Researchers have created a model to suggest which rheumatology visits can be effectively carried out via telehealth and which should remain in-person visits. The model not only could help to alleviate the decision burden on providers but also help to navigate how to incorporate telehealth into daily rheumatology practice as the COVID-19 pandemic subsides, experts say.
Dr David Leverenz
The beginning of the pandemic quickly drove rheumatology practices to adopt telehealth ― which in the past has been studied only in selective groups of relatively stable patients. In Duke University’s Department of Rheumatology, nearly 90% of visits were conducted via telehealth for several weeks in April and May 2020, said David L. Leverenz, MD, an assistant professor of medicine at Duke University School of Medicine in Durham, North Carolina.
Since then, the practice has continued providing telehealth for a wider variety of patients: patiaents with high disease activity, those with low disease activity, people living 3 hours away or just 5 minutes from the medical center.
“Although the pandemic has really improved, and certainly we feel very safe providing in-person care, we’ve realized that it’s actually really possible to provide telehealth care to a lot of patients,” he said.
But there is no guidance on which visits may be more appropriate for telemedicine. To address this question, Leverenz and colleagues developed a model using an EASY score (Encounter Appropriate Score for You). After a patient encounter, providers rate whether that appointment would have been appropriate for telehealth. Using over 12,300 EASY scores from clinic visits throughout 2021, the investigators then built a model that uses patient sociodemographics, clinical and encounter characteristics, and provider experience with and preference for telemedicine to predict whether follow-up patient encounters could be conducted via telehealth or should be in person.
The model was then piloted with four Duke University rheumatologists for visits that took place from March through July 2022. A month prior to visits, researchers provided the rheumatologists a list of in-patient visits that the model suggested could be changed to telehealth visits. The list also included false changes, in which randomly selected visits were presented as being recommended for telehealth, although the model had not actually recommended the changes.
Leverenz and colleagues measured how frequently providers agreed with changes recommended by the model and the false changes. If a provider accepted a model-suggested change, the patient was contacted to either accept or reject the proposed switch. If a provider agreed with a false change, no additional action was taken.
Of the 201 patient encounters, there were 101 model-recommended changes and 100 artificial changes. Nearly all (99%) model-recommended changes suggested moving from in-person visits to telehealth visits. There was one recommendation to change a scheduled telehealth encounter to an in-person encounter. Of the false recommendations, 90 were for scheduled in-person visits, and 10 were for scheduled telehealth appointments.
Providers agreed with 71% of the model-recommended changes and with 21% of the false changes. Leverenz and colleagues reported “moderate agreement” between providers and model recommendations, reflected by a Cohen kappa coefficient of 0.45.
“This lined up pretty well with what providers thought was appropriate for the patient when presented with the information, and it also showed that on top of current practice, that the model really does have the ability to expand appropriate telehealth care, which is really what we’re looking to do,” Leverenz said.
The article describing these results was published in Arthritis Care and Research on October 2.
Perhaps unsurprisingly, clinician preference for and experience with telemedicine were the biggest predictors of whether changes suggested by the model were accepted.
Focus Model to Specific Diagnoses or Many?
Currently, clinicians who are already juggling many other responsibilities throughout the day must use their own judgment to determine whether telemedicine may be appropriate. A model such as this could help alleviate that decision burden, said Kathleen Fear, PhD, the director of data and analytics at the University of Rochester Medical Center Health Lab, in New York.
“A model that can help with scheduling or prompt a provider or patient for when a visit is appropriate for telemedicine seems like a really effective way to make the most of telemedicine while reducing potential burden on providers,” she said.
Leverenz imagines that this model could be embedded into electronic health records as a “decision support tool” to aid discussions between patients and providers on whether telehealth might be appropriate for upcoming visits.
Dr Christine Peoples
But developing a model that can generate predictions for the wide variety of conditions seen in daily rheumatology practice can be a challenge, said Christine Peoples, MD, a clinical associate professor of medicine and director of the tele-rheumatology program at the University of Pittsburgh in Pennsylvania.
“If you focus the model to certain diagnoses, at least the beginning, that’s very helpful, because it’s too difficult to have one model for every single reason that folks see a rheumatologist,” she said.
Dr Daniel Albert
Daniel A. Albert, MD, a professor of medicine and pediatrics at the Geisel School of Medicine at Dartmouth in Hanover, New Hampshire, agreed. The model is “a good start,” he said, and highlights that tele-rheumatology continues to be underutilized in practice. But he argued that the moderate agreement found with the model was relatively low.
A more focused algorithm that targets a single or several more common conditions may be more accurate, he said. “You probably want to break it down,” Albert said.
But Leverenz argued that the novelty of this model is that it incorporates the many different conditions seen in daily rheumatology practice, whereas previous programs utilizing telehealth focused on specific conditions and patients with low disease activity.
In addition, the model is currently provider centric and does not take patient preference into account, Albert added. Leverenz said that that is the next step in further developing this model. He is currently conducting qualitative analyses with patients to better understand what patients think and how often their views on telehealth differ from that of their care providers.
“Hopefully, we can expand appropriate telehealth visits by teaching providers not just to do what they think is right for the patient but also meet the patient’s expectations and needs, based on what we learn,” he said.
The study was funded by a grant from Pfizer. Leverenz has received grant from Pfizer and has served as a consultant for Sanofi. None of the study’s other authors report relevant financial relationships. Peoples is an educational consultant on telehealth for Pfizer. Alberts has previously received grant funding from Pfizer. Fear has disclosed no relevant financial relationships.
Arthritis Care Res. Published online October 2, 2023. Abstract
For more news, follow Medscape on Facebook, X, Instagram, and YouTube.
Source: Read Full Article