I had the pleasure of interviewing Mark Cowen, who is in the Department of Medicine and is the Chief of Clinical Decision Services at St. Joseph Mercy Hospital in Ann Arbor, MI. He and his team sought to analyze the relationship between patient’s admission risk of mortality and their ratings of hospitals on the Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) survey. Their paper was recently published in the Journal of Hospital Medicine.
How did you become interested in the relationship between mortality risk and HCAHPS?
When our teams were thinking about how to organize our care to maximize our patient outcomes, we realized that the risk of mortality at admit was a high predictor of several poor patient outcomes (such as long length of stay and readmissions). At some point, we also had the notion that these may be the same patients that are providing lower scores on HCAHPS survey, and that we could use this information proactively (at the time of admission) to prevent a bad experience.
What method did you use to determine which patients had a high risk of mortality at admission?
The method we use to determine the risk score consists of a variety of clinical and lab values, demographics (such as age and gender), past utilization of services, and key diagnoses. Through an iterative approach, we determine the prediction model, which produces an actual score for the patients risk.
How does the score help you take better care of the patient?
We actually generate the score on admitted patients before a bed can be assigned. We then use the score in a variety of ways to ensure the highest risk patients get the care they need. For example, we use the score to influence an appropriate bed assignment (where high risk patients are assigned an intermediate versus regular bed). We also use the score to ensure the highest risk patients have orders on the chart before the patient arrives to the unit. We also have these patients screened by palliative care. At the time of discharge, we ensure these patients have early follow up (within 7 days), and for those going to skilled nursing facilities, we work with the facility to ensure they know what to expect when the patients get there.
Why do you think higher risk patients rated hospitals lower in physician communication and hospital responsiveness?
It is hard to say for sure, but it seems these high-risk patients have very different needs than lower risk patients. Without any targeted interventions (such as those we described above), we do not seem to be recognizing or meeting their needs within the domains of physician communication and responsiveness.
Where there any unexpected findings in your study?
We were surprised at the finding that higher risk patients rated us higher in the domain of discharge instructions. It may be that these patients are recognized as needing extra help and planning at the time of discharge, and that we have been able to recognize and meet their discharge needs better than their needs during the hospital stay.
How should hospitalists use the information from your study to change their practice?
Even if you don’t have a fancy regression model to determine the risk of patients, some formal or informal risk assessment at the time of admission would be good for hospitalists to put in place (e.g. asking the question, “Would you be surprised if this patient died in the next year?”). This would help them more objectively determine which patients need extra help and time, to improve their experience and their outcomes.
Where does this take you with respect to future research efforts?
Our teams are very interested in using this risk of mortality to identify patients and develop effective interventions that improve outcomes and reduce unnecessary utilization of services.
Mark Cowen, MD is an Internist and the Chief of Clinical Decision Services at St. Joseph Mercy Hospital in Ann Arbor, MI.