Dataset Information

Ascertaining Framingham heart failure phenotype from inpatient electronic health record data using natural language processing: a multicentre Atherosclerosis Risk in Communities (ARIC) validation study.

ABSTRACT:

Objectives

Using free-text clinical notes and reports from hospitalised patients, determine the performance of natural language processing (NLP) ascertainment of Framingham heart failure (HF) criteria and phenotype.

Study design

A retrospective observational study design of patients hospitalised in 2015 from four hospitals participating in the Atherosclerosis Risk in Communities (ARIC) study was used to determine NLP performance in the ascertainment of Framingham HF criteria and phenotype.

Setting

Four ARIC study hospitals, each representing an ARIC study region in the USA.

Participants

A stratified random sample of hospitalisations identified using a broad range of International Classification of Disease, ninth revision, diagnostic codes indicative of an HF event and occurring during 2015 was drawn for this study. A randomly selected set of 394 hospitalisations was used as the derivation dataset and 406 hospitalisations was used as the validation dataset.

Intervention

Use of NLP on free-text clinical notes and reports to ascertain Framingham HF criteria and phenotype.

Primary and secondary outcome measures

NLP performance as measured by sensitivity, specificity, positive-predictive value (PPV) and agreement in ascertainment of Framingham HF criteria and phenotype. Manual medical record review by trained ARIC abstractors was used as the reference standard.

Results

Overall, performance of NLP ascertainment of Framingham HF phenotype in the validation dataset was good, with 78.8%, 81.7%, 84.4% and 80.0% for sensitivity, specificity, PPV and agreement, respectively.

Conclusions

By decreasing the need for manual chart review, our results on the use of NLP to ascertain Framingham HF phenotype from free-text electronic health record data suggest that validated NLP technology holds the potential for significantly improving the feasibility and efficiency of conducting large-scale epidemiologic surveillance of HF prevalence and incidence.

SUBMITTER: Moore CR

PROVIDER: S-EPMC8204176 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:BackgroundAbstracting cerebrovascular disease (CeVD) from inpatient electronic medical records (EMRs) through natural language processing (NLP) is pivotal for automated disease surveillance and improving patient outcomes. Existing methods rely on coders' abstraction, which has time delays and under-coding issues. This study sought to develop an NLP-based method to detect CeVD using EMR clinical notes.MethodsCeVD status was confirmed through a chart review on randomly selected hospitalized patients who were 18 years or older and discharged from 3 hospitals in Calgary, Alberta, Canada, between January 1 and June 30, 2015. These patients' chart data were linked to administrative discharge abstract database (DAD) and Sunrise™ Clinical Manager (SCM) EMR database records by Personal Health Number (a unique lifetime identifier) and admission date. We trained multiple natural language processing (NLP) predictive models by combining two clinical concept extraction methods and two supervised machine learning (ML) methods: random forest and XGBoost. Using chart review as the reference standard, we compared the model performances with those of the commonly applied International Classification of Diseases (ICD-10-CA) codes, on the metrics of sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV).ResultOf the study sample (n = 3036), the prevalence of CeVD was 11.8% (n = 360); the median patient age was 63; and females accounted for 50.3% (n = 1528) based on chart data. Among 49 extracted clinical documents from the EMR, four document types were identified as the most influential text sources for identifying CeVD disease ("nursing transfer report," "discharge summary," "nursing notes," and "inpatient consultation."). The best performing NLP model was XGBoost, combining the Unified Medical Language System concepts extracted by cTAKES (e.g., top-ranked concepts, "Cerebrovascular accident" and "Transient ischemic attack"), and the term frequency-inverse document frequency vectorizer. Compared with ICD codes, the model achieved higher validity overall, such as sensitivity (25.0% vs 70.0%), specificity (99.3% vs 99.1%), PPV (82.6 vs. 87.8%), and NPV (90.8% vs 97.1%).ConclusionThe NLP algorithm developed in this study performed better than the ICD code algorithm in detecting CeVD. The NLP models could result in an automated EMR tool for identifying CeVD cases and be applied for future studies such as surveillance, and longitudinal studies.

Project description:The Atherosclerosis Risk in Communities (ARIC) Study, sponsored by the National Heart, Lung and Blood Institute (NHLBI), is a prospective epidemiologic study conducted in four U.S. communities. The four communities are Forsyth County, NC; Jackson, MS; the northwest suburbs of Minneapolis, MN; and Washington County, MD. ARIC is designed to investigate the etiology and natural history of atherosclerosis, the etiology of clinical atherosclerotic diseases, and variation in cardiovascular risk factors, medical care and disease by race, gender, location, and date. ARIC includes two parts: the Cohort Component and the Community Surveillance Component. The Cohort Component began in 1987, and each ARIC field center randomly selected and recruited a cohort sample of approximately 4,000 individuals aged 45-64 from a defined population in their community. A total of 15,792 participants received an extensive examination, including medical, social, and demographic data. These participants were reexamined every three years with the first screen (baseline) occurring in 1987-89, the second in 1990-92, the third in 1993-95, and the fourth and last exam was in 1996-98. Follow-up occurs yearly by telephone to maintain contact with participants and to assess health status of the cohort. In the Community Surveillance Component, currently ongoing, these four communities are investigated to determine the community-wide occurrence of hospitalized myocardial infarction and coronary heart disease deaths in men and women aged 35-84 years. Hospitalized stroke is investigated in cohort participants only. Starting in 2006, the study conducts community surveillance of inpatient (ages 55 years and older) and outpatient heart failure (ages 65 years and older) for heart failure events beginning in 2005. ARIC is currently funded through January 31, 2012. This study is part of the Gene Environment Association Studies initiative (GENEVA, <a href="http://www.genevastudy.org" target="_blank">http://www.genevastudy.org</a>) funded by the trans-NIH Genes, Environment, and Health Initiative (GEI). The overarching goal is to identify novel genetic factors that contribute to atherosclerosis and cardiovascular disease through large-scale genome-wide association studies of well-characterized cohorts of adults in four defined populations. Genotyping was performed at the Broad Institute of MIT and Harvard, a GENEVA genotyping center. Data cleaning and harmonization were done at the GEI-funded GENEVA Coordinating Center at the University of Washington.

Project description:BACKGROUND:Previous studies suggest that heart failure (HF) is an independent risk factor for cognitive decline. A better understanding of the relationship between HF, cognitive status, and cognitive decline in a community-based sample may help clinicians understand disease risk. OBJECTIVE:To examine whether persons with HF have a higher prevalence of cognitive impairment and whether persons developing HF have more rapid cognitive decline. DESIGN:This observational cohort study of American adults in the Atherosclerosis Risk in Communities (ARIC) study has two components: cross-sectional analysis examining the association between prevalent HF and cognition using multinomial logistic regression, and change over time analysis detailing the association between incident HF and change in cognition over 15 years. PARTICIPANTS:Among visit 5 (2011-2013) participants (median age 75 years), 6495 had neurocognitive information available for cross-sectional analysis. Change over time analysis examined the 5414 participants who had cognitive scores and no prevalent HF at visit 4 (1996-1998). MEASUREMENTS:The primary outcome was cognitive status, classified as normal, mild cognitive impairment [MCI], and dementia on the basis of standardized cognitive tests (delayed word recall, word fluency, and digit symbol substitution). Cognitive change was examined over a 15-year period. Control variables included socio-demographic, vascular, and smoking/drinking measures. RESULTS:At visit 5, participants with HF had a higher prevalence of dementia (adjusted relative risk ratio [RRR] = 1.60 [95% CI 1.13, 2.25]) and MCI (RRR = 1.36 [1.12, 1.64]) than those without HF. A decline in cognition between visits 4 and 5 was - 0.07 standard deviation units [- 0.13, - 0.01] greater among persons who developed HF compared to those who did not. Results did not differ by ejection fraction. CONCLUSION:HF is associated with neurocognitive dysfunction and decline independent of other co-morbid conditions. Further study is needed to determine the underlying pathophysiology.

Project description:Carotid intima-media thickness has been reported to predict kidney function decline. However, whether carotid intima-media thickness is associated with a hard kidney end point, ESRD, has not been investigated.We studied 13,197 Atherosclerosis Risk in Communities participants at visit 1 (1987-1989) without history of cardiovascular disease, including coronary heart disease, stroke, and heart failure, at baseline and assessed whether carotid intima-media thickness measured by B-mode ultrasound is associated with ESRD risk using Cox proportional hazards models. Regarding carotid intima-media thickness parameters, we investigated the mean and maximum values of overall and segment-specific (common, bifurcation, and internal carotid arteries) measurements.Mean age was 54.0 (SD=5.7) years old, and there were 3373 (25.6%) blacks and 7370 (55.8%) women. During a median follow-up of 22.7 years, 433 participants developed ESRD (1.4/1000 person-years). After adjusting for shared risk factors for atherosclerosis and kidney disease, including baseline kidney function, carotid intima-media thickness was significantly associated with ESRD risk (hazard ratio [HR] between quartiles 4 and 1, 1.46; 95% confidence interval [95% CI], 1.02 to 2.08 for overall mean intima-media thickness and HR between quartiles 4 and 1, 1.75; 95% CI, 1.24 to 2.48 for overall maximum intima-media thickness). The associations were largely consistent in demographic and clinical subgroups. When we explored segment-specific intima-media thicknesses, the associations with ESRD were most robust for bifurcation carotid (e.g., adjusted HR between quartiles 4 and 1 of mean intima-media thickness, 1.49; 95% CI, 1.04 to 2.13 for bifurcation; adjusted HR between quartiles 4 and 1 of mean intima-media thickness, 1.36; 95% CI, 0.94 to 1.97 for common; and adjusted HR between quartiles 4 and 1 of mean intima-media thickness, 0.93; 95% CI, 0.67 to 1.29 for internal).Carotid intima-media thickness was independently associated with incident ESRD in the general population, suggesting the shared etiology of atherosclerosis and ESRD.