Blue Oceans for Health   Smarter, Better Healthcare Through New Products, Services, and Ideas

One of the recent cost-control measures that Medicare has been experimenting with is a planned penalty for hospital systems with high readmissions. For example, if the reimbursement data a hospital files with Medicare shows a higher 30-day readmission rate for patients it previously treated, also called “bouncebacks,” a percentage deduction will be made from all future Medicare payments to that hospital. The basis of this new rule stems from a belief that hospitals with high readmission rates are the result of inadequate care continuity practices and not the result of skewed populations being served. For this post, I will leave aside the many criticisms (e.g., for indigent care hospitals, for population outliers) of the new policy and focus on the innovation trends for helping individual hospitals lower their readmission rates.

Leaving so soon? Most quality experts believe readmissions could be reduced if high-risk patients remained as inpatients longer. (Courtesy Hospital & Health Networks)

The research group that I currently work with at Emory University’s Department of Surgery and Georgia State University’s Andrew Young School of Policy Studies view excessive readmissions as the first signs of  correctable errors in the discharge process. These errors can be broadly grouped together as systems-based and decision-related.  Systems-based errors are when a patient is not adequately prepared for discharge because of an internal system failure. For example, the process for discharge at a hospital may not properly instruct a patient on the use of home-oxygen prior to discharge. Decision-related errors are when lack of information or external pressure lead to a patient being discharged too early.

Systems-based discharge errors are currently being addressed through traditional quality improvement mechanisms now being applied in the healthcare setting. However, decision-related discharge errors represent an under-explored opportunity for hospitals to reduce their readmission rates. The general thinking is that if physicians can have a more accurate sense of the likelihood of readmission, patients can be discharged at a more appropriate time while not wasting resources by simply holding on to every patient for a longer time period.

Although approaches have varied, the common wisdom to address decision-related discharge errors has been to take advantage of the latest advances in bioinformatics (i.e., healthcare IT) and apply them in real-time to patient discharge decisions. Currently, the most developed commercial solution is Microsoft’s Amalga healthcare information management platform (3M has a similar IT product oriented more toward quality improvement offices). The basic principle of these systems is for algorithm-based analysis of existing patient data to develop and refine predictive tools for use by a physician at the time of discharge of a future patient. For example, as the system collects data on patients who ahad gallbladder surgery it will become increasingly better at predicting which future gallbladder patients will most likely be readmitted. With such information in hand, a surgeon could potentially flag certain patients as high-risk for readmission and manage their discharge more conservatively.

It is important to note that product offerings like Amalga have not been readily adopted by the mainstream healthcare information management community. Critics note that Microsoft has been struggling to establish itself in healthcare IT due to its late entry and lack of a comprehensive product line. Recent moves by Microsoft signal that the company recognizes these vulnerabilities. A 50/50 joint venture called “Caradigm” between Microsoft (an IT and platform leader) and GE Healthcare (an electronic health record industry veteran) aims to capture many of Microsoft’s latest clinical informatics innovations and package them into existing health system platforms.

Currently, these uses of predictive data analysis are in their infancy. To use a term from business innovation theory, we’re in an “era of ferment.” What I find even more interesting than the technical hurdles firms are currently struggling with is the foreseeable problem on the horizon of how we pair technical expertise (healthcare providers) with these predictive tools. This man-machine interface is easy to dismiss, but I believe that successfully addressing it will be the determinant of a successful dominant design.

Disclosure: I currently receive a graduate research stipend from the National Institutes of Health (1RC4AG039071) for work related to surgical patient readmissions and discharge decision-making.