Recent Publications by CFE Educators

BACKGROUND

Frailty is prevalent among older adults with asthma or chronic obstructive pulmonary disease (obstructive lung diseases [OLDs]). Frailty and OLD's co-occurrence is associated with increased hospitalization/mortality. Chemosensory dysfunction is closely connected to both OLD and frailty. We evaluated the utility of olfactory decline as a biomarker of frailty in the setting of OLD.

METHODS

We performed a prospective, longitudinal, nationally representative study of community-dwelling older US adults in the National Social Life, Health and Aging Project, an omnibus in-home survey. Respondents reported a physician's diagnosis of OLD. Decline in odor identification and sensitivity over 5 years and frailty (adapted fried frailty phenotype criteria) were measured using standard tools. Multivariate logistic regressions evaluated the association between OLD status, olfactory decline, and frailty.

RESULTS

We compared individuals with OLD (n = 98; mean age 71.2 years, 59.2% women) and those without OLD (n = 1036; mean age 69.5 years, 58.9% women). Olfactory identification decline was associated with developing frailty over the 5-year follow-up period in individuals with OLD (odds ratio [OR] = 9.1, 95% confidence interval [CI] = 2.1-38.6, p = 0.003). Olfactory decline predicted incidence of frailty in individuals with OLD (identification: OR = 4.8, 95% CI = 1.3-17.5, P = 0.018; sensitivity: OR = 6.1, 95%CI = 1.2-31.0, p = 0.030) but not in those without OLD adjusting for demographics, heavy alcohol use, current smoking, and comorbidity. Results were robust to different thresholds for olfactory decline and frailty development.

CONCLUSIONS

Older adults with OLD who experience olfactory decline face higher odds of developing frailty. Use of olfactory decline as a biomarker to identify frailty could allow earlier intervention and decrease adverse outcomes for high-risk older adults with OLD.

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, 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.