Non-literacy biased, culturally fair cognitive detection tool in primary care patients with cognitive concerns: a randomized controlled trial

Dementia is a significant public health concern, affecting a substantial portion of the older adult population, with rates increasing globally. Despite the availability of numerous cognitive assessment tools, a considerable number of dementia cases remain undiagnosed in primary care settings, particularly among racial and ethnic minority groups and socioeconomically disadvantaged individuals. This underdiagnosis is attributed to various factors at the levels of the assessment instruments, patients, clinicians, and the healthcare system itself. Many existing cognitive assessment tools are lengthy, expensive, require specialized administration and equipment, and often lack guidance on subsequent steps after test results. Furthermore, many tools were developed and validated primarily in white populations, limiting their cultural appropriateness and potentially leading to biased results in diverse populations. Studies have demonstrated that existing cognitive tests may have different cutoff points for detecting dementia in diverse racial and ethnic populations, leading to missed or delayed diagnoses. Primary care providers (PCPs) play a critical role in dementia diagnosis and care, given their frequent patient contact and established trust. Cognitive assessment endorsed by PCPs is generally well-accepted by patients. To address these limitations, the researchers developed the 5-Cog paradigm, a non-literacy biased, culturally appropriate, and user-friendly 5-minute cognitive assessment paired with an electronic medical record (EMR)-embedded clinical decision tree to aid PCPs in dementia care decisions. This study aimed to evaluate the efficacy of the 5-Cog paradigm in improving dementia-related care actions in a diverse primary care population experiencing health disparities.

The existing literature highlights significant challenges in dementia detection and management in primary care, especially among underserved populations. Many studies have focused on the validity of brief cognitive assessments, the impact of physician education, or dementia care management strategies. However, these elements have largely been examined in isolation, and there's a lack of comprehensive evidence on which approach most effectively improves dementia care, particularly in populations facing health disparities. The literature shows a disproportionate burden of dementia among historically marginalized communities, yet these communities are underrepresented in research studies. Existing cognitive tests often fail to account for cultural differences and health inequities, resulting in inaccurate diagnoses and delayed interventions. The use of brief cognitive assessments in primary care has shown promise but requires integration with a clinical decision-support system to guide appropriate follow-up actions, which are often lacking. The use of EMR-embedded clinical decision support tools has been advocated in various primary care settings to facilitate efficient decision making and guide referrals.

This study employed a single-blind, randomized controlled trial (RCT) design to assess the efficacy of the 5-Cog paradigm. 4538 patients were screened for eligibility, with 1201 enrolled from May 29, 2019, to September 15, 2022. Participants (mean age 72.8 years, 72% women, and 94% Black, Hispanic, or Latino) were randomly assigned (1:1 ratio) to either the 5-Cog intervention or an active control group. The 5-Cog paradigm consisted of four tests: the Picture-based Memory Impairment Screen (PMIS), the Motoric Cognitive Risk Syndrome (MCR) assessment, Symbol-Match, and clinical decision-making support within the EMR. The active control involved equivalent testing time, including the Short Assessment of Health Literacy in Medicine (SAHL) and grip strength measurement, followed by an EMR note with recommendations. The primary outcome was the occurrence of at least one of the following within 90 days: new diagnosis of mild cognitive impairment (MCI) or dementia by the PCP; ordering of imaging or laboratory tests; initiation of new dementia medications; or referral to a specialist. Secondary outcomes included 12-month hospitalization and emergency department visit rates. Data were analyzed using intention-to-treat principles with multivariable logistic regression, adjusted for age, sex, and education. Sensitivity analyses were conducted for subgroups defined by sex, ethnicity, education, and language. Post-hoc analyses compared participants enrolled before and after a COVID-19-related research suspension.

A total of 1201 participants were enrolled in the study. The 5-Cog paradigm showed a statistically significant improvement in the primary outcome (improved dementia care actions) compared to the control group (18.5% vs. 6.8%, P < 0.001). The adjusted odds ratio for the primary outcome was 3.43 (95% CI 2.32–5.07). Individual components of the primary outcome also showed significantly higher rates in the 5-Cog arm: new MCI or dementia diagnoses (OR 6.48, P < 0.001), imaging ordered (OR 4.80, P < 0.001), tests ordered (OR 7.64, P < 0.001), and specialist referrals (OR 2.38, P < 0.001). New prescriptions showed a non-significant trend (OR 3.24, P = 0.15). There were no significant differences between groups in the secondary outcomes (hospitalizations and emergency department visits) at 12 months. Sensitivity analyses revealed statistically significant results across subgroups defined by sex, ethnicity, education, and language. In the 5-Cog arm, the 5-Cog battery had a sensitivity of 96% and specificity of 71% for PCP diagnoses of MCI or dementia. In a subsample that underwent additional cognitive testing, the 5-Cog sensitivity was 70%, and specificity was 71% for detecting MCI or dementia using the MoCA test cutoff of <17.

This RCT provides strong evidence for the efficacy of the 5-Cog paradigm in improving dementia care actions in a diverse primary care population. The significantly higher rates of dementia-related diagnoses, investigations, and referrals in the 5-Cog arm demonstrate the paradigm's effectiveness in enhancing the detection and management of cognitive impairment. The findings address the critical need for culturally appropriate and efficient tools to improve dementia care in primary care settings, particularly among underserved populations. The study's strengths include its large sample size, diverse population, real-world setting, and focus on clinically relevant outcomes. While the lack of a significant effect on new prescriptions may be due to factors such as awaiting further investigations, clinician experience, or concerns about medication side effects, the overall impact on other dementia care actions is substantial. This contrasts with previous studies that showed no effect of solely education-based interventions on dementia care. The use of EMR-embedded recommendations clearly added value to the assessment.

The 5-Cog paradigm, a brief, culturally sensitive cognitive assessment integrated with EMR-based clinical decision support, significantly improved dementia care actions in primary care patients with cognitive concerns. This study provides compelling evidence to support the widespread implementation of this approach to enhance early diagnosis and improve the management of dementia in primary care, particularly among underserved populations. Future research could focus on broader implementation studies across various primary care settings and explorations into integrating the 5-Cog paradigm into routine clinical workflows, such as the Medicare annual wellness visit.

The study was conducted in a single primary care clinic, limiting generalizability. MCI and dementia diagnoses were based on PCP assessments rather than standardized neuropsychological evaluations. The primary outcome was limited to 90 days of follow-up, while the secondary outcome data only included hospitalizations and emergency department visits. The study sample included participants with pre-existing cognitive concerns, limiting generalizability to asymptomatic individuals. Finally, some participants did not provide racial information, limiting detailed subgroup comparisons.