Abstract
In this study, telehealth appropriateness was most associated with provider preference, the current and previous encounter modality, and RA disease activity. Other factors like patient demographics, RA medications, and comorbid noninflammatory causes of joint pain were not associated with telehealth appropriateness.
Introduction
The COVID-19 pandemic catalyzed a paradigm shift in rheumatology, with telehealth adoption jumping from nearly zero to over 90% almost overnight. While remote care offers significant convenience, Rheumatoid Arthritis (RA) is a clinical specialty that traditionally relies on the "laying on of hands." Physical examinations are critical for detecting synovitis—subtle joint swelling that patients may not always perceive. As healthcare transitions into a long-term hybrid care model, a critical question remains: how do we determine which encounters are truly appropriate for telehealth and which require an in-person clinical assessment?
This study addresses this gap by identifying the clinical and demographic factors that influence a provider's perception of telehealth appropriateness using a novel assessment tool.
Methods
Researchers conducted a prospective cohort study at Duke Health between January and August 2021. To measure the primary outcome, they utilized the Encounter Appropriateness Score for You (EASY). This 3-point scale allows providers to rate an encounter immediately following its conclusion:
- In-person or telehealth acceptable
- In-person preferred
- Telehealth preferred
For the purposes of statistical analysis, these scores were collapsed into a binomial variable: Telehealth Appropriate versus In-person Preferred. The team employed Robust Poisson regression with Generalized Estimating Equations (GEE). This approach was chosen to account for repeated measures, acknowledging that individual patients may have multiple visits and that specific providers see many different patients, creating nested correlations within the dataset.
Datasets and Scope
The analysis was based on a comprehensive clinical dataset extracted from Duke Health’s Electronic Health Records (EHR):
- Encounters: 1,823 outpatient visits.
- Participants: 1,177 unique adult RA patients and 25 rheumatology providers.
- Clinical Metrics: Data included ICD-10 codes, DMARD (disease-modifying anti-rheumatic drug) prescriptions, and RAPID3 scores—a validated patient-reported outcome measure used to gauge RA disease activity.
Key Insights and Results
The Clinical Ceiling
Disease activity emerged as the primary clinical driver of modality preference. When a patient’s RA was flaring (moderate to high disease activity), the perceived appropriateness of telehealth dropped significantly (RR 0.57 for high activity). This confirms that providers view the physical exam as indispensable when the disease is not well-controlled.
Provider Bias and the "Rubber-Stamping" Effect
One of the strongest predictors of appropriateness was the provider’s own "average preference" from previous encounters. Some clinicians are inherently more comfortable with remote care than others, suggesting that triage tools must account for clinician-specific comfort levels rather than just patient data. Furthermore, providers were much more likely to rate a visit as telehealth-appropriate if the visit was currently occurring via telehealth (RR 2.27), indicating a potential cognitive bias where clinicians justify the current modality.
The Hybrid Rhythm
Interestingly, if the previous encounter was conducted via telehealth, providers were less likely to find telehealth appropriate for the current visit. This suggests a "hybrid rhythm" in rheumatology care, where clinicians feel comfortable with remote check-ins provided they can see the patient in person at alternating intervals.
Demographic Neutrality
Encouragingly, factors such as race, sex, and insurance type were not associated with appropriateness ratings. This suggests that providers are not making modality decisions based on socio-economic stereotypes, supporting health equity in the transition to digital care. Additionally, while age was slightly associated with increased telehealth appropriateness (RR 1.05 per 5-year increase), this likely reflects a clinical desire to reduce the physical burden of travel for older patients.
The Modality Gap
Overall, providers deemed telehealth appropriate in only 32.2% of all RA encounters. There was significant discordance in actual practice: in 34.7% of telehealth visits, the provider later reported they would have preferred an in-person encounter. This gap was more pronounced in telephone visits (45%) compared to video visits (27.3%).
Limitations and Future Directions
The study's findings are subject to a few constraints. As a single-center study at an academic institution (Duke Health), the results may not generalize to private practices or rural settings. Furthermore, the EASY score only captures the provider's perspective; the study did not measure patient satisfaction or whether patients felt their care quality was impacted by the modality. Finally, the data was collected during a period of reimbursement parity, which may change as insurance policies evolve.
Future research will focus on translating these findings into predictive clinical algorithms. The goal is to develop an "appropriateness tool" that helps schedulers and patients determine the optimal visit modality before the appointment is even booked, ensuring that high-acuity patients receive the in-person care they require.
Summary updated