The science of statistical analysis is an empirical one, which is to say that a given question will yield a reliable number based on a formula. In business, such numbers are used to measure performance, identify shortcomings or uncover opportunities. Math, however, doesn't always cleanly reconcile itself to desired outcomes Take the health care industry, where the recently adopted buzz phrase "pay for performance" has been applied to measure not just the timely application of care to a patient, but the impact of that treatment on the human condition. "The model in managed care today says I'm not going to pay you for doing the work, I'm going to pay you for getting the right outcome," says Meg Aranow, CIO at Boston Medical Center, a 540-bed hospital and teaching facility affiliated with Boston University. "We won't pay you to just see the patient, we'll pay you to make the patient better. That's rational. But in health care, there's a lot of argument about what 'better' means."
The very same issue is under examination within many mature business intelligence initiatives where participants are asking whether accepted performance metrics are actually serving desired business outcomes. Does desirable cost takeout have a larger negative impact on the quality of our goods and services? Are the goals of executives aligned with the values customers associate with our business? Such questions can be surveyed, but in the end, the business needs to know why it is in business in the first place. (Aranow will address this question and others in her upcoming presentation at BI Forum, June 4-6 at the Argent Hotel in San Francisco.)
Often, it boils down to managing expectations, both internally and externally. In health care, Aranow says, as a nation we are forever upping the ante because of constant advances in technology, and everyone logically expecting the best care the system has to offer. "Like any new technology that hits the marketplace before there's significant market share and before the R&D has been recouped, it's pretty crazy." There are ways to make the effect of health care technology qualitative, barometers and proxies, but these involve imperfect human inferences about what measurable event equals "health." Boston Medical Center uses such proxies -- such as avoidance of a future hospital stay -- but it's an imperfect equation.
"I think the biggest problem in the industry right now is that there are no standards for standards," says Aranow. "We have data standards, but we don't have standard measures of quality. That's very hard for us as an institution that answers to many different masters, all of whom measure quality in a different way." It's very expensive for Aranow's people to respond to each of many measures that face the clinical community as well as regulators and payers, all of whom have their own algorithms for quality. To underscore this, Aronow has seen Boston Medical's own publicly Internet-available Medicare data used by private hospitals to benchmark their own performance. "They combine our Medicare claims data with what they call 'proprietary algorithms' and magically turn it into what they call 'quality data.' We have no control over it, we have no idea how they're using it, it's really kind of wild and wooly out there."
In the end, companies make their own standards, which is appropriate, but not necessarily in the comparative sense. Aranow has even had to come up with a defense of another for-profit institution's quality data benchmarked subjectively against her own. "Even without saying anyone is ill-intended, it obviously will be expensive if every health care institution has to try to respond to everything that's out there on the Internet representing quality."
But managing expectations also means being ahead of the operational curve. Aranow's presentation at BI Forum will also look at the way health care -- admittedly late to the BI table -- is coming to grips with the issues common to all industries. "We're generally very data rich, but we have not necessarily focused our attention on organizing and structuring that data in a way that would give us the kind of predictive values that other industries use. Predictive values for us is less about when we will run out of inventory, and more about, 'if many patients present symptoms, what kinds of predictions can I make about the next person who presents with those same issues.'" In health care, this kind of predictive modeling is called evidence-based medicine. "I look back and say, 'under these circumstances, what's my predictable outcome? If I want that outcome to change, what do I have to do to change that outcome?'" To do this, Aranow's team will look at lab results, medications, patient height and weight, past history and potentially, a host of other things. "The things we want to do, because we're interested in improving health and improving health outcomes, are the same things that the managed care companies want us to do. So right now, in the beginnings of the industry I'd say, the incentives are really aligned between the payers and the providers and there's lots of fertile territory to both reduce expense and improve outcomes at the same time."
That is not to say all parties will measure it in the same way. The best defense, Aranow says, is a good offense. "We've built this data warehouse to make us as nimble as possible to respond to what is really an emerging market for us. We are not always sure where it's going to go, who's going to ask what. We want to be as well positioned as possible to respond to all legitimate inquiries and requests and we also want to inform ourselves so that we can kind of influence policy when we have that opportunity."
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
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access