Implementation of a Machine-Learning Algorithm in the Electronic Health Record for Targeted Screening for Familial Hypercholesterolemia: A Quality Improvement Study.
Ontology highlight
ABSTRACT:
SUBMITTER: Sheth S
PROVIDER: S-EPMC8204766 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:BackgroundFamilial hypercholesterolemia (FH) is a condition characterized by high cholesterol levels and increased risk for coronary heart disease (CHD) that often goes undiagnosed. The Dutch Lipid Network Criteria (DLNC) are used to identify FH in clinical settings via physical examination, personal and family history of CHD, in addition to the presence of deleterious mutations of the LDLR, ApoB, and PCSK9 genes. Agreement between clinical and genetic diagnosis of FH varies. While an ICD diagnosis code was not available for coding FH until 2016, Systematized Nomenclature of Medicine (SNOMED) clinical concept codes, including genetic diagnoses, for FH have been utilized in electronic health records (EHRs).ObjectiveTo evaluate the concordance of identifying FH via SNOMED and ICD-10 CM codes vs the DLNC in an EHR database.MethodsUsing the Practice Fusion EHR database, the sensitivity, specificity, positive predictive value (PPV), and negative predictive value were calculated comparing an FH cohort identified via SNOMED and ICD-10 CM codes to one identified via the DLNC.ResultsAmong 907,616 patients with hypercholesterolemia, 2,180 were identified as FH via SNOMED code (zero were identified via ICD-10 CM), 259 had a DLNC score 6-8 (probable FH), and 45 had a DLNC score >8 (definite FH). Compared to DLNC score >8, the sensitivity, specificity, and PPV of the FH SNOMED code were 84.4%, 99.4%, and 6.4%, respectively. Compared to DLNC score ?6, the sensitivity was 36.8% and the specificity was 99.5% with a PPV of 18.7%.ConclusionCompared to the clinical criteria for FH, identification of FH patients via SNOMED diagnosis codes had high sensitivity and specificity, but low PPV. The discordance of these two techniques in identifying FH patients speaks to the challenges in identifying FH patients in large electronic databases such as administrative claims and EHR.
| S-EPMC6241698 | biostudies-literature
Project description:BackgroundFamilial hypercholesterolemia (FH) is a genetic disorder characterized by a high level of low-density lipoprotein cholesterol (LDL-C) and is an important cause for premature cardiovascular disease. Because of underdiagnoses, an acute event is often the first clinical manifestation of FH. There are limited data on the prevalence and treatment of FH among adults admitted for treatment of acute cardiovascular events in Bulgaria. Our objective was to assess the proportion and management of FH patients from those admitted to hospital for treatment of acute symptomatic acute atherosclerotic cardiovascular events (ASCVD), the achievement of LDL-C targets of European Society of Cardiology/European Atherosclerosis Society guidelines and related public healthcare resources.ObjectiveDigitalized healthcare records for patients admitted for treatment of symptomatic ASCVD acute events between August 2018 and August 2019 were used for the analysis. Five cardiology hospitals provided data for hospitalizations, laboratory tests, and ambulatory follow-ups up to February 2020. Patients' hospital and ambulatory records were linked, and medical histories were extracted via a specifically developed algorithm, and analyzed. Outcomes included the proportion of patients classified as FH as defined by the Dutch Lipid Network Criteria (DLNC), use of lipid-lowering therapy, LDL-C achieved by 1, 3, 6, and 12 months post-index event, and public resources spent on hospital and ambulatory treatment.ResultsWe reviewed 11,090 hospital records of patients admitted for treatment of acute events in the period August 2018-August 2019 with ICD codes for ASCVD (Supplementary Table S3). FH was identified in 731 (6.6%) patients, with DLNC score ≥ 3, (682 with coronary artery disease, 32 with cerebrovascular disease, and 17 with peripheral artery disease). We did not find the criteria for FH in 5797 patients. The remaining 4562 records were inconclusive due to lack of data in the hospital dossier. Less than half of FH patients (274/731, 37%) were discharged on high-intensity statin therapy prescribed (34/731, 5%) with combination therapy. The vast majority (96.2% with LDL-C ≥ 1.8 mmol/l) had poorly controlled LDL-C during the first year after discharge. Patients with a probable/definite DLNC score ≥ 6 points and those with recurrent events contributed to the higher cost paid both by the healthcare system and the patients themselves.ConclusionThese findings reinforce the need for more aggressive lipid-lowering therapy, and underline the efficiency of using an electronic medical records search tool to support physicians in improving early FH diagnosis, aiming to minimize residual and future ASCVD events among FH patients and their family members. Supplementary file1 (MP4 21838 KB).
| S-EPMC8107160 | biostudies-literature
Project description:Background Familial hypercholesterolemia ( FH ), is a historically underdiagnosed, undertreated, high-risk condition that is associated with a high burden of cardiovascular morbidity and mortality. In this study, we use a population-based approach using electronic health record ( EHR )-based algorithms to identify FH . We report the major adverse cardiovascular events, mortality, and cost of medical care associated with this diagnosis. Methods and Results In our 1.18 million EHR- eligible cohort, International Classification of Diseases, Ninth Revision ( ICD -9) code-defined hyperlipidemia was categorized into FH and non- FH groups using an EHR algorithm designed using the modified Dutch Lipid Clinic Network criteria. Major adverse cardiovascular events, mortality, and cost of medical care were analyzed. A priori associated variables/confounders were used for multivariate analyses using binary logistic regression and linear regression with propensity score-based weighted methods as appropriate. EHR FH was identified in 32 613 individuals, which was 2.7% of the 1.18 million EHR cohort and 13.7% of 237 903 patients with hyperlipidemia. FH had higher rates of myocardial infarction (14.77% versus 8.33%; P<0.0001), heart failure (11.82% versus 10.50%; P<0.0001), and, after adjusting for traditional risk factors, significantly correlated to a composite major adverse cardiovascular events variable (odds ratio, 4.02; 95% CI, 3.88-4.16; P<0.0001), mortality (odds ratio, 1.20; CI, 1.15-1.26; P<0.0001), and higher total revenue per-year (incidence rate ratio, 1.30; 95% CI, 1.28-1.33; P<0.0001). Conclusions EHR -based algorithms discovered a disproportionately high prevalence of FH in our medical cohort, which was associated with worse outcomes and higher costs of medical care. This data-driven approach allows for a more precise method to identify traditionally high-risk groups within large populations allowing for targeted prevention and therapeutic strategies.
| S-EPMC6662375 | biostudies-literature
Project description:Familial hypercholesterolemia (FH) is an autosomal dominant disorder characterized by abnormally high concentrations of low-density lipoprotein (LDL) cholesterol in the blood, which predisposes affected persons to premature coronary heart disease (CHD) and death. FH is one of the most common inherited disorders and the most common one known to cause premature CHD in people of European descent. The vast majority of people with FH have inherited a single mutation from one parent in either the LDL receptor (LDLR), apolipoprotein B (APOB), or proprotein convertase subtilisin/kexin type 9 (PCSK9) genes. Despite their greatly elevated risk of coronary heart disease, most individuals with FH remain undiagnosed, untreated, or inadequately treated. Cascade screening is a mechanism for identifying people at risk for a genetic condition by a process of systematic family tracing. The National Institute for Health and Clinical Excellence in the United Kingdom recommends cascade screening of close biological relatives of people with a clinical diagnosis of FH in order to effectively identify additional FH patients. The ultimate goal of this testing is to reduce morbidity and mortality from heart disease in persons with FH through early diagnosis and effective disease management. The goal of this article is to outline the available evidence on the clinical validity and utility of cascade screening for FH, while emphasizing the availability, usefulness, and recommendation for including DNA testing (if the disease-causing mutation has been identified).
| S-EPMC3102597 | biostudies-literature
Project description:BackgroundA major problem in treating acute kidney injury (AKI) is that clinical criteria for recognition are markers of established kidney damage or impaired function; treatment before such damage manifests is desirable. Clinicians could intervene during what may be a crucial stage for preventing permanent kidney injury if patients with incipient AKI and those at high risk of developing AKI could be identified.ObjectiveIn this study, we evaluate a machine learning algorithm for early detection and prediction of AKI.DesignWe used a machine learning technique, boosted ensembles of decision trees, to train an AKI prediction tool on retrospective data taken from more than 300 000 inpatient encounters.SettingData were collected from inpatient wards at Stanford Medical Center and intensive care unit patients at Beth Israel Deaconess Medical Center.PatientsPatients older than the age of 18 whose hospital stays lasted between 5 and 1000 hours and who had at least one documented measurement of heart rate, respiratory rate, temperature, serum creatinine (SCr), and Glasgow Coma Scale (GCS).MeasurementsWe tested the algorithm's ability to detect AKI at onset and to predict AKI 12, 24, 48, and 72 hours before onset.MethodsWe tested AKI detection and prediction using the National Health Service (NHS) England AKI Algorithm as a gold standard. We additionally tested the algorithm's ability to detect AKI as defined by the Kidney Disease: Improving Global Outcomes (KDIGO) guidelines. We compared the algorithm's 3-fold cross-validation performance to the Sequential Organ Failure Assessment (SOFA) score for AKI identification in terms of area under the receiver operating characteristic (AUROC).ResultsThe algorithm demonstrated high AUROC for detecting and predicting NHS-defined AKI at all tested time points. The algorithm achieves AUROC of 0.872 (95% confidence interval [CI], 0.867-0.878) for AKI detection at time of onset. For prediction 12 hours before onset, the algorithm achieves an AUROC of 0.800 (95% CI, 0.792-0.809). For 24-hour predictions, the algorithm achieves AUROC of 0.795 (95% CI, 0.785-0.804). For 48-hour and 72-hour predictions, the algorithm achieves AUROC values of 0.761 (95% CI, 0.753-0.768) and 0.728 (95% CI, 0.719-0.737), respectively.LimitationsBecause of the retrospective nature of this study, we cannot draw any conclusions about the impact the algorithm's predictions will have on patient outcomes in a clinical setting.ConclusionsThe results of these experiments suggest that a machine learning-based AKI prediction tool may offer important prognostic capabilities for determining which patients are likely to suffer AKI, potentially allowing clinicians to intervene before kidney damage manifests.
| S-EPMC6080076 | biostudies-literature
Project description:Familial hypercholesterolemia (FH) is a common inherited disorder that results in premature atherosclerosis. Diagnosis of FH is suspected on the basis of clinical criteria, but confirmation requires genetic testing. In the era of statins, early diagnosis and initiation of treatment can modify disease progression and outcomes. Therefore, cascade screening with a combination of lipid concentration measurements and DNA testing should be used to identify relatives of index cases with a clinical diagnosis of FH. Autosomal dominant FH is related to mutations in the low-density lipoprotein receptor (LDLR), apolipoprotein B-100 (APOB), or proprotein convertase subtilisin/kexin type 9 (PCSK9) genes. Genetic screening of the LDLR gene is challenging to achieve at a feasible cost, especially in people who do not have a founder effect. Nucleotide sequencing of all exons and flanking splicing regions in combination with multiplex ligation probe amplification to detect large insertions or deletions is considered the gold-standard approach to screen for LDLR mutations. Alternatively, the cDNA can be sequenced; however, this procedure is not suitable for use in large populations, because of the need of RNA extraction. Multiplex analysis can be appropriate for population with founder effects or a low number of different mutations. Finally, there are many techniques for a mutation scanning approach, which have some benefits over sequencing, and also with the potential for detecting known and novel mutations. Familial defective Apo B is amenable to genetic diagnosis by screening for a few mutations. Recently, gain-of-function mutations in PCSK9 gene have been demonstrated to cause FH phenotype. Strategies for population screening, cost-effectiveness of genetic screening, ethical aspects, and insurance policies are discussed and need implementation worldwide.
| S-EPMC3681171 | biostudies-literature
Project description:BackgroundFinancial codes are often used to extract diagnoses from electronic health records. This approach is prone to false positives. Alternatively, queries are constructed, but these are highly center and language specific. A tantalizing alternative is the automatic identification of patients by employing machine learning on format-free text entries.ObjectiveThe aim of this study was to develop an easily implementable workflow that builds a machine learning algorithm capable of accurately identifying patients with rheumatoid arthritis from format-free text fields in electronic health records.MethodsTwo electronic health record data sets were employed: Leiden (n=3000) and Erlangen (n=4771). Using a portion of the Leiden data (n=2000), we compared 6 different machine learning methods and a naïve word-matching algorithm using 10-fold cross-validation. Performances were compared using the area under the receiver operating characteristic curve (AUROC) and the area under the precision recall curve (AUPRC), and F1 score was used as the primary criterion for selecting the best method to build a classifying algorithm. We selected the optimal threshold of positive predictive value for case identification based on the output of the best method in the training data. This validation workflow was subsequently applied to a portion of the Erlangen data (n=4293). For testing, the best performing methods were applied to remaining data (Leiden n=1000; Erlangen n=478) for an unbiased evaluation.ResultsFor the Leiden data set, the word-matching algorithm demonstrated mixed performance (AUROC 0.90; AUPRC 0.33; F1 score 0.55), and 4 methods significantly outperformed word-matching, with support vector machines performing best (AUROC 0.98; AUPRC 0.88; F1 score 0.83). Applying this support vector machine classifier to the test data resulted in a similarly high performance (F1 score 0.81; positive predictive value [PPV] 0.94), and with this method, we could identify 2873 patients with rheumatoid arthritis in less than 7 seconds out of the complete collection of 23,300 patients in the Leiden electronic health record system. For the Erlangen data set, gradient boosting performed best (AUROC 0.94; AUPRC 0.85; F1 score 0.82) in the training set, and applied to the test data, resulted once again in good results (F1 score 0.67; PPV 0.97).ConclusionsWe demonstrate that machine learning methods can extract the records of patients with rheumatoid arthritis from electronic health record data with high precision, allowing research on very large populations for limited costs. Our approach is language and center independent and could be applied to any type of diagnosis. We have developed our pipeline into a universally applicable and easy-to-implement workflow to equip centers with their own high-performing algorithm. This allows the creation of observational studies of unprecedented size covering different countries for low cost from already available data in electronic health record systems.
| S-EPMC7735897 | biostudies-literature
Project description:To examine the impact of electronic health record (EHR) deployment on Surgical Care Improvement Project (SCIP) measures in a tertiary-care teaching hospital.SCIP Core Measure dataset from the CMS Hospital Inpatient Quality Reporting Program (March 2010 to February 2012).One-group pre- and post-EHR logistic regression and difference-in-differences analyses.Statistically significant short-term declines in scores were observed for the composite, postoperative removal of urinary catheter and post-cardiac surgery glucose control measures. A statistically insignificant improvement in scores for these measures was noted 3 months after EHR deployment.The transition to an EHR appears to be associated with a short-term decline in quality. Implementation strategies should be developed to preempt or minimize this initial decline.
| S-EPMC4319882 | biostudies-other
Project description:In familial hypercholesterolemia (FH), carriers profit from presymptomatic diagnosis and early treatment. Due to the autosomal dominant pattern of inheritance, first degree relatives of patients are at 50% risk. A program to identify healthy relatives at risk of premature cardiovascular problems, funded by the Netherlands government until 2014, raised questions on privacy and autonomy in view of the chosen active approach of family members. Several countries are building cascade screening programs inspired by Dutch experience, but meanwhile, the Netherlands' screening program itself is in transition. Insight in stakeholders' views on approaching family members is lacking. Literature and policy documents were studied, and stakeholders were interviewed on pros and cons of actively approaching healthy relatives. Sociotechnical analysis explored new roles and responsibilities, with uptake, privacy, autonomy, psychological burden, resources, and awareness as relevant themes. Stakeholders agree on the importance of early diagnosis and informing the family. Dutch healthcare typically focuses on cure, rather than prevention. Barriers to cascade screening are paying an own financial contribution, limited resources for informing relatives, and privacy regulation. To benefit from predictive, personalized, and preventive medicine, the roles and responsibilities of stakeholders in genetic testing as a preventive strategy, and informing family members, need to be carefully realigned.
| S-EPMC6163326 | biostudies-literature
Project description:BackgroundCardiovascular outcomes for people with familial hypercholesterolaemia can be improved with diagnosis and medical management. However, 90% of individuals with familial hypercholesterolaemia remain undiagnosed in the USA. We aimed to accelerate early diagnosis and timely intervention for more than 1·3 million undiagnosed individuals with familial hypercholesterolaemia at high risk for early heart attacks and strokes by applying machine learning to large health-care encounter datasets.MethodsWe trained the FIND FH machine learning model using deidentified health-care encounter data, including procedure and diagnostic codes, prescriptions, and laboratory findings, from 939 clinically diagnosed individuals with familial hypercholesterolaemia (395 of whom had a molecular diagnosis) and 83 136 individuals presumed free of familial hypercholesterolaemia, sampled from four US institutions. The model was then applied to a national health-care encounter database (170 million individuals) and an integrated health-care delivery system dataset (174 000 individuals). Individuals used in model training and those evaluated by the model were required to have at least one cardiovascular disease risk factor (eg, hypertension, hypercholesterolaemia, or hyperlipidemia). A Health Insurance Portability and Accountability Act of 1996-compliant programme was developed to allow providers to receive identification of individuals likely to have familial hypercholesterolaemia in their practice.FindingsUsing a model with a measured precision (positive predictive value) of 0·85, recall (sensitivity) of 0·45, area under the precision-recall curve of 0·55, and area under the receiver operating characteristic curve of 0·89, we flagged 1 331 759 of 170 416 201 patients in the national database and 866 of 173 733 individuals in the health-care delivery system dataset as likely to have familial hypercholesterolaemia. Familial hypercholesterolaemia experts reviewed a sample of flagged individuals (45 from the national database and 103 from the health-care delivery system dataset) and applied clinical familial hypercholesterolaemia diagnostic criteria. Of those reviewed, 87% (95% Cl 73-100) in the national database and 77% (68-86) in the health-care delivery system dataset were categorised as having a high enough clinical suspicion of familial hypercholesterolaemia to warrant guideline-based clinical evaluation and treatment.InterpretationThe FIND FH model successfully scans large, diverse, and disparate health-care encounter databases to identify individuals with familial hypercholesterolaemia.FundingThe FH Foundation funded this study. Support was received from Amgen, Sanofi, and Regeneron.
| S-EPMC8086528 | biostudies-literature