Hospitals routinely implement processes aimed at increasing provision of high quality care for patients, such as calling patients to encourage follow-up visits or providing doctors and nurses with alerts in electronic health records to prompt reminders for vaccinations. But most of the time, hospitals cannot tell whether these processes are working optimally. A program at NYU Langone Health, led by Leora Horwitz, MD, an associate professor of population health and medicine and director of the Center for Healthcare Innovation and Delivery Science, addresses this question by implementing randomized quality improvement projects. She and her colleagues reported their findings in the September 19, 2019 issue of the New England Journal of Medicine.
While companies in the private sector such as Google and Amazon routinely incorporate randomized continuous quality improvement methods into their workflow, this is new terrain for health systems. Randomized quality improvement projects, as used by Horwitz and her colleagues, evaluate the effectiveness of routine processes that hospital staff perform on a daily basis, and enable staff to rapidly improve their practices.
This program is important because there are always better ways to do things, but unless we have some data to show us that what we're doing is not fully effective, we have no incentive or inclination to find a better way to do it. Unless we study whether what we're doing is working, we cannot allocate the resources that we have most effectively. And that means that we're not necessarily providing the best possible care to our patients."
Leora Horwitz, MD, associate professor of population health and medicine and director of the Center for Healthcare Innovation and Delivery Science, NYU Langone Health
Horwitz implemented randomized quality improvement projects throughout NYU Langone-- spanning inpatient units, outpatient offices and the emergency department--and included efforts to improve care after hospitalization, increase receipt of recommended preventive screening, capture patient-reported outcomes, and increase smoking cessation counseling rates, among other topics. In one such project, which tested a new program to telephone patients after discharge from the hospital, patients with odd-numbered electronic records received calls, while those with even-numbered records did not receive calls. Another project, which looked at phone scripts used to remind patients of annual well visits, randomized patients to script A or B, and then each script was used in alternating weeks for several weeks.
Among the findings was that changing the text of a prompt to provide tobacco cessation counseling in the office produced a statistically significant increase in medication prescription rates. Changing a few sentences in telephone outreach scripts shortened phone calls and increased annual visit appointments.
The researchers also found that the post-discharge phone calls to patients were largely ineffective. Patients who received phone calls returned to the hospital at the same rate as those who didn't receive calls. Patients receiving calls also had the same satisfaction ratings with the hospital as those who didn't receive calls.
With this knowledge, hospital staff now have options to determine where to focus their resources, Horwitz says. Call scripts may need to be changed or staff might only need to call the high-risk patients instead of calling every patient.
"At NYU Langone Health, the well-being of our patients always comes first," says Andrew W. Brotman, MD, senior vice president and vice dean for Clinical Affairs and Strategy. "To fulfill our role as a learning health system, we must demonstrate the agility to adapt or modify any procedure so that it delivers maximum benefit to patients. Dr. Horwitz's work ensures we are allocating staff resources in the most effective manner possible."
Horwitz emphasizes that projects that test hospital procedures do not deprive patients of interventions that are effective. Rather, randomized testing is designed to benefit patients by quickly determining what works and then implementing the effective arm or changing an ineffective process to work better. She also acknowledges notes that there are important ethical considerations in doing this work.
"I believe we have an ethical responsibility to rigorously assess whether our operational interventions are effective, even when they may seem trivial, such as scripts for calls or mailings that we send to people to get them to get their colonoscopy," Horwitz says. "If we don't, we can't be sure we are doing the best by our patients."
The randomization programs were designed to be easy to implement, Horwitz says. Staff do not need to use special randomization tools or maintain separate databases. People doing the work only need to make small changes in their workflow, such as changing the script they use from one week to the next.
"This can be done quite easily. It doesn't have to be a $10 million, 10 year NIH trial," Horwitz says. "We can build these things into the routine way we do our work."