Automated extraction of ejection fraction for quality measurement using regular expressions in Unstructured Information Management Architecture (UIMA) for heart failure

Computer-assisted patient identification tool in inborn errors of metabolism - potential for rare disease patient registry and big data analysis

BACKGROUND

Patient registries are crucial for rare disease management. However, manual registry construction is labor-intensive and often not user-friendly. Our goal is to establish Hong Kong's first computer-assisted patient identification tool for rare diseases, starting with inborn errors of metabolism (IEM).

METHODS

Patient data from 2010 to 2019 was retrieved from electronic databases. Through big data analytics, patient data were filtered based on specific IEM-related biochemical and genetic tests. Clinical notes were analyzed using a rule-based natural language processing technique called regular expression. The algorithm classified each extracted paragraph as "IEM-related" or "not IEM-related." Pathologists reviewed the paragraphs for curation, and the algorithm's performance was evaluated.

RESULTS

Out of 46,419 patients with IEM-related tests, the algorithm identified 100 as "IEM-related." After pathologists' validation, 96 cases were confirmed as true IEM, with 1 uncertain case and 3 false positives. A secondary ascertainment yielded a sensitivity of 92.3% compared to our previously published IEM cohort.

CONCLUSIONS

Our artificial intelligence approach provides a novel method to identify IEM patients, facilitating the creation of a centralized, computer-assisted rare disease patient registry at the local and national levels. This data can potentially be accessed by multiple stakeholders for collaborative research and to enhance healthcare management for rare diseases.

A general text mining method to extract echocardiography measurement results from echocardiography documents

BACKGROUND

In everyday medical practice, the results of cardiac ultrasound examinations are generally recorded in unstructured text, from which extracting relevant information is an important and challenging task. This paper presents a generally applicable language and corpus-independent text mining method for extracting and structuring numerical measurement results and their descriptions from echocardiography reports.

METHOD

The developed method is based on generally applicable text mining preprocessing activities, it automatically identifies and standardizes the descriptions of the cardiac ultrasound measures, and it stores the extracted and standardized measurement descriptions with their measurement results in a structured form for later usage. The method does not contain any regular expression-based search and does not rely on information about the structure of the document.

RESULTS

The method has been tested on a document set containing more than 20,000 echocardiographic reports by examining the efficiency of extracting 12 echocardiography parameters considered important by experts. The method extracted and structured the echocardiography parameters under the study with good sensitivity (lowest value: 0.775, highest value: 1.0, average: 0.904) and excellent specificity (for all cases 1.0). The F1 score ranged between 0.873 and 1.0, and its average value was 0.948.

CONCLUSION

The presented case study has shown that the proposed method can extract measurement results from echocardiography documents with high confidence without performing a direct search or having detailed information about the data recording habits. Furthermore, it effectively handles spelling errors, abbreviations and the highly varied terminology used in descriptions. As it does not rely on any information related to the structure or the language of the documents or data recording habits, it can be applied for processing any free-text written medical texts.

Developing a Parsimonious Frailty Index for Older, Multimorbid Adults With Heart Failure Using Machine Learning

Frailty is associated with adverse outcomes in heart failure (HF). A parsimonious frailty index (FI) that predicts outcomes of older, multimorbid patients with HF could be a useful resource for clinicians. A retrospective study of veterans hospitalized from October 2015 to October 2018 with HF, aged ≥50 years, and discharged home developed a 10-item parsimonious FI using machine learning from diagnostic codes, laboratory results, vital signs, and ejection fraction (EF) from outpatient encounters. An unsupervised clustering technique identified 5 FI strata: severely frail, moderately frail, mildly frail, prefrail, and robust. We report hazard ratios (HRs) of mortality, adjusting for age, gender, race, and EF and odds ratios (ORs) for 30-day and 1-year emergency department visits and all-cause hospitalizations after discharge. We identified 37,431 veterans (age, 73 ± 10 years; co-morbidity index, 5 ± 3; 43.5% with EF ≤40%). All frailty groups had a higher mortality than the robust group: severely frail (HR 2.63, 95% confidence interval [CI] 2.42 to 2.86), moderately frail (HR 2.04, 95% CI 1.87 to 2.22), mildly frail (HR 1.60, 95% CI 1.47 to 1.74), and prefrail (HR 1.18, 95% CI: 1.07 to 1.29). The associations between frailty and mortality remained unchanged in the stratified analysis by age or EF. The combined (severely, moderately, and mildly) frail group had higher odds of 30-day emergency visits (OR 1.62, 95% CI 1.43 to 1.83), all-cause readmission (OR, 1.75, 95% CI 1.52 to 2.02), 1-year emergency visits (OR 1.70, 95% CI 1.53 to 1.89), rehospitalization (OR 2.18, 95% CI 1.97 to 2.41) than the robust group. In conclusion, a 10-item FI is associated with postdischarge outcomes among patients discharged home after a hospitalization for HF. A parsimonious FI may aid clinical prediction at the point of care.

Validation of a Natural Language Processing Algorithm for the Extraction of the Sleep Parameters from the Polysomnography Reports

Background: There is a need to better understand the association between sleep and chronic diseases. In this study we developed a natural language processing (NLP) algorithm to mine polysomnography (PSG) free-text notes from electronic medical records (EMR) and evaluated the performance. Methods: Using the Veterans Health Administration EMR, we identified 46,093 PSG studies using CPT code 95,810 from 1 October 2000−30 September 2019. We randomly selected 200 notes to compare the accuracy of the NLP algorithm in mining sleep parameters including total sleep time (TST), sleep efficiency (SE) and sleep onset latency (SOL), wake after sleep onset (WASO), and apnea-hypopnea index (AHI) compared to visual inspection by raters masked to the NLP output. Results: The NLP performance on the training phase was >0.90 for precision, recall, and F-1 score for TST, SOL, SE, WASO, and AHI. The NLP performance on the test phase was >0.90 for precision, recall, and F-1 score for TST, SOL, SE, WASO, and AHI. Conclusions: This study showed that NLP is an accurate technique to extract sleep parameters from PSG reports in the EMR. Thus, NLP can serve as an effective tool in large health care systems to evaluate and improve patient care.

Racial Differences in Burdensome Transitions in Heart Failure Patients with Palliative Care: A Propensity-Matched Analysis

Background: Examining racial disparities in the treatment of heart failure (HF) patients and the effects of palliative care (PC) consultation is important to developing culturally competent clinical behaviors. Objective: To compare burdensome transitions for Black and White Veterans hospitalized with HF after PC consultation. Participants: This retrospective study evaluated Veterans admitted for HF to Veterans Administration hospitals who received PC consultation from October 2010 through August 2017. Methods: We propensity-matched Black to White Veterans using demographic, comorbidity, clinical, hospital, and survival time data. Results: Propensity matching of our cohort (n = 5638) yielded 796 Black and White Veterans (total n = 1592) who were well-matched on observed variables (standard mean difference <0.15 for all variables). Matched Black Veterans had more burdensome transitions than White Veterans (n = 218, 27.4% vs. n = 174, 21.9%; p = 0.011) over the six-month follow-up period. Conclusions: This propensity-matched cohort found racial differences in burdensome transitions among admitted HF patients after PC consultation.

Automatic extraction of 12 cardiovascular concepts from German discharge letters using pre-trained language models

Objective

A vast amount of medical data is still stored in unstructured text documents. We present an automated method of information extraction from German unstructured clinical routine data from the cardiology domain enabling their usage in state-of-the-art data-driven deep learning projects.

Methods

We evaluated pre-trained language models to extract a set of 12 cardiovascular concepts in German discharge letters. We compared three bidirectional encoder representations from transformers pre-trained on different corpora and fine-tuned them on the task of cardiovascular concept extraction using 204 discharge letters manually annotated by cardiologists at the University Hospital Heidelberg. We compared our results with traditional machine learning methods based on a long short-term memory network and a conditional random field.

Results

Our best performing model, based on publicly available German pre-trained bidirectional encoder representations from the transformer model, achieved a token-wise micro-average F1-score of 86% and outperformed the baseline by at least 6%. Moreover, this approach achieved the best trade-off between precision (positive predictive value) and recall (sensitivity).

Conclusion

Our results show the applicability of state-of-the-art deep learning methods using pre-trained language models for the task of cardiovascular concept extraction using limited training data. This minimizes annotation efforts, which are currently the bottleneck of any application of data-driven deep learning projects in the clinical domain for German and many other European languages.

Harnessing the Potential of Primary Care Pharmacists to Improve Heart Failure Management

BACKGROUND

Improved utilization of guideline-directed medical therapy (GDMT) in the management of heart failure with reduced ejection fraction (HFrEF) can reduce mortality, reduce heart failure hospitalizations, and improve quality of life. Despite well-established clinical guidelines, these therapies remain significantly underprescribed. The goal of this intervention was to increase prescribing of angiotensin-converting enzyme inhibitor (ACEI)/angiotensin II receptor blocker (ARB), angiotensin receptor neprilysin inhibitor (ARNI), and beta-blockers at ≥ 50% target doses.

METHODS

The study team identified key drivers to adequate dosing of GDMT: (1) frequent and reliable follow-up visits for titration opportunities, (2) identification of actionable patients for therapy initiation and titration, and (3) reduction in prescribing practice variability. The interventions were implemented at an outpatient clinical site and consisted of three main components: (1) establishing a pharmacist-led heart failure medication titration clinic, (2) creation of a standardized titration protocol, and (3) utilization of a patient dashboard to identify eligible patients.

RESULTS

For patients seen in the titration clinic, in 14 months, the mean dose per patient increased from 31.3% to 70.5% of target dose for ACEI/ARB/ARNI, and from 45.8% to 85.4% for beta-blockers. At this clinical site, the percentage of HFrEF patients receiving > 50% of targeted dose increased from 39.7% to 46.7% for ACEI/ARB/ARNI, and from 39.5% to 42.9% for beta-blockers. For ACEI/ARB/ARNI, use of target doses was 5.9% higher (95% confidence interval [CI] = 3.6%-8.3%, p < 0.0001) for the intervention site, 0.2% higher (95% CI = -2.2%-2.5%, p = 0.89) during the intervention period, and 10.4% higher (95% CI = 6.9%-13.9%, p < 0.0001) for the interaction (intervention site during the intervention time period). For beta-blockers, use of target doses was 1.0% higher (95% CI = -0.6%-2.6%, p = 0.20) for the intervention site, 0.8% lower (95% CI = -2.4%-0.8%, p = 0.29) for the intervention period, and 5.8% higher (95% CI = 3.5%-8.1%, p < 0.0001) for the interaction (intervention site during the intervention time period).

CONCLUSION

Through this project's interventions, the prescribing of ACEI/ARB/ARNI and beta-blocker therapy at ≥ 50% target doses for patients with HFrEF was increased. This study demonstrates the value of a multifaceted, team-based approach that integrates population-level interventions such as clinical dashboard management with a pharmacist-led heart failure medication titration clinic.

Racial/Ethnic Differences in 30-Day Mortality for Heart Failure and Pneumonia in the Veterans Health Administration Using Claims-based, Clinical, and Social Risk-adjustment Variables

BACKGROUND

Prior studies have identified lower mortality in Black Veterans compared with White Veterans after hospitalization for common medical conditions, but these studies adjusted for comorbid conditions identified in administrative claims.

OBJECTIVES

The objectives of this study were to compare mortality for non-Hispanic White (hereafter, "White"), non-Hispanic Black (hereafter, "Black"), and Hispanic Veterans hospitalized for heart failure (HF) and pneumonia and determine whether observed mortality differences varied according to whether claims-based comorbid conditions and/or clinical variables were included in risk-adjustment models.

RESEARCH DESIGN

This was an observational study.

SUBJECTS

The study cohort included 143,520 admissions for HF and 127,782 admissions for pneumonia for Veterans hospitalized in 132 Veterans Health Administration (VA) Medical Centers between January 2009 and September 2015.

MEASURES

The primary independent variable was racial/ethnic group (ie, Black, Hispanic, and non-Hispanic White), and the outcome was all-cause mortality 30 days following admission. To compare mortality by race/ethnicity, we used logistic regression models that included different combinations of claims-based, clinical, and sociodemographic variables. For each model, we estimated the average marginal effect (AME) for Black and Hispanic Veterans relative to White Veterans.

RESULTS

Among the 143,520 (127,782) hospitalizations for HF (pneumonia), the average patient age was 71.6 (70.9) years and 98.4% (97.1%) were male. The unadjusted 30-day mortality rates for HF (pneumonia) were 7.2% (11.0%) for White, 4.1% (10.4%) for Black and 8.4% (16.9%) for Hispanic Veterans. Relative to White Veterans, when only claims-based variables were used for risk adjustment, the AME (95% confidence interval) for the HF [pneumonia] cohort was -2.17 (-2.45, -1.89) [0.08 (-0.41, 0.58)] for Black Veterans and 1.32 (0.49, 2.15) [4.51 (3.65, 5.38)] for Hispanic Veterans. When clinical variables were incorporated in addition to claims-based ones, the AME, relative to White Veterans, for the HF [pneumonia] cohort was -1.57 (-1.88, -1.27) [-0.83 (-1.31, -0.36)] for Black Veterans and 1.50 (0.71, 2.30) [3.30 (2.49, 4.11)] for Hispanic Veterans.

CONCLUSIONS

Compared with White Veterans, Black Veterans had lower mortality, and Hispanic Veterans had higher mortality for HF and pneumonia. The inclusion of clinical variables into risk-adjustment models impacted the magnitude of racial/ethnic differences in mortality following hospitalization. Future studies examining racial/ethnic disparities should consider including clinical variables for risk adjustment.

Identification of Five Quality Needs for Rheumatology (Text Analysis and Literature Review)

Background: While the use of the term "quality" in industry relates to the basic idea of making processes measurable and standardizing processes, medicine focuses on achieving health goals that go far beyond the mere implementation of diagnostic and therapeutic processes. However, the quality management systems used are often simple, self-created concepts that concentrate on administrative processes without considering the quality of the results, which is essential for the patient. For several rheumatic diseases, both outcome and treatment goals have been defined. This work summarizes current mainstreams of strategies with published quality efforts in rheumatology. Methods: PubMed, Cochrane Library, and Web of Science were used to search for studies, and additional manual searches were carried out. Screening and content evaluation were carried out using the PRISMA-P 2015 checklist. After duplicate search in the Endnote reference management software (version X9.1), the software Rayyan QCRI (https://rayyan.qcri.org) was applied to check for pre-defined inclusion and exclusion criteria. Abstracts and full texts were screened and rated using Voyant Tools (https://voyant-tools.org/). Key issues were identified using the collocate analysis. Results: The number of selected publications was small but specific (14 relevant correlations with coefficients >0.8). Using trend analysis, 15 publications with relative frequency of keywords >0.0125 were used for content analysis, revealing 5 quality needs. The treat to target (T2T) initiative was identified as fundamental paradigm. Outcome parameters required for T2T also allow quality assessments in routine clinical work. Quality care by multidisciplinary teams also focusing on polypharmacy and other quality aspects become essential, A global software platform to assess quality aspects is missing. Such an approach requires reporting of multiple outcome parameters according to evidence-based clinical guidelines and recommendations for the different rheumatic diseases. All health aspects defined by the WHO (physical, mental, and social health) have to be integrated into the management of rheumatic patients. Conclusion: For the future, quality projects need goals defined by T2T based initiatives in routine clinical work, secondary quality goals include multidisciplinary cooperation and reduction of polypharmacy. Quality indicators and standards in different health systems will provide new information to optimize patients' care in different health systems.

Age at Diagnosis of Heart Failure in United States Veterans With and Without HIV Infection

Background

Although HIV is associated with increased risk of heart failure (HF), it is not known if people living with HIV develop HF at a younger age compared with individuals without HIV. Crude comparisons of age at diagnosis of HF between individuals with and without HIV does not account for differences in underlying age structures between the populations.

Methods and Results

We used Veterans Health Administration data to compare the age at HF diagnosis between veterans with and without HIV, with adjustment for difference in population age structure. Statistical weights, calculated for each 1‐year strata of veterans with HIV in each calendar year from 2000 to 2018, were applied to the veterans without HIV to standardize the age structure. We identified 5093 veterans with HIV (98% men, 34% White) with first HF episode recorded after HIV diagnosis (median age at incidence of HF, 58 years), and 1 425 987 veterans without HIV (98% men, 78% White) with HF (corresponding age, 72 years), with an absolute difference of 14 years. After accounting for difference in age structure, the adjusted median age at HF diagnosis for veterans without HIV was 63 years, 5 years difference with veterans with HIV (P<0.001). The age differences were consistent across important subgroups such as preserved versus reduced ejection fraction and inpatient versus outpatient index HF.

Conclusions

Veterans with HIV are diagnosed with HF at a significantly younger age compared with veterans without HIV. These findings may have implications for HF prevention in individuals with HIV. Future studies are needed to make the findings more generalizable.

Natural Language Processing and Machine Learning for Identifying Incident Stroke From Electronic Health Records: Algorithm Development and Validation

Background

Stroke is an important clinical outcome in cardiovascular research. However, the ascertainment of incident stroke is typically accomplished via time-consuming manual chart abstraction. Current phenotyping efforts using electronic health records for stroke focus on case ascertainment rather than incident disease, which requires knowledge of the temporal sequence of events.

Objective

The aim of this study was to develop a machine learning–based phenotyping algorithm for incident stroke ascertainment based on diagnosis codes, procedure codes, and clinical concepts extracted from clinical notes using natural language processing.

Methods

The algorithm was trained and validated using an existing epidemiology cohort consisting of 4914 patients with atrial fibrillation (AF) with manually curated incident stroke events. Various combinations of feature sets and machine learning classifiers were compared. Using a heuristic rule based on the composition of concepts and codes, we further detected the stroke subtype (ischemic stroke/transient ischemic attack or hemorrhagic stroke) of each identified stroke. The algorithm was further validated using a cohort (n=150) stratified sampled from a population in Olmsted County, Minnesota (N=74,314).

Results

Among the 4914 patients with AF, 740 had validated incident stroke events. The best-performing stroke phenotyping algorithm used clinical concepts, diagnosis codes, and procedure codes as features in a random forest classifier. Among patients with stroke codes in the general population sample, the best-performing model achieved a positive predictive value of 86% (43/50; 95% CI 0.74-0.93) and a negative predictive value of 96% (96/100). For subtype identification, we achieved an accuracy of 83% in the AF cohort and 80% in the general population sample.

Conclusions

We developed and validated a machine learning–based algorithm that performed well for identifying incident stroke and for determining type of stroke. The algorithm also performed well on a sample from a general population, further demonstrating its generalizability and potential for adoption by other institutions.

Application of optical character recognition with natural language processing for large-scale quality metric data extraction in colonoscopy reports

BACKGROUND AND AIMS

Colonoscopy is commonly performed for colorectal cancer screening in the United States. Reports are often generated in a non-standardized format and are not always integrated into electronic health records. Thus, this information is not readily available for streamlining quality management, participating in endoscopy registries, or reporting of patient- and center-specific risk factors predictive of outcomes. We aim to demonstrate the use of a new hybrid approach using natural language processing of charts that have been elucidated with optical character recognition processing (OCR/NLP hybrid) to obtain relevant clinical information from scanned colonoscopy and pathology reports, a technology co-developed by Cleveland Clinic and eHealth Technologies (West Henrietta, NY, USA).

METHODS

This was a retrospective study conducted at Cleveland Clinic, Cleveland, Ohio, and the University of Minnesota, Minneapolis, Minnesota. A randomly sampled list of outpatient screening colonoscopy procedures and pathology reports was selected. Desired variables were then collected. Two researchers first manually reviewed the reports for the desired variables. Then, the OCR/NLP algorithm was used to obtain the same variables from 3 electronic health records in use at our institution: Epic (Verona, Wisc, USA), ProVation (Minneapolis, Minn, USA) used for endoscopy reporting, and Sunquest PowerPath (Tucson, Ariz, USA) used for pathology reporting.

RESULTS

Compared with manual data extraction, the accuracy of the hybrid OCR/NLP approach to detect polyps was 95.8%, adenomas 98.5%, sessile serrated polyps 99.3%, advanced adenomas 98%, inadequate bowel preparation 98.4%, and failed cecal intubation 99%. Comparison of the dataset collected via NLP alone with that collected using the hybrid OCR/NLP approach showed that the accuracy for almost all variables was >99%.

CONCLUSIONS

Our study is the first to validate the use of a unique hybrid OCR/NLP technology to extract desired variables from scanned procedure and pathology reports contained in image format with an accuracy >95%.

Cohort selection for clinical trials using hierarchical neural network

OBJECTIVE

Cohort selection for clinical trials is a key step for clinical research. We proposed a hierarchical neural network to determine whether a patient satisfied selection criteria or not.

MATERIALS AND METHODS

We designed a hierarchical neural network (denoted as CNN-Highway-LSTM or LSTM-Highway-LSTM) for the track 1 of the national natural language processing (NLP) clinical challenge (n2c2) on cohort selection for clinical trials in 2018. The neural network is composed of 5 components: (1) sentence representation using convolutional neural network (CNN) or long short-term memory (LSTM) network; (2) a highway network to adjust information flow; (3) a self-attention neural network to reweight sentences; (4) document representation using LSTM, which takes sentence representations in chronological order as input; (5) a fully connected neural network to determine whether each criterion is met or not. We compared the proposed method with its variants, including the methods only using the first component to represent documents directly and the fully connected neural network for classification (denoted as CNN-only or LSTM-only) and the methods without using the highway network (denoted as CNN-LSTM or LSTM-LSTM). The performance of all methods was measured by micro-averaged precision, recall, and F1 score.

RESULTS

The micro-averaged F1 scores of CNN-only, LSTM-only, CNN-LSTM, LSTM-LSTM, CNN-Highway-LSTM, and LSTM-Highway-LSTM were 85.24%, 84.25%, 87.27%, 88.68%, 88.48%, and 90.21%, respectively. The highest micro-averaged F1 score is higher than our submitted 1 of 88.55%, which is 1 of the top-ranked results in the challenge. The results indicate that the proposed method is effective for cohort selection for clinical trials.

DISCUSSION

Although the proposed method achieved promising results, some mistakes were caused by word ambiguity, negation, number analysis and incomplete dictionary. Moreover, imbalanced data was another challenge that needs to be tackled in the future.

CONCLUSION

In this article, we proposed a hierarchical neural network for cohort selection. Experimental results show that this method is good at selecting cohort.

Underserved populations with missing race ethnicity data differ significantly from those with structured race/ethnicity documentation

OBJECTIVE

We aimed to address deficiencies in structured electronic health record (EHR) data for race and ethnicity by identifying black and Hispanic patients from unstructured clinical notes and assessing differences between patients with or without structured race/ethnicity data.

MATERIALS AND METHODS

Using EHR notes for 16 665 patients with encounters at a primary care practice, we developed rule-based natural language processing (NLP) algorithms to classify patients as black/Hispanic. We evaluated performance of the method against an annotated gold standard, compared race and ethnicity between NLP-derived and structured EHR data, and compared characteristics of patients identified as black or Hispanic using only NLP vs patients identified as such only in structured EHR data.

RESULTS

For the sample of 16 665 patients, NLP identified 948 additional patients as black, a 26%increase, and 665 additional patients as Hispanic, a 20% increase. Compared with the patients identified as black or Hispanic in structured EHR data, patients identified as black or Hispanic via NLP only were older, more likely to be male, less likely to have commercial insurance, and more likely to have higher comorbidity.

DISCUSSION

Structured EHR data for race and ethnicity are subject to data quality issues. Supplementing structured EHR race data with NLP-derived race and ethnicity may allow researchers to better assess the demographic makeup of populations and draw more accurate conclusions about intergroup differences in health outcomes.

CONCLUSIONS

Black or Hispanic patients who are not documented as such in structured EHR race/ethnicity fields differ significantly from those who are. Relatively simple NLP can help address this limitation.

Evaluation of Changes in Veterans Affairs Medical Centers' Mortality Rates After Risk Adjustment for Socioeconomic Status

IMPORTANCE

Socioeconomic factors are associated with worse outcomes after hospitalization, but neither the Centers for Medicare & Medicaid Services (CMS) nor the Veterans Affairs (VA) health care system adjust for socioeconomic factors in profiling hospital mortality.

OBJECTIVE

To evaluate changes in Veterans Affairs medical centers' (VAMCs') risk-standardized mortality rates among veterans hospitalized for heart failure and pneumonia after adjusting for socioeconomic factors.

DESIGN, SETTING, AND PARTICIPANTS

In this cross-sectional study, retrospective data were used to assess 131 VAMCs' risk-standardized 30-day mortality rates with or without adjustment for socioeconomic covariates. The study population included 42 892 veterans hospitalized with heart failure and 39 062 veterans hospitalized with pneumonia from January 1, 2012, to December 31, 2014. Data were analyzed from March 1, 2019, to April 1, 2020.

MAIN OUTCOMES AND MEASURES

The primary outcome was 30-day mortality after admission. Socioeconomic covariates included neighborhood disadvantage, race/ethnicity, homelessness, rurality, nursing home residence, reason for Medicare eligibility, Medicaid and Medicare dual eligibility, and VA priority.

RESULTS

The study population included 42 892 veterans hospitalized with heart failure (98.2% male; mean [SD] age, 71.9 [11.4] years) and 39 062 veterans hospitalized with pneumonia (96.8% male; mean [SD] age, 71.0 [12.4] years). The addition of socioeconomic factors to the CMS models modestly increased the C statistic from 0.77 (95% CI, 0.77-0.78) to 0.78 (95% CI, 0.78-0.78) for 30-day mortality after heart failure and from 0.73 (95% CI, 0.72-0.73) to 0.74 (95% CI, 0.73-0.74) for 30-day mortality after pneumonia. Mortality rates were highly correlated (Spearman correlations of ≥0.98) in models that included or did not include socioeconomic factors. With the use of the CMS model for heart failure, VAMCs in the lowest quintile had a mean (SD) mortality rate of 6.0% (0.4%), those in the middle 3 quintiles had a mean (SD) mortality rate of 7.2% (0.4%), and those in the highest quintile had a mean (SD) mortality rate of 8.8% (0.6%). After the inclusion of socioeconomic covariates, the adjusted mean (SD) mortality was 6.1% (0.4%) for hospitals in the lowest quintile, 7.2% (0.4%) for those in the middle 3 quintiles, and 8.6% (0.5%) for those in the highest quintile. The mean absolute change in rank after socioeconomic adjustment was 3.0 ranking positions (interquartile range, 1.0-4.0) among hospitals in the highest quintile of mortality after heart failure and 4.4 ranking positions (interquartile range, 1.0-6.0) among VAMCs in the lowest quintile. Similar findings were observed for mortality rankings in pneumonia and after inclusion of clinical covariates.

CONCLUSIONS AND RELEVANCE

This study suggests that adjustments for socioeconomic factors did not meaningfully change VAMCs' risk-adjusted 30-day mortality rates for veterans hospitalized for heart failure and pneumonia. The implications of such adjustments should be examined for other quality measures and health systems.

Trends in Left Ventricular Ejection Fraction for Patients With a New Diagnosis of Heart Failure

BACKGROUND

The left ventricular ejection fraction (LVEF) guides treatment of heart failure, yet this data has not been systematically collected in large data sets. We sought to characterize the epidemiology of incident heart failure using the initial LVEF.

METHODS

We identified 219 537 patients in the Veterans Affairs system between 2011 and 2017 who had an LVEF documented within 365 days before and 30 days after the heart failure diagnosis date. LVEF was obtained from natural language processing from imaging and provider notes. In multivariate analysis, we assessed characteristics associated with having an initial LVEF <40%.

RESULTS

Most patients were male and White; a plurality were within the 60 to 69 year age decile. A majority of patients had ischemic heart disease and a high burden of co-morbidities. Over time, presentation with an LVEF <40% became slightly less common, with a nadir in 2015. Presentation with an initial LVEF <40% was more common in younger patients, men, Black and Hispanic patients, an inpatient presentation, lower systolic blood pressure, lower pulse pressure, and higher heart rate. Ischemic heart disease, alcohol use disorder, peripheral arterial disease, and ventricular arrhythmias were associated with an initial LVEF <40%, while most other comorbid conditions (eg, atrial fibrillation, chronic obstructive pulmonary disease, malignancy) were more strongly associated with an initial LVEF >50%.

CONCLUSIONS

For patients with heart failure, particularly at the extremes of age, an initial preserved LVEF is common. In addition to clinical characteristics, certain races (Black and Hispanic) were more likely to present with a reduced LVEF. Further studies are needed to determine if racial differences are due to patient or health systems issues such as access to care.

How Do You Measure Up: Quality Measurement for Improving Patient Care and Establishing the Value of Infectious Diseases Specialists

The shift from volume-based to value-based reimbursement has created a need for quantifying clinical performance of infectious diseases (ID) physicians. Nationally recognized ID specialty-specific quality measures will allow stakeholders, such as patients and payers, to determine the value of care provided by ID physicians and will promote clinical quality improvement. Few ID-specific measures have been developed; herein, we provide an overview of the importance of quality measurement for ID, discuss issues in quality measurement specific to ID, and describe standards by which candidate quality measures can be evaluated. If ID specialists recognize the need for quality measurement, then ID specialists can direct ID-related quality improvement, quantify the impact of ID physicians on patient outcomes, compare their performance to that of peers, and convey to stakeholders the value of the specialty.

Association Between Heart Failure and Postoperative Mortality Among Patients Undergoing Ambulatory Noncardiac Surgery

Importance

Heart failure is an established risk factor for postoperative mortality, but how heart failure is associated with operative outcomes specifically in the ambulatory surgical setting is not well characterized.

Objective

To assess the risk of postoperative mortality and complications in patients with vs without heart failure at various levels of echocardiographic (left ventricular systolic dysfunction) and clinical (symptoms) severity who were undergoing ambulatory surgery.

Design, Setting, and Participants

In this US multisite retrospective cohort study of all adult patients undergoing ambulatory, elective, noncardiac surgery in the Veterans Affairs Surgical Quality Improvement Project database during fiscal years 2009 to 2016, a total of 355 121 patient records were identified and analyzed with 1 year of follow-up after surgery (final date of follow-up September 1, 2017).

Exposures

Heart failure, left ventricular ejection fraction, and presence of signs or symptoms of heart failure within 30 days of surgery.

Main Outcomes and Measures

The primary outcomes were postoperative mortality at 90 days and any postoperative complication at 30 days.

Results

Among 355 121 total patients, outcome data from 19 353 patients with heart failure (5.5%; mean [SD] age, 67.9 [10.1] years; 18 841 [96.9%] male) and 334 768 patients without heart failure (94.5%; mean [SD] age, 57.2 [14.0] years; 301 198 [90.0%] male) were analyzed. Compared with patients without heart failure, patients with heart failure had a higher risk of 90-day postoperative mortality (crude mortality risk, 2.00% vs 0.39%; adjusted odds ratio [aOR], 1.95; 95% CI, 1.69-2.44), and risk of mortality progressively increased with decreasing systolic function. Compared with patients without heart failure, symptomatic patients with heart failure had a greater risk of mortality (crude mortality risk, 3.57%; aOR, 2.76; 95% CI, 2.07-3.70), as did asymptomatic patients with heart failure (crude mortality risk, 1.85%; aOR, 1.85; 95% CI, 1.60-2.15). Patients with heart failure had a higher risk of experiencing a 30-day postoperative complication than did patients without heart failure (crude risk, 5.65% vs 2.65%; aOR, 1.10; 95% CI, 1.02-1.19).

Conclusions and Relevance

In this study, among patients undergoing elective, ambulatory surgery, heart failure with or without symptoms was significantly associated with 90-day mortality and 30-day postoperative complications. These data may be helpful in preoperative discussions with patients with heart failure undergoing ambulatory surgery.

Demystifying artificial intelligence in pharmacy

PURPOSE

To provide pharmacists and other clinicians with a basic understanding of the underlying principles and practical applications of artificial intelligence (AI) in the medication-use process.

SUMMARY

"Artificial intelligence" is a general term used to describe the theory and development of computer systems to perform tasks that normally would require human cognition, such as perception, language understanding, reasoning, learning, planning, and problem solving. Following the fundamental theorem of informatics, a better term for AI would be "augmented intelligence," or leveraging the strengths of computers and the strengths of clinicians together to obtain improved outcomes for patients. Understanding the vocabulary of and methods used in AI will help clinicians productively communicate with data scientists to collaborate on developing models that augment patient care. This primer includes discussion of approaches to identifying problems in practice that could benefit from application of AI and those that would not, as well as methods of training, validating, implementing, evaluating, and maintaining AI models. Some key limitations of AI related to the medication-use process are also discussed.

CONCLUSION

As medication-use domain experts, pharmacists play a key role in developing and evaluating AI in healthcare. An understanding of the core concepts of AI is necessary to engage in collaboration with data scientists and critically evaluating its place in patient care, especially as clinical practice continues to evolve and develop.

Heart Failure Outcomes and Associated Factors Among Veterans With Human Immunodeficiency Virus Infection

OBJECTIVES

This study sought to investigate outcomes of heart failure (HF) in veterans living with human immunodeficiency virus (HIV).

BACKGROUND

Data on outcomes of HF among people living with human immunodeficiency virus (PLHIV) are limited.

METHODS

We performed a retrospective cohort study of Veterans Health Affairs data to investigate outcomes of HF in PLHIV. We identified 5,747 HIV+ veterans with diagnosis of HF from 2000 to 2018 and 33,497 HIV- frequency-matched controls were included. Clinical outcomes included all-cause mortality, HF hospital admission, and all-cause hospital admission.

RESULTS

Compared with HIV- veterans with HF, HIV+ veterans with HF were more likely to be black (56% vs. 14%), be smokers (52% vs. 29%), use alcohol (32% vs. 13%) or drugs (37% vs. 8%), and have a higher comorbidity burden (Elixhauser comorbidity index 5.1 vs. 2.6). The mean ejection fraction (EF) (45 ± 16%) was comparable between HIV+ and HIV- veterans. HIV+ veterans with HF had a higher age-, sex-, and race-adjusted 1-year all-cause mortality (30.7% vs. 20.3%), HF hospital admission (21.2% vs. 18.0%), and all-cause admission (50.2% vs. 38.5%) rates. Among veterans with HIV and HF, those with low CD4 count (<200 cells/ml) and high HIV viral load (>75 copies/μl) had worse outcomes. The associations remained statistically significant after adjusting for extensive list of covariates. The incidence of all-cause mortality and HF admissions was higher among HIV+ veterans with ejection fraction <45% CONCLUSIONS: HIV+ veterans with HF had higher risk of hospitalization and mortality compared with their HIV- counterparts, with worse outcomes reported for individuals with lower CD4 count, higher viral load, and lower ejection fraction.

Evaluating the Portability of an NLP System for Processing Echocardiograms: A Retrospective, Multi-site Observational Study

While natural language processing (NLP) of unstructured clinical narratives holds the potential for patient care and clinical research, portability of NLP approaches across multiple sites remains a major challenge. This study investigated the portability of an NLP system developed initially at the Department of Veterans Affairs (VA) to extract 27 key cardiac concepts from free-text or semi-structured echocardiograms from three academic edical centers: Weill Cornell Medicine, Mayo Clinic and Northwestern Medicine. While the NLP system showed high precision and recall easurements for four target concepts (aortic valve regurgitation, left atrium size at end systole, mitral valve regurgitation, tricuspid valve regurgitation) across all sites, we found moderate or poor results for the remaining concepts and the NLP system performance varied between individual sites.

Mortality Trends for Veterans Hospitalized With Heart Failure and Pneumonia Using Claims-Based vs Clinical Risk-Adjustment Variables

IMPORTANCE

Prior studies have reported declines in mortality for patients admitted to Veterans Health Administration (VA) and non-VA hospitals using claims-based risk adjustment. These apparent mortality reductions may be influenced by changes in coding practices.

OBJECTIVE

To compare trends in the VA for 30-day mortality following hospitalization for heart failure (HF) and pneumonia using claims-based and clinical risk-adjustment models.

DESIGN, SETTING, AND PARTICIPANTS

This observational time-trend study analyzed admissions to a VA Medical Center with a principal diagnosis of HF, pneumonia, or sepsis/respiratory failure (RF) with a secondary diagnosis of pneumonia. Exclusion criteria included having less than 12 months of VA enrollment, being discharged alive within 24 hours, leaving against medical advice, and hospice utilization.

EXPOSURES

Admission to a VA hospital from January 2009 through September 2015.

MAIN OUTCOMES AND MEASURES

The primary outcome was 30-day, all-cause mortality. All models included age and sex. Claims-based covariates included 22 (30) comorbidities for HF (pneumonia). Clinical covariates included vital signs, laboratory values, and ejection fraction.

RESULTS

Among the 146 924 HF admissions, the mean (SD) age was 71.6 (11.4) years and 144 502 (98.4%) were men; among the 131 325 admissions for pneumonia, the mean (SD) age was 70.8 (12.3) years and 127 491 (97.1%) were men. Unadjusted 30-day mortality rates were 6.45% (HF) and 11.22% (pneumonia). Claims-based models showed an increased predicted risk of 30-day mortality over time (0.019 percentage points per quarter for HF [95% CI, 0.015 to 0.023]; 0.053 percentage points per quarter for pneumonia [95% CI, 0.043 to 0.063]). Clinical models showed declines or no change in predicted risk (-0.014 percentage points per quarter for HF [95% CI, -0.020 to -0.008]; -0.004 percentage points per quarter for pneumonia [95% CI, -0.017 to 0.008]). Claims-based risk adjustment yielded declines in 30-day mortality of 0.051 percentage points per quarter for HF (95% CI, -0.074 to -0.027) and 0.084 percentage points per quarter for pneumonia (95% CI, -0.111 to -0.056). Models adjusting for clinical covariates attenuated or eliminated these changes for HF (-0.017 percentage points per quarter; 95% CI, -0.039 to 0.006) and for pneumonia (-0.026 percentage points per quarter; 95% CI, -0.052 to 0.001). Compared with the claims-based models, the clinical models for HF and pneumonia more accurately differentiated between patients who died after 30 days and those who did not.

CONCLUSIONS AND RELEVANCE

Among HF and pneumonia hospitalizations, adjusting for clinical covariates attenuated declines in mortality rates identified using claims-based models. Assessments of temporal trends in 30-day mortality using claims-based risk adjustment should be interpreted with caution.

Determining post-test risk in a national sample of stress nuclear myocardial perfusion imaging reports: Implications for natural language processing tools

BACKGROUND

Reporting standards promote clarity and consistency of stress myocardial perfusion imaging (MPI) reports, but do not require an assessment of post-test risk. Natural Language Processing (NLP) tools could potentially help estimate this risk, yet it is unknown whether reports contain adequate descriptive data to use NLP.

METHODS

Among VA patients who underwent stress MPI and coronary angiography between January 1, 2009 and December 31, 2011, 99 stress test reports were randomly selected for analysis. Two reviewers independently categorized each report for the presence of critical data elements essential to describing post-test ischemic risk.

RESULTS

Few stress MPI reports provided a formal assessment of post-test risk within the impression section (3%) or the entire document (4%). In most cases, risk was determinable by combining critical data elements (74% impression, 98% whole). If ischemic risk was not determinable (25% impression, 2% whole), inadequate description of systolic function (9% impression, 1% whole) and inadequate description of ischemia (5% impression, 1% whole) were most commonly implicated.

CONCLUSIONS

Post-test ischemic risk was determinable but rarely reported in this sample of stress MPI reports. This supports the potential use of NLP to help clarify risk. Further study of NLP in this context is needed.

EXTraction of EMR numerical data: an efficient and generalizable tool to EXTEND clinical research

Background

Electronic medical records (EMR) contain numerical data important for clinical outcomes research, such as vital signs and cardiac ejection fractions (EF), which tend to be embedded in narrative clinical notes. In current practice, this data is often manually extracted for use in research studies. However, due to the large volume of notes in datasets, manually extracting numerical data often becomes infeasible. The objective of this study is to develop and validate a natural language processing (NLP) tool that can efficiently extract numerical clinical data from narrative notes.

Results

To validate the accuracy of the tool EXTraction of EMR Numerical Data (EXTEND), we developed a reference standard by manually extracting vital signs from 285 notes, EF values from 300 notes, glycated hemoglobin (HbA1C), and serum creatinine from 890 notes. For each parameter of interest, we calculated the sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and F₁ score of EXTEND using two metrics.

(1) completion of data extraction, and (2) accuracy of data extraction compared to the actual values in the note verified by chart review. At the note level, extraction by EXTEND was considered correct only if it accurately detected and extracted all values of interest in a note.

Using manually-annotated labels as the gold standard, the note-level accuracy of EXTEND in capturing the numerical vital sign values, EF, HbA1C and creatinine ranged from 0.88 to 0.95 for sensitivity, 0.95 to 1.0 for specificity, 0.95 to 1.0 for PPV, 0.89 to 0.99 for NPV, and 0.92 to 0.96 in F₁ scores. Compared to the actual value level, the sensitivity, PPV, and F₁ score of EXTEND ranged from 0.91 to 0.95, 0.95 to 1.0 and 0.95 to 0.96.

Conclusions

EXTEND is an efficient, flexible tool that uses knowledge-based rules to extract clinical numerical parameters with high accuracy. By increasing dictionary terms and developing new rules, the usage of EXTEND can easily be expanded to extract additional numerical data important in clinical outcomes research.

Association of Sex With Postoperative Mortality Among Patients With Heart Failure Who Underwent Elective Noncardiac Operations

IMPORTANCE

Sex differences in postoperative outcomes in patients with heart failure (HF) have not been well characterized. Women generally experience a lower postoperative mortality risk after noncardiac operations. It is unclear if this pattern holds among patients with HF.

OBJECTIVES

To determine if the risk of postoperative mortality is associated with sex among patients with HF who underwent noncardiac operations and to determine if sex is associated with the relationship between HF and postoperative mortality.

DESIGN, SETTING, AND PARTICIPANTS

This multisite cohort study used data from the US Department of Veterans Affairs Surgical Quality Improvement Project database for all patients who underwent elective noncardiac operations from October 1, 2009, to September 30, 2016, with a minimum of 1 year follow-up. The data analysis was conducted from May 1, 2018, to August 31, 2018.

EXPOSURES

Heart failure, left ventricular ejection fraction, and sex.

MAIN OUTCOMES AND MEASURES

Postoperative mortality at 90 days.

RESULTS

Among 609 735 patients who underwent elective noncardiac operations from 2009 to 2016, 47 997 patients had HF (7.9%; mean [SD] age, 68.6 [10.1] years; 1391 [2.9%] women) and 561 738 patients did not have HF (92.1%; mean [SD] age, 59.4 [13.4] years; 50 862 [9.1%] women). Among patients with HF, female sex was not independently associated with 90-day postoperative mortality (adjusted odds ratio [aOR], 0.97; 95% CI, 0.71-1.32). Although HF was associated with increased odds of postoperative mortality in both sexes compared with their peers without HF, the odds of postoperative mortality were higher among women with HF (aOR, 2.44; 95% CI, 1.73-3.45) than men with HF (aOR, 1.64; 95% CI, 1.54-1.74), suggesting that HF may negate the general protective association of female sex with postoperative mortality (P for interaction of HF × sex = .03). This pattern was consistent across all levels of left ventricular ejection fraction.

CONCLUSIONS AND RELEVANCE

Although HF was associated with increased odds of postoperative mortality in both sexes compared with their peers without HF, the odds of postoperative mortality were higher among women with HF than men with HF, suggesting that HF may negate the general protective association of female sex with postoperative mortality risk in noncardiac operations.

Comparison of Accessibility, Cost, and Quality of Elective Coronary Revascularization Between Veterans Affairs and Community Care Hospitals

Question

Does the Veterans Affairs Community Care Program, which allows veterans to receive care at non–Veterans Affairs sites, increase the accessibility and value of their elective coronary revascularization procedures?

Findings

Among 13 237 elective percutaneous coronary interventions and 5818 elective coronary artery bypass graft procedures in this veteran cohort study, use of the Community Care Program reduced aggregate veteran travel distance for revascularization. Community Care Program hospitals had higher mortality and costs for percutaneous coronary intervention and had equivalent mortality and lower costs for coronary artery bypass graft surgery.

Meaning

In our veteran cohort, use of Community Care Program hospitals improved overall access for revascularization; Community Care Program hospitals provided lower-value percutaneous coronary intervention procedures but higher-value coronary artery bypass graft procedures.

Importance

The Veterans Affairs (VA) Community Care (CC) Program supplements VA care with community-based medical services. However, access gains and value provided by CC have not been well described.

Objectives

To compare the access, cost, and quality of elective coronary revascularization procedures between VA and CC hospitals and to evaluate if procedural volume or publicly reported quality data can be used to identify high-value care.

Design, Setting, and Participants

Observational cohort study of veterans younger than 65 years undergoing an elective coronary revascularization, controlling for differences in risk factors using propensity adjustment. The setting was VA and CC hospitals. Participants were veterans undergoing elective percutaneous coronary intervention (PCI) and veterans undergoing coronary artery bypass graft (CABG) procedures between October 1, 2008, and September 30, 2011. The analysis was conducted between July 2014 and July 2017.

Exposures

Receipt of an elective coronary revascularization at a VA vs CC facility.

Main Outcomes and Measures

Access to care as measured by travel distance, 30-day mortality, and costs.

Results

In the 3 years ending on September 30, 2011, a total of 13 237 elective PCIs (79.1% at the VA) and 5818 elective CABG procedures (83.6% at the VA) were performed in VA or CC hospitals among veterans meeting study inclusion criteria. On average, use of CC was associated with reduced net travel by 53.6 miles for PCI and by 73.3 miles for CABG surgery compared with VA-only care. Adjusted 30-day mortality after PCI was higher in CC compared with VA (1.54% for CC vs 0.65% for VA, P < .001) but was similar after CABG surgery (1.33% for CC vs 1.51% for VA, P = .74). There were no differences in adjusted 30-day readmission rates for PCI (7.04% for CC vs 7.73% for VA, P = .66) or CABG surgery (8.13% for CC vs 7.00% for VA, P = .28). The mean adjusted PCI cost was higher in CC ($22 025 for CC vs $15 683 for VA, P < .001). The mean adjusted CABG cost was lower in CC ($55 526 for CC vs $63 144 for VA, P < .01). Neither procedural volume nor publicly reported mortality data identified hospitals that provided higher-value care with the exception that CABG mortality was lower in small-volume CC hospitals.

Conclusions and Relevance

In this veteran cohort, PCIs performed in CC hospitals were associated with shorter travel distance but with higher mortality, higher costs, and minimal travel savings compared with VA hospitals. The CABG procedures performed in CC hospitals were associated with shorter travel distance, similar mortality, and lower costs. As the VA considers expansion of the CC program, ongoing assessments of value and access gains are essential to optimize veteran outcomes and VA spending.

Enhancing ontology-driven diagnostic reasoning with a symptom-dependency-aware Naïve Bayes classifier

Background

Ontology has attracted substantial attention from both academia and industry. Handling uncertainty reasoning is important in researching ontology. For example, when a patient is suffering from cirrhosis, the appearance of abdominal vein varices is four times more likely than the presence of bitter taste. Such medical knowledge is crucial for decision-making in various medical applications but is missing from existing medical ontologies. In this paper, we aim to discover medical knowledge probabilities from electronic medical record (EMR) texts to enrich ontologies. First, we build an ontology by identifying meaningful entity mentions from EMRs. Then, we propose a symptom-dependency-aware naïve Bayes classifier (SDNB) that is based on the assumption that there is a level of dependency among symptoms. To ensure the accuracy of the diagnostic classification, we incorporate the probability of a disease into the ontology via innovative approaches.

Results

We conduct a series of experiments to evaluate whether the proposed method can discover meaningful and accurate probabilities for medical knowledge. Based on over 30,000 deidentified medical records, we explore 336 abdominal diseases and 81 related symptoms. Among these 336 gastrointestinal diseases, the probabilities of 31 diseases are obtained via our method. These 31 probabilities of diseases and 189 conditional probabilities between diseases and the symptoms are added into the generated ontology.

Conclusion

In this paper, we propose a medical knowledge probability discovery method that is based on the analysis and extraction of EMR text data for enriching a medical ontology with probability information. The experimental results demonstrate that the proposed method can effectively identify accurate medical knowledge probability information from EMR data. In addition, the proposed method can efficiently and accurately calculate the probability of a patient suffering from a specified disease, thereby demonstrating the advantage of combining an ontology and a symptom-dependency-aware naïve Bayes classifier.

Association Between HIV Infection and the Risk of Heart Failure With Reduced Ejection Fraction and Preserved Ejection Fraction in the Antiretroviral Therapy Era: Results From the Veterans Aging Cohort Study

Importance

With improved survival, heart failure (HF) has become a major complication for individuals with human immunodeficiency virus (HIV) infection. It is unclear if this risk extends to different types of HF in the antiretroviral therapy (ART) era. Determining whether HIV infection is associated with HF with reduced ejection fraction (HFrEF), HF with preserved ejection fraction (HFpEF), or both is critical because HF types differ with respect to underlying mechanism, treatment, and prognosis.

Objectives

To investigate whether HIV infection increases the risk of future HFrEF and HFpEF and to assess if this risk varies by sociodemographic and HIV-specific factors.

Design, Setting, and Participants

This study evaluated 98 015 participants without baseline cardiovascular disease from the Veterans Aging Cohort Study, an observational cohort of HIV-infected veterans and uninfected veterans matched by age, sex, race/ethnicity, and clinical site, enrolled on or after April 1, 2003, and followed up through September 30, 2012. The dates of the analysis were October 2015 to November 2016.

Exposure

Human immunodeficiency virus infection.

Main Outcomes and Measures

Outcomes included HFpEF (EF≥50%), borderline HFpEF (EF 40%-49%), HFrEF (EF<40%), and HF of unknown type (EF missing).

Results

Among 98 015 participants, the mean (SD) age at enrollment in the study was 48.3 (9.8) years, 97.0% were male, and 32.2% had HIV infection. During a median follow-up of 7.1 years, there were 2636 total HF events (34.6% were HFpEF, 15.5% were borderline HFpEF, 37.1% were HFrEF, and 12.8% were HF of unknown type). Compared with uninfected veterans, HIV-infected veterans had an increased risk of HFpEF (hazard ratio [HR], 1.21; 95% CI, 1.03-1.41), borderline HFpEF (HR, 1.37; 95% CI, 1.09-1.72), and HFrEF (HR, 1.61; 95% CI, 1.40-1.86). The risk of HFrEF was pronounced in veterans younger than 40 years at baseline (HR, 3.59; 95% CI, 1.95-6.58). Among HIV-infected veterans, time-updated HIV-1 RNA viral load of at least 500 copies/mL compared with less than 500 copies/mL was associated with an increased risk of HFrEF, and time-updated CD4 cell count less than 200 cells/mm3 compared with at least 500 cells/mm3 was associated with an increased risk of HFrEF and HFpEF.

Conclusions and Relevance

Individuals who are infected with HIV have an increased risk of HFpEF, borderline HFpEF, and HFrEF compared with uninfected individuals. The increased risk of HFrEF can manifest decades earlier than would be expected in a typical uninfected population. Future research should focus on prevention, risk stratification, and identification of the mechanisms for HFrEF and HFpEF in the HIV-infected population.

Natural Language Processing of Clinical Notes on Chronic Diseases: Systematic Review

Background

Novel approaches that complement and go beyond evidence-based medicine are required in the domain of chronic diseases, given the growing incidence of such conditions on the worldwide population. A promising avenue is the secondary use of electronic health records (EHRs), where patient data are analyzed to conduct clinical and translational research. Methods based on machine learning to process EHRs are resulting in improved understanding of patient clinical trajectories and chronic disease risk prediction, creating a unique opportunity to derive previously unknown clinical insights. However, a wealth of clinical histories remains locked behind clinical narratives in free-form text. Consequently, unlocking the full potential of EHR data is contingent on the development of natural language processing (NLP) methods to automatically transform clinical text into structured clinical data that can guide clinical decisions and potentially delay or prevent disease onset.

Objective

The goal of the research was to provide a comprehensive overview of the development and uptake of NLP methods applied to free-text clinical notes related to chronic diseases, including the investigation of challenges faced by NLP methodologies in understanding clinical narratives.

Methods

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed and searches were conducted in 5 databases using “clinical notes,” “natural language processing,” and “chronic disease” and their variations as keywords to maximize coverage of the articles.

Results

Of the 2652 articles considered, 106 met the inclusion criteria. Review of the included papers resulted in identification of 43 chronic diseases, which were then further classified into 10 disease categories using the International Classification of Diseases, 10th Revision. The majority of studies focused on diseases of the circulatory system (n=38) while endocrine and metabolic diseases were fewest (n=14). This was due to the structure of clinical records related to metabolic diseases, which typically contain much more structured data, compared with medical records for diseases of the circulatory system, which focus more on unstructured data and consequently have seen a stronger focus of NLP. The review has shown that there is a significant increase in the use of machine learning methods compared to rule-based approaches; however, deep learning methods remain emergent (n=3). Consequently, the majority of works focus on classification of disease phenotype with only a handful of papers addressing extraction of comorbidities from the free text or integration of clinical notes with structured data. There is a notable use of relatively simple methods, such as shallow classifiers (or combination with rule-based methods), due to the interpretability of predictions, which still represents a significant issue for more complex methods. Finally, scarcity of publicly available data may also have contributed to insufficient development of more advanced methods, such as extraction of word embeddings from clinical notes.

Conclusions

Efforts are still required to improve (1) progression of clinical NLP methods from extraction toward understanding; (2) recognition of relations among entities rather than entities in isolation; (3) temporal extraction to understand past, current, and future clinical events; (4) exploitation of alternative sources of clinical knowledge; and (5) availability of large-scale, de-identified clinical corpora.

Association of Left Ventricular Ejection Fraction and Symptoms With Mortality After Elective Noncardiac Surgery Among Patients With Heart Failure

IMPORTANCE

Heart failure is an established risk factor for postoperative mortality, but how left ventricular ejection fraction and heart failure symptoms affect surgical outcomes is not fully described.

OBJECTIVES

To determine the risk of postoperative mortality among patients with heart failure at various levels of echocardiographic (left ventricular systolic dysfunction) and clinical (symptoms) severity compared with those without heart failure and to evaluate how risk varies across levels of surgical complexity.

DESIGN, SETTING, AND PARTICIPANTS

US multisite retrospective cohort study of all adult patients receiving elective, noncardiac surgery in the Veterans Affairs Surgical Quality Improvement Project database from 2009 through 2016. A total of 609 735 patient records were identified and analyzed with 1 year of follow-up after having surgery (final study follow-up: September 1, 2017).

EXPOSURES

Heart failure, left ventricular ejection fraction, and presence of signs or symptoms of heart failure within 30 days of surgery.

MAIN OUTCOME AND MEASURE

The primary outcome was postoperative mortality at 90 days.

RESULTS

Outcome data from 47 997 patients with heart failure (7.9%; mean [SD] age, 68.6 [10.1] years; 1391 women [2.9%]) and 561 738 patients without heart failure (92.1%; mean [SD] age, 59.4 [13.4] years; 50 862 women [9.1%]) were analyzed. Compared with patients without heart failure, those with heart failure had a higher risk of 90-day postoperative mortality (2635 vs 6881 90-day deaths; crude mortality risk, 5.49% vs 1.22%; adjusted absolute risk difference [RD], 1.03% [95% CI, 0.91%-1.15%]; adjusted odds ratio [OR], 1.67 [95% CI, 1.57-1.76]). Compared with patients without heart failure, symptomatic patients with heart failure (n = 5906) had a higher risk (597 deaths [10.11%]; adjusted absolute RD, 2.37% [95% CI, 2.06%-2.57%]; adjusted OR, 2.37 [95% CI, 2.14-2.63]). Asymptomatic patients with heart failure (n = 42 091) (2038 deaths [crude risk, 4.84%]; adjusted absolute RD, 0.74% [95% CI, 0.63%-0.87%]; adjusted OR, 1.53 [95% CI, 1.44-1.63]), including the subset with preserved left ventricular systolic function (1144 deaths [4.42%]; adjusted absolute RD, 0.66% [95% CI, 0.54%-0.79%]; adjusted OR, 1.46 [95% CI, 1.35-1.57]), also experienced elevated risk.

CONCLUSIONS AND RELEVANCE

Among patients undergoing elective noncardiac surgery, heart failure with or without symptoms was significantly associated with 90-day postoperative mortality. These data may be helpful in preoperative discussions with patients with heart failure undergoing noncardiac surgery.

Accuracy of claim data in the identification and classification of adults with congenital heart diseases in electronic medical records

BACKGROUND

The content of electronic medical records (EMRs) encompasses both structured data, such as billing codes, and unstructured data, including free-text reports. Epidemiological and clinical research into adult congenital heart disease (ACHD) increasingly relies on administrative claim data using the International Classification of Diseases (9th revision) (ICD-9). In France, administrative databases use ICD-10, the reliability of which is largely unknown in this context.

AIMS

To assess the accuracy of ICD-10 codes retrieved from administrative claim data in the identification and classification of ACHD.

METHODS

We randomly included 6000 patients hospitalized at least once in 2000-2014 in a cardiology department with a dedicated specialized ACHD Unit. For each patient, the clinical diagnosis extracted from the EMR was compared with the assigned ICD-10 codes. Performance of ICD-10 codes in the identification and classification of ACHD was assessed by estimating sensitivity, specificity and positive predictive value.

RESULTS

Among the 6000 patients included, 780 (13%) patients with ACHD were manually identified from EMRs (107,092 documents). ICD-10 codes correctly categorized 629 as having ACHD (sensitivity 0.81, 95% confidence interval 0.78-0.83), with a specificity of 0.99 (95% confidence interval 0.99-1). The performance of ICD-10 codes in correctly categorizing the ACHD defect subtype depended on the defect, with sensitivity ranging from 0 (e.g. unspecified congenital malformation of tricuspid valve) to 1 (e.g. common arterial trunk), and specificity ranging from 0.99 to 1.

CONCLUSIONS

Administrative data using ICD-10 codes is a precise tool for detecting ACHD, and may be used to establish a national cohort. Mining free-text reports in addition to coded administrative data may offset the lack of sensitivity and accuracy when describing the spectrum of congenital heart disease using ICD-10 codes.

Identifying Patients With Relapsing-Remitting Multiple Sclerosis Using Algorithms Applied to US Integrated Delivery Network Healthcare Data

BACKGROUND

Relapsing-remitting multiple sclerosis (RRMS) has a major impact on affected patients; therefore, improved understanding of RRMS is important, particularly in the context of real-world evidence.

OBJECTIVES

To develop and validate algorithms for identifying patients with RRMS in both unstructured clinical notes found in electronic health records (EHRs) and structured/coded health care claims data.

METHODS

US Integrated Delivery Network data (2010-2014) were queried for study inclusion criteria (possible multiple sclerosis [MS] base cohort): one or more MS diagnosis code, patients aged 18 years or older, 1 year or more baseline history, and no other demyelinating diseases. Sets of algorithms were developed to search narrative text of unstructured clinical notes (EHR clinical notes-based algorithms) and structured/coded data (claims-based algorithms) to identify adult patients with RRMS, excluding patients with evidence of progressive MS. Medical records were reviewed manually for algorithm validation. Positive predictive value was calculated for both EHR clinical notes-based and claims-based algorithms.

RESULTS

From a sample of 5308 patients with possible MS, 837 patients with RRMS were identified using only the EHR clinical notes-based algorithms and 2271 patients were identified using only the claims-based algorithms; 779 patients were identified using both algorithms. The positive predictive value was 99.1% (95% confidence interval [CI], 94.2%-100%) for the EHR clinical notes-based algorithms and 94.6% (95% CI, 89.1%-97.8%) to 94.9% (95% CI, 89.8%-97.9%) for the claims-based algorithms.

CONCLUSIONS

The algorithms evaluated in this study identified a real-world cohort of patients with RRMS without evidence of progressive MS that can be studied in clinical research with confidence.

Ascertaining Depression Severity by Extracting Patient Health Questionnaire-9 (PHQ-9) Scores from Clinical Notes

The Patient Health Questionnaire-9 (PHQ-9) is a validated instrument for assessing depression severity. While some electronic health record (EHR) systems capture PHQ-9 scores in a structured format, unstructured clinical notes remain the only source in many settings, which presents data retrieval challenges for research and clinical decision support. To address this gap, we extended the open-source Leo natural language processing (NLP) platform to extract PHQ-9 scores from clinical notes and evaluated performance using EHR data for n=123,703 patients who were prescribed antidepressants. Compared to a reference standard, the NLP method exhibited high accuracy (97%), sensitivity (98%), precision (97%), and F-score (97%). Furthermore, of patients with PHQ-9 scores identified by the NLP method, 31% (n=498) had at least one PHQ-9 score clinically indicative of major depressive disorder (MDD), but lacked a structured ICD-9/10 diagnosis code for MDD. This NLP technique may facilitate accurate identification and stratification of patients with depression.

Extraction of Ejection Fraction from Echocardiography Notes for Constructing a Cohort of Patients having Heart Failure with reduced Ejection Fraction (HFrEF)

Left ventricular ejection fraction (LVEF) is an important prognostic indicator of cardiovascular outcomes. It is used clinically to determine the indication for several therapeutic interventions. LVEF is most commonly derived using in-line tools and some manual assessment by cardiologists from standardized echocardiographic views. LVEF is typically documented in free-text reports, and variation in LVEF documentation pose a challenge for the extraction and utilization of LVEF in computer-based clinical workflows. To address this problem, we developed a computerized algorithm to extract LVEF from echocardiography reports for the identification of patients having heart failure with reduced ejection fraction (HFrEF) for therapeutic intervention at a large healthcare system. We processed echocardiogram reports for 57,158 patients with coded diagnosis of Heart Failure that visited the healthcare system over a two-year period. Our algorithm identified a total of 3910 patients with reduced ejection fraction. Of the 46,634 echocardiography reports processed, 97% included a mention of LVEF. Of these reports, 85% contained numerical ejection fraction values, 9% contained ranges, and the remaining 6% contained qualitative descriptions. Overall, 18% of extracted numerical LVEFs were ≤ 40%. Furthermore, manual validation for a sample of 339 reports yielded an accuracy of 1.0. Our study demonstrates that a regular expression-based approach can accurately extract LVEF from echocardiograms, and is useful for delineating heart-failure patients with reduced ejection fraction.

Incorporating Guideline Adherence and Practice Implementation Issues into the Design of Decision Support for Beta-Blocker Titration for Heart Failure

BACKGROUND

The recognition of and response to undertreatment of heart failure (HF) patients can be complicated. A clinical reminder can facilitate use of guideline-concordant β-blocker titration for HF patients with depressed ejection fraction. However, the design must consider the cognitive demands on the providers and the context of the work.

OBJECTIVE

This study's purpose is to develop requirements for a clinical decision support tool (a clinical reminder) by analyzing the cognitive demands of the task along with the factors in the Cabana framework of physician adherence to guidelines, the health information technology (HIT) sociotechnical framework, and the Promoting Action on Research Implementation in Health Services (PARIHS) framework of health services implementation. It utilizes a tool that extracts information from medical records (including ejection fraction in free text reports) to identify qualifying patients at risk of undertreatment.

METHODS

We conducted interviews with 17 primary care providers, 5 PharmDs, and 5 Registered Nurses across three Veterans Health Administration outpatient clinics. The interviews were based on cognitive task analysis (CTA) methods and enhanced through the inclusion of the Cabana, HIT sociotechnical, and PARIHS frameworks. The analysis of the interview data led to the development of requirements and a prototype design for a clinical reminder. We conducted a small pilot usability assessment of the clinical reminder using realistic clinical scenarios.

RESULTS

We identified organizational challenges (such as time pressures and underuse of pharmacists), knowledge issues regarding the guideline, and information needs regarding patient history and treatment status. We based the design of the clinical reminder on how to best address these challenges. The usability assessment indicated the tool could help the decision and titration processes.

CONCLUSION

Through the use of CTA methods enhanced with adherence, sociotechnical, and implementation frameworks, we designed a decision support tool that considers important challenges in the decision and execution of β-blocker titration for qualifying HF patients at risk of undertreatment.

Prognostic Significance of Baseline Serum Sodium in Heart Failure With Preserved Ejection Fraction

BACKGROUND

The purpose of this study was to evaluate the relationship between serum sodium at the time of diagnosis and long term clinical outcomes in a large national cohort of patients with heart failure with preserved ejection fraction.

METHODS AND RESULTS

We studied 25 440 patients with heart failure with preserved ejection fraction treated at Veterans Affairs medical centers across the United States between 2002 and 2012. Serum sodium at the time of heart failure diagnosis was analyzed as a continuous variable and in categories as follows: low (115.00-134.99 mmol/L), low-normal (135.00-137.99 mmol/L), referent group (138.00-140.99 mmol/L), high normal (141.00-143.99 mmol/L), and high (144.00-160.00 mmol/L). Multivariable Cox regression and negative binomial regression were performed to estimate hazard ratios (95% confidence interval [CI]) and incidence density ratios (95% CI) for the associations of serum sodium with mortality and hospitalizations (heart failure and all-cause), respectively. The average age of patients was 70.8 years, 96.2% were male, and 14% were black. Compared with the referent group, low, low-normal, and high sodium values were associated with 36% (95% CI, 28%-44%), 6% (95% CI, 1%-12%), and 9% (95% CI, 1%-17%) higher risk of all-cause mortality, respectively. Low and low-normal serum sodium were associated with 48% (95% CI, 10%-100%) and 38% (95% CI, 8%-77%) higher risk of number of days of heart failure hospitalizations per year, and with 44% (95% CI, 32%-56%) and 18% (95% CI, 10%-27%) higher risk of number of days of all-cause hospitalizations per year, respectively.

CONCLUSIONS

Both elevated and reduced serum sodium, including values currently considered within normal range, are associated with adverse outcomes in patients with heart failure with preserved ejection fraction.

From Sour Grapes to Low-Hanging Fruit: A Case Study Demonstrating a Practical Strategy for Natural Language Processing Portability

Natural Language Processing (NLP) holds potential for patient care and clinical research, but a gap exists between promise and reality. While some studies have demonstrated portability of NLP systems across multiple sites, challenges remain. Strategies to mitigate these challenges can strive for complex NLP problems using advanced methods (hard-to-reach fruit), or focus on simple NLP problems using practical methods (low-hanging fruit). This paper investigates a practical strategy for NLP portability using extraction of left ventricular ejection fraction (LVEF) as a use case. We used a tool developed at the Department of Veterans Affair (VA) to extract the LVEF values from free-text echocardiograms in the MIMIC-III database. The approach showed an accuracy of 98.4%, sensitivity of 99.4%, a positive predictive value of 98.7%, and F-score of 99.0%. This experience, in which a simple NLP solution proved highly portable with excellent performance, illustrates the point that simple NLP applications may be easier to disseminate and adapt, and in the short term may prove more useful, than complex applications.

Defining Phenotypes from Clinical Data to Drive Genomic Research

The rise in available longitudinal patient information in electronic health records (EHRs) and their coupling to DNA biobanks has resulted in a dramatic increase in genomic research using EHR data for phenotypic information. EHRs have the benefit of providing a deep and broad data source of health-related phenotypes, including drug response traits, expanding the phenome available to researchers for discovery. The earliest efforts at repurposing EHR data for research involved manual chart review of limited numbers of patients but now typically involve applications of rule-based and machine learning algorithms operating on sometimes huge corpora for both genome-wide and phenome-wide approaches. We highlight here the current methods, impact, challenges, and opportunities for repurposing clinical data to define patient phenotypes for genomics discovery. Use of EHR data has proven a powerful method for elucidation of genomic influences on diseases, traits, and drug-response phenotypes and will continue to have increasing applications in large cohort studies.

Automating Quality Measures for Heart Failure Using Natural Language Processing: A Descriptive Study in the Department of Veterans Affairs

Background

We developed an accurate, stakeholder-informed, automated, natural language processing (NLP) system to measure the quality of heart failure (HF) inpatient care, and explored the potential for adoption of this system within an integrated health care system.

Objective

To accurately automate a United States Department of Veterans Affairs (VA) quality measure for inpatients with HF.

Methods

We automated the HF quality measure Congestive Heart Failure Inpatient Measure 19 (CHI19) that identifies whether a given patient has left ventricular ejection fraction (LVEF) <40%, and if so, whether an angiotensin-converting enzyme inhibitor or angiotensin-receptor blocker was prescribed at discharge if there were no contraindications. We used documents from 1083 unique inpatients from eight VA medical centers to develop a reference standard (RS) to train (n=314) and test (n=769) the Congestive Heart Failure Information Extraction Framework (CHIEF). We also conducted semi-structured interviews (n=15) for stakeholder feedback on implementation of the CHIEF.

Results

The CHIEF classified each hospitalization in the test set with a sensitivity (SN) of 98.9% and positive predictive value of 98.7%, compared with an RS and SN of 98.5% for available External Peer Review Program assessments. Of the 1083 patients available for the NLP system, the CHIEF evaluated and classified 100% of cases. Stakeholders identified potential implementation facilitators and clinical uses of the CHIEF.

Conclusions

The CHIEF provided complete data for all patients in the cohort and could potentially improve the efficiency, timeliness, and utility of HF quality measurements.

Clinical information extraction applications: A literature review

BACKGROUND

With the rapid adoption of electronic health records (EHRs), it is desirable to harvest information and knowledge from EHRs to support automated systems at the point of care and to enable secondary use of EHRs for clinical and translational research. One critical component used to facilitate the secondary use of EHR data is the information extraction (IE) task, which automatically extracts and encodes clinical information from text.

OBJECTIVES

In this literature review, we present a review of recent published research on clinical information extraction (IE) applications.

METHODS

A literature search was conducted for articles published from January 2009 to September 2016 based on Ovid MEDLINE In-Process & Other Non-Indexed Citations, Ovid MEDLINE, Ovid EMBASE, Scopus, Web of Science, and ACM Digital Library.

RESULTS

A total of 1917 publications were identified for title and abstract screening. Of these publications, 263 articles were selected and discussed in this review in terms of publication venues and data sources, clinical IE tools, methods, and applications in the areas of disease- and drug-related studies, and clinical workflow optimizations.

CONCLUSIONS

Clinical IE has been used for a wide range of applications, however, there is a considerable gap between clinical studies using EHR data and studies using clinical IE. This study enabled us to gain a more concrete understanding of the gap and to provide potential solutions to bridge this gap.

Clinical Data Reuse or Secondary Use: Current Status and Potential Future Progress

Objective: To perform a review of recent research in clinical data reuse or secondary use, and envision future advances in this field. Methods: The review is based on a large literature search in MEDLINE (through PubMed), conference proceedings, and the ACM Digital Library, focusing only on research published between 2005 and early 2016. Each selected publication was reviewed by the authors, and a structured analysis and summarization of its content was developed. Results: The initial search produced 359 publications, reduced after a manual examination of abstracts and full publications. The following aspects of clinical data reuse are discussed: motivations and challenges, privacy and ethical concerns, data integration and interoperability, data models and terminologies, unstructured data reuse, structured data mining, clinical practice and research integration, and examples of clinical data reuse (quality measurement and learning healthcare systems). Conclusion: Reuse of clinical data is a fast-growing field recognized as essential to realize the potentials for high quality healthcare, improved healthcare management, reduced healthcare costs, population health management, and effective clinical research.

Text mining applied to electronic cardiovascular procedure reports to identify patients with trileaflet aortic stenosis and coronary artery disease

BACKGROUND

Interrogation of the electronic health record (EHR) using billing codes as a surrogate for diagnoses of interest has been widely used for clinical research. However, the accuracy of this methodology is variable, as it reflects billing codes rather than severity of disease, and depends on the disease and the accuracy of the coding practitioner. Systematic application of text mining to the EHR has had variable success for the detection of cardiovascular phenotypes. We hypothesize that the application of text mining algorithms to cardiovascular procedure reports may be a superior method to identify patients with cardiovascular conditions of interest.

METHODS

We adapted the Oracle product Endeca, which utilizes text mining to identify terms of interest from a NoSQL-like database, for purposes of searching cardiovascular procedure reports and termed the tool "PennSeek". We imported 282,569 echocardiography reports representing 81,164 individuals and 27,205 cardiac catheterization reports representing 14,567 individuals from non-searchable databases into PennSeek. We then applied clinical criteria to these reports in PennSeek to identify patients with trileaflet aortic stenosis (TAS) and coronary artery disease (CAD). Accuracy of patient identification by text mining through PennSeek was compared with ICD-9 billing codes.

RESULTS

Text mining identified 7115 patients with TAS and 9247 patients with CAD. ICD-9 codes identified 8272 patients with TAS and 6913 patients with CAD. 4346 patients with AS and 6024 patients with CAD were identified by both approaches. A randomly selected sample of 200-250 patients uniquely identified by text mining was compared with 200-250 patients uniquely identified by billing codes for both diseases. We demonstrate that text mining was superior, with a positive predictive value (PPV) of 0.95 compared to 0.53 by ICD-9 for TAS, and a PPV of 0.97 compared to 0.86 for CAD.

CONCLUSION

These results highlight the superiority of text mining algorithms applied to electronic cardiovascular procedure reports in the identification of phenotypes of interest for cardiovascular research.

Unlocking echocardiogram measurements for heart disease research through natural language processing

BACKGROUND

In order to investigate the mechanisms of cardiovascular disease in HIV infected and uninfected patients, an analysis of echocardiogram reports is required for a large longitudinal multi-center study.

IMPLEMENTATION

A natural language processing system using a dictionary lookup, rules, and patterns was developed to extract heart function measurements that are typically recorded in echocardiogram reports as measurement-value pairs. Curated semantic bootstrapping was used to create a custom dictionary that extends existing terminologies based on terms that actually appear in the medical record. A novel disambiguation method based on semantic constraints was created to identify and discard erroneous alternative definitions of the measurement terms. The system was built utilizing a scalable framework, making it available for processing large datasets.

RESULTS

The system was developed for and validated on notes from three sources: general clinic notes, echocardiogram reports, and radiology reports. The system achieved F-scores of 0.872, 0.844, and 0.877 with precision of 0.936, 0.982, and 0.969 for each dataset respectively averaged across all extracted values. Left ventricular ejection fraction (LVEF) is the most frequently extracted measurement. The precision of extraction of the LVEF measure ranged from 0.968 to 1.0 across different document types.

CONCLUSIONS

This system illustrates the feasibility and effectiveness of a large-scale information extraction on clinical data. New clinical questions can be addressed in the domain of heart failure using retrospective clinical data analysis because key heart function measurements can be successfully extracted using natural language processing.

Congestive heart failure information extraction framework for automated treatment performance measures assessment

OBJECTIVE

This paper describes a new congestive heart failure (CHF) treatment performance measure information extraction system - CHIEF - developed as part of the Automated Data Acquisition for Heart Failure project, a Veterans Health Administration project aiming at improving the detection of patients not receiving recommended care for CHF.

DESIGN

CHIEF is based on the Apache Unstructured Information Management Architecture framework, and uses a combination of rules, dictionaries, and machine learning methods to extract left ventricular function mentions and values, CHF medications, and documented reasons for a patient not receiving these medications.

MEASUREMENTS

The training and evaluation of CHIEF were based on subsets of a reference standard of various clinical notes from 1083 Veterans Health Administration patients. Domain experts manually annotated these notes to create our reference standard. Metrics used included recall, precision, and the F 1 -measure.

RESULTS

In general, CHIEF extracted CHF medications with high recall (>0.990) and good precision (0.960-0.978). Mentions of Left Ventricular Ejection Fraction were also extracted with high recall (0.978-0.986) and precision (0.986-0.994), and quantitative values of Left Ventricular Ejection Fraction were found with 0.910-0.945 recall and with high precision (0.939-0.976). Reasons for not prescribing CHF medications were more difficult to extract, only reaching fair accuracy with about 0.310-0.400 recall and 0.250-0.320 precision.

CONCLUSION

This study demonstrated that applying natural language processing to unlock the rich and detailed clinical information found in clinical narrative text notes makes fast and scalable quality improvement approaches possible, eventually improving management and outpatient treatment of patients suffering from CHF.

A new era of quality measurement in rheumatology: electronic clinical quality measures and national registries

PURPOSE OF REVIEW

This article reviews the evolution of quality measurement in rheumatology, highlighting new health-information technology infrastructure and standards that are enabling unprecedented innovation in this field.

RECENT FINDINGS

Spurred by landmark legislation that ties physician payment to value, the widespread use of electronic health records, and standards such as the Quality Data Model, quality measurement in rheumatology is rapidly evolving. Rather than relying on retrospective assessments of care gathered through administrative claims or manual chart abstraction, new electronic clinical quality measures (eCQMs) allow automated data capture from electronic health records. At the same time, qualified clinical data registries, like the American College of Rheumatology's Rheumatology Informatics System for Effectiveness registry, are enabling large-scale implementation of eCQMs across national electronic health record networks with real-time performance feedback to clinicians. Although successful examples of eCQM development and implementation in rheumatology and other fields exist, there also remain challenges, such as lack of health system data interoperability and problems with measure accuracy.

SUMMARY

Quality measurement and improvement is increasingly an essential component of rheumatology practice. Advances in health information technology are likely to continue to make implementation of eCQMs easier and measurement more clinically meaningful and accurate in coming years.

Ensembles of NLP Tools for Data Element Extraction from Clinical Notes

Natural Language Processing (NLP) is essential for concept extraction from narrative text in electronic health records (EHR). To extract numerous and diverse concepts, such as data elements (i.e., important concepts related to a certain medical condition), a plausible solution is to combine various NLP tools into an ensemble to improve extraction performance. However, it is unclear to what extent ensembles of popular NLP tools improve the extraction of numerous and diverse concepts. Therefore, we built an NLP ensemble pipeline to synergize the strength of popular NLP tools using seven ensemble methods, and to quantify the improvement in performance achieved by ensembles in the extraction of data elements for three very different cohorts. Evaluation results show that the pipeline can improve the performance of NLP tools, but there is high variability depending on the cohort.

Extraction of left ventricular ejection fraction information from various types of clinical reports

Efforts to improve the treatment of congestive heart failure, a common and serious medical condition, include the use of quality measures to assess guideline-concordant care. The goal of this study is to identify left ventricular ejection fraction (LVEF) information from various types of clinical notes, and to then use this information for heart failure quality measurement. We analyzed the annotation differences between a new corpus of clinical notes from the Echocardiography, Radiology, and Text Integrated Utility package and other corpora annotated for natural language processing (NLP) research in the Department of Veterans Affairs. These reports contain varying degrees of structure. To examine whether existing LVEF extraction modules we developed in prior research improve the accuracy of LVEF information extraction from the new corpus, we created two sequence-tagging NLP modules trained with a new data set, with or without predictions from the existing LVEF extraction modules. We also conducted a set of experiments to examine the impact of training data size on information extraction accuracy. We found that less training data is needed when reports are highly structured, and that combining predictions from existing LVEF extraction modules improves information extraction when reports have less structured formats and a rich set of vocabulary.

Feasibility of Population Health Analytics and Data Visualization for Decision Support in the Infectious Diseases Domain: A pilot study

OBJECTIVE

Big data or population-based information has the potential to reduce uncertainty in medicine by informing clinicians about individual patient care. The objectives of this study were: 1) to explore the feasibility of extracting and displaying population-based information from an actual clinical population's database records, 2) to explore specific design features for improving population display, 3) to explore perceptions of population information displays, and 4) to explore the impact of population information display on cognitive outcomes.

METHODS

We used the Veteran's Affairs (VA) database to identify similar complex patients based on a similar complex patient case. Study outcomes measures were 1) preferences for population information display 2) time looking at the population display, 3) time to read the chart, and 4) appropriateness of plans with pre- and post-presentation of population data. Finally, we redesigned the population information display based on our findings from this study.

RESULTS

The qualitative data analysis for preferences of population information display resulted in four themes: 1) trusting the big/population data can be an issue, 2) embedded analytics is necessary to explore patient similarities, 3) need for tools to control the view (overview, zoom and filter), and 4) different presentations of the population display can be beneficial to improve the display. We found that appropriateness of plans was at 60% for both groups (t9=-1.9; p=0.08), and overall time looking at the population information display was 2.3 minutes versus 3.6 minutes with experts processing information faster than non-experts (t8= -2.3, p=0.04).

CONCLUSION

A population database has great potential for reducing complexity and uncertainty in medicine to improve clinical care. The preferences identified for the population information display will guide future health information technology system designers for better and more intuitive display.