Bielinski, Suzette J.

Higher levels of hepatocyte growth factor (HGF) have been associated with the presence of peripheral arterial disease (PAD), but prospective associations are unknown. We examined the association of circulating HGF levels with incident PAD. Between 2000 and 2002, HGF was measured in 6742 Multi-Ethnic Study of Atherosclerosis participants without PAD. Incident clinical PAD, adjudicated on the basis of a positive history for the presence of disease-related symptoms or treatment, was ascertained through 2015. Incident low ankle-brachial index (ABI), defined as an ABI < 0.9 and a decline of ≥ 0.15, was assessed among 5736 individuals who had an ABI > 0.9 at baseline and ≥1 follow-up ABI measurement 3 to 10 years later. There were 116 clinical PAD and 197 low ABI events that occurred over a median follow-up of 14 and 9 years, respectively. After adjustment for demographic and clinical variables, a standard deviation increment of HGF (303 ng/L) was associated with an increased risk of clinical PAD (hazard ratio: 1.21; 95% confidence interval [CI]: 1.05-1.39) but not a low ABI (rate ratio: 1.03; 95% CI: 0.85-1.25). In conclusion, higher HGF levels were modestly associated with an increased risk of developing clinical PAD.

Background: The effective self-management of heart failure (HF) is highly dependent on cognitive function. Dementia frequently coexists with HF, but its exact prevalence and impact on health care utilization and death is not well understood. Methods: Residents from 7 southeast Minnesota counties with a first-ever code for HF (ICD-9 428 and ICD-10 I50) between 1/1/2012 and 12/31/2017 were identified. Dementia before and after the HF diagnosis was ascertained with codes recommended by the US Department of Health and Human Services. Patients were followed through 12/31/2018 for death and hospitalizations. Cox and Andersen-Gill models were used to examine associations between dementia (prior and post HF) and death and hospitalizations, respectively. Dementia after HF was analyzed as a time-dependent variable. Results: Among 6,312 patients with HF (mean age 75 years, 49% female), 636 (10%) had a prior diagnosis of dementia. After a mean (SD) follow-up of 3.1 (1.8) years, 601 (9.5%) patients were diagnosed with dementia post HF. The median time from HF to dementia was 1.3 years, with 43% of dementia cases occurring within 1 year and 64% within 2 years post HF. During follow-up, 2,678 deaths and 15,095 hospitalizations occurred. After adjustment, patients with dementia before HF had nearly a 2-times increased risk of death and 10% increased risk of hospitalization compared to those that did not have dementia (Table). Dementia after HF was associated with a 78% increased risk of death and 63% increased risk of hospitalization compared to those that did not have dementia. Conclusions: Approximately 10% of patients had dementia prior to HF and 10% develop dementia shortly after HF diagnosis. Patients with HF and dementia have an increased risk of death and hospitalization compared to dementia-free counterparts. These data underscore the importance of recognizing the presence and impact of dementia on outcomes in HF to optimize care delivery.

Background: Current cardiovascular disease (CVD) risk scores are derived from research cohorts and are particularly inaccurate in women, older adults, and those with missing data. To overcome these limitations, we aimed to develop a cohort to capitalize on the depth and breadth of clinical data within electronic health record (EHR) systems in order to develop next-generation sex-specific risk prediction scores for incident CVD. Methods: All individuals 30 years of age or older residing in Olmsted County, Minnesota on 1/1/2006 were identified. We developed and validated algorithms to define a variety of risk factors, thus building a comprehensive risk profile for each patient. Outcomes including myocardial infarction (MI), percutaneous intervention (PCI), coronary artery bypass graft (CABG), and CVD death were ascertained through 9/30/2017. Results: We identified 73,069 individuals without CVD (Table). We retrieved a total of 14,962,762 lab results; 14,534,466 diagnoses; 17,062,601 services/procedures; 1,236,998 outpatient prescriptions; 1,079,065 heart rate measurements; and 1,320,115 blood pressure measurements. The median number of blood pressure and heart rate measurements ascertained per individuals were 11 and 9, respectively. The five most prevalent conditions were: hypertension, hyperlipidemia, arthritis, depression, and cardiac arrhythmias. During follow-up 1,455 MIs, 1,581 PCI, 652 CABG, and 2,161 CVD-related deaths occurred. Conclusions: We developed a cohort with comprehensive risk profiles and follow-up for each patient. Using sophisticated machine learning approaches, this electronic cohort will be utilized to develop next-generation sex-specific CVD risk prediction scores. These approaches will allow us to address several challenges with use of EHR data including the ability to 1) deal with missing values, 2) assess and utilize a large number of variables without over-fitting, 3) allow non-linear relationships, and 4) use time-to-event data.

Background: The focus of most existing phenotyping algorithms based on electronic health record (EHR) data has been to accurately identify cases and non-cases of specific diseases. However, a more challenging task is to accurately identify disease incidence, as identifying the first occurrence of disease is more important for efficient and valid clinical and epidemiological research. Moreover, stroke is a challenging phenotype due to diagnosis difficulty and common miscoding. This task generally requires utilization of multiple types of EHR data (e.g., diagnoses and procedure codes, unstructured clinical notes) and a more robust algorithm integrating both natural language processing and machine learning. In this study, we developed and validated an EHR-based classifier to accurately identify stroke incidence among a cohort of atrial fibrillation (AF) patients Methods: We developed a stroke phenotyping algorithm using International Classification of Diseases, Ninth Revision (ICD-9) codes, Current Procedural Terminology (CPT) codes, and expert-provided keywords as model features. Structured data was extracted from Rochester Epidemiology Project (REP) database. Natural Language Processing (NLP) was used to extract and validate keyword occurrence in clinical notes. A window of ±30 days was considered when including/excluding keywords/codes into the input vector. Frequencies of keywords/codes were used as input feature sets for model training. Multiple competing models were trained using various combinations of feature sets and two machine learning algorithms: logistic regression and random forest. Training data were provided by two nurse abstractors and included validated stroke incidences from a previously established atrial fibrillation cohort. Precision, recall, and F-score of the algorithm were calculated to assess and compare model performances. Results: Among 4,914 patients with atrial fibrillation, 1,773 patients were screened. 3,141 patients had no stroke-related codes or keywords and were presumed to be free of stroke during follow-up. Among the screened patients, 740 had validated strokes and 1,033 did not have a stroke based on review of the EHR by trained nurse abstractors. The best performing stroke incidence phenotyping classifier utilized Keywords+ICD-9+CPT features using a random forest classifier, achieving a precision of 0.942, recall of 0.943, and F-score of 0.943. Conclusion: In conclusion, we developed and validated a stroke algorithm that performed well for identifying stroke incidence in an enriched population (AF cohort), which extends beyond the typical binary case/non-case stroke identification problem. Future work will involve testing the generalizability of this algorithm in a general population.

Purpose The Mayo Clinic Biobank was established to provide a large group of patients from which comparison groups (ie, controls) could be selected for case–control studies, to create a prospective cohort with sufficient power for common outcomes and to support electronic health record (EHR) studies. Participants A total of 56 862 participants enrolled (21% response rate) into the Mayo Clinic Biobank from Rochester, Minnesota (77%, n=43 836), Jacksonville, Florida (18%, n=10 368) and La Crosse, Wisconsin (5%, n=2658). Participants were all Mayo Clinic patients, 18 years of age or older and US residents. Findings to date Overall, 43% of participants were 65 years of age or older and female participants were more frequent (59%) than males at all sites. Most participants resided in the Upper Midwest regions of the USA (Minnesota, Iowa, Illinois or Wisconsin), Florida or Georgia. Self-reported race among Biobank participants was 90% white. Here we provide examples of the types of studies that have successfully utilised the resource, including (1) investigations of the population itself, (2) provision of controls for case–control studies, (3) genotype-driven research, (4) EHR-based research and (5) prospective recruitment to other studies. Over 270 projects have been approved to date to access Biobank data and/or samples; over 200 000 sample aliquots have been approved for distribution. Future plans The data and samples in the Mayo Clinic Biobank can be used for various types of epidemiological and clinical studies, especially in the setting of case–control studies for which the Biobank samples serve as control samples. We are planning cohort studies with additional follow-up and acquisition of genetic information on a large scale.

Abstract Pharmacogenomics (PGx) clinical decision support integrated into the electronic health record (EHR) has the potential to provide relevant knowledge to clinicians to enable individualized care. However, past experience implementing PGx clinical decision support into multiple EHR platforms has identified important clinical, procedural, and technical challenges. Commercial EHRs have been widely criticized for the lack of readiness to implement precision medicine. Herein, we share our experiences and lessons learned implementing new EHR functionality charting PGx phenotypes in a unique repository, genomic indicators, instead of using the problem or allergy list. The Gen-Ind has additional features including a brief description of the clinical impact, a hyperlink to the original laboratory report, and links to additional educational resources. The automatic generation of genomic indicators from interfaced PGx test results facilitates implementation and long-term maintenance of PGx data in the EHR and can be used as criteria for synchronous and asynchronous CDS.

Background: The decline in cardiovascular disease (CVD) mortality in the United States noted since the mid 1970s has recently leveled off. This has generated concern and a need for further analyses to understand geographic variation, cause-specific deaths, location of death, and age-specific trends. Methods: Death certificate data obtained from the Minnesota Department of Health were used for analysis. The American Heart Association (AHA) classification was applied to the underlying cause of death to generate the following CVD categories: coronary heart disease (CHD), non-CHD diseases of the heart, and non-cardiac circulatory diseases. Temporal trends in death rates were explored by negative binomial regression, which enables estimates of age- and sex-adjusted average annual percentage changes (AAPCs). Differences in AAPCs between 2000-2009 and 2010-2017 were tested overall and across demographic, CVD, and location of death categories. Results: CVD mortality rates, which were declining in 2000-2009, have leveled off in 2010-2017 (Table). These trends were observed across all subgroups. After adjusting for sex, the age group that drove the decline in rates from 2000-2009 was the 65-84 year old group, which experienced no change in rates from 2010-2017. Across CVD categories, after adjusting for age and sex, the decline in rates from 2000-2009 was largely attributable to steep reductions in CHD and non-cardiac circulatory diseases, with lesser reduction in non-CHD diseases of the heart. In 2010-2017, CHD and non-cardiac circulatory diseases rates showed no change, whereas the rate of non-CHD diseases of the heart increased. The rates for in- and out-of-hospital deaths improved in 2000-2009 with no further improvement observed in 2010-2017. Conclusions: CVD mortality rates are no longer declining in Minnesota, paralleling national trends. These trends are particularly concerning for non-CHD diseases of the heart.

Background: Accurate identification of stroke cases from electronic health record (EHR) is needed for efficient and valid clinical and epidemiology research. A number of studies have evaluated the validity of using ICD-9 and ICD-10 codes for stroke identification. Previous research results show that sensitivity and positive predictive value (PPV) for stroke identification is the lowest among all other cerebrovascular diseases, with notable differences observed by stroke pathological subtypes. While most prevalent cerebrovascular disease cases can be detected using 430-438/I60-I69 collectively, more accurate and comprehensive stroke phenotyping algorithm is needed to identify incident stroke cases from EHR. In our work, we compared case identification results using ICD-codes exclusively with ICD plus a natural language processing (NLP) algorithm. Methods: We developed the NLP part of our stroke algorithm using a list of expert provided stroke-related keywords, which covers transient ischemic attack (TIA), ischemic stroke, and hemorrhagic stroke. We validated the hybrid (ICD+NLP) algorithm and compared it with ICD-exclusive algorithm in a previously established atrial fibrillation cohort (n=5,062). Clinical notes and ICD codes of all patients after the incident AF event were reviewed by two nurse abstracters to confirm new stroke incidences. All past clinical notes and ICD codes of a subset of patients (n=402) were reviewed to confirm lifetime (prior and current) stroke cases. Manual abstraction results were considered the gold standard for evaluation of ICD only and ICD+NLP automatic extraction. Sensitivity and positive predictive values (PPV) of both algorithms were calculated. Results: Among 5,062 patients, 593 patients were confirmed to have suffered a stroke after atrial fibrillation while 4,469 patients were confirmed with no EHR evidence of stroke after AF. The ICD-exclusive algorithm had a sensitivity of 47.2% and a PPV of 47.8% for detecting new stroke incidences. The hybrid stroke algorithm achieved a sensitivity of 92.4% and a PPV of 60.6% for extraction of incident stroke. For extraction of lifetime stroke cases, the hybrid approach achieved a sensitivity of 93.9% and a PPV of 88.7%. Performance of new stroke incidence extraction is limited because past/current stroke mentions in clinical notes are difficult to distinguish by our NLP algorithm (may be a future direction). Conclusions: We developed and validated a stroke algorithm that performed well for identifying incident and lifetime stroke cases. The addition of NLP into the stroke algorithm improved the sensitivity and PPV of compared to an ICD-exclusive algorithm.