Reducing readmissions in IBD patients: A predictive model

Hospital readmission rates have been under scrutiny as a key driver of excessive healthcare spending. Why are so many people sent home from the hospital only to return within 30 days?

“The Centers for Medicare and Medicaid Services (CMS) believes a significant portion of 30-day readmissions stem from a lack of care coordination and from poor transitions of care from inpatient to outpatient settings,” says Maged Rizk, MD, a gastroenterologist at Cleveland Clinic. “And they’re right. There are some patient factors, but there are definitely processes where we can do better by our patients.”

Many patients in Dr. Rizk’s practice who are readmitted within 30 days suffer from inflammatory bowel disease (IBD). So Dr. Rizk set out to create a predictive model that would help his team identify which IBD patients were most likely to be readmitted.

His research, which was presented at ACG 2016, lays the groundwork for improving discharge and follow-up processes for this population with the goal of reducing preventable readmissions.

Using prospective data, Dr. Rizk designed a predictive model that is 73 percent accurate.

“We collected lots of different information, including disease-specific, health severity and process oriented indicators,” he explains.

The (somewhat surprising) results

In Dr. Rizk’s study, the most predictive factors for IBD readmission within 30 days were:

• Whether the patient was admitted for pain
• If there was a documented IBD flare during inpatient stay
• The number of medications a patient was on
• Whether or not a patient was prescribed narcotics
• Length of stay
• Whether or not a patient was discharged on a pain management regimen

“Two findings surprised me quite a bit,” says Dr. Rizk. “First, patients who were on a pain management regimen were actually more likely to be readmitted than patients with no such regimen. And second, patients who were prescribed narcotics during the admission were less likely to be readmitted. In a way, it speaks to the idea that opioid therapy should not be part of the maintenance therapy a patient receives, and that those who do have flares do need pain management when admitted to the hospital.”

Putting the data to work

Dr. Rizk’s model provides new guidance for discharge planning.

For example, “if a patient is being discharged on a pain control plan, I would make sure he or she has an appointment with a pain management doctor,” says Dr. Rizk. “Also, I would have the practice case manager call this patient more frequently.”

“A care coordinator for a large practice such as ours can’t check-in with every single patient,” explains Dr. Rizk, “but we can now target the high-risk patients and give them the attention they need. In the future, health enabling technology may also help with care coordination in between regularly scheduled appointments.

“Readmissions have definitely put a strain on our resources,” Dr. Rizk affirms. “To address this problem, we need to look at specific patient populations and attend to their unique needs.”