We asked AcademyHealth member and 2012 Presidential Scholar Robert Lieberthal, Ph.D., Assistant Professor, Jefferson School of Population Health, to share some thoughts on his recently published work on predicting hospital outcomes and the degree to which his findings aligned with patient reported data. His guest post is below.
There is currently a focus by many healthcare stakeholders in disseminating evidence on the quality of healthcare directly to patients in order to improve quality, and ultimately improve the value of the care delivered in our healthcare system. In 2001, in Crossing the Quality Chasm, the IOM identified “transparency of system performance… (and) …patient satisfaction” as one key to the redesign of our health system. Since the publication of that report, there is a great deal more data available regarding quality of care, such as CMS’s Hospital Compare website . This data is so rich and so detailed that there is now a need to find ways to organize and make use of the large number of quality measures in this publicly available data source. That was the impetus for a study I undertook, in collaborations with Dominique Comer, that was recently published in the journal Risk Management and Insurance Review.
Given the data available, the setting we had in mind was health insurance. We used a statistical method, Pridit, to aggregate data on hospital characteristics, process measures of care, outcomes of care, and patient satisfaction scores. We then determined whether scores using data from 2008-2010 were correlated with risk adjusted mortality outcomes from 2011. Our study was designed to mirror the dilemma health insurers face when setting reimbursement rates, creating preferred provider networks, and implementing pay for performance programs. For example, health insurers working in 2013 to set hospital reimbursement rates for 2014 typically use data from 2012 and earlier, since claims for care lag the delivery of care. Thus, health insurers want to ensure that any methodology for determining hospital quality is both predictive of important outcomes such as risk-adjusted mortality and stable over a number of years.
One of the key findings from our study was about the relationship between patient satisfaction and outcomes. We found that the highest quality hospitals (based on the aggregate data outlined above) do not receive the highest patient satisfaction ratings as measured by HCAHPS data. In fact, those hospitals that received patient satisfaction scores “in the middle” in a given year, rather than the highest or lowest ratings, were more likely to be of high quality as determined by our methodology. That is despite the fact that the patient satisfaction measures, such as cleanliness and ability of doctors and nurses to communicate well, would seem like common sense ways to improve the quality of care. High quality hospitals as measured in 2008-2010 were also more likely to have lower than average risk-adjusted mortality rates in 2011, which speaks to the validity of our methodology for predicting hospital outcomes.
Upon reflection, our study evoked a fundamental question about what health insurers should do with patient reported data: what if the providers that patients prefer are not the ones that produce the best outcomes? In that case, a health insurer that tried to improve outcomes by aggressively “steering” patients to high quality hospitals could itself receive lower satisfaction scores from plan members. Other researchers have identified this tension between patient satisfaction and quality. For example, Glyn Elwyn and colleagues, writing in the British Medical Journal, pointed out that “Patients’ preferences do not exactly overlap with good quality care…indeed, some of their wishes, particularly when uninformed or ill informed can be detrimental.”  Thus achieving higher value through improved outcomes alone will not work if it ignores the importance of the patient experience. Indeed, the greatest challenge for improving healthcare outcomes may lie in communicating the results of peer reviewed research for patients, empowering them to use evidence on outcomes to make important healthcare decisions.
1. Elwyn, G., Buetow, S., Hibbard, J., & Wensing, M. (2007). Measuring Quality through Performance: Respecting the subjective: quality measurement from the patient’s perspective. BMJ: British Medical Journal, 335(7628), 1021.