The role of a specialized memory clinic supporting primary care providers in a safety net health system

Abstract

Background

Although most dementia care occurs in primary care, consultation with dementia specialty care is sometimes indicated. Access to dementia specialists is limited, particularly in resource-limited environments such as the public health safety net, which may require triaging referrals to preserve access for patients with needs that can not be met in a primary care setting.

Methods

The eConsult system for primary care providers to refer patients to a subspecialty memory clinic is described for a large safety net health system. Demographic and clinical characteristics are presented for patients evaluated within the memory clinic setting compared to the health system overall. ICD-10-CM codes were used to identify cognitive diagnoses and medical comorbidities. Chi-squared tests were used to compare categorical variables and t-tests for continuous variables.

Results

94 individuals age 50 or older were seen in the memory clinic in 2019, of whom 43 were new evaluations. The most common visit diagnoses for new memory clinic patients were Alzheimer’s disease (33%), no cognitive diagnosis (28%), unspecified dementia (19%), and mild cognitive impairment (12%); for follow up patients, the most common diagnoses were Alzheimer’s disease (49%), unspecified dementia (18%), no cognitive diagnosis (14%), and mild cognitive impairment (10%). For those without a cognitive diagnosis, common visit diagnoses included cognitive symptoms, mood or sleep disorders, and metabolic disturbances. Of the 11 new internal referrals with a prior coded diagnosis of dementia, median time from first diagnosis to their initial memory clinic visit was 224 days.

Conclusions

Despite clear systemwide parameters for referral and extensive pre-referral screening via an eConsult system, the most common diagnosis for memory clinic patients was Alzheimer’s disease. Direct studies of eConsult are needed to determine primary care providers’ needs when referring patients with dementia to a memory clinic setting.

Background

Although most dementia care occurs in the primary care setting [1, 2], consultation with specialty dementia care is generally indicated in a number of clinical situations, including atypical or young onset presentation, management of behavioral symptoms, and access to novel treatments. However, there is a persistent and significant undersupply of dementia care specialists (including neurologists, psychiatrists, and geriatricians) to support primary care clinicians [3, 4], a problem that can be especially challenging in health systems that serve vulnerable populations, such as those who are low-income and have Medicaid or are uninsured [5].

The Los Angeles County Department of Health Services (DHS) is the second largest municipal health system in the country, providing primary care for an empaneled population of 568,000 and more than 900,000 annual specialty visits [6]. As a safety net healthcare system, DHS serves a diverse and socioeconomically disadvantaged patient population, the majority of whom are from minority race/ethnicity groups and have a preferred language other than English; the majority of patients > 65 have either Medicaid alone or are uninsured and do not qualify for Medicare. Like other safety net providers, DHS has historically experienced challenges in timely access to specialty care, with reported wait times of over 9 months for certain specialties in 2011 [5]. DHS uses two primary systemwide strategies to manage access to specialty care. First, there is a library of Expected Practices or guidance for common clinical scenarios that might result in a specialty referral, with suggested diagnostic workup, management, and indications for specialty consultation. Second, DHS uses a secure, HIPAA-compliant eConsult system to manage all non-emergent clinical situations that might require specialty consultation. After a specialty reviewer is notified of an open eConsult, they review the provided clinical information and may recommend additional testing or management that can be performed within the primary care environment; the eConsult is either closed with the primary care provider’s concerns addressed, or recommended for a specialty appointment [5]. The adoption of eConsult was associated with streamlined primary-specialist communication and reduction in wait times for specialty appointments [5, 7] but at the cost of increased administrative burden for primary care clinicians [7]. Providers receive required training on the use of the eConsult system. Although there is no direct training on individual Expected Practices, eConsults may require specific clinical information before review as outlined in the relevant Expected Practice (either via structured fields, or by request of the specialty reviewer), thus reinforcing their use. For dementia, additional training is provided through Grand Rounds and other conferences by memory clinic staff as requested on an ad hoc basis.

The DHS Expected Practice for dementia provides a concise overview of standard of care dementia evaluation and management, aimed at assisting primary care providers to complete the basic evaluation and to manage typical late onset dementia [8]. The Expected Practice outlines core components of a basic dementia evaluation, including (1) history elements such as impact on activities of daily living, behavioral changes, and use of medications that can affect cognition; (2) use of standardized screening instruments for cognition and mood, as well as a neurologic examination; (3) laboratory studies including thyroid studies and vitamin B12 levels; and (4) brain imaging with MRI as the preferred modality if available. The Expected Practice also includes indications, standard dosing, and common side effects for acetylcholinesterase inhibitors and memantine, as well as recommendations for advanced care planning, driving, and social services. Referral to neurology for dementia care is recommended in a limited number of situations, such as early onset disease (< 60yrs), uncertainty about the diagnosis of dementia or underlying etiology, abnormal neurologic exam findings, behavioral symptoms beyond the capacity of primary care to manage, or rapidly progressive cognitive decline. Dementia specialists within DHS include general neurologists, geriatricians, and psychiatrists depending on geographic location; however, DHS includes a subspecialty memory clinic within neurology at Rancho Los Amigos hospital that also receives referrals from other specialty providers. In this paper we describe the memory clinic objectives and the demographic and clinical characteristics of patients seen in this setting several years after the DHS-wide implementation of eConsult.

Methods

Setting This study is a retrospective review of characteristics of patients seen at the memory clinic at Rancho Los Amigos hospital in 2019 compared to DHS patients overall. The memory clinic was established in 1985 with a mission to improve health care delivery and quality of life for people with dementia as well as their families and caregivers. This memory clinic is a clinical site for the USC-Rancho California Alzheimers Disease Center (CADC), one of 10 centers that receives supplemental funding through a state grant to support additional services, and is the only CADC situated within a county health system. Although other specialized consultation is available for dementia at other sites within DHS through neurology and geriatrics, these other clinics typically lack the dedicated social worker and integrated psychiatry available within the memory clinic.

The memory clinic operates one half-day session a week with 3 providers per clinic, approximately 44 weeks per year; and typically 1 or 2 new patients are scheduled per clinic session. Providers include behavioral neurologists, a geriatric psychiatrist, and a PA (physician associate/assistant, FSG), as well as rotating medical students, residents, and fellows; the clinic is also supported by a clinical social worker. The memory clinic receives referrals via eConsult from both DHS and non-DHS providers, typically from community partners that serve a comparable patient population; eConsult requests are screened by one of the authors (FSG) to ensure that workup and management have been guided by the Expected Practice for dementia, and typically more than 50% are closed for further management in primary care, without requiring a specialist evaluation. The clinic also sees patients who self-refer, are referred from community-based organizations and a small number from private practice providers in the surrounding community. Newly referred patients receive a comprehensive diagnostic workup, which typically includes cognitive assessments and an MRI brain, and may also include additional neuropsychological testing and FDG-PET if indicated in the opinion of the treating clinician; CSF and plasma biomarkers for AD and APOE genotyping are also currently used but were not readily available within DHS during the time frame of the study. Clinical findings, neuropsychological testing, and imaging and biomarker data are reviewed in a biweekly multidisciplinary conference to reach a consensus diagnosis and determine a treatment plan. For patients who continue follow up at the memory clinic, ongoing management primarily include treatment with acetylcholinesterase inhibitors and memantine, medication and nonpharmacological strategies for managing behavioral symptoms of dementia, and connecting caregivers with available community resources through a dedicated social worker. Patients may also be invited to participate in observational research or clinical trials. Amyloid PET scans and anti-amyloid therapies are not currently available within the DHS system.

Study design Deidentified data was extracted by the Clinical Research Informatics team from the DHS Cerner electronic health record (EHR) for all adults aged 50 or older with at least one ambulatory encounter within the health system in 2019. Available data included patient demographics and encounter information, including visit location, service, insurance status, and associated diagnoses. Although separate EHR fields, race and ethnicity were combined for analysis because the majority of patients identifying as Hispanic in this system have “Unknown” as their reported race in the EHR. Data were excluded for non-standard encounters, such as those for clinical research or employee health. Medical service was used to identify clinic encounters at the Rancho memory clinic. Patients with at least one 2019 memory clinic encounter were stratified into new versus follow up patients based on presence or absence of a clinic encounter prior to 2019 within the current EHR system. Because the health system also accepts referrals for specialty care from non-DHS clinics (most commonly other safety net providers, such as Federally Qualified Health Centers), patients were further stratified based on whether their first encounter in the health system occurred at the memory clinic (suggesting an external referral) or had a prior DHS encounter. Medical comorbidities, including diabetes mellitus, were defined based on Quan et al.’s adaptation of the Charlson Comorbidity Index (CCI) for ICD-10 [9,10,11]; for hypertension and hyperlipidemia, which are not included in the CCI, the Medicare Chronic Condition Warehouse (CCW) definitions were used.

The Medicare CCW definitions were used to identify ICD-10 codes for Alzheimer’s disease and non-Alzheimer’s dementia. For patients with a dementia diagnosis, we used the first encounter associated with a dementia diagnosis code as their date of incident dementia, and then calculated the difference in days between the incident date and the date of their first visit at the memory clinic. We subdivided Rancho memory clinic patients by age category (50–64 vs. 65 and older) and by new vs. follow up encounters, and then categorized patients by the cognitive diagnosis given at their first 2019 memory clinic encounter as (1) any specified dementia diagnosis, (2) unspecified dementia; (3) no dementia diagnosis but diagnosis of mild cognitive impairment; or (4) no cognitive diagnosis. For patients with any specified diagnosis, we further separated non-Alzheimer’s dementia into those for dementia with Lewy bodies, frontotemporal dementia, vascular dementia, and other specified dementia. A full list of categorized ICD-10 codes is presented in Table S1. This study was determined to be exempt from 45 CFR 46 according to § 46.104(d) as category [4] by USC’s Institutional Review Board, and the requirement for informed consent was therefore waived. All research was conducted with ethical standards in accordance with the Declaration of Helsinki.

Statistical analysis Demographic and clinical variables were compared for memory clinic patients compared with older adults with DHS overall using 2-tailed t-tests for continuous variables and Chi-square tests for categorical variables. Analysis was done using R version 4.4.1, RStudio 2024.04.2, and the tidyverse and vtable packages.

Results

Demographics of the study population are presented in Table 1. There were 94 unique individuals age 50 or older seen in the memory clinic in 2019; of these, 43 were new evaluations, and 51 were follow ups from prior years. Of the new patients, 23 had a prior encounter within DHS (“DHS patients”, presumed to have been referred via eConsult) and 20 had no prior encounter (“non-DHS patients”). There were statistically significant differences between memory clinic patients and the overall population of DHS adults in terms of age (72.52 vs. 60.25, p < 0.001), race/ethnicity (higher proportion non-Hispanic white or unknown; lower proportion Hispanic, non-Hispanic Black or Asian, p < 0.001), and marital status (higher proportion married/partnered, widowed, or unknown; lower proportion single or divorced/separated, p = 0.04). Memory clinic patients had fewer chronic medical conditions (CCI 2.36 vs. 3.04, p = 0.04), a lower proportion of hyperlipidemia (28.3% vs. 42.8%, p = 0.008), and fewer total visits (8.62 vs. 10.68, p < 0.001).

Table 1 Demographics of rancho memory clinics compared to older adults receiving ambulatory care at DHS sites in 2019

The most common visit diagnoses for new memory clinic patients were Alzheimer’s disease (14 individuals, 33%), no cognitive diagnosis (12 individuals, 28%), unspecified dementia (8 individuals, 19%), and mild cognitive impairment (5 individuals, 12%); for follow up patients, the most common diagnoses were dementia due to Alzheimer’s disease (25 individuals, 49%), unspecified dementia (9 individuals, 18%), no cognitive diagnosis (7 individuals, 14%), and mild cognitive impairment (5 individuals, 10%). Figure 1 compares the proportion of new and follow up memory clinic patients who were diagnosed with a specific dementia etiology, with unspecified dementia, with MCI, and with no cognitive diagnosis. Figure 2 shows the diagnosed dementia etiology for those memory clinic patients who had a specified dementia type; most patients with a specified diagnosis were diagnosed with Alzheimer’s disease. Non-cognitive ICD-10 codes for the 12 new patients without a coded cognitive diagnosis are presented in Table 2. Of the 23 new evaluations of DHS patients, 11 had a prior coded diagnosis of dementia; 5 had their first coded dementia diagnosis at their initial memory clinic visit; 2 received a dementia diagnosis subsequent to the initial visit, and 5 did not receive a dementia diagnosis. For those with a coded dementia diagnosis prior to their first memory clinic visit, the median time from their first coded dementia diagnosis to their initial memory clinic visit was 224 days.

Fig. 1

Memory clinic diagnoses at first 2019 encounter for new vs. follow up patients

Fig. 2

Specified diagnoses at first 2019 memory clinic encounter

Table 2 ICD-10 codes associated with new patient encounters without an associated cognitive diagnosis

Discussion

In this study we describe the demographic and clinical characteristics of patients evaluated at a memory clinic embedded within a large safety net health system in Los Angeles county. Compared to DHS patients overall, patients evaluated at the memory clinic had fewer documented comorbidities and fewer visits within the system, likely reflecting that about half of referrals are external from the health system and receive the majority of their healthcare elsewhere. Most patients evaluated by a dementia specialist were diagnosed with a cognitive disorder, with greater specificity of diagnosis seen at follow up visits, presumably after the initial diagnostic workup was complete. This pattern is similar to the pattern observed in Medicare claims data that a higher proportion of patients are diagnosed with a specific etiology of their dementia over time [1].

The majority of patients evaluated at the Rancho memory clinic by a dementia specialist were diagnosed with Alzheimer’s disease (AD). This finding is somewhat unexpected given that health system’s Expected Practice is that patients with amnestic dementia without parkinsonism or other atypical features are typically to be evaluated, diagnosed and managed in a primary care setting. Although some patients may have had atypical presentations of AD and presented a diagnostic challenge for the primary care provider, it is likely that many of these patients were referred for care management. This finding is also supported by the fact that approximately half of new memory clinic internal referrals had a pre-existing diagnosis of dementia, in some instances years prior, suggesting that patients are referred not primarily for diagnostic evaluation but when the symptoms of dementia become more difficult to manage in a primary care setting, potentially because of difficult to manage behavioral symptoms of dementia, which is an accepted reason for referral to specialty care in the Expected Practice. In addition, early onset dementia is another reason for referral specified in the Expected Practice and patients under age 65 represented 14 of the 43 new evaluations.

Our study has several limitations. We used a definition of dementia requiring a single encounter with an associated diagnosis, which may be associated with false positive diagnoses [12] and indeed, a minority of new patients evaluated were not ultimately given a cognitive diagnosis by a memory disorders specialist. In the memory clinic setting, dementia etiology was determined by subspecialists in memory disorders generally supported by structural MRI of the brain and sometimes additional neuropsychological testing and FDG-PET; however, the frequency of which these additional tests were ordered was not available in our data set. Amyloid PET is not currently available within DHS, and fluid biomarkers were not widely available at the time of the study, which may limit the accuracy of Alzheimer’s disease diagnosis. We were also not able to review the eConsults themselves, including the total annual volume, proportion referred versus closed, or specific reasons primary care providers requested specialty referral to the memory disorders clinic. Nonetheless, we believe our study also has several strengths, most notably that the patient population evaluated at the memory clinic is significantly more ethnically diverse than most memory clinics [13], which are typically associated with academic medical centers, reflecting the demographics of patients who receive care in safety net settings in LA County.

Although the use of e-consultation systems has been reported for various specialties, including geriatrics [14] and neurology [15], to our knowledge none have specifically focused on memory disorders and dementia. The majority of the published literature on e-consults comes from large health systems including the public health safety net, as in this study [5, 14, 16], supporting the concept that this approach is appropriate and adaptable to needs of patients across the full spectrum of health care delivery systems. However, the majority of patients evaluated in the specialty memory clinic setting in our study were diagnosed with Alzheimer’s disease despite extensive pre-referral screening via eConsult, suggesting that the combination of Expected Practice and eConsult is likely insufficient to meet the needs of overworked primary care providers in a resource-constrained system. Future work reviewing the eConsult data themselves, linked to patient outcomes, could provide additional insight into reasons for referral and reinforce the need for additional support of providers in the primary care setting.

With the increasing availability of plasma biomarkers and new anti-amyloid therapies for Alzheimer’s disease, there may be increased pressure on health systems to re-evaluate how they approach diagnosis and management of people with cognitive disorders. Although the present study predates the FDA approval of anti-amyloid therapies, these therapies are increasingly available in many settings and may become available within DHS in the future. As these therapies would likely be managed in a specialty care setting within DHS, this change would likely require a reconsideration of the Expected Practice and eConsult referral criteria to ensure that primary care clinicians can readily identify and refer those patients who might be eligible for treatment. While existing referrals likely appear appropriate based on health system criteria, the apparently high number for individuals with an existing dementia diagnosis has implications for specialty access, as referrals for care management may require more longitudinal follow up than referrals purely for diagnosis. Interventions at the health system level, such as template changes to accommodate additional time for patients with dementia, and increased access to social work with dementia-specific expertise, may be needed to empower primary care clinicians to care for these patients within their own clinics, thereby preserving access to specialty care for the subset that need it most. Partnership between primary care and the memory clinic can ensure that primary care providers receive the support they need to care for patients while ensuring specialty care is available for selected patients who would most benefit.

Conclusions

In this study of a subspecialty memory clinic located within a safety net health system, the most common diagnosis was Alzheimer’s disease. Direct studies of eConsult are needed to determine primary care providers’ needs when referring patients with dementia to a memory clinic setting.