If ordinary DRGs don’t give you enough detail, try this set of codes The University of Michigan Health System used APR-DRGs to pinpoint problems with performance
Published in the December 2004 issue of Today’s Hospitalist
Aileen Sedman, MD, associate chief for clinical affairs at the University of Michigan Health System, was sifting through a two-foot high stack of reports that had volumes of information about her hospital’s length of stay and costs, but little in the way of detail that she could apply to the task at hand.
The problem? The data, which tracked her hospital’s performance via diagnosis-related groups (DRGs), gave her a big-picture view on how physicians fared when treating diseases like asthma. The information, however, provided little to no detail on the severity of illness
The APR-DRG system divides DRGs into four levels of severity to identify those patients who are receiving good care.
of those patients. As a result, the data were nearly useless in comparing length of stay, readmission rates and costs to other hospitals.
“Because there was no acuity adjustment,” explained Dr. Sedman, “the data weren’t appropriate. If someone had told me that our length of stay was lower than that of another hospital, it wouldn’t really help me.”
Then she learned about a different kind of DRG, one that could give her exactly the type of detail she was looking for. Known as all-patient refined DRGs, or APR-DRGs, the code set offers remarkably more detail than traditional DRGs. And in Dr. Sedman’s case, those codes would prove to be the key to not only unlocking information about her hospital’s performance, but improving it.
Under APR-DRGs, a condition like asthma is broken down into four levels of severity: minor, moderate, major and extreme severity of illness. The codes allow clinician/ administrators like Dr. Sedman to pinpoint exactly which patients with a certain disease—low-complexity level 1 patients or high-complexity level 4 patients—are experiencing longer stays or driving up costs.
Using an 18-step algorithm based on an analysis of all the ICD-9 codes listed for each patient, the APR-DRG system scrutinizes all the diagnoses physicians use to document their care. The algorithm assigns a level of acuity to patients based on that analysis.
“If a new procedure shows up in a physician’s documentation,” Dr. Sedman explains, “the system takes that into account. Or if another diagnosis that’s higher acuity than your previous diagnosis comes in with your ICD-9 codes, the system reassigns the acuity.”
While most hospital administrators are familiar with APR-DRGs, which were created in the 1980s, Dr. Sedman says that relatively few physicians are aware of this system and how it can be used to redesign clinical care processes. In part, that’s what motivated Dr. Sedman, a pediatric nephrologist who is currently professor emeritus at the medical school, to use APR-DRGs as the basis for a quality improvement project.
Dr. Sedman started out by examining how well the University of Michigan Mott Children’s Hospital cared for a single condition: noncomplicated asthma (APR-DRG 141.1). She then compared the hospital’s results to data from the National Association of Children’s Hospitals and Related Institutions. The group’s database tracks the performance of children’s hospitals across the country, analyzing data from 1.5 million discharges a year.
Dr. Sedman and her colleagues were surprised to learn that while the hospital performed very well when it came to complicated asthma cases (levels 2 through 4), there was room for improvement on level 1 patients. Mott posted a slightly longer length of stay, slightly higher costs and a slightly higher readmission rate.
Dr. Sedman says the explanation for this difference quickly became apparent: Level 1 cases are admitted to the pediatric service, which uses general pediatricians, not pulmonologists, as attendings. As a result, pulmonologists are often called in to consult and don’t see the patient immediately. “That first 24 hours of care is crucial,” she explains.
An obvious solution would be to admit all noncomplicated asthma patients directly to pulmonology, but Dr. Sedman says that would clash with Mott’s teaching mission. The hospital must have a general pediatric service to give its housestaff access to a wide range of patients, she explains, and that includes level 1 cases of noncomplicated asthma.
As a result, hospital officials decided to design a system to change care for level 1 asthma patients. Staff from the office of clinical affairs worked with pulmonologists and hospitalists to create standardized orders that took into account everything they would want done to a level 1 asthma patient. One of the key features was a standardized protocol to appropriately wean patients from oxygen and nebulized medications.
“Instead of requiring that a physician come and stand by the bedside two or three times a day to check on the patient before any orders were changed,” Dr. Sedman explains, “we set up a standard protocol. It says if the patient’s respiratory rate drops to a certain level or the oxygen saturation rate is a certain level, you can turn down the oxygen or decrease the nebulization treatments. This allows nurses and respiratory therapists to wean patients according to what pulmonologists feel are the appropriate criteria. You don’t have to reinvent the wheel every time a patient comes in.”
The hospital also created a system to make sure that asthma educators would automatically see patients upon admission. Instead of waiting for a call from a physician—which would sometimes happen hours after a patient was admitted—the admission itself now triggers a visit from an asthma educator.
“All of a sudden,” Dr. Sedman says, “our asthma educator was hearing about these patients immediately, instead of toward the end of the first day of admission.”
She explains that a timely visit is critical because it sets many different processes in motion. “If a child ends up being hospitalized for asthma, in 80 percent of the cases,” Dr. Sedman says, the family needs help. Some families may not have the appropriate medications or the right nebulizer, or they may not have received the right education on how to administer medications.
“Having the asthma educator sit down with them and go over all the issues very early in the admission made us aware of what needed to be done before the patient went home,” Dr. Sedman says. “We’ve moved that process up 24 hours.”
After implementing the changes, the hospital tracked its performance data from 1999 to 2002. Dr. Sedman used APR-DRG codes to measure the impact of the changes and found that care for level 1 noncomplicated cases of asthma improved in three major ways:
• Length of stay. The hospital reduced its length of stay for level 1 cases of the disease from 2.16 days in 1999 to 1.75 days in 2002. The average length of stay for the national database in 2002, by comparison, was two days.
• Costs. Between 1999 and 2002, costs for treating level 1 cases of noncomplicated asthma rose 12 percent at Mott, compared to an 18 percent increase nationally.
• Readmission rates. Readmission rates dropped from 2.97 percent in 1999 to 0.80 percent in 2002. The national average was about 2 percent.
(The complete results of the research project were published in the October 2004 issue of Pediatrics.)
While the project improved care and reduced costs, Dr. Sedman, who today serves as a medical advisor to the National Association of Children’s Hospitals, says another important result was demonstrating that clinicians can find the detail they need in APR-DRGs to identify problems.
“An important message of this paper is that we are capable of dissecting out these acuity levels,” she explains. And although she used pediatric versions of the codes, she says that the same thing could easily be done in adult care settings.
Discerning severity of illness was particularly important in her study, Dr. Sedman notes. If she had simply looked at standard DRGs for all cases of noncomplicated asthma, she explains, she would never have realized there was room for improvement in level 1 cases. The hospital’s superior performance on level 2, 3 and 4 cases of the disease, she says, would have hidden the increased length of stay, costs and readmission rates for the level 1 cases.
While most hospital administrators use APR-DRGs in their work, Dr. Sedman says, many clinicians don’t even know that this methodology exists. She adds that for the system to be used to its full potential, clinicians need to get involved.
As an example of just how powerful these codes can be in tracking and improving performance, Dr. Sedman points to what may be her biggest sign of success: Physicians sometimes approach her asking for data on their performance. “In the years I’ve been doing this,” she says, “we’ve started to have people actually ask for their APR-DRG reports.”