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.

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.

Negativity and Positivity in the ICU: Exploratory Development of Automated Sentiment Capture in the Electronic Health Record.

OBJECTIVES: To develop proof-of-concept algorithms using alternative approaches to capture provider sentiment in ICU notes. DESIGN: Retrospective observational cohort study. SETTING: The Multiparameter Intelligent Monitoring of Intensive Care III (MIMIC-III) and the University of California, San Francisco (UCSF) deidentified notes databases. PATIENTS: Adult (≥18 yr old) patients admitted to the ICU. MEASUREMENTS AND MAIN RESULTS: We developed two sentiment models: 1) a keywords-based approach using a consensus-based clinical sentiment lexicon comprised of 72 positive and 103 negative phrases, including negations and 2) a Decoding-enhanced Bidirectional Encoder Representations from Transformers with disentangled attention-v3-based deep learning model (keywords-independent) trained on clinical sentiment labels. We applied the models to 198,944 notes across 52,997 ICU admissions in the MIMIC-III database. Analyses were replicated on an external sample of patients admitted to a UCSF ICU from 2018 to 2019. We also labeled sentiment in 1,493 note fragments and compared the predictive accuracy of our tools to three popular sentiment classifiers. Clinical sentiment terms were found in 99% of patient visits across 88% of notes. Our two sentiment tools were substantially more predictive (Spearman correlations of 0.62-0.84, p values < 0.00001) of labeled sentiment compared with general language algorithms (0.28-0.46). CONCLUSION: Our exploratory healthcare-specific sentiment models can more accurately detect positivity and negativity in clinical notes compared with general sentiment tools not designed for clinical usage.

The Epidemiology of Smoking in Older Adults: A National Cohort Study.

BACKGROUND: Older smokers account for the greatest tobacco-related morbidity and mortality in the USA, while quitting smoking remains the single most effective preventive health intervention for reducing the risk of smoking-related illness. Yet, knowledge about patterns of smoking and smoking cessation in older adults is lacking. OBJECTIVE: Assess trends in prevalence of cigarette smoking between 1998 and 2018 and identify patterns and predictors of smoking cessation in US older adults. DESIGN: Retrospective cohort study PARTICIPANTS: Individuals aged 55+ enrolled in the nationally representative Health and Retirement Study, 1998-2018 MAIN MEASURES: Current smoking was assessed with the question: "Do you smoke cigarettes now?" Quitting smoking was defined as having at least two consecutive waves (between 2 and 4 years) in which participants who were current smokers in 1998 reported they were not currently smoking in subsequent waves. KEY RESULTS: Age-adjusted smoking prevalence decreased from 15.9% in 1998 (95% confidence interval (CI) 15.2, 16.7) to 11.2% in 2018 (95% CI 10.4, 12.1). Among 2187 current smokers in 1998 (mean age 64, 56% female), 56% of those living to age 90 had a sustained period of smoking cessation. Smoking less than 10 cigarettes/day was strongly associated with an increased likelihood of quitting smoking (subdistribution hazard ratio 2.3; 95% CI 1.9, 2.8), compared to those who smoked more than 20 cigarettes/day. CONCLUSIONS: Smoking prevalence among older persons has declined and substantial numbers of older smokers succeed in quitting smoking for a sustained period. These findings highlight the need for continued aggressive efforts at tobacco cessation among older persons.

Attitudes toward deprescribing among older adults with dementia in the United States.

BACKGROUND: People with dementia (PWD) take medications that may be unnecessary or harmful. This problem can be addressed through deprescribing, but it is unclear if PWD would be willing to engage in deprescribing with their providers. Our goal was to investigate attitudes toward deprescribing among PWD. METHODS: This was a cross-sectional study of 422 PWD aged ≥65 years who completed the medications attitudes module of the National Health and Aging Trends Study (NHATS) in 2016. Proxies provided responses when a participant was unable to respond due to health or cognitive problems. Attitudinal outcomes comprised responses to two statements from the patients' attitudes toward deprescribing questionnaire and its revised version (representing belief about the necessity of one's medications and willingness to deprescribe); another elicited the maximum number of pills that a respondent would be comfortable taking. RESULTS: The weighted sample represented over 1.8 million PWD; 39% were 75 to 84 years old and 38% were 85 years or older, 60% were female, and 55% reported six or more regular medications. Proxies provided responses for 26% of PWD. Overall, 22% believed that they may be taking one or more medicines that they no longer needed, 87% were willing to stop one or more of their medications, and 50% were uncomfortable taking five or more medications. Attitudinal outcomes were similar across sociodemographic and clinical factors. PWD taking ≥6 medications were more likely to endorse a belief that at least one medication was no longer necessary compared to those taking <6 (adjusted probability 29% [95% confidence interval (CI), 22%-38%] vs. 13% [95% CI, 8%-20%]; p = 0.004); the same applied for willingness to deprescribe (92% [95% CI, 87%-95%] vs. 83% [95% CI, 76%-89%]; p = 0.04). CONCLUSIONS: A majority of PWD are willing to deprescribe, representing an opportunity to improve quality of life for this vulnerable population.

Incidence of potentially disruptive medical and social events in older adults with and without dementia.

BACKGROUND: Potentially disruptive medical, surgical, and social events-such as pneumonia, hip fracture, and widowhood-may accelerate the trajectory of decline and impact caregiving needs in older adults, especially among people with dementia (PWD). Prior research has focused primarily on nursing home residents with dementia. We sought to assess the incidence of potentially disruptive events in community-dwelling people with and without dementia. METHODS: Retrospective cohort study of participants aged 65+ enrolled in the Health and Retirement Study between 2010 and 2018 (n = 9346), including a subset who were married-partnered at baseline (n = 5105). Dementia was defined with a previously validated algorithm. We calculated age-adjusted and gender-stratified incidence per 1000 person-years and incidence rate ratios of: 1) hospitalization for pneumonia, 2) hip fracture, and 3) widowhood in people with and without dementia. RESULTS: PWD (n = 596) were older (mean age 84 vs. 75) and a higher proportion were female (67% vs. 57%) than people without dementia (PWoD) (n = 8750). Age-adjusted incidence rates (per 1000 person-years) of pneumonia were higher in PWD (113.1; 95% CI 94.3, 131.9) compared to PWoD (62.1; 95% CI 54.7, 69.5), as were hip fractures (12.3; 95% CI 9.1, 15.6 for PWD compared to 8.1; 95% CI 6.9, 9.2 in PWoD). Point estimates of widowhood incidence were slightly higher for PWD (25.3; 95% CI 20.1, 30.5) compared to PWoD (21.9; 95% CI 20.3, 23.5), but differences were not statistically significant. The association of dementia with hip fracture-but not pneumonia or widowhood-was modified by gender (male incidence rate ratio [IRR] 2.24, 95% CI 1.34, 3.75 versus female IRR 1.31 95% CI 0.92,1.86); interaction term p = 0.02). CONCLUSIONS: Compared to PWoD, community-dwelling PWD had higher rates of pneumonia and hip fracture, but not widowhood. Knowing how often PWD experience these events can aid in anticipatory guidance and care planning for this growing population.