The Glaring Omission in Healthcare: Patient Satisfaction and Outcome Data
The New England Journal of Medicine recently published an excellent article on patient satisfaction and outcomes data. (“The Patient Experience and Health Outcomes“).
My Toyota maintenance guy sends me a customer satisfaction email automatically after each “clinical encounter” with my cars. He asks me to rate the quality of the service he provided as well as the quality of the outcome (“Did we fix your problem?”) and the cost effectiveness (“Do you feel that our prices were fair, clearly explained beforehand, and understandable?”). Toyota corporate offices review these results in detail and they hold those dealerships totally accountable, with consequences for bad numbers. You would think that the functionality of EMRs that costs millions of dollars could at least match my Toyota maintenance guy.
But in healthcare, we’re different. For one thing, our software is designed around 1970s engineering practices and technology, but more importantly, we hold ourselves in a self-appointed position of superiority. Our patients are not medically qualified to understand the “true” quality of our God-like work. In the afterlife, the mystery and complexity of our services will be revealed to you, so for now, don’t bother your simple little head. Just trust us that we are great, regardless of what you might think.
As a business person and a CIO, the only two metrics that really matter to me are employee satisfaction and customer satisfaction. As fellow CIOs can attest, we are inundated with metrics. Managing a complex IT environment in a healthcare setting is like surfing in a hurricane of metrics, at every layer of technology that we manage, from the data center to the software application. But… the only two metrics that really matter are employee satisfaction and customer satisfaction. Every other metric is a means to those two ends.
The reality is, in healthcare, we’re chickens. We’re afraid to ask the patient what they think of our services and treatments, and we veil that fear in false claims of complexity and scientific validity. In the world of psychology, we perpetuate the “illusion of validity“. In the words of Jack Nicholson in a Few Good Men, we can’t handle the truth, so we avoid the vulnerability.
On a practical level, the current (and misplaced) love affair we have with predictive analytics in healthcare is little more than a teenage romance novel without patient satisfaction and outcomes data. Why are readmissions the current focus of predictive analytics? It’s a bona fide problem in healthcare, true, but it’s also the only patient outcome that we can reasonably measure—we know with data as evidence that the patient has been readmitted. We don’t have the clinical outcomes data to predict anything else. In terms of knowledge based software design, we don’t have a “training set” to teach our predictive algorithms. Good luck to good old Watson at IBM. Without patient outcomes, Watson won’t know what works and what doesn’t work in healthcare. And the satisfaction surveys that sample a portion of our patients 90 days after an encounter? Those don’t count. They are anonymous and almost never tied back to specific care providers and treatment protocols.
Credit to the NEJM for bringing this topic to the front of discussions. Let’s hope that healthcare CEOs in clinics and hospitals will face the truth and start following the culture of Toyota. Ignore the pundits who claim that patient satisfaction is too complicated to measure. Build a survey that passes the common sense test to you, and then insist that your EMR vendor enable its automatic dissemination and integration with your clinical data.
What do you think we have to gain from making better use to patient satisfaction data?