When you hear the terms “clinical data warehouse” and “predictive modeling,” you probably don’t immediately think of home healthcare agencies. But over the last 10 years, the nonprofit Visiting Nurse Service of New York (VNSNY) has developed some fairly sophisticated ways of providing relevant and timely data to its clinicians.
Healthcare Informatics recently interviewed Robert Rosati, Ph.D., vice president of clinical informatics at the Center for Home Care Policy and Research at VNSNY, about these initiatives.
Many home health agencies have electronic systems to gather patient information and handle billing, but using that data for patient safety and quality improvement requires a level of system integration and data analysis capability that most agencies don’t have, Rosati notes. But VNSNY, which serves 30,000 patients a day in more than a dozen different programs, has worked diligently to integrate data from those programs.
“From a technical perspective, we couldn’t have done this until we first built a clinical data warehouse,” Rosati explains. The next step was to design Web-based reporting tools so that anyone in the organization could have access to reports and analyses.
With input from agency executives, VNSNY Research created a monthly Quality Scorecard to monitor goals and actual performance. The scorecard compiles data from several sources, including electronic health records, payment systems, and a survey on patient satisfaction conducted by an external organization. The scorecard is broken down into four categories: process measures, outcome measures, cost measures and patient satisfaction. “To further capitalize on the clinical data we had, we built predictive models to identify patients at higher levels of risk for re-hospitalization,” Rosati explains.
Rosati and colleagues analyzed the VNSNY database and found several factors that put home-care patients at increased risk of hospitalization such as urinary incontinence, respiratory symptoms and congestive heart failure. They developed a computer model to place patients in one of seven risk categories, so that appropriate clinical steps could be taken.
Identifying patients at high risk allows nurses to take steps such ordering remote monitors or seeking the assistance of an advanced practice nurse. In a paper recently published in the Journal for Healthcare Quality, Rosati and colleagues noted that these and other health IT initiatives have been instrumental in helping VNSNY achieve a 12-percent decrease in the overall patient re-hospitalization rate between 2001 and 2009. “Of course the improvements can’t be solely attributed to the risk models, because we have made other clinical changes,” he adds, “but many of those changes have been driven by access to good data as well.”
Looking ahead, Rosati says the modeling can be used to optimize which type of program patients should consider. “For instance, many patients in acute care might benefit from hospice, so if we can identify those patients’ characteristics, it can lead to earlier discussions about hospice, which often happens too late.”
VNSNY has clearly been a pioneer in the use of health IT. With the nationwide trend toward accountable-care organizations and patient-centered medical homes requiring their participation, more home health agencies will have to follow its lead.