Impact of hospitalizations on problematic medication use among community dwelling persons with dementia.

BACKGROUND: Hospitalizations are frequently disruptive for persons with dementia (PWD) in part due to the use of potentially problematic medications for complications such as delirium, pain, and insomnia. We sought to determine the impact of hospitalizations on problematic medication prescribing in the months following hospitalization. METHODS: We included community-dwelling PWD in the Health and Retirement Study aged ≥66 with a hospitalization from 2008-2018. We characterized problematic medications as medications that negatively affect cognition (strongly anticholinergics/sedative-hypnotics), medications from 2019 Beers criteria, and medications from STOPP-V2. To capture durable changes, we compared problematic medications 4 weeks pre-hospitalization (baseline) to 4 months post-hospitalization period. We used a generalized linear mixed model with Poisson distribution adjusting for age, sex, comorbidity count, pre-hospital chronic medications, and timepoint. RESULTS: Among 1,475 PWD, 504 had a qualifying hospitalization (median age 84 (IQR=79-90), 66% female, 17% Black). There was a small increase in problematic medications from the baseline to post-hospitalization timepoint that did not reach statistical significance (adjusted mean 1.28 vs. 1.40, difference 0.12 (95% CI -0.03, 0.26), p=0.12). Results were consistent across medication domains and certain subgroups. In one pre-specified subgroup, individuals on <5 pre-hospital chronic medications showed a greater increase in post-hospital problematic medications compared to those on ≥5 medications (p=0.04 for interaction, mean increase from baseline to post-hospitalization of 0.25 for those with <5 medications (95% CI 0.05, 0.44) vs. 0.06 (95% CI -0.12, 0.25) for those with ≥5 medications). CONCLUSIONS: Hospitalizations had a small, non-statistically significant effect on longer-term problematic medication use among PWD.

Prognoses Associated With Palliative Performance Scale Scores in Modern Palliative Care Practice.

Importance: The Palliative Performance Scale (PPS) is one of the most widely used prognostic tools for patients with serious illness. However, current prognostic estimates associated with PPS scores are based on data that are over a decade old. Objective: To generate updated prognostic estimates by PPS score, care setting, and illness category, and examine how well PPS predicts short- and longer-term survival. Design, Setting, and Participants: This prognostic study was conducted at a large academic medical center with robust inpatient and outpatient palliative care practices using electronic health record data linked with data from California Vital Records. Eligible participants included patients who received a palliative care consultation between January 1, 2018, and December 31, 2020. Data analysis was conducted from November 2022 to February 2024. Exposure: Palliative care consultation with a PPS score documented. Main Outcomes and Measures: The primary outcomes were predicted 1-, 6-, and 12-month mortality and median survival of patients by PPS score in the inpatient and outpatient settings, and performance of the PPS across a range of survival times. In subgroup analyses, mortality risk by PPS score was estimated in patients with cancer vs noncancer illnesses and those seen in-person vs by video telemedicine in the outpatient setting. Results: Overall, 4779 patients (mean [SD] age, 63.5 [14.8] years; 2437 female [51.0%] and 2342 male [49.0%]) had a palliative care consultation with a PPS score documented. Of these patients, 2276 were seen in the inpatient setting and 3080 were seen in the outpatient setting. In both the inpatient and outpatient settings, 1-, 6-, and 12-month mortality were higher and median survival was shorter for patients with lower PPS scores. Prognostic estimates associated with PPS scores were substantially longer (2.3- to 11.7-fold) than previous estimates commonly used by clinicians. The PPS had good ability to discriminate between patients who lived and those who died in the inpatient setting (integrated time-dependent area under the curve [iAUC], 0.74) but its discriminative ability was lower in the outpatient setting (iAUC, 0.67). The PPS better predicted 1-month survival than longer-term survival. Mortality rates were higher for patients with cancer than other serious illnesses at most PPS levels. Conclusions and Relevance: In this prognostic study, prognostic estimates associated with PPS scores were substantially longer than previous estimates commonly used by clinicians. Based on these findings, an online calculator was updated to assist clinicians in reaching prognostic estimates that are more consistent with modern palliative care practice and specific to the patient's setting and diagnosis group.

High-risk medication use among older adults with cognitive impairment living alone in the United States.

BACKGROUND: More than one-fourth of older adults with cognitive impairment (CI) live alone; these individuals often lack support for medication management and face a high risk of adverse drug events. We characterized the frequency and types of high-risk medications used by older adults with CI living alone and, for context, compared patterns with those in older adults with CI living with others. METHODS: This was a cross-sectional study of National Health and Aging Trends Study (NHATS) data and Medicare claims (2015-2017). We ascertained cognitive status from NHATS and medication use with Part D claims. We compared high-risk medication use (those with adverse cognitive effects or low tolerance for misuse) among older adults with CI living alone versus living with others using logistic regression models adjusted for demographic/clinical factors. RESULTS: The unweighted sample included 1569 older adults with CI, of whom 491 (weighted national estimate, 31%) were living alone. In the living-alone group, the mean age was 79.9 years and 66% were female, 64% reported managing medications on their own without difficulty, 14% reported managing medications on their own with difficulty, and 18% received total support with medication management. Older adults with CI living alone used a median of 5 medications (IQR, 3-8), 16% took ≥10 medications, and 46% took ≥1 high-risk medication (anticholinergic/sedating: 24%; opioid: 13%; anticoagulant: 10%; sulfonylurea: 10%; insulin: 9%). Compared with those living with others, the use of high-risk medications was similar (p > 0.05 for unadjusted/adjusted comparisons). Those living alone were more likely both to take at least one high-risk medication and not receive help with medication management: 34% in those living alone versus 23% living with others (p < 0.05 for unadjusted/adjusted comparisons). CONCLUSIONS: Older adults with CI living alone use many medications; nearly half use high-risk medications. Our findings can inform medication optimization interventions supporting this vulnerable population.

Re-thinking all-cause COVID-19 hospitalizations as a surrogate measure for severe illness in observational surveillance studies.

All-cause COVID-19 hospitalization ≤ 30 days of infection is a common outcome for severe illness in observational/surveillance studies. Milder COVID-19 disease and COVID-19-specific measurements calls for an evaluation of this endpoint. This was a descriptive, retrospective cohort study of adults ≥ 18 who were established in primary care at Veteran Health Administration (VHA) facilities. The outcome was hospitalization within 30 days of a laboratory-confirmed, symptomatic SARS-CoV-2 infection. Between December 15, 2021 and May 1, 2022, a simple random sample of all VA facilities, excluding Puerto Rico or Philippines, was drawn to identify these hospitalized cases and determine whether hospitalization was due to COVID-19-specific causes. A chart review was conducted to record the inpatient clinical team's diagnosis and whether the inpatient team classified the diagnosis as COVID-19 related or not. These data were used to classify hospitalizations as either due to COVID-19-specific causes (direct manifestations of SARS-CoV-2 infection) or non-COVID-19-specific hospitalizations (incidental SARS-CoV-2 infection), A simple random sample of 9966 (12.3%) all-cause hospitalizations (95% CI: 12.1%, 12.5%) was used to select 300 representative patients. Of these, 226/300 (75.3%) were determined to be COVID-19-specific. COVID-19 pneumonia was most common (147/226, 65.0%). The highest proportion of COVID-19-specific hospitalizations occurred among unvaccinated (85.0%), followed by vaccinated but not boosted (73.7%) and boosted (59.4%) (p < 0.001). The proportion of non-COVID-19-specific hospitalizations was higher in the later period (15-30 days: 55.0%) than the early (0-15 days: 22.5%) (p = 0.003). This study supports the outcome of COVID-19-specific hospitalization instead of all-cause hospitalization in observational studies. The earlier outcome period (0-15 days) was less susceptible to potential measurement bias.

Geriatric conditions and healthcare utilization among older adults living in subsidized housing.

BACKGROUND: Nearly 2.9 million older Americans with lower incomes live in subsidized housing. While regional and single-site studies show that this group has higher rates of healthcare utilization compared to older adults in the general community, little is known about healthcare utilization nationally nor associated risk factors. METHODS: We conducted a retrospective cohort study of Medicare beneficiaries aged ≥65 enrolled in the National Health and Aging Trends Study in 2011, linked to Medicare claims data, including individuals living in subsidized housing and the general community. Participants were followed annually through 2020. Outcomes were hospitalization, short-term skilled nursing facility (SNF) utilization, long-term care utilization, and death. Fine-Gray competing risks regression analysis was used to assess the association of subsidized housing residence with hospitalization and nursing facility utilization, and Cox proportional hazards regression analysis was used to assess the association with death. RESULTS: Among 6294 participants (3600 women, 2694 men; mean age, 75.5 years [SD, 7.0]), 295 lived in subsidized housing at baseline and 5999 in the general community. Compared to older adults in the general community, those in subsidized housing had a higher adjusted subdistribution hazard ratio [sHR] of hospitalization (sHR 1.21; 95% CI, 1.03-1.43), short-term SNF utilization (sHR 1.49; 95% CI, 1.15-1.92), and long-term care utilization (sHR 2.72; 95% CI, 1.67-4.43), but similar hazard of death (HR, 0.86; 95% CI, 0.69-1.08). Individuals with functional impairment had a higher adjusted subdistribution hazard of hospitalization and short-term SNF utilization and individuals with dementia and functional impairment had a higher hazard of long-term care utilization. CONCLUSIONS: Older adults living in subsidized housing have higher hazards of hospitalization and nursing facility utilization compared to those in the general community. Housing-based interventions to optimize aging in place and mitigate risk of nursing facility utilization should consider risk factors including functional impairment and dementia.

Development of the UroARC Surgical Calculator: A Novel Risk Calculator for Older Adults Undergoing Surgery for Bladder Outlet Obstruction.

PURPOSE: Bladder outlet obstruction (BOO) is common in older adults. Many older adults who pursue surgery have additional vulnerabilities affecting surgical risk, including frailty. A clinical tool that builds on frailty to predict surgical outcomes for the spectrum of BOO procedures would be helpful to aid in surgical decision-making but does not currently exist. MATERIALS AND METHODS: Medicare beneficiaries undergoing BOO surgery from 2014 to 2016 were identified and analyzed using the Medicare MedPAR, Outpatient, and Carrier files. Eight different BOO surgery categories were created. Baseline frailty was calculated for each beneficiary using the Claims-Based Frailty Index (CFI). All 93 variables in the CFI and the 17 variables in the Charlson Comorbidity Index were individually entered into stepwise logistic regression models to determine variables most highly predictive of complications. Similar and duplicative variables were combined into categories. Calibration curves and tests of model fit, including C statistics, Brier scores, and Spiegelhalter P values, were calculated to ensure the prognostic accuracy for postoperative complications. RESULTS: In total, 212,543 beneficiaries were identified. Approximately 42.5% were prefrail (0.15 ≤ CFI < 0.25), 8.7% were mildly frail (0.25 ≤ CFI < 0.35), and 1.2% were moderately-to-severely frail (CFI ≥0.35). Using stepwise logistic regression, 13 distinct prognostic variable categories were identified as the most reliable predictors of postoperative outcomes. Most models demonstrated excellent model discrimination and calibration with high C statistic and Spiegelhalter P values, respectively, and high accuracy with low Brier scores. Calibration curves for each outcome demonstrated excellent model fit. CONCLUSIONS: This novel risk assessment tool may help guide surgical prognostication among this vulnerable population.

Incidence of Traumatic Brain Injury in a Longitudinal Cohort of Older Adults.

Importance: Traumatic brain injury (TBI) occurs at the highest rate in older adulthood and increases risk for cognitive impairment and dementia. Objectives: To update existing TBI surveillance data to capture nonhospital settings and to explore how social determinants of health (SDOH) are associated with TBI incidence among older adults. Design, Setting, and Participants: This nationally representative longitudinal cohort study assessed participants for 18 years, from August 2000 through December 2018, using data from the Health and Retirement Study (HRS) and linked Medicare claims dates. Analyses were completed August 9 through December 12, 2022. Participants were 65 years of age or older in the HRS with survey data linked to Medicare without a TBI prior to HRS enrollment. They were community dwelling at enrollment but were retained in HRS if they were later institutionalized. Exposures: Baseline demographic, cognitive, medical, and SDOH information from HRS. Main Outcomes and Measures: Incident TBI was defined using inpatient and outpatient International Classification of Diseases, Ninth or Tenth Revision, diagnosis codes received the same day or within 1 day as the emergency department (ED) visit code and the computed tomography (CT) or magnetic resonance imaging (MRI) code, after baseline HRS interview. A cohort with TBI codes but no ED visit or CT or MRI scan was derived to capture diagnoses in nonhospital settings. Descriptive statistics and bivariate associations of TBI with demographic and SDOH characteristics used sample weights. Fine-Gray regression models estimated associations between covariates and TBI, with death as a competing risk. Imputation considering outcome and complex survey design was performed by race and ethnicity, sex, education level, and Area Deprivation Index percentiles 1, 50, and 100. Other exposure variables were fixed at their weighted means. Results: Among 9239 eligible respondents, 5258 (57.7%) were female and 1210 (9.1%) were Black, 574 (4.7%) were Hispanic, and 7297 (84.4%) were White. Mean (SD) baseline age was 75.2 (8.0) years. During follow-up (18 years), 797 (8.9%) of respondents received an incident TBI diagnosis with an ED visit and a CT code within 1 day, 964 (10.2%) received an incident TBI diagnosis and an ED code, and 1148 (12.9%) received a TBI code with or without an ED visit and CT scan code. Compared with respondents without incident TBI, respondents with TBI were more likely to be female (absolute difference, 7.0 [95% CI, 3.3-10.8]; P < .001) and White (absolute difference, 5.1 [95% CI, 2.8-7.4]; P < .001), have normal cognition (vs cognitive impairment or dementia; absolute difference, 6.1 [95% CI, 2.8-9.3]; P = .001), higher education (absolute difference, 3.8 [95% CI, 0.9-6.7]; P < .001), and wealth (absolute difference, 6.5 [95% CI, 2.3-10.7]; P = .01), and be without baseline lung disease (absolute difference, 5.1 [95% CI, 3.0-7.2]; P < .001) or functional impairment (absolute difference, 3.3 [95% CI, 0.4-6.1]; P = .03). In adjusted multivariate models, lower education (subdistribution hazard ratio [SHR], 0.73 [95% CI, 0.57-0.94]; P = .01), Black race (SHR, 0.61 [95% CI, 0.46-0.80]; P < .001), area deprivation index national rank (SHR 1.00 [95% CI 0.99-1.00]; P = .009), and male sex (SHR, 0.73 [95% CI, 0.56-0.94]; P = .02) were associated with membership in the group without TBI. Sensitivity analyses using a broader definition of TBI yielded similar results. Conclusions and Relevance: In this longitudinal cohort study of older adults, almost 13% experienced incident TBI during the 18-year study period. For older adults who seek care for TBI, race and ethnicity, sex, and SDOH factors may be associated with incidence of TBI, seeking medical attention for TBI in older adulthood, or both.

Synaptic injury in the inner plexiform layer of the retina is associated with progression in multiple sclerosis.

While neurodegeneration underlies the pathological basis for permanent disability in multiple sclerosis (MS), predictive biomarkers for progression are lacking. Using an animal model of chronic MS, we find that synaptic injury precedes neuronal loss and identify thinning of the inner plexiform layer (IPL) as an early feature of inflammatory demyelination-prior to symptom onset. As neuronal domains are anatomically segregated in the retina and can be monitored longitudinally, we hypothesize that thinning of the IPL could represent a biomarker for progression in MS. Leveraging our dataset with over 800 participants enrolled for more than 12 years, we find that IPL atrophy directly precedes progression and propose that synaptic loss is predictive of functional decline. Using a blood proteome-wide analysis, we demonstrate a strong correlation between demyelination, glial activation, and synapse loss independent of neuroaxonal injury. In summary, monitoring synaptic injury is a biologically relevant approach that reflects a potential driver of progression.

One-Year Recovery Among Survivors of Prolonged Severe COVID-19: A National Multicenter Cohort.

OBJECTIVES: Understanding the long-term effects of severe COVID-19 illness on survivors is essential for effective pandemic recovery planning. Therefore, we investigated impairments among hospitalized adults discharged to long-term acute care hospitals (LTACHs) for prolonged severe COVID-19 illness who survived 1 year. DESIGN: The Recovery After Transfer to an LTACH for COVID-19 (RAFT COVID) study was a national, multicenter, prospective longitudinal cohort study. SETTING AND PATIENTS: We included hospitalized English-speaking adults transferred to one of nine LTACHs in the United States between March 2020 and February 2021 and completed a survey. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: Validated instruments for impairments and free response questions about recovering. Among 282 potentially eligible participants who provided permission to be contacted, 156 (55.3%) participated (median age, 65; 38.5% female; 61.3% in good prior health; median length of stay of 57 d; 77% mechanically ventilated for a median of 26 d; 42% had a tracheostomy). Approximately two-thirds (64%) had a persistent impairment, including physical (57%), respiratory (49%; 19% on supplemental oxygen), psychiatric (24%), and cognitive impairments (15%). Nearly half (47%) had two or more impairment types. Participants also experienced persistent debility from hospital-acquired complications, including mononeuropathies and pressure ulcers. Participants described protracted recovery, attributing improvements to exercise/rehabilitation, support, and time. While considered life-altering with 78.7% not returning to their usual health, participants expressed gratitude for recovering; 99% returned home and 60% of previously employed individuals returned to work. CONCLUSIONS: Nearly two-thirds of survivors of among the most prolonged severe COVID-19 illness had persistent impairments at 1 year that resembled post-intensive care syndrome after critical illness plus debility from hospital-acquired complications.

Patterns of peripartum depression and anxiety during the pre-vaccine COVID-19 pandemic.

BACKGROUND: Pregnant people are vulnerable to new or worsening mental health conditions. This study aims to describe prevalence and course of depression and anxiety symptoms in pregnancy during the pre-vaccine COVID-19 pandemic. METHODS: This is a prospective cohort study of pregnant individuals with known or suspected COVID-19. Participants completed Edinburgh Postnatal Depression Scale (EPDS) and Generalized-Anxiety Disorder-7 (GAD-7) questionnaires, screening tools for depression and anxiety, at 34weeks gestational age, 6-8weeks postpartum, and 6months postpartum. Prevalence of elevated depressive and anxiety symptoms at each visit was described. Univariable logistic regression analysis was used to determine the association between demographic and clinical factors and those with elevated depression or anxiety symptoms. RESULTS: 317 participants were included. The prevalence of elevated antepartum depression symptoms was 14.6%, 10.3%, and 20.6% at 34weeks gestational age, 6-8weeks postpartum, and 6months postpartum, respectively. The rate of elevated anxiety symptoms was 15.1%, 10.0%, and 17.3% at 34weeks gestational age, 6-8weeks postpartum, and 6months postpartum, respectively. A prior history of depression and/or anxiety (p's < 0.03), as well as higher EPDS and GAD-7 scores at enrollment (p's < 0.04) associated with elevated depression and anxiety symptoms throughout pregnancy and the postpartum period. Quarantining during pregnancy was associated with elevated anxiety symptoms at 34weeks gestational age in univariate (P = 0.027) analyses. COVID-19 diagnosis and hospitalization were not associated with elevated depression or anxiety symptoms. CONCLUSIONS: Elevated depression and anxiety symptoms were prevalent throughout pregnancy and the postpartum period, particularly in those with prior depression and/or anxiety and who quarantined. Strategies that target social isolation may mitigate potential adverse consequences for pregnant people, and continued vigilance in recognition of depression and anxiety in pregnancy should be considered.

Association of Race and Ethnicity With Postoperative Gabapentinoid and Opioid Prescribing Trends for Older Adults.

BACKGROUND: Disparities in opioid prescribing by race/ethnicity have been described in many healthcare settings, with White patients being more likely to receive an opioid prescription than other races studied. As surgeons increase prescribing of nonopioid medications in response to the opioid epidemic, it is unknown whether postoperative prescribing disparities also exist for these medications, specifically gabapentinoids. METHODS: We conducted a retrospective cohort study using a 20% Medicare sample for 2013-2018. We included patients ≥66 years without prior gabapentinoid use who underwent one of 14 common surgical procedures. The primary outcome was the proportion of patients prescribed gabapentinoids at discharge among racial and ethnic groups. Secondary outcomes were days' supply of gabapentinoids, opioid prescribing at discharge, and oral morphine equivalent (OME) of opioid prescriptions. Trends over time were constructed by analyzing proportion of postoperative prescribing of gabapentinoids and opioids for each year. For trends by year by racial/ethnic groups, we ran a multivariable logistic regression with an interaction term of procedure year and racial/ethnic group. RESULTS: Of the 494,922 patients in the cohort (54% female, 86% White, 5% Black, 5% Hispanic, mean age 73.7 years), 3.7% received a new gabapentinoid prescription. Gabapentinoid prescribing increased over time for all groups and did not differ significantly among groups (P = 0.13). Opioid prescribing also increased, with higher proportion of prescribing to White patients than to Black and Hispanic patients in every year except 2014. CONCLUSIONS: We found no significant prescribing variation of gabapentinoids in the postoperative period between racial/ethnic groups. Importantly, we found that despite national attention to disparities in opioid prescribing, variation continues to persist in postoperative opioid prescribing, with a higher proportion of White patients being prescribed opioids, a difference that persisted over time.

Which older adults are at highest risk of prescribing cascades? A national study of the gabapentinoid-loop diuretic cascade.

BACKGROUND: Prescribing cascades are important contributors to polypharmacy. Little is known about which older adults are at highest risk of experiencing prescribing cascades. We explored which older veterans are at highest risk of the gabapentinoid (including gabapentin and pregabalin)-loop diuretic (LD) cascade, given the dramatic increase in gabapentinoid prescribing in recent years. METHODS: Using Veterans Affairs and Medicare claims data (2010-2019), we performed a prescription sequence symmetry analysis (PSSA) to assess loop diuretic initiation before and after gabapentinoid initiation among older veterans (≥66 years). To identify the cascade, we calculated the adjusted sequence ratio (aSR), which assesses the temporality of LD relative to gabapentinoid initiation. To explore high-risk groups, we used multivariable logistic regression with prescribing order modeled as a binary dependent variable. We calculated adjusted odds ratios (aORs), measuring the extent to which factors are associated with one prescribing order versus another. RESULTS: Of 151,442 veterans who initiated a gabapentinoid, there were 1,981 patients who initiated a LD within 6 months after initiating a gabapentinoid compared to 1,599 patients who initiated a LD within 6 months before initiating a gabapentinoid. In the gabapentinoid-LD group, the mean age was 73 years, 98% were male, 13% were Black, 5% were Hispanic, and 80% were White. Patients in each group were similar across patient and health utilization factors (standardized mean difference <0.10 for all comparisons). The aSR was 1.23 (95% CI: 1.13, 1.34), strongly suggesting the cascade's presence. People age ≥85 years were less likely to have the cascade (compared to 66-74 years; aOR 0.74, 95% CI: 0.56-0.96), and people taking ≥10 medications were more likely to have the cascade (compared to 0-4 drugs; aOR 1.39, 95% CI: 1.07-1.82). CONCLUSIONS: Among older adults, those who are younger and taking many medications may be at higher risk of the gabapentinoid-LD cascade, contributing to worsening polypharmacy and potential drug-related harms. We did not identify strong predictors of this cascade, suggesting that prescribing cascade prevention efforts should be widespread rather than focused on specific subgroups.

Long-term cognitive outcome after elective hip or knee total joint arthroplasty: A population-based observational study.

BACKGROUND: One year after elective hip or knee total joint arthroplasty (TJA), >30% of older adults meet criteria for postoperative neurocognitive disorder. However, this is not contextualized with long-term cognitive outcomes in comparable surgical and nonsurgical controls. We analyzed population-based data to compare long-term cognitive outcomes in older adults after TJA, other surgeries, and with and without arthritis pain. METHODS: This was a retrospective observational analysis of United States older adults in the Health and Retirement Study (HRS) who underwent elective TJA, or elective surgery without expected functional benefits (e.g., cholecystectomy; inguinal herniorrhaphy), between 1998 and 2018 at aged 65 or older. TJA recipients were also age- and sex-matched to nonsurgical controls who reported moderate-severe arthritic pain or denied pain, so that comparison groups included surgical and nonsurgical (pain-suffering and pain-free) controls. We modeled biennially-assessed memory performance, a measure of direct and proxy cognitive assessments, before and after surgery, normalized to the rate of memory decline ("cognitive aging") in controls to express effect size estimates as excess, or fewer, months of memory decline. We used linear mixed effects models adjusted for preoperative health and demographic factors, including frailty, flexibly capturing time before/after surgery (knots at -4, 0, 8 years; discontinuity at surgery). RESULTS: There were 1947 TJA recipients (average age 74; 63% women; 1358 knee, 589 hip) and 1631 surgical controls (average age 76; 38% women). Memory decline 3 years after TJA was similar to surgical controls (5.2 [95% confidence interval, CI -1.2 to 11.5] months less memory decline in the TJA group, p = 0.11) and nonsurgical controls. At 5 years, TJA recipients experienced 5.0 [95% CI -0.9 to 10.9] months less memory decline than arthritic pain nonsurgical controls. CONCLUSION: There is no systematic accelerated memory decline at 3 years after TJA compared with surgical or nonsurgical controls.

Measuring Implicit Bias in ICU Notes Using Word-Embedding Neural Network Models.

BACKGROUND: Language in nonmedical data sets is known to transmit human-like biases when used in natural language processing (NLP) algorithms that can reinforce disparities. It is unclear if NLP algorithms of medical notes could lead to similar transmissions of biases. RESEARCH QUESTION: Can we identify implicit bias in clinical notes, and are biases stable across time and geography? STUDY DESIGN AND METHODS: To determine whether different racial and ethnic descriptors are similar contextually to stigmatizing language in ICU notes and whether these relationships are stable across time and geography, we identified notes on critically ill adults admitted to the University of California, San Francisco (UCSF), from 2012 through 2022 and to Beth Israel Deaconess Hospital (BIDMC) from 2001 through 2012. Because word meaning is derived largely from context, we trained unsupervised word-embedding algorithms to measure the similarity (cosine similarity) quantitatively of the context between a racial or ethnic descriptor (eg, African-American) and a stigmatizing target word (eg, nonco-operative) or group of words (violence, passivity, noncompliance, nonadherence). RESULTS: In UCSF notes, Black descriptors were less likely to be similar contextually to violent words compared with White descriptors. Contrastingly, in BIDMC notes, Black descriptors were more likely to be similar contextually to violent words compared with White descriptors. The UCSF data set also showed that Black descriptors were more similar contextually to passivity and noncompliance words compared with Latinx descriptors. INTERPRETATION: Implicit bias is identifiable in ICU notes. Racial and ethnic group descriptors carry different contextual relationships to stigmatizing words, depending on when and where notes were written. Because NLP models seem able to transmit implicit bias from training data, use of NLP algorithms in clinical prediction could reinforce disparities. Active debiasing strategies may be necessary to achieve algorithmic fairness when using language models in clinical research.

Synthetic Health Data Can Augment Community Research Efforts to Better Inform the Public During Emerging Pandemics.

The COVID-19 pandemic had disproportionate effects on the Veteran population due to the increased prevalence of medical and environmental risk factors. Synthetic electronic health record (EHR) data can help meet the acute need for Veteran population-specific predictive modeling efforts by avoiding the strict barriers to access, currently present within Veteran Health Administration (VHA) datasets. The U.S. Food and Drug Administration (FDA) and the VHA launched the precisionFDA COVID-19 Risk Factor Modeling Challenge to develop COVID-19 diagnostic and prognostic models; identify Veteran population-specific risk factors; and test the usefulness of synthetic data as a substitute for real data. The use of synthetic data boosted challenge participation by providing a dataset that was accessible to all competitors. Models trained on synthetic data showed similar but systematically inflated model performance metrics to those trained on real data. The important risk factors identified in the synthetic data largely overlapped with those identified from the real data, and both sets of risk factors were validated in the literature. Tradeoffs exist between synthetic data generation approaches based on whether a real EHR dataset is required as input. Synthetic data generated directly from real EHR input will more closely align with the characteristics of the relevant cohort. This work shows that synthetic EHR data will have practical value to the Veterans' health research community for the foreseeable future.