With each issue, Trib+Health brings you an interview with experts on issues related to health care. Here is this week's subject:
Indranil Bardhan is the area coordinator and professor of management information systems in the Naveen Jindal School of Management at the University of Texas at Dallas. He previously served as visiting professor in the Department of Clinical Sciences at UT Southwestern Medical School. Bardhan’s research seeks to evaluate the impact of health IT initiatives on the cost and quality of healthcare delivery, including the development of predictive models for readmissions of congestive heart failure patients.
Editor's note: This interview has been edited for length and clarity.
Trib+Health: Why is patient readmission an issue in healthcare?
Indranil Bardhan: Readmission is a big policy issue today and has been for the last few years because the Centers for Medicare and Medicaid Services, called CMS, instituted a policy in 2012. This stated that specific chronic conditions, one of them being congestive heart failure, which we looked at, they were going to start imposing penalties on hospitals whose readmissions rate were greater than a certain threshold, compared to hospitals in their region. They would basically get a lower rate or reimbursement on their Medicare reimbursements.
Readmission is a significant cost burden because every time a patient comes back for the same diagnosis and has to get treated, that imposes additional costs to the payers. Medicare being the single largest payer that we have in this country, for them because they are paying primarily for elderly people, minimizing readmission rate even by a few percentage points has significant cost benefits. For congestive heart failure alone, it has been estimated in the range of $30 billion and $50 billion a year.
Trib+Health: What did you learn from those rates?
Bardhan: We were trying to predict who was going to be readmitted. We were looking for those who were at the highest risk of being readmitted. We were also trying to estimate many times someone would be readmitted, assuming a patient is likely to be readmitted. We wanted to know the frequency of admission.
If we can predict who is most likely to be readmitted and when those are likely to occur, then hospitals can make appropriate capacity planning decisions such as having the right mix of staff and resources available for those patients.
Trib+Health: What factors were you taking into account to determine who is at risk?
Bardhan: This was a very unique and rich data source, which prior to our study really had not been used to look at readmissions. We actually partnered with UT Southwestern Medical Center as well as the Dallas Fort Worth Hospital Council. The council has a research foundation. They collect data from approximately 72 hospitals in the Dallas/Fort Worth region. If a patient goes to any of these 72 hospitals in the region, their claims data, based on their visit and charges, all of that gets sent to the council. As result, they create a database in which you can track where the patients go. A particular patient may go into a Baylor hospital today and get discharged tomorrow, then 15 days from now they might be admitted to a Dallas Presbyterian Hospital.
Prior to our research, researchers would only be able to track patient visits back to the same hospital or health system. They had no visibility at all as to what was going on outside of the hospital or health system. The uniqueness of the data was that we could track patient visits practically to any of the other hospitals in the region.
Trib+Health: Did you see what you expected in the data?
Bardhan: We did find that some of our research was consistent with earlier work. African American males, for instance, were one of the highest at-risk groups. Medicare patients were highly likely to be readmitted as well.
We had a couple of very interesting findings. One was that we looked very closely at the role of health information technology. Hospitals have spent hundreds of millions of dollars implementing health information systems. In general, cardiology information systems and even hospital scheduling systems have a beneficial impact on readmission rates. Hospitals that have implemented these systems are likely to see a reduction in their readmission rates over time.
Medicare patients are more likely to be readmitted compared to patients who are younger and compared to private payer patients. But once Medicare patients are readmitted, their risk for future readmission goes down significantly. The data was telling us that once these patients were readmitted the first time, after that the hospitals do a really good job of taking care of these patients, so that they are not hit with these penalties from CMS.
The hospitals are actually reacting to the CMS rules. They have actually become more careful in terms of treating Medicare patients because they know they could get hit with these penalties and lose out on reimbursement money.
Trib+Health: What could the practical use of these kinds of prediction be?
Bardhan: For policymakers, this helps to provide an insight into what types of investments they need to make, such as health information technology. Clearly, the results show that this technology does play a really useful role in reducing readmission.
The model we are using can also be developed into a decision support system in the form of software that could be used by physicians and hospital staff to monitor patients and thereby predict the probability of patient readmission.