The Bronx (N.Y.) Collaborative, a consortium of three hospital systems serving
Patients at
The Bronx Collaborative's readmissions intervention program involved 500 hospital patients over the age of 50 covered by Medicare, Medicaid or commercial insurance plans offered by EmblemHealth and Healthfirst, the two insurers in the Collaborative. The patients were all admitted to the medicine services of the hospitals and were expected to be discharged home; they were also required to have a working phone number. Patients who were cognitively impaired without a regular caregiver, and who were homeless, or had psychiatric conditions and/or HIV/AIDs, were excluded from the program.
Data collected from the patients at admission was fed into a predictive model created by Treo Solutions, which ranked patients based on their calculated probabilities for readmission. The resulting ranked list directed the program’s staff interventional triage for patients at the highest risk for readmission.
The model ranked patients using a system of weighted factors, including previous inpatient history, previous readmissions, age, gender, race, disease condition and zip code, an important variable for understanding access to care, according to Herb Fillmore, vice president for strategic innovation at Treo Solutions. Predictive variables were selected through logistic regressions of variables used in previously constructed models as well as through consultations with physicians.
"The goal was to determine, among all people who come into the hospital that fit a certain criteria, who you should approach first. There are lots of people and only so many nurses," says Mr. Fillmore.
Once patients had been identified, nurse care transition managers and care transition analysts worked with the patients to implement components of the intervention. Key components included patient education, a follow-up appointment with a physician that was scheduled prior to discharge for one to two weeks after discharge, a post-discharge follow-up call to address patients' lingering concerns and another follow-up call after the scheduled follow-up appointment to ensure patient attendance.
After this process, nurses continued checking in with the patient until the end of the 60-day period. Patient data was stored in a centralized data collection system residing in the
In the end, the group that received two or more interventions had a 33 percent lower 60-day readmission rate than a comparison group that did not receive the interventions. Patients with whom care transition managers were only able to make contact once did not experience a similar decline in readmissions.
"The intervention only worked when we got to patients at least twice," says Anne Meara, RN, associate vice president of Montefiore Care Management. "We need to work on the 'once' group to identify sub-populations within it and to determine the right type of intervention for those people. Not the same type of intervention works for everyone."
She suspects that psychosocial indicators such as literacy levels, mental state and living situation play an important role in determining which types of interventions may be most useful for the group with which caregivers were only able to intervene once.
"The biggest lesson learned here is 'don't wait'. We [The Bronx Collaborative and Treo Solutions] were able we identify people at risk for readmissions on the day of admission. Start the process of care management on the day of admission, and if you think a person is at risk for readmission, make sure you get contact information with multiple sources so you can follow up with that person in the community,” says Mr. Fillmore. "You have to start on the day of admission knowing a person coming into hospital is carrying a lot of baggage in terms of life experiences as well as disease burden. It's not good enough to say a person is high risk, we have to say 'why is that person at high risk.'"
"This is not a check-off box on a discharge summary; it's really important work," says Ms. Meara. "We have to ask if we have the systemic ability to make seamless transitions. Are we staffed to do it? Are staff members trained to do it? And do patients understand why we need what we need from them to keep them out of the hospital? There is still low-hanging fruit here, but to be ultimately successful we have to make these questions part of the standard of care."
More Articles on Quality:
Consumer Reports Sheds Light on Hospitals' Surgical Outcomes With New Ratings
Study: Averse Event Reduction Led to Greater Revenue Loss Than Savings
Patient Safety Tool: Implementing Best Practices for Inpatient Medicine Reconciliation