Feasibility and Acceptability of a Multidisciplinary Academic Telemedicine System for Memory Care in Response to COVID-19

Discussion

The COVID-19 pandemic presented new challenges that telemedicine was uniquely situated to solve. Changes that would typically encompass months of planning and pilot testing were compressed into weeks. We demonstrate that the rapid implementation and expansion of telemedicine services at the UCSF Memory and Aging Center outpatient clinic was feasible and acceptable to providers and did not significantly affect the clinic's utility and accessibility metrics.

After the implementation and transition to telemedicine, our clinic reached and surpassed previous clinic utility metrics. Despite an overall drop in scheduled appointments during the COVID-19 pandemic, canceled appointments declined, whereas the number of completed physician encounters per month increased. Multiple factors drove this increase in clinic accessibility. Implementation of VTC reduced the rate of follow-up appointment cancellations for patients who lived in midrange disparity neighborhoods and those who lived greater than 50 miles from the clinic. Furthermore, the implementation of VTC was associated with a reduction in the number of new patients who canceled their appointments, possibly suggesting that VTC reduced an unmeasured barrier to care. Multiple features of VTC increased the opportunity for patients from mid and high disparity neighborhoods to attend in-person or VTC follow-up appointments and increased the probability of new patients attending a follow-up clinical encounter within 1 calendar year of their first appointment. Overall, these results favor the use of VTC for ambulatory ADRD evaluations in physician clinical encounters.

The United States health care system has adopted telemedicine with remarkable speed for COVID-19–related care and chronic disease management. Given that telemedicine has become the default means of delivery of care during the COVID-19 pandemic, it is imperative that we proactively evaluate and address telemedicine's potential impact on health disparities in our vulnerable ADRD population. The observations mentioned above suggest that the implementation of telemedicine services at our clinic did not significantly affect the diversity of our clinic's population based on ethnic/racial status, estimated household income, age, gender, or ADI. Regarding our clinic's accessibility to non-English speakers, the available data can only comment on accessibility related to the use, but not the request for, interpreting services. As such, our data provide a limited window into whether language is a barrier to telemedicine services. There was a decrease in the total number of patients scheduled and seen in the clinic with an interpreter compared with the pretelemedicine era. However, established patients who required an interpreter maintained the same access to our clinic. As our clinic's capacity to provide in-person/virtual interpreters did not change after the implementation of telemedicine, the decline in new patients who required an interpreter was likely related to an unmeasured accessibility variable before or during the scheduling of new patients. These results suggest that ADRD telemedicine services did not exacerbate health disparities in vulnerable populations. At the same time, these results reinforce the observation that, nationally, patients with preexisting disparities in healthcare, such as language barriers, have experienced increased barriers to care during the COVID-19 pandemic.^31,32 Further analysis is warranted after the lifting of COVID-19 restrictions to confirm the pandemic's influence on patient referrals and scheduling.

Finally, we gained valuable insights by surveying physician and neuropsychologist experiences with telemedicine. First, physicians reported more previous telemedicine experience than neuropsychologists, likely given the lack of a well-established teleneuropsychology model.^33(p19) Before COVID-19, the American Psychological Association endorsed limited guidelines for remote services that were not readily in use or relevant to the challenges imposed by the pandemic.^33,-,35 It would not be until July 2020 that the Inter Organizational Practice Committee would issue a new set of guidelines for formal teleneuropsychological evaluation applicable to pandemic limitations.³⁶ At this point, the UCSF MAC clinic had already changed its clinic model and reestablished pretelemedicine patient volume. This new model fundamentally disrupted our neuropsychologist's workflow, which we suspect was a major component of our neuropsychologists' negative experience compared with physicians. Of particular interest, although physicians reported that telemedicine was more appropriate for follow-up visits than new encounters, neuropsychologists reported the opposite. We suspect the contrast is likely because of nuanced differences in each group's clinical evaluation. Physicians reported considerably less confidence in their physical examination, which is often necessary for diagnosing new patients with dementia syndromes defined by subtle motoric features. Neuropsychologists conversely lacked a reliable battery of teleneurocognitive tasks with proven equivalence to in-person evaluations. Accordingly, follow-up evaluations that relied on comparison to previous in-person assessments proved more problematic than new evaluations dependent on the new measures. Both groups reported similar key perceived benefits (improved patient access, family participation, and patient convenience) with minor differences related to the nuanced work experiences of the subspecialties. Finally, both groups reported concerns that the quality of their evaluations was diminished through the modality of VTC. Examples include VTC magnifying the impact of a patient's sensory (hearing/visual) deficit or a lack of technology literacy, causing another barrier to effective care. These results are promising and suggest that key stakeholders in ADRD-related care overall found telemedicine services to be a reasonable and acceptable alternative to in-person evaluations with notable caveats. Ongoing use and future deployment of ADRD telemedicine services must address and validate these quality concerns related to diagnosis and use nuanced triage protocols to determine an individual patient's suitability for telemedicine evaluations.

Telemedicine has historically been embroiled in controversies related to the sanctity of face-to-face visits and the metaphysical relationship between patients and physicians. However, over 20 years of international experience with ADRD telemedicine programs have demonstrated that telemedicine can be an equivalent and acceptable alternative to providers and patients alike that may be a more practical solution to care in the modern world. Regardless, there remain critical limitations that have not yet been addressed. Future research must address our lack of knowledge related to the diagnostic quality and economics of VTC encounters.

Although previous research has investigated the capacity of VTC to diagnose MCI vs unspecified ADRD, to the author's knowledge, no study has addressed the specificity of diagnosis beyond the MCI-dementia dichotomy nor appraised ADRD VTC clinical outcomes. The survey of providers' telemedicine experience highlighted this concern. Neuropsychologists reported concerns that VTC assessment of executive function abilities was limited by the testing methods available and the uncontrollable environment of the VTC encounter. Physicians reported concerns focused on the limitations of the physical evaluations, specifically the ability to differentiate Parkinson spectrum syndromes (dementia with Lewy body, Parkinson disease dementia, etc.) and motor neuron disease spectrum (frontotemporal dementia and amyotrophic lateral sclerosis), further evidenced by their ranking of least appropriate clinical syndromes to be seen via VTC. Given the nature of our retrospective review and the limitations of available data, we cannot reliably comment on whether the implementation of telemedicine services objectively affected the ability of providers to make certain diagnoses accurately. Future research must address the shortcomings of video-based physical examination and limited reliable measures of executive function through video telemedicine. We suspect that given the current limitations of ADRD telemedicine, new triage and preclinical screening protocols will be necessary to determine which patients are clinically appropriate for telehealth evaluations. Neuropsychologists will need to create new digital neurocognitive assessments that can be performed remotely without close supervision and verify their sensitivity and specificity relative to gold standard in-person assessments.^33(p19) ADRD specializing physicians will need to explore the use and test the validity of new digital surrogates for physical examination components such as eye movement, muscle tone, and cortical sensory deficits.³⁷ Should we be unable to replace certain aspects of the physical examination reliably, it will be necessary to support the creation of remote VTC clinic sites where an in-person medical staff and facilitate the examination of a remote ADRD specialist. Comparing the effectiveness of remote patient monitoring using technologies such as telemedicine with standard in-person care has been listed as one of the top 50 priorities for health care by the National Institute of Medicine in the United States,³⁸ highlighting the issue of access to health care as a global one.³⁹ If ADRD telemedicine is unable to reliably differentiate certain conditions, the potential benefits of this technology may not be worth the shortcomings of inaccurate neurodegenerative phenotyping.

Determining the socioeconomic savings of ADRD telemedicine is another challenge. Our study only touched the surface of these complex issues and found that reducing travel barriers increased clinic accessibility. From the patient's perspective, the continued availability of appropriate telemedicine services provides the opportunity to access healthcare services while avoiding travel-related costs and decreasing time spent away from work or other obligations. However, for providers, the socioeconomic impact is less clear. Telemedicine can be efficiently delivered at a cost lower than in-person services, given the decreased need for physical space and staff support. However, the incremental resources and associated costs of electronic platforms and staff required to help patients remotely have increased the overall cost of telemedicine infrastructure and may negate any potential savings. Future research must address the need to balance the opposing market forces while facilitating collaborative efforts with our partners in public health, insurance providers, and primary care to prevent bottlenecking the referral process for patients already at risk for health disparities. We will need to begin streamlining our referral and care management networks to find and fill gaps in clinical care. Furthermore, we will need to begin standardizing state laws and coding/billing practices that would further facilitate ADRD teleneurology. If such implementation is successful, financial and, care-related performance improvements could be significant.

There are several key limitations of this study that need to be acknowledged. We did not evaluate the qualitative or clinical outcomes of telemedicine encounters, including patient satisfaction or a cost-benefit analysis. The study is from a single, large academic center, and our findings may not be generalizable to other specialties, practices, or locations. In particular, minority populations who may have lacked VTC technology may have been underrepresented in this study. Furthermore, this study has compared telemedicine implementation data between pandemic and prepandemic epochs. Accordingly, unmeasured psychosocial and economic disturbances experienced by our patients during the COVID-19 pandemic may have created significant biases that have affected the observed rate of cancellations and no-shows. ADRD telemedicine, despite 20 years of experience, remains an underdeveloped domain of neurology for which there remain many questions of accuracy, equity, and practicality. Regardless, this study provides valuable insights into the experience of a memory clinic transition to ADRD telemedicine services in response to COVID-19 and is the first study to describe how this transition affected clinic utility metrics, patient populations, and provider experiences relative to clinic performance before the COVID-19 pandemic.

In the UCSF Memory and Aging Center, we have demonstrated that the integration of telemedicine into dementia care is a feasible and acceptable alternative to in-person care leading to an increase in follow-up encounters and a decrease in cancelations without negatively affecting clinic demographics. The COVID-19 pandemic has forced us to radically rethink and change our memory care delivery models. We anticipate that evidence will demonstrate that virtual visits are valuable in certain clinical contexts and, in time, will be integrated alongside in-person evaluations into a new model of care. How our profession adapts to this technology in the coming years will determine whether we shape its deployment to meet our needs or adjust our practice to meet its demand.