Most healthcare organizations don't need further reminders about the struggle they face implementing information technology. What appear to be routine transactions for retailers and financial services are still often based on paper [in the healthcare industry.] Despite all the costs involved in clinical and administrative process automation, the consensus is that this transformation produces significant efficiency gains. But there's a secondary benefit too, also widely used in other industries, which involves using that transaction data to improve business performance.
That arena is known as business intelligence (BI) and its goal is generally to help managers easily see key performance indicators (KPI). Many healthcare organizations have begun using these techniques very effectively. Aetna and Wellpoint have both priced their products more profitably by better understanding their customers' likely actuarial risk. However, as with transactional systems, the implementation of the applications and data warehousing needed for BI is complex.
Matt Quinn, Healthcare Program Manager at Teradata, a unit of NCR Inc., describes four stages of evolution. In the first stage a single snapshot of data is taken from the online transaction processing (OLTP) system, and typically a report is generated. But creating those reports typically slows the performance of the OLTP. Many different groups within an organization tend to want reports using the same data, so in the second stage separate datamarts are created. Within a health insurer, separate groups looking into fraud, medical management, actuarial risk and pricing, and marketing might all have separate datamarts with separate reporting tools.
While this second stage helps improve the operations problem, it raises some other issues. Multiple datasets may not all reflect the latest information. It's hard to integrate data for analytic purposes across departments. In addition, because only some data is loaded into each data mart, some of the key details may not be available. Quinn says that for insurers, "most organizations understand the problem with datamarts, but no organization doesn't have them."
The next stage is the creation of an enterprise data warehouse (EDW), in which all useful transactional data is warehoused in one physical place, which can be accessed by multiple BI applications and different users. Ultimately the fourth stage is the "active data warehouse," which is used by staff across the organization to make real-time decisions. For example, some airlines now look at customer information to decide whether they'll hold an outbound flight if enough high-value passengers are coming in on a late connection.
But in the real world, Chris Christy, director for healthcare industry marketing, Business Objects, politely suggests there is "broad variation" in EDW adoption, particularly among healthcare providers. In his view, "you need a robust business intelligence tool, including the extraction transformation and load (ETL) capabilities, data quality components, and the ability to access feeds from all the transactional systems."
In other words, BI tools can help improve healthcare performance by taking feeds from whatever sources they need to, while healthcare organizations develop the financial and organizational resources to get to that elusive EDW goal.