Press Ganey Looks to Science to Understand Patients’ Feelings

Press Ganey CEO Rick Siegrist and Vice President of Product Analytics Susan Madden told attendees at the National Client Conference in Orlando this morning about their work in the area of sentiment analysis – the science of analyzing and categorizing patients' written and spoken comments about their care experiences.

Siegrist said that using natural language software to categorize and assess the emotions found in patient comments is “the secret sauce to really moving the needle on satisfaction from good to very good. It drives competitive advantage in health care.”

While other industries have been using sentiment analysis for years, health care has done little in this area, a curious fact given that emotions in health care are stronger than in any other industry, a result of life and death issues and the complexity of the interactions, he said.

Analyzing the intensity of emotions in comments gives providers of care a way to trace what really drives dissatisfaction with a care encounter. With thousands of comments analyzed, trends emerge that can pinpoint precisely the root cause of problems in service.

In the era of social media, there is new urgency to uncovering what's behind emotional reactions, Siegrist said. “People who exhibit strongly positive or negative sentiments are much more likely to express those views to others. Through Facebook and blogs, people aren’t reaching five friends anymore, but 50 people or more,” he said.

Madden echoed that finding, saying that research shows that people who say they are unlikely to recommend a hospital or medical practice for care are far more likely to comment on surveys.

That’s why Press Ganey is hustling to bring a sentiment analysis product to market “very soon,” Siegrist said. He said that the results of that work may turn out to be more powerful even than patient satisfaction scores in assessing the patient experience of care and providing actionable information for improvement.

Madden described the results of Press Ganey's ongoing, 16-hospital pilot project on sentiment analysis. Using the natural language software, Madden and her team analyzed thousands of patient comments from inpatient and emergency department surveys. The software takes keywords in comments and places them in categories. The team placed sentiments in a value scale from negative 5 to plus 5, arriving at sentiment scores.

There are four models for slicing and dicing the results, she said, but so far none has been settled on. Press Ganey is looking to clients to help determine what model will be used, she said.

Madden illustrated the root cause analysis of patient sentiment by looking at pain management. Through the sentiment rankings, Press Ganey has looked at where pain management was a particular issue, isolating at one hospital several units with chronic and intense pain control problems. The same analysis can be done by DRG and insurance status, among a number of other analyses, she said.

Notably, the pilot project has found that positively rated comments are associated with higher mean patient satisfaction scores, and a positive correlation has been found between sentiment scores and HCAHPS scores. “Those findings may seem to be predictable, but they validated our work on sentiment analysis,” she said.