Year: 2023
Type: article
Abstract: Alzheimer's disease and Alzheimer's disease related dementias (ADRD) cause disability and death in older adults.1 Specialty palliative care improves outcomes for other serious illnesses but is rarely used for ADRD.2 To address this gap, we are conducting the ADRD-PC multi-site clinical trial of specialty palliative care for hospitalized people with ADRD and their caregivers.3 To facilitate timely trial enrollment, we designed and implemented an electronic health record (EHR) algorithm to identify people with ADRD at the time of hospitalization at a single site.4 Across four sites, we now report utility of this algorithm identifying people with late-stage ADRD. We (1) examine the positive predictive value (PPV) of the EHR algorithm compared to a chart diagnosis of ADRD confirmed by research assistant (RA), and (2) the PPV of the EHR algorithm compared to physician confirmed diagnosis of late-stage ADRD. We also provide descriptive data on the efficiencies for participant enrollment. The ADRD-PC trial enrolls dyads of hospitalized patients with late-stage ADRD and their caregivers at medical centers in North Carolina (NC), Massachusetts (MA), Indiana (IN), and Colorado (CO). Dyads are randomized to either the intervention or control (usual care). Those randomized to the intervention received a dementia-specific specialty palliative care consultation with transitional care up to 2 weeks after discharge. The ADRD-PC trial began enrollment in July 2021 and is ongoing. The Institutional Review Board of record was at The University of North Carolina-Chapel Hill (UNC)—all external sites ceded oversight to UNC's IRB. Eligible hospitalized patients are over age 55, have late-stage dementia (e.g., Global Deterioration Scale5 (GDS) score of 6 or 7, or GDS 5 plus Charlson Comorbidity Index6 (CCI) score of 5 or greater), are not receiving hospice or palliative care, and have an English-speaking surrogate decision maker. Each site implemented an EHR-based algorithm.4 The EHR algorithm identifies admitted patients aged 55 or older with one of 34 ICD-10 dementia-related diagnosis codes (Appendix of original publication). Each site's RA receives a real-time notification of potentially eligible patients, and then reviews the EHR for trial eligibility, particularly the presence of late-stage dementia. Physicians who are site PIs confirm the ADRD diagnosis and stage using EHR review and consultation with the patient's attending physician (Supplementary Figure S1). We used descriptive statistics to describe participant characteristics. The PPV of the EHR algorithm for ADRD diagnosis is the probability that a participant with algorithm-identified ADRD has an established diagnosis of ADRD upon chart review by the RA. The PPV for late-stage ADRD is the total number of dyads confirmed as having late-stage dementia over all suspected cases of late-stage dementia identified by the algorithm and RA chart review. Participant characteristics are outlined in Table 1. For the n = 3786 potential cases identified by the algorithm, 2644 true positives were confirmed by RA chart review. Thus, the PPV of the algorithm to detect people with ADRD was 69.8%, and site-based PPV ranged from 54.5% to 83.0% (Table 2). The RA chart reviews identified n = 805 potential cases of late-stage ADRD, and site-PIs confirmed 688 as people with late stage (GDS 5–7) ADRD. Thus, the PPV of the EHR algorithm combined with RA chart review was 85.5%, and site-based PPV for late-stage ADRD ranged from 75.1% to 96.1%. An EHR algorithm to identify hospitalized people with ADRD facilitated rapid identification and enrollment in a clinical trial across multiple sites. Furthermore, the EHR algorithm combined with brief chart review efficiently identified people with late-stage dementia, thus facilitating enrollment in a multi-site clinical trial. A historical barrier to the success of pragmatic trials for people with ADRD has been the inability to identify participants who may benefit from the intervention.7 In 2020, investigators from the National Institute on Aging funded, IMbedded Pragmatic Alzheimer's disease and AD-Related Dementias Clinical Trials (IMPACT) collaboratory encouraged the development of accurate and complete case-finding algorithms that could be used in diverse populations and across settings.7 Currently, there is no “gold standard” algorithm for identifying people with ADRD. A major strength of this study is our successful implementation of this algorithm in multiple sites. As with similar algorithms, we likely miss potential participants who lack dementia-specific ICD10 codes, though this is more likely in early-stage ADRD. Algorithms under development using natural language processing may be more inclusive, or may permit greater specificity such as a focus on people with behavioral symptom distress.8, 9 Future adaptations of this method may facilitate ADRD clinical trials or quality improvement across various clinical settings. All listed authors had full access to all the data in the study, take responsibility for the integrity of the data and the accuracy of the data analysis, and had authority over manuscript preparation, the decision to submit the manuscript for publication, and approved its current contents. All authors meet the criteria for authorship stated in the Uniform Requirements for Manuscripts Submitted to Biomedical Journals. The authors declare no conflicts of interest. Laura Hanson's research reported in this publication was supported in part by the National Institute on Aging of the National Institutes of Health (R01 AG065394). The content is solely the authors' responsibility and does not necessarily represent the official views of the National Institutes of Health. Funding to support this work: AG065394. Supplementary Figure S1. ADRD-PC consort diagram: Screening to enrollment. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.
DOI: 10.1111/jgs.18558
Institution: University of North Carolina at Chapel Hill
Language: English
Subfield: Psychiatry and Mental health
Field: Medicine
Domain: Health Sciences
Open Access status: bronze
Funder: National Institute on Aging
Awards: G1612345590, G2737324056
OpenAlex ID: W4385969535