June 23, 2008

Buried Treasure
By Elizabeth S. Roop
For The Record
Vol. 20 No. 13 P. 14

Many hospitals are not taking advantage of one of their most precious commodities: data. Unearthing these nuggets of knowledge can lend an up-close perspective on how to improve operations and efficiencies.

Performance and quality-based initiatives, along with a push for transparent pricing and broad implementation of clinical and financial information systems, have provided hospitals with vast amounts of data on every aspect of their business—data that could be harnessed to gain unprecedented insights into clinical and administrative operations.

The sticking point is that the data are often dispersed across multiple databases with little standardization among them. As a result, many facilities are left in the dark about just how powerful that information could be were it aggregated, mined, and proactively analyzed at the enterprise level.

“The aggregation of that data in such a way so they can manage it across the boundaries of the silos is really the trick,” says Ivo Nelson, vice president of IBM Global Healthcare Provider at IBM Healthcare and Life Sciences. “We’re coming into an era now where [hospitals] have to be accountable for information they are not used to being accountable for. They’re starting to get paid based on quality. They have a lot of data that they need in order to better manage their organizations—to be able to manage not just quality but also processes [to provide] better information that allows clinicians to make better decisions.”

Mining for Gold
For that reason, many hospitals have begun taking advantage of their data through the application of basic business intelligence reporting. By tapping into the data contained within individual systems, facilities are provided with a retrospective view of their operations, which provides insights into how they can improve quality and safety, reduce costs, increase revenues, and remain compliant.

For example, facilities and providers utilizing IntelliDose, an oncology clinical information solution from IntrinsiQ that automates chemotherapy order writing and integrates with electronic medical records (EMRs) and other systems, are able to generate highly specific reports that can guide multiple clinical and administrative decisions.

“Some users are looking at the drug utilization report. They can have it show them what number of vials of a particular drug they will need for their next day’s chemotherapy administration, so they can order from their suppliers,” says Joe Cooper, vice president of research and analysis and senior analyst with IntrinsiQ. “There is no point in having $1 million worth of products on the shelf that there is no demand for.”

Others use the data to evaluate their business practices. The numbers reveal everything from utilization and reimbursement rates for a particular drug to how quickly they are moving patients through the process.

While there is significant value in the reports IntelliDose users are currently able to generate, the real power comes when the data it contains are integrated with other systems, so it can be aggregated to provide a systemwide or consortiumwide view.

“When you can look at what drugs are being prescribed for what tumors and what lines of therapy across a whole range of hospitals, practices, and academic centers, that’s where it gets really powerful and useful,” says IntrinsiQ Chief Operating Officer Ted Owens.

It is that linkage across systems that gives facilities the ability to delve deep into their data to reveal a wealth of information about processes and operations on both the clinical and administrative sides of the house. The aggregation of disparate data sets the stage for the next generation of data mining: the application of predictive modeling and other analytics to drive enterprise-level decision making to improve overall performance and operations.

“Transaction systems are just helping isolated processes work better and faster. The analytics is where you get the real insight into some of the major change that needs to take place,” says Nelson.

On the clinical side, analytics exposes the good and the bad in terms of practice patterns, so facilities can gain a clear understanding of the primary drivers around treatment and efficacy outcomes. It can be used to forecast diagnoses and subsequent increases in consumption of services based on drug utilization and patient demographics.

Analytics allows facilities to track massive amounts of data generated from clinical activities to identify trends and irregularities and analyze risk, which helps identify the most efficient practice patterns. It also provides the foundation for disease management protocols, which require significant information sharing and analysis between provider and payer to achieve maximum outcomes.

“There is also a good case to be made for using clinical information for risk and cost management in that you can find the determinants of primary care outcomes and also primary care risks,” says Jason Burke, worldwide director of health and life sciences at SAS, a provider of business intelligence and analytical software and services.

On the administrative side, analytics can be applied to financial and claims data to identify areas for cost savings and  enhance reimbursement strategies. By analyzing practice and payment patterns, facilities can address utilization issues, establish more accurate rate structures, and identify practice patterns within profitable service lines that can be applied to underperforming units.

Adding claims data to the picture allows for more accurate forecasting of collections and revenues, detection, and elimination of errors and fraudulent claims, as well as improved claims management and payer response times.

“We see in a lot of other market segments a very proactive approach to looking at operational data in the context of forecasting, not just revenue performance but things like supply-related issues and better inventory control,” says Burke. “The more forward-looking companies are being more aggressive in taking on analytics as part of their own culture and part of their own business operations. There is also a growing recognition from a lot of our customers that analytics is not about creating reports. It’s about obtaining sustainable business or clinical advantages through an ongoing application of statistical sciences.”

Missing Pieces
When consideration is given to all the benefits that could be derived from the application of analytics to the data hospitals already have, the question is why it is not happening with more frequency.

As is often the case, there is no single answer. The complex relationship between hospitals and the data they collect and how they are collected is one problem.

Nelson notes that IBM engineers, many of whom come into healthcare from other industries, say that the data models within healthcare are five to 10 times more complex because of the number of data elements and the relationship of the data. That complexity has put healthcare years behind other industries, such as retail, that have refined their analytics to the point where they can predict customer behaviors with pinpoint accuracy.

“The fact that we’re behind isn’t because we have a bunch of stupid people out there working in hospitals. It really is very complex. Hospitals are complex from a data management perspective. That is one of the obstacles. They have to realize that this isn’t easy. But on the other hand, it is absolutely necessary,” Nelson says. “[Other] industries really have it nailed because they’ve done it for so long that they know how to use data to optimize their operations. That is what we need in healthcare, but the complexity means it is going to take us a little longer to get there.”

Feeding the complexity is first and foremost the disparity of the systems used to collect data throughout hospitals. With each unit collecting its own data for its own purposes, there tends to be a lack of standardization among specific data elements, even within hospitals that have integrated information systems.

What is meaningful to the financial side of the house is meaningless to the clinical side and vice versa. Thus, the first step for many facilities is to find a way to integrate the data so that they can be transformed into intelligent information for use at the enterprise level.

In this regard, vendors are stepping up to the plate. IBM recently launched Enterprise Health Analytics, a suite of services, infrastructure, and tools that converts discrete data from multiple sources so that they can be combined and analyzed from the enterprise perspective.

“If they don’t have a foundation built that sits on top of all these databases so that they are viewing it as an enterprise project, they just never get there. It’s almost like they are trying to implement a different application every time they pull a report as opposed to creating an infrastructure so they can aggregate, standardize, and really manage the integrity of the data,” says Nelson.

SAS offers SAS Enterprise Miner, which incorporates a suite of integrated data-mining tools with a user-friendly interface to enable even those without extensive programming experience to quickly perform sophisticated analysis.

“The reality is that there are software and data standards available that facilitate bringing that information together now in a way that maintains a good level of context and quality and allows decision making to occur,” says Burke.

However, because many hospitals are still working through the business intelligence aspects of data mining, they have not yet reached a point where they have the skills or knowledge to transition to enterprisewide analytics. Even when in-house talent is available, they are often overwhelmed with the labor-intensive process of generating reports from so many different systems, which delays the ability to advance into analytics.

“By and large, a lot of organizations are still struggling with basic business intelligence applications like generating reports, which are an interesting and necessary component of gaining business insight but, unfortunately, it’s very retrospective instead of proactive,” says Burke.

Many also believe that without fully integrated EMRs in place, it is not possible to perform enterprise analytics. Burke notes that while there are certain insights that can only come through the adoption of EMRs, there is a “whole raft of business insights and organizational improvements that are not contingent upon the clinical information. They’re contingent upon having good enterprise systems in place that have good analytics flowing through them.”

Nelson goes one step further, saying that not only do hospitals have ample electronic information available even without EMRs but also that the information they do have and how they want to use it should play a role in system selection.

“You don’t have to wait for the holy grail of the EMR to happen,” he says. “Almost every hospital in the country has some activity right now where they are taking [information] off paper and putting it into computers. We’re going through a massive modernization process, and it is important that they understand what type of information they have, what type of reporting they have, and how they are going to use that to improve decision making. It is important that they understand that right now, before they implement their systems, or they’ll just have to start over.”

Taking Baby Steps
Whether a hospital is ready for full-scale enterprise analytics or simply wants to start the process with a single project, the key is to start with a clear definition of objectives and secure top-level support.

“In the absence of that, what typically happens is we see pockets of analytics emerging throughout the organization and there isn’t any scale,” says Burke. “The organization struggles to figure out how to incorporate those insights into the day-to-day rhythm of the business.”

In addition to executive support, priorities must be aligned to the business process and goals to ensure that the appropriate resources are dedicated to the initiative. Because analytics is a cross-functional initiative, it requires diverse representation, including clinical, administrative, human resources, and IT. By bringing different perspectives to the table, the right resources can be mobilized to achieve the desired outcomes, whether clinical, administrative, or both.

Putting an enterprise analytics plan in place, including identifying the priorities, determining the human and technical resources it will require, and evaluating the type and quality of data necessary to achieve objectives, is something every facility can and should be doing now.

Without at least that level of understanding, “They won’t be positioned to be successful in the future world of analytics,” says Nelson. “I don’t think it is a question of if; it is a question of when. I believe analytics is going to be the next big thing in healthcare. We’re coming out of an era that has been going on for the past 15 years, which has been the automation of the medical record. Before that, it was revenue cycles and billing systems. The next wave is going to be aggregating all of that data and using analytics to better run the organization. That’s when we’ll really see true transformation. That is when we’ll see an opportunity for dramatic change in healthcare.”

— Elizabeth S. Roop is a Tampa, Fla.-based freelance writer specializing in healthcare and HIT.