measures).
In order to produce useful data, a BI program must incorporate both curve 1 and curve 2 metrics. Tracking downstream revenue from an office visit with a diabetic patient over a six-month time period can’t be accomplished until data from financial and clinical systems is incorporated. Real progress, in fact, begins at the intersection of these metrics.
Data must be extracted and aggregated from these systems in such a way that the data can be easily manipulated in order to prepare it for analysis (i.e., using a single identifier for a patient throughout the data).
|
Beyond employing data that is trusted, timely and trended, a business intelligence program should include the following features that can optimize results: Dashboard metrics. Financial and clinical leaders often don’t have the time to dig into vast amounts of data to determine what part of operations requires immediate attention and what areas are operating smoothly. An effective BI tool will be able to provide dashboard-type metrics that quickly indicate performance in key areas. Because such data will be reviewed frequently, as often as daily in some cases, having this instant analysis is critical to managing effectively. At the same time, the BI tool should also allow for easily drilling down into the data in those areas that do require additional attention in order to identify root causes of problems. Customized reports. Whether with social media, professional networking or personal websites, technology users want information that is customized to their interests and needs. Likewise, stakeholders within a physician organization have differing areas of focus. A CEO may want a dashboard with performance indicators on volume trends or operating margins, while a CMO’s dashboard may have more quality-centric metrics, and an individual physician may want to see compensation trends. An effective analytical tool will be able to sort through the vast amounts of data according to stakeholder needs and provide customized dashboards for all levels of users. Easy accessibility. Just as technology users have become accustomed to customization, they also expect to be able to access information anywhere, anytime. Reports, therefore, should have online accessibility ideally from wherever users are and with whatever kind of computer they’re using — desktop, tablet, smart phones. |
What makes BI successful?
The ability of an organization to successfully extract data and prepare it for use, in turn, affects the strength of the data — a critical factor in the success of any BI program. Strong data must be:
• Trusted. To create reliable data, organizations must establish data ownership by creating a single source of data, or data warehouse. Multiple sources of data may hinder access or cause “My data is right, your data is wrong” type arguments over data authenticity instead of strategic discussions about performance. Users, physicians in particular, will mistrust or even ignore reports showing discrepancies because the data is derived from different sources. The warehouse is developed by one designated group that extracts core metrics (e.g., the number of relative value units for an individual physician) from an organization’s various information management systems. Reports are then developed using data from this warehouse exclusively, not the original source of the data.
• Timely. Timelines for action have narrowed tremendously in recent years. In the past, it was sufficient to benchmark against data that was a year old; today, it is much more important to use real-time data whenever possible. Data, therefore, must be updated regularly. Physicians who receive poor patient satisfaction scores, for example, will find no value in months-old data. They want to be able to respond to negative feedback as quickly as possible. The same goes for costs. Costs can escalate quickly when physicians order supplies that haven’t been budgeted. Timely data can change behavior faster.
• Trended. Data needs to be “sliced and diced” in many more ways to provide the kind of analytics required in today’s more complicated, competitive environment. Data can be trended by individual physician, specialty or region, for example. Trending at the individual physician level is particularly important to drive changes in behavior that in turn drive performance improvement. Once data moves closer to a real-time environment, it also becomes more important to compare historical internal performance as opposed to benchmarking against industry standards. Certain standards, such as MGMA and peer comparisons — e.g., a medical center compared against other medical centers, or a cardiologist compared against other cardiologists — will always remain relevant. However, for larger organizations with several hundred physicians, internal peer comparison is often more useful than external standards and are what can truly drive best practices.
The complement to strong data is successful adoption — which can be equally complex. If the data is trusted, timely and trended, it becomes more meaningful to users. But these features alone do not guarantee adoption. Organizations must take the time, whether through seminars or informal meetings, to educate users on the processes that were used to generate the data and make sure they understand how to interpret the data.
Finally, once users have access to such data, through such vehicles as dashboard reports, and understand it, they can be held accountable for their performance. If, for example, a primary care physician knows what his target productivity rate is, he can better gauge his own performance and recognize when improvement is required.
Successful adoption of a BI program also means addressing the individual needs of the physician organization, recognizing that each brings different market demographics and physicians, and therefore unique challenges, to the table. Because each physician organization is in a different place in the journey to incorporating curve 1 and curve 2 metrics, its BI program must be customized to meet individual needs.
If the importance of either strong data or user adoption is ignored, a BI program will only absorb precious human and financial resources without providing any real value.
Realizing the benefits of BI
When the appropriate BI tools are used effectively, physician organizations will have repeatable, reliable data that can be used to gauge current performance and drive future needs in order to achieve planned financial and quality results. In essence, an effective BI program ties together the clinical, operational and financial parts of a physician organization to gain a better understanding of where opportunities for improvement exist and where the solutions may be found.
For example, financial data may show that a physician’s productivity, measured in work RVUs, is lower than that of his peers in the practice. The practice may want the physician to see more patients to improve productivity, but the physician may resist, insisting that he does not have time in his schedule for more patient visits.
A review of clinical information on the physician’s patients may reveal the real source of the physician’s subpar productivity. The physician’s distribution of E&M coding, which indicates the acuity of patient cases, may be outside the national norm. The reason may be insufficient documentation. Perhaps the physician is documenting too many visits at a lower acuity than the patient’s true level of illness. The lower acuity is reflected in the physician’s work RVUs, which lowers not only his productivity, but also his compensation and reimbursement for the practice. The real solution may not be increasing the physician’s patient panel, but getting the physician the support and education he needs to improve his documentation and coding to accurately reflect the level of service he provides — all of which is revealed through the BI reporting.
On the near horizon are applications of BI that will drive physician workflow and care outcomes. For instance, BI can be used to identify patterns within a specific patient population, such as diabetics, and develop care protocol decisions based on these findings — all with the intention of improving quality and reducing costs.
If, for instance, data shows that diabetic patients who see their PCPs quarterly have fewer visits to the emergency department than patients with annual or more infrequent visits, care providers can encourage patients to visit their PCPs at least quarterly. This change in patient care would then reduce costs for payers and enable physician organizations to meet quality incentives.
In effect, BI and predictive analytics exchange the “rearview mirror” approach to performance improvement that depends upon the past for the “windshield” view that offers insight and direction into the future.
The process is not without its challenges. Building an effective BI program first and foremost requires understanding the scope of the initiative. The willingness to put forth the investment will be rewarded with outcomes that are not accidental but designed to succeed in meeting the strategic goals of the organization.
