Hospitals appear to have verydifferent philosophies on where to put patients for post-acutecare.

According to a new study by a team of UCLA researchers that examined 112,620 patientsat 217 hospitals, some hospitals refer patients to in-patientfacilities far more than others.

Read: PPACA reduces uncompensated hospitalcare

While the average hospital referred 20.7 percent of patients tosuch facilities after acute care, the range spanned from a low of2.7 percent to a high of 39.7 percent.

Similarly, some hospitals refer only about 3 percent of patientsto home care, while others refer 40 percent to home care.

The study did not find compelling evidence to suggest thatcertain patterns were associated with higher mortality rates.

Read: Medicare patients pay far more at criticalaccess hospitals

But it did find that hospitals that refer a higher proportion ofpatients to in-patient facilities tend to have higher 30-dayreadmission rates. And the patients typically stay for shorterperiods of time in the hospital.

That suggests to researchers that hospitals may be referringthose recovering from surgeries or illnesses to in-patientfacilities too soon.

Read: Hospital mergers drive upcosts

“These findings suggest that some hospitals may be usingpost-acute care as a substitute for inpatient care,” said Dr. GregSacks, the study’s lead investigator, in a university news release.“This might lead to patients being discharged from the hospitalprematurely, which then results in higher readmission rates.”

The study further suggests that health care payment formulas maybe driving hospitals’ decisions to discharge patients earlier thanthey should.

But the study does not propose any specific solutions, insteadsaying the issue needs to be studied further in an effort to strikea balance between ensuring patients receive adequate care andavoiding unnecessary care.

“Our findings suggest that there is an urgent need to study theappropriate use of post-acute care to develop guidelines to assistpostoperative discharge planning,” it said.