Health care institutions are always hotbeds of recordkeeping, or if one prefers, keeping up with records. Floor-walking doctors, clinicians, pharmacists, technicians, researchers, payers, regulators and administrators are all faced with information needs that today require technology infrastructure and a keen understanding of human goals and needs. To dive into one or two of these health care roles is instant education on the challenges faced today, so we'll approach this story from both the technical and practitioner points of view.

Children's Hospital Boston is a working and teaching 300-bed hospital that treats patients from New England and referrals from around the world. Danny Shaw was hired in 1998 with the then-fashionable title of chief knowledge officer, a role reporting to a CIO who wanted to start a knowledge management (KM) program enterprise-wide. This was Shaw's initial assignment eight years ago, much like the rest of his 30 years of IT work in data warehousing and project management on the fringes of KM.

Shaw tells a story that is familiar to most hospitals, and harks back to when IT was strictly focused on operational aspects, keeping the lights on, running the business. "There wasn't a whole lot of focus and attention given to the back-end components that are my world now. The technology mix was old IBM, a lot of best of breed vendor applications, a mix of the clinical and back office things that were available at the time." Now the business is more consolidated around a smaller collection of larger vendors: PeopleSoft for finance, administration and operations; Cerner for clinical components; and bits of other large vendor apps for ancillary uses by radiology, imaging and the like. "From a technology perspective we're much cleaner if not necessarily leaner. You don't get thinner buying things like PeopleSoft, but we are leaner in the sense there are fewer vendors involved."

Before and during this consolidation, Shaw was tasked with creating a data warehouse to connect disparate front-end systems, which involved building interfaces for each. By doing so, Shaw's team aimed to centralize all of the hospital's clinical, administrative and delivery information, which would help to lower cost while improving care and boosting the hospital's research efforts as progress allowed. "At the start there was not a whole lot of senior level focus on potential business outcomes," he says. "My boss was certainly looking toward those things, but from our [IT] perspective, the intent was to build systems and repositories that could provide answers for outcome questions, whether it's on the business side, the operations side or the clinical side."

The first time around, in 1998, building interfaces was a manual and labor-heavy procedure, which took about 18 months. Shaw attributes some of the pain to less than optimal project planning, but the greater issue was common to technology pioneers, in this case immature extract/transform/load (ETL) tools used to discover and align disparate sources in a data repository. "We looked at most of the available ETL tools in 1998 and they just seemed to be as much trouble to configure as developing our own interfaces," Shaw says. "It's exactly the opposite now, the available toolsets are much better and much easier to use and install and configure." By 2002, as the infrastructure evolved and another data renovation was required, Shaw's group chose to implement Informatica's Power-Center platform for data integration. The second iteration of interfaces, reconstructed in PowerCenter, took weeks instead of months to build and gave Shaw much more confidence to address new projects with still more interfaces. "It was important since PeopleSoft financials were becoming a big part of our world and we had a lot of new clinical systems coming online."

Though he talks technology, Shaw and his group were ultimately preparing themselves to answer to people, the primary customers being clinical researchers and the administration side of the house. "Senior management and finance come to us a lot for scorecard-type reporting, operational outcomes, that sort of thing," he says. "But we probably spend the most time helping clinical researchers sift through something on the order of 20 years of clinical data in the context of whatever research protocol they have going on at the moment." This is where the combination of integrated data infrastructure and bright minds on the medical side come together, and where it's worthwhile to look at the process from both vantage points.

The Researchers

As an infectious disease physician and assistant professor of pediatrics, Dr. Grace Lee spends about 85 percent of her time on research work. Though a joint appointment between the hospital, Harvard Medical School and Harvard Pilgrim Health Care, Lee does population-based research on cost-effectiveness and vaccine policy, (which looks at the implications of whether to implement a vaccine and what happens after a vaccine is recommended). "A lot of the work is about children with cancer or transplants that are compromised in some way and get severe infections. What we haven't done in the past is come up with a comprehensive understanding of how to manage these patients." Traditionally, the needed knowledge has been confined to the physician on service at the time, what they remember of the case and the medical literature of the time. Lee is working with data systems to look at what happened to these patients, their outcomes, and how management may have affected their outcomes in a good or bad way. "First, we had to identify the population of kids who were compromised and that probably took six months. You'd think we'd know who the patients are but we often don't, so we cross-matched billing codes for different kinds of transplants. That was actually very hard to do, but we came up with a master list of patients." A separate effort required pulling from different data systems to a comprehensive database that contains lab studies, microbiological studies, radiological and pathological studies, outpatient and inpatient visits and surgical procedures, something that could never be done with paper records. The interplay with Shaw's team centered on what information was available and how it could be made useful. This was more difficult than it might appear, since the systems involved were not really built for research or clinical purposes. In some cases the quantity and accuracy of data suggest the direction for Lee's research.

In other cases, data is more closely aligned to immediate care. Dr. Maryanne Quinn is a pediatric endocrinologist who does some clinical research, but her primary job as a floor-walking doctor is taking care of children with Type I diabetes. In Quinn's line of work, it is extremely important to track the ongoing progress of diabetes, which makes it crucial to know when the disease first presented itself in the patient. "We have a person who goes back and checks the record, electronic or paper, of when and where the child was diagnosed." Since 2004, every hospital visit is accompanied by a survey of recent symptoms to further build the data set and identify adverse outcomes. "The way we work with [the IS group] is just to get the results every six months so we can know as a group how we're doing in caring for the patient," says Quinn. In support of this goal, a statistician analyzes and shares the data with other clinicians, nutritionists, nurses, social workers and physicians at Children's.

The Gatekeepers

The value of such programs is self-evident, and absolutely dependent on the information systems group. While all sides are enthusiastic about their collaboration, all of the research vetting is done through an investigative review board (IRB), which signs off on any access to patient records. HIPAA regulations were part of that, but even before it came into effect, a Committee on Clinical Investigation was the gatekeeper of all research efforts at the hospital. Neither IT nor researchers have any complaint with the regulatory aspect. "It probably forced us to escalate things in terms of policy, things we would have done anyway but it certainly put those things on the front burner," says Shaw. Today the systems and processes used to generate reports creates a consistent audit trail. "We can tell when Johnny's clinical record was touched, it was because of this IRB protocol for doctors so and so." Even though approval can take months, Quinn agrees that the IRB is a good firewall for protecting families, and protecting researchers from not going where they should without permission.

At the same time, Shaw would like to offer more self-service reporting to researchers to speed the process and allow clinicians to act on their instincts. Children's is in the process of implementing reporting tools from Business Objects to give more self-service, partially for a selfish reason. "It's manageable and part of the job, but we could free up a lot of time by getting out of the loop of trying to build every little report for every person in the hospital who needs to know something," he says. "But with the tools we have available minus Business Objects, there's too much access to the data warehouse to control."

This creates an interesting ownership issue, since Maryanne Quinn finds herself too dependant on IS to want the responsibility of self-service. "In many ways it's [reporting] such a specialty area that it would be difficult to come up with all the parameters you'd set if you didn't have somebody with a technical background. I have found the people in ISD to be incredibly helpful in defining the parameters of what we want to search for." In this case, the clinical and data approach are not always the same. "It might be as simple as gender, you might forget to ask if it's a boy or a girl. I have made that mistake myself."

But Grace Lee likes the idea of self-service. "For a lot of researchers it would be a huge help just to know whether an idea is feasible or not, whether you have the numbers to be able to do a study," she says. "We have built an all-purpose database that can be used for many reasons, which made it a little easier to say, 'sure, we'll take everything and not worry about the actual numbers until later.' But if people have a specific question in mind, it would be so much easier to take a quick look and see how many patients we've seen with diabetes or whatever."

The discussion is generational in part, since knowledge workers who are newer to the workplace are dependent on different tools than their predecessors. At the same time, the discussion is distinctly cultural. "Historically, knowledge sharing has not been done well within hospitals especially," Shaw says. "It has little to do with the lack of need or desire, I don't think I've run into a clinician who is unwilling to do just about anything to maximize their quality of care, but they are very busy people and they are not willing to risk their workflow on new technology if they think it's just for the sake of technology." For Shaw and his peers, the future is about delivering tools to clinicians that make workflow move more smoothly; on the back-end, it will be about building systems that actually work for their customers without burdening them with overhead.

This article was originally published in the August 2006 issue of BI Review.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access