By Tripp Underwood

Government agencies in charge of determining what constitutes efficient, quality health care have taken to looking at hospital readmission rates. On the surface, this makes perfect sense: If patients are continually being readmitted to a hospital, that hospital must not be doing enough to treat patients appropriately on the first go-round. But new research indicates that relying too heavily on readmissions as an efficiency metric may wrongly put some health care institutions – particularly pediatric hospitals – at a disadvantage.

At the American Academy of Pediatrics (AAP) meeting this week, a team led by James Gay, MD, medical director of Utilization and Case Management at Monroe Carell Jr. Children’s Hospital at Vanderbilt, presented research involving more than 1 million patients cared for at children’s hospitals across the country. The team, which also included Boston Children’s Mark Neuman, MD, MPH, posed this question: If hospital ratings are going to be tied so strongly to readmission rates, shouldn’t that rating system recognize the difference between potentially preventable readmissions (PPRs) and those that are unavoidable?

Currently, some state Medicaid programs use software such as 3M PPR, developed for this exact purpose. Like the basic idea that inspired it, the 3M PPR system works well on principle. However, according to Gay and colleagues, it doesn’t capture all the nuances of what makes a readmission preventable or not.  That’s especially true for pediatric hospitals, where children often require multiple hospitalizations throughout the course of treatment. This is troubling, considering that some states are levying financial penalties when readmission rates are deemed too high.

"There’s no doubt high readmission rates can be an indication of poor care," says Neuman. "But by no means is it the only quality measure. If readmission rates are going to hold so much water in terms of financially penalizing health care institutions, we need to be sure the metrics in place are as accurate as possible."

The team reviewed the records of 58 children’s hospitals, looking at more than 1.5 million hospitalizations for 1.2 million unique patients. They examined readmission rates for 314 diagnostic categories, using criteria defined by 3M PPR, and compared them to all readmission cases in the country to see how the "preventable" label was being applied. (Normal newborns, deaths and patients who left the hospital against the advice of their doctors were excluded.) While they were unable to prove whether a readmission was preventable or not, their intent was to document potential shortcomings of the PPR software that could unduly penalize some hospitals.

The overall 30-day readmission rate in the study was 13 percent. The 3M PPR algorithm deemed 60 percent of these readmissions to be unpreventable; this included all readmissions for chemotherapy, acute leukemia, and cystic fibrosis, as well as more than 40 percent of readmissions for seizures, gastroenteritis, central line and urinary tract infections and failure to thrive. Of the readmissions labeled preventable, 80 percent were related to sickle cell crisis, bronchiolitis, ventricular shunt procedures, asthma, appendectomy and pneumonia.

Given the data, the team concluded that 3M PPR does a decent job but is far from perfect. The software repeatedly excluded from the "preventable" category some diagnoses that may benefit from quality improvements; for instance, it did not identify any readmissions for cystic fibrosis as potentially preventable, although it stands to reason that some could have been. Likewise, 3M PPR deemed readmissions for some common conditions like bronchiolitis to be potentially preventable—but according to Gay, most pediatricians would argue strongly that the need for readmission can’t be predicted that accurately for bronchiolitis.

Because of these discrepancies, using a single PPR standard isn’t accurate enough for all hospitals, especially pediatric institutions that typically care for children who are critically ill or have several comorbidities. This will become a real concern if more and more state payers use PPRs as a metric.

Gay, Neuman and the rest of their team are calling for more research to validate 3M PPR and similar systems, especially in situations where pediatric institutions are facing financial penalties.

"Boston Children’s and many of the other U.S. pediatric hospitals care for some of the sickest children in the world," Neuman says. "These kids often require far more care and resources than a typical patient cared for at a non-pediatric hospital. Given that disparity, holding children’s hospitals accountable to a universal readmission metric with clear imperfections isn’t fair."