Experts urge use of data for guidance, intelligent action’

Displays enhance frequent measurement, trend charts

Pilots have innumerable gauges in a cockpit to receive constant information, but by comparison, physicians are generally "flying blind" when they see patients, two leading outcomes experts say.

Physicians often don’t know how many of their diabetic patients have their blood sugar under control. They don’t know how long the average wait is. They don’t know how many patients wanted a same-day appointment but couldn’t get one.

To turn data into a kind of "instrument panel," the Dartmouth-Hitchcock Clinic in Lebanon, NH, has created data walls with charts showing key indicators over time. "We have just really begun to use measures for learning in health care — learning about the processes and the systems," says Paul B. Batalden, MD. Baltalden is director of Health Care Improvement Leadership Development at Dartmouth Medical School and one of the first to adapt Edward Deming’s principles of quality management to health care.

Recent advances in outcomes management have focused on using performance indicators to report to consumers. But Batalden notes that those retrospective, broad-scale measures don’t help physicians in day-to-day decision making.

"The purpose of report cards is for judgment and not for learning," says Batalden. "Sometimes there’s a belief that the measures used for performance assessment and judgment will be good measures for learning. They can sometimes wake us up or call our attention [to problems], but they often don’t guide us about the actions we should consider taking."

His colleague, Eugene Nelson, PhD, director of quality education measurement and research for the clinic, helped set up the measurement and analysis that would help various departments guide their care. The key is timeliness and usefulness of the data, he says.

"It’s meant to provide them with information that will help them monitor the vital functions of the system they’re running," he says. "The purpose of the information is to gain insight and take intelligent action."

Are you hitting your goals?

Outside the office of the chairman of cardiovascular surgery at the Dartmouth-Hitchcock Clinic, charts are updated weekly with new data. They show trend lines such as observed vs. expected mortality and complication rates such as sternal wound infections.

The charts also have lines at the upper and lower control values, a statistical measure that shows whether the variation is within a normal range or a true outlier. For example, if sternal wound infections rose above the upper bar, that could indicate a problem for physicians to investigate immediately. Charts could also include a line for a target goal.

"They’ve cut their infection rate from the mid-30% range to around the 15% range," says Nelson. "They’ve looked at a number of quality- and cost-related issues."

How often to collect data and what to collect varies based on the goals of the medical group, he notes. For some measures, such as financial ones, quarterly information might be adequate, he says.

But at Nashua Internal Medicine, which implemented an open access system to allow same-day appointments, daily measurements helped assess progress. Schedulers used a mini-survey with small samples, such as five patients at 10 a.m. and five patients at 2 p.m.

"You can actually see on this morning how many open slots they have at the beginning of the day and how many people didn’t get seen when they wanted to be seen," says Nelson. Meanwhile, overall access data, including satisfaction with access and third next available appointment, was plotted on graphs and displayed on the data wall. (See sample charts, above and p. 99.)

"What we’ve really learned is that there’s no one data element that can truly measure the work that people do," says Vicki Kahn, MHA, senior information consultant with Nashua Internal Medicine. "You have to track a number of them."

Instrument panels’ for doctors and patients

For further detailed information, the department uses a patient satisfaction survey that is mailed to randomly selected patients and tabulated in the clinic’s central office. That system has a three-month time lag.

All of the displayed data reflects a team approach, with information from all clinicians. But medical directors, operations managers, and department heads receive the same trend data by clinician. And soon, physicians will receive their personal "instrument panels," as well.

"What we found is our physicians want to see their individual detail and many of them want to see it against their peer group," says Kahn, who notes that the data are adjusted for such differences as patient age and gender and full-time equivalency of the physician.

The Spine Center has taken the novel approach of using instrument panels to monitor individual patients or groups of patients with a similar diagnosis. Patients answer questions before a visit, and a quickly produced chart shows such items as spine pain, symptoms, functional status, and satisfaction with treatment.

"You can see the change in status from first visit to second visit to third visit," says Nelson.

No computer skills necessary

It may sound onerous to collect, analyze, and present the continuous data. But Nelson insists that the process is not difficult.

The data collection can take place when a receptionist asks patients questions over the phone or when they register, then marks responses on a sheet. Or the practice can use a scannable survey form with outcomes management software.

"Just making a hand-done tally is fine," says Nelson. "That doesn’t take any computer programming. If you have four questions you ask people, you can build little bar charts just by hand, one by one. You can do all of this without touching a computer. Or you an do it in a sophisticated way."

In fact, the display of the data is more important than the method of collection, he says.

"Start with a question that you really want to have answered, [such as] How many people see their own primary care physician? What’s the observed mortality rate vs. the expected mortality rate? How many people have the phone answered in three rings or less?’" he says.

"Provide the answer to that question and use visual methods to answer the question," he says. "You may lay out a clinical process in a simplified version — in six steps — and literally hang the measures off each of the key steps, thereby helping people see how discrete elements in a larger process are performing."

And that is what turns data into an instrument panel that can guide care.