Natalie C. Ernecoff

Background: Electronic frailty indices (eFIs) derived from structured electronic health record (EHR) data may incompletely capture functional deficits. Large language models (LLMs) can extract frailty-related information from clinical notes, but whether improved frailty ascertainment improves prognostic performance is uncertain. Objective: To compare frailty ascertainment and prognostic performance between standard and LLM-enhanced eFIs among hospitalized adults. Design: Retrospective cohort study Participants: Adults aged ≥50 years enrolled in two studies at an academic medical center and affiliated community hospital between February 2022 and April 2024. Main Measures: A validated 35-deficit diagnosis-based eFI and an LLM-enhanced eFI incorporating deficits extracted from physical therapy (PT) documentation using GPT-4o. Frailty was defined using an index threshold of ≥0.20. The primary outcome was a composite of discharge to a facility, readmission, or death within 90 days. Multivariable logistic regression models assessed associations with the outcome, and discrimination was compared using area under the receiver operating characteristic curve (AUROC). Secondary analyses examined threshold-based classification and PT-documentation availability; prespecified sensitivity analyses evaluated population and outcome restrictions. Results: Among 616 hospitalizations, mean (SD) age was 70.9 (10.2) years, 329 (53.4%) had PT documentation, and 320 (51.9%) experienced the composite outcome. Standard and LLM-enhanced eFIs classified 216 (35.1%) and 362 (58.8%) hospitalizations as frail, respectively. Both were independently associated with the outcome (standard eFI: adjusted OR, 1.98 [95% CI, 1.39–2.83]; LLM-enhanced eFI: adjusted OR, 2.46 [95% CI, 1.75–3.47]). Discrimination did not differ significantly (AUROC, 0.636 vs 0.660; difference, 0.024 [95% CI, −0.006 to 0.053]; p =0.12). LLM enhancement increased sensitivity (42.5% to 69.4%) but decreased specificity (73.0% to 52.7%). Conclusions: LLM enhancement substantially increased frailty ascertainment and sensitivity for adverse outcomes but did not significantly improve prognostic discrimination. Thus, LLM-enhanced frailty assessment may depend on whether the intended use prioritizes broad identification of functional vulnerability or prediction of adverse post-hospital outcomes.

Background: Advance care planning (ACP) is a process that involves discussing a person's goals, values, and preferences; it is particularly important for persons living with dementia (PLWD) given that dementia is incurable and progressive. To ensure results that will impact real‐world practices, ACP outcome measures must be psychometrically strong, meaningful to key partners, and pragmatic to collect. Therefore, we conducted a scoping review of outcome measures utilized in ACP randomized controlled clinical trials (RCTs) enrolling PLWD or their care partners and evaluated their pragmatic characteristics. Methods: We searched MEDLINE/PubMed, EMBASE, CINAHL, PsycINFO, and Web of Science for peer‐reviewed ACP RCTs enrolling PLWD or their care partners from 2011 to 2021. We abstracted characteristics of primary and secondary outcome measures, including pragmatic characteristics using an adapted Psychometric and Pragmatic Evidence Rating Scale and ACP outcome domains using the standardized ACP Outcome Framework (i.e., process, action, healthcare, or quality of care). Results: We included 21 ACP RCTs. Trials included 103 outcome measures (39 primary and 64 secondary), of which 11% measured process, 14% measured action, 49% measured healthcare, and 26% measured quality of care. Twenty‐four (23%) outcome measures were highly pragmatic, the majority of which (67%) reflected healthcare outcome measures. Sixty‐one (59%) outcomes were assessed as highly relevant to PLWD or their care partners. Only 20% ( n = 21) of outcome measures were embedded into clinical practice. Most (62%) RCTs were conducted in nursing homes, and 33% were focused PLWD with advanced stage disease. Conclusions: In RCTs testing ACP interventions to support PLWD, only 23% of outcome measures were highly pragmatic, and most of these measured healthcare utilizations. Outcome assessments were rarely integrated into the EHR during routine clinical care. New outcome measures that address the lived experience of PLWD and their care partners plus have high pragmatic characteristics are needed for embedded pragmatic clinical trials.

Key Points: Integrating palliative care into kidney care can lead to improvements in symptom management, advance care planning, and hospice utilization. An embedded kidney palliative care model can enable delivery of active medical management without dialysis to patients whose values align with forgoing dialysis. Background: Patients with CKD have high symptom burden, low rates of advance care planning (ACP), and frequently receive care that is not goal concordant. Improved integration of palliative care into nephrology and access to active medical management without dialysis (AMMWD) have the potential to improve outcomes through better symptom management and enhanced shared decision making. Methods: We describe the development of a kidney palliative care (KPC) clinic and how palliative care practices are integrated within an academic nephrology clinic. We performed a retrospective electronic health record (EHR) review for patients seen in this clinic between January 2015 and February 2019 to describe key clinical activities and delivery of AMMWD. Results: A total of 165 patients were seen in the KPC clinic (139 with CKD and 26 who were already receiving dialysis). Fatigue, mobility issues, and pain were the three most prevalent symptoms (85%, 66%, 58%, respectively). Ninety-one percent of patients had a surrogate decision maker documented in the EHR; 87% of patients had a goals-of-care conversation documented in the EHR. Of the 139 patients with CKD, 67 (48%) chose AMMWD as their disease progressed. Sixty-eight percent (41 of 60) of patients who died during the study were referred to hospice. Conclusions: Our findings suggest that the integration of palliative care into nephrology can assist in identification of symptoms, lead to high rates of ACP, and provide a mechanism for patients to choose and receive AMMWD. The percentage of patients choosing AMMWD in our study suggests that increased shared decision making may lower rates of dialysis initiation in the United States. Additional prospective research and registries for assessing the effects of AMMWD have the potential to improve care for people living with CKD.

Background: Specialty palliative care (SPC) provides patient-centered care to people with serious illness and may reduce costs. Specific cost-saving functions of SPC remain unclear. Objectives: (1) To assess the effect of SPC on inpatient costs and length of stay (LOS) and (2) to evaluate differences in costs by indication and timing of SPC. Design: Case–control with in patients who received an SPC consultation and propensity matched controls. Setting: One large U.S. integrated delivery finance system. Measurements: Using administrative data, we assessed costs associated with inpatient stays, a subset of whom received an SPC consultation. Consultations were stratified by reasons based on physician discretion: goals of care, pain management, hospice evaluation, nonpain symptom management, or support. The primary outcome was total operating costs and the secondary outcome was hospital LOS. Results: In total, 1404 patients with SPC consultations associated with unique hospital encounters were matched with 2806 controls. Total operating costs were lower for patients who received an SPC consultation when the consultation was within 0 to 1 days of admission ($6,924 vs. $7,635, p = 0.002). Likewise, LOS was shorter (4.3 vs. 4.7 days, p < 0.001). Upon stratification by reason, goals-of-care consultations early in the hospital stay (days 0–1) were associated with reduced total operating costs ($7,205 vs. $8,677, p < 0.001). Costs were higher for pain management consultations ($7,727 vs. $6,914, p = 0.047). Consultation for hospice evaluation was associated with lower costs, particularly when early (hospital days 0–1: $4,125 vs. $7,415, p < 0.001). Conclusions: SPC was associated with significant cost saving and decreased LOS when occurring early in a hospitalization and used for goals-of-care and hospice evaluation.

Key Points: The Surprise Question can be successfully integrated into the electronic health record for routine collection to aid in clinical practice. A low response rate indicates additional implementation efforts are needed to encourage integration of the Surprise Question into clinical practice. Assessment of reasons for nonuptake highlighted improving technical implementation and providing additional decision making support. Background: The Surprise Question (SQ; “Would you be surprised if this patient died in the next 12 months?”) is a validated prognostication tool for mortality and hospitalization among patients with advanced CKD. Barriers in clinical workflows have slowed SQ implementation in practice. Objectives: The aims of this study were: (1) to evaluate implementation outcomes after the use of electronic health record (EHR) decision support to automate the collection of the SQ; and (2) to assess the prognostic utility of the SQ for mortality and hospitalization/emergency room (ER) visits. Methods: We developed and implemented a best practice alert (BPA) in the EHR to identify nephrology outpatients ≥60 years of age with an eGFR 2. At appointment, the BPA prompted the physician to answer the SQ. We assessed the rate and timeliness of provider responses. We conducted a post-hoc open-ended survey to assess physician perceptions of SQ implementation. We assessed the SQ’s prognostic utility in survival and time-to-hospital encounter (hospitalization/ER visit) analyses. Results: Among 510 patients for whom the BPA triggered, 95 (19%) had the SQ completed by 16 physicians. Among those completed, nearly all (98%) were on appointment day, and 61 (64%) the first time the BPA fired. Providers answered “no” for 27 (28%) and “yes” for 68 (72%) patients. By 12 months, six (22%) “no” patients died; three (4%) “yes” patients died (hazard ratio [HR] 2.86, ref: yes, 95% CI, 1.06 to 7.69). About 35% of “no” patients and 32% of “yes” patients had a hospital encounter by 12 months (HR, 1.85, ref: yes, 95% CI, 0.93 to 3.69). Physicians noted (1) they had goals-of-care conversations unprompted; (2) EHR-based interventions alone for goals-of-care are ineffective; and (3) more robust engagement is necessary. Conclusions: We successfully integrated the SQ into the EHR to aid in clinical practice. Additional implementation efforts are needed to encourage further integration of the SQ in clinical practice.

Background: Written clinical communication regarding patients’ disease understanding and values may facilitate goal-concordant care, yet little is known about the quality of electronic health record (EHR) documentation. We sought to (1) describe frequency of communication best practices in EHR-documented goals-of-care discussions, and (2) assess whether templated notes improve quality of documentation. Methods: Researchers pulled text of EHR-documented goals-of-care discussions for hospitalized patients with Stage IV cancer from admission to 60-days follow-up. Text was included when in a single encounter the clinician addressed: (a) prognosis and/or illness understanding; and (b) goals and/or treatment options. Researchers qualitatively coded text based on guidelines for communication best practices, and noted if an EHR template was used. Results: Forty-two percent (206/492) of patients had EHR-documented goals-of-care discussions. Text frequently described communication of cancer progression (89%), though rarely included prognosis (22%). Text often included patients’ goals and values (83%), and at least on specific treatment decision (82%). Communication about treatments was included for 98% of patients; common examples included cancer treatment (62%), hospice (62%), resuscitation (51%), or intensive care (38%). Clinicians documented making recommendations for 40% of patients. Text addressing patient emotional and spiritual concerns was uncommon (15%). Compared to free text, use of a template was associated with increased documentation of goals and values (80% vs. 61%, p < 0.01), but not other best practices. Conclusion: Insights from the study can be used to guide future training and research to study and improve the quality of documentation about goal of care, and its impact on goal-concordant care.