A systematic review of fall prediction models for community-dwelling older adults: comparison between models based on research cohorts and models based on routinely collected data.

BACKGROUND: Prediction models can identify fall-prone individuals. Prediction models can be based on either data from research cohorts (cohort-based) or routinely collected data (RCD-based). We review and compare cohort-based and RCD-based studies describing the development and/or validation of fall prediction models for community-dwelling older adults. METHODS: Medline and Embase were searched via Ovid until January 2023. We included studies describing the development or validation of multivariable prediction models of falls in older adults (60+). Both risk of bias and reporting quality were assessed using the PROBAST and TRIPOD, respectively. RESULTS: We included and reviewed 28 relevant studies, describing 30 prediction models (23 cohort-based and 7 RCD-based), and external validation of two existing models (one cohort-based and one RCD-based). The median sample sizes for cohort-based and RCD-based studies were 1365 [interquartile range (IQR) 426-2766] versus 90 441 (IQR 56 442-128 157), and the ranges of fall rates were 5.4% to 60.4% versus 1.6% to 13.1%, respectively. Discrimination performance was comparable between cohort-based and RCD-based models, with the respective area under the receiver operating characteristic curves ranging from 0.65 to 0.88 versus 0.71 to 0.81. The median number of predictors in cohort-based final models was 6 (IQR 5-11); for RCD-based models, it was 16 (IQR 11-26). All but one cohort-based model had high bias risks, primarily due to deficiencies in statistical analysis and outcome determination. CONCLUSIONS: Cohort-based models to predict falls in older adults in the community are plentiful. RCD-based models are yet in their infancy but provide comparable predictive performance with no additional data collection efforts. Future studies should focus on methodological and reporting quality.

Stroke Risk Reduction in Migraine Patients Using Propranolol: Evidence from Two Large-Scale Real-World Data Analyses.

Background: Propranolol, a non-selective beta-blocker, is commonly used for migraine prevention, but its impact on stroke risk among migraine patients remains controversial. Using two large electronic health records-based datasets, we examined stroke risk differences between migraine patients with- and without- documented use of propranolol. Methods: This retrospective case-control study utilized EHR data from the Vanderbilt University Medical Center (VUMC) and the All of Us Research Program. Migraine patients were first identified based on the International Classification of Headache Disorders, 3rd edition (ICHD-3) criteria using diagnosis codes. Among these patients, cases were defined as those with a primary diagnosis of stroke following the first diagnosis of migraine, while controls had no stroke after their first migraine diagnosis. Logistic regression models, adjusted for potential factors associated with stroke risk, assessed the association between propranolol use and stroke risk, stratified by sex and migraine subtype. A Cox proportional hazards regression model was used to estimate the hazard ratio (HR) for stroke risk at 1, 2, 5, and 10 years from baseline. Results: In the VUMC database, 378 cases and 15,209 controls were identified, while the All of Us database included 267 cases and 6,579 controls. Propranolol significantly reduced stroke risk in female migraine patients (VUMC: OR=0.52, p=0.006; All of Us: OR=0.39, p=0.007), but not in males. The effect was more pronounced for ischemic stroke and in females with migraines without aura (MO) (VUMC: OR=0.60, p=0.014; All of Us: OR=0.28, p=0.006). The Cox model showed lower stroke rates in propranolol-treated female migraine patients at 1, 2, 5, and 10 years (VUMC: HR=0.06-0.55, p=0.0018-0.085; All of Us: HR=0.23, p=0.045 at 10 years). Conclusions: Propranolol is associated with a significant reduction in stroke risk, particularly ischemic stroke, among female migraine without aura patients. These findings suggest that propranolol may benefit stroke prevention in high-risk populations.

Identification of an ANCA-Associated Vasculitis Cohort Using Deep Learning and Electronic Health Records.

Background: ANCA-associated vasculitis (AAV) is a rare but serious disease. Traditional case-identification methods using claims data can be time-intensive and may miss important subgroups. We hypothesized that a deep learning model analyzing electronic health records (EHR) can more accurately identify AAV cases. Methods: We examined the Mass General Brigham (MGB) repository of clinical documentation from 12/1/1979 to 5/11/2021, using expert-curated keywords and ICD codes to identify a large cohort of potential AAV cases. Three labeled datasets (I, II, III) were created, each containing note sections. We trained and evaluated a range of machine learning and deep learning algorithms for note-level classification, using metrics like positive predictive value (PPV), sensitivity, F-score, area under the receiver operating characteristic curve (AUROC), and area under the precision and recall curve (AUPRC). The deep learning model was further evaluated for its ability to classify AAV cases at the patient-level, compared with rule-based algorithms in 2,000 randomly chosen samples. Results: Datasets I, II, and III comprised 6,000, 3,008, and 7,500 note sections, respectively. Deep learning achieved the highest AUROC in all three datasets, with scores of 0.983, 0.991, and 0.991. The deep learning approach also had among the highest PPVs across the three datasets (0.941, 0.954, and 0.800, respectively). In a test cohort of 2,000 cases, the deep learning model achieved a PPV of 0.262 and an estimated sensitivity of 0.975. Compared to the best rule-based algorithm, the deep learning model identified six additional AAV cases, representing 13% of the total. Conclusion: The deep learning model effectively classifies clinical note sections for AAV diagnosis. Its application to EHR notes can potentially uncover additional cases missed by traditional rule-based methods.

COPD Exacerbations Before and During COVID-19 in France, Germany, Italy, the UK and the US.

Background: Exacerbations of chronic obstructive pulmonary disease (COPD) were reported less frequently during the COVID-19 pandemic. We report real-world data on COPD exacerbation rates before and during this pandemic. Methods: Exacerbation patterns were analysed using electronic medical records or claims data of patients with COPD before (2017-2019) and during the COVID-19 pandemic (2020 through early 2022) in France, Germany, Italy, the United Kingdom and the United States. Data from each country were analysed separately. The proportions of patients with COPD receiving maintenance treatment were also estimated. Results: The proportion of patients with exacerbations fell 45-78% across five countries in 2020 versus 2019. Exacerbation rates in most countries were reduced by >50% in 2020 compared with 2019. The proportions of patients with an exacerbation increased in most countries in 2021. Across each country, seasonal exacerbation increases seen during autumn and winter in pre-pandemic years were absent during the first year of the pandemic. The percentage of patients filling COPD prescriptions across each country increased by 4.53-22.13% in 2019 to 9.94-34.17% in 2021. Conclusion: Early, steep declines in exacerbation rates occurred in 2020 versus 2019 across all five countries and were accompanied by a loss of the seasonal pattern of exacerbation.

Agreement between patient-reported and clinically documented symptoms of acute myeloid leukemia: Study protocol.

AIM: To describe our methods to compare patient-reported symptoms of acute myeloid leukemia and the corresponding documentation by healthcare providers in the electronic health record. BACKGROUND: Patients with acute myeloid leukemia experience many distressing symptoms, particularly related to chemotherapy. The timely recognition and provision of evidence-based interventions to manage these symptoms can improve outcomes. However, lack of standardized formatting for symptom documentation within electronic health records leads to challenges for clinicians when accessing and comprehending patients' symptom information, as it primarily exists in narrative forms in various parts of the electronic health record. This variability raises concerns about over- or under-reporting of symptoms. Consistency between patient-reported symptoms and clinician's symptom documentation is important for patient-centered symptom management, but little is known about the degree of agreement between patient reports and their documentation. This is a detailed description of the study's methodology, procedures and design to determine how patient-reported symptoms are similar or different from symptoms documented in electronic health records by clinicians. DESIGN: Exploratory, descriptive study. METHODS: Forty symptoms will be assessed as patient-reported outcomes using the modified version of the Memorial Symptom Assessment Scale. The research team will annotate symptoms from the electronic health record (clinical notes and flowsheets) corresponding to the 40 symptoms. The degree of agreement between patient reports and electronic health record documentation will be analyzed using positive and negative agreement, kappa statistics and McNemar's test. CONCLUSION: We present innovative methods to comprehensively compare the symptoms reported by acute myeloid leukemia patients with all available electronic health record documentation, including clinical notes and flowsheets, providing insights into symptom reporting in clinical practice. IMPACT: Findings from this study will provide foundational understanding and compelling evidence, suggesting the need for more thorough efforts to assess patients' symptoms. Methods presented in this paper are applicable to other symptom-intensive diseases.

Social Determinants of Health Mediate Racial Disparities in Cardiovascular Disease in Men With Prostate Cancer.

Background: Cardiovascular disease (CVD) is a significant cause of morbidity and mortality in men with prostate cancer; however, data on racial disparities in CVD outcomes are limited. Objectives: We quantified the disparities in CVD according to self-identified race and the role of the structural social determinants of health in mediating disparities in prostate cancer patients. Methods: A retrospective cohort study of 3,543 prostate cancer patients treated with systemic androgen deprivation therapy (ADT) between 2008 and 2021 at a quaternary, multisite health care system was performed. The multivariable adjusted association between self-reported race (Black vs White) and incident major adverse cardiovascular events (MACE) after ADT initiation was evaluated using cause-specific proportional hazards. Mediation analysis determined the role of theme-specific and overall social vulnerability index (SVI) in explaining the racial disparities in CVD outcomes. Results: Black race was associated with an increased hazard of MACE (HR: 1.38; 95% CI: 1.16-1.65; P < 0.001). The association with Black race was strongest for incident heart failure (HR: 1.79; 95% CI: 1.32-2.43), cerebrovascular disease (HR: 1.98; 95% CI: 1.37-2.87), and peripheral artery disease (HR: 1.76; 95% CI: 1.26-2.45) (P < 0.001). SVI, specifically the socioeconomic status theme, mediated 98% of the disparity in MACE risk between Black and White patients. Conclusions: Black patients are significantly more likely to experience adverse CVD outcomes after systemic ADT compared with their White counterparts. These disparities are mediated by socioeconomic status and other structural determinants of health as captured by census tract SVI. Our findings motivate multilevel interventions focused on addressing socioeconomic vulnerability.

Revolutionizing urogynecology: Machine learning application with patient-centric technology: Promise, challenges, and future directions.

In an epoch where digital innovation is redefining the medical landscape, electronic health records (EHRs) stand out as a pivotal transformative force. Urogynecology, a discipline anchored in intricate patient histories and meticulous follow-ups, is on the brink of profound transformation due to these digital strides. While EHRs have unified patient data, challenges related to data privacy, interoperability, and access persist. In response, we present Pelvic Health Place (PHPlace) - a multilingual, patient-centric application. Purposefully designed to bolster patient engagement, PHPlace provides clinicians with essential pre-consultation insights, streamlines the consent process, vividly delineates surgical pathways, and assures comprehensive long-term monitoring. This platform also establishes a foundation for global data amalgamation, promising to invigorate research and potentially harness artificial intelligence (AI) capabilities. With AI integration, we anticipate a more tailored treatment approach and enriched patient education, signaling a pivotal shift in urogynecology and emphasizing the imperative for ongoing academic inquiry.

Accurate prediction of all-cause mortality in patients with metabolic dysfunction-associated steatotic liver disease using electronic health records.

INTRODUCTION AND OBJECTIVES: Despite the huge clinical burden of MASLD, validated tools for early risk stratification are lacking, and heterogeneous disease expression and a highly variable rate of progression to clinical outcomes result in prognostic uncertainty. We aimed to investigate longitudinal electronic health record-based outcome prediction in MASLD using a state-of-the-art machine learning model. PATIENTS AND METHODS: n = 940 patients with histologically-defined MASLD were used to develop a deep-learning model for all-cause mortality prediction. Patient timelines, spanning 12 years, were fully-annotated with demographic/clinical characteristics, ICD-9 and -10 codes, blood test results, prescribing data, and secondary care activity. A Transformer neural network (TNN) was trained to output concomitant probabilities of 12-, 24-, and 36-month all-cause mortality. In-sample performance was assessed using 5-fold cross-validation. Out-of-sample performance was assessed in an independent set of n = 528 MASLD patients. RESULTS: In-sample model performance achieved AUROC curve 0.74-0.90 (95 % CI: 0.72-0.94), sensitivity 64 %-82 %, specificity 75 %-92 % and Positive Predictive Value (PPV) 94 %-98 %. Out-of-sample model validation had AUROC 0.70-0.86 (95 % CI: 0.67-0.90), sensitivity 69 %-70 %, specificity 96 %-97 % and PPV 75 %-77 %. Key predictive factors, identified using coefficients of determination, were age, presence of type 2 diabetes, and history of hospital admissions with length of stay >14 days. CONCLUSIONS: A TNN, applied to routinely-collected longitudinal electronic health records, achieved good performance in prediction of 12-, 24-, and 36-month all-cause mortality in patients with MASLD. Extrapolation of our technique to population-level data will enable scalable and accurate risk stratification to identify people most likely to benefit from anticipatory health care and personalized interventions.

Recurrent urinary tract infections and prophylactic antibiotic use in women: cross-sectional study.

BACKGROUND: Despite the considerable morbidity caused by recurrent UTIs (rUTIs), and the wider personal and public health implications from frequent antibiotic use, few studies adequately describe the prevalence and characteristics of women with rUTIs or those who use prophylactic antibiotics. AIM: To describe the prevalence, characteristics, and urine profiles of women with rUTIs with and without prophylactic antibiotic use in Welsh primary care. DESIGN AND SETTING: Retrospective cross-sectional study in Welsh General Practice using the SAIL Databank. METHOD: We describe the characteristics of women aged ≥18 years with rUTIs or using prophylactic antibiotics from 2010-2020, and associated urine culture results from 2015 - 2020. RESULTS: 6.0% of women (n=92,213) had rUTIs, and 1.7% (n=26,862) were prescribed prophylactic antibiotics. Only 49% of prophylactic antibiotic users met the definition of rUTIs before initiation. 81% of women with rUTIs had a urine culture result in the preceding 12 months with high rates of resistance to trimethoprim and amoxicillin. 64% of women taking prophylactic antibiotics had a urine culture result before initiation, and 18% (n=320) of women prescribed trimethoprim had resistance to it on the antecedent sample. CONCLUSION: A substantial proportion of women had rUTIs or incident prophylactic antibiotic use. However, 64% of women had urine cultured before starting prophylaxis. There was a high proportion of cultured bacteria resistant to two antibiotics used for rUTI prevention and evidence of resistance to the prescribed antibiotic. More frequent urine cultures for rUTI diagnosis and before prophylactic antibiotic initiation could better inform antibiotic choices.

Assessing the quality of emergency department data for injury surveillance in Victoria, Australia: a comparative analysis of two Victorian hospital data sources.

OBJECTIVE: The emergency department (ED) is pivotal in treating serious injuries, making it a valuable source for population-based injury surveillance. In Victoria, information that is relevant to injury surveillance is collected in the Victorian Emergency Minimum Dataset (VEMD). This study aims to assess the data quality of the VEMD as an injury data source by comparing it with the Victorian Admitted Episodes Dataset (VAED). DESIGN: A retrospective observational study of administrative healthcare data. SETTING AND PARTICIPANTS: VEMD and VAED data from July 2014 to June 2019 were compared. Including only hospitals contributing to both datasets, cases that (1) arrived at the ED and (2) were subsequently admitted, were selected. RESULTS: While the overall number of cases was similar, VAED outnumbered VEMD cases (414 630 vs 404 608), suggesting potential under-reporting of injuries in the ED. Age-related differences indicated a relative under-representation of older individuals in the VEMD. Injuries caused by falls or transport, and intentional injuries were relatively under-reported in the VEMD. CONCLUSIONS: Injury cases were more numerous in the VAED than in the VEMD even though the number is expected to be equal based on case selection. Older patients were under-represented in the VEMD; this could partly be attributed to patients being admitted for an injury after they presented to the ED with a non-injury ailment. The patterns of under-representation described in this study should be taken into account in ED-based injury incidence reporting.

Experiences from patients in mental healthcare accessing their electronic health records: results from a cross-national survey in Estonia, Finland, Norway, and Sweden.

BACKGROUND: Patients' online record access (ORA) enables patients to read and use their health data through online digital solutions. One such solution, patient-accessible electronic health records (PAEHRs) have been implemented in Estonia, Finland, Norway, and Sweden. While accumulated research has pointed to many potential benefits of ORA, its application in mental healthcare (MHC) continues to be contested. The present study aimed to describe MHC users' overall experiences with national PAEHR services. METHODS: The study analysed the MHC-part of the NORDeHEALTH 2022 Patient Survey, a large-scale multi-country survey. The survey consisted of 45 questions, including demographic variables and questions related to users' experiences with ORA. We focused on the questions concerning positive experiences (benefits), negative experiences (errors, omissions, offence), and breaches of security and privacy. Participants were included in this analysis if they reported receiving mental healthcare within the past two years. Descriptive statistics were used to summarise data, and percentages were calculated on available data. RESULTS: 6,157 respondents were included. In line with previous research, almost half (45%) reported very positive experiences with ORA. A majority in each country also reported improved trust (at least 69%) and communication (at least 71%) with healthcare providers. One-third (29.5%) reported very negative experiences with ORA. In total, half of the respondents (47.9%) found errors and a third (35.5%) found omissions in their medical documentation. One-third (34.8%) of all respondents also reported being offended by the content. When errors or omissions were identified, about half (46.5%) reported that they took no action. There seems to be differences in how patients experience errors, omissions, and missing information between the countries. A small proportion reported instances where family or others demanded access to their records (3.1%), and about one in ten (10.7%) noted that unauthorised individuals had seen their health information. CONCLUSIONS: Overall, MHC patients reported more positive experiences than negative, but a large portion of respondents reported problems with the content of the PAEHR. Further research on best practice in implementation of ORA in MHC is therefore needed, to ensure that all patients may reap the benefits while limiting potential negative consequences.

Extraction of sleep information from clinical notes of Alzheimer's disease patients using natural language processing.

OBJECTIVES: Alzheimer's disease (AD) is the most common form of dementia in the United States. Sleep is one of the lifestyle-related factors that has been shown critical for optimal cognitive function in old age. However, there is a lack of research studying the association between sleep and AD incidence. A major bottleneck for conducting such research is that the traditional way to acquire sleep information is time-consuming, inefficient, non-scalable, and limited to patients' subjective experience. We aim to automate the extraction of specific sleep-related patterns, such as snoring, napping, poor sleep quality, daytime sleepiness, night wakings, other sleep problems, and sleep duration, from clinical notes of AD patients. These sleep patterns are hypothesized to play a role in the incidence of AD, providing insight into the relationship between sleep and AD onset and progression. MATERIALS AND METHODS: A gold standard dataset is created from manual annotation of 570 randomly sampled clinical note documents from the adSLEEP, a corpus of 192 000 de-identified clinical notes of 7266 AD patients retrieved from the University of Pittsburgh Medical Center (UPMC). We developed a rule-based natural language processing (NLP) algorithm, machine learning models, and large language model (LLM)-based NLP algorithms to automate the extraction of sleep-related concepts, including snoring, napping, sleep problem, bad sleep quality, daytime sleepiness, night wakings, and sleep duration, from the gold standard dataset. RESULTS: The annotated dataset of 482 patients comprised a predominantly White (89.2%), older adult population with an average age of 84.7 years, where females represented 64.1%, and a vast majority were non-Hispanic or Latino (94.6%). Rule-based NLP algorithm achieved the best performance of F1 across all sleep-related concepts. In terms of positive predictive value (PPV), the rule-based NLP algorithm achieved the highest PPV scores for daytime sleepiness (1.00) and sleep duration (1.00), while the machine learning models had the highest PPV for napping (0.95) and bad sleep quality (0.86), and LLAMA2 with finetuning had the highest PPV for night wakings (0.93) and sleep problem (0.89). DISCUSSION: Although sleep information is infrequently documented in the clinical notes, the proposed rule-based NLP algorithm and LLM-based NLP algorithms still achieved promising results. In comparison, the machine learning-based approaches did not achieve good results, which is due to the small size of sleep information in the training data. CONCLUSION: The results show that the rule-based NLP algorithm consistently achieved the best performance for all sleep concepts. This study focused on the clinical notes of patients with AD but could be extended to general sleep information extraction for other diseases.

Lithium and the risk of fractures in patients with bipolar disorder: A population-based cohort study.

Lithium is considered to be the most effective mood stabilizer for bipolar disorder. Evolving evidence suggested lithium can also regulate bone metabolism which may reduce the risk of fractures. While there are concerns about fractures for antipsychotics and mood stabilizing antiepileptics, very little is known about the overall risk of fractures associated with specific treatments. This study aimed to compare the risk of fractures in patients with bipolar disorder prescribed lithium, antipsychotics or mood stabilizing antiepileptics (valproate, lamotrigine, carbamazepine). Among 40,697 patients with bipolar disorder from 1993 to 2019 identified from a primary care electronic health record database in the UK, 13,385 were new users of mood stabilizing agents (lithium:2339; non-lithium: 11,046). Lithium was associated with a lower risk of fractures compared with non-lithium treatments (HR 0.66, 95 % CI 0.44-0.98). The results were similar when comparing lithium with prolactin raising and sparing antipsychotics, and individual antiepileptics. Lithium use may lower fracture risk, a benefit that is particularly relevant for patients with serious mental illness who are more prone to falls due to their behaviors. Our findings could help inform better treatment decisions for bipolar disorder, and lithium's potential to prevent fractures should be considered for patients at high risk of fractures.

Leveraging Large Language Models for Knowledge-free Weak Supervision in Clinical Natural Language Processing.

The performance of deep learning-based natural language processing systems is based on large amounts of labeled training data which, in the clinical domain, are not easily available or affordable. Weak supervision and in-context learning offer partial solutions to this issue, particularly using large language models (LLMs), but their performance still trails traditional supervised methods with moderate amounts of gold-standard data. In particular, inferencing with LLMs is computationally heavy. We propose an approach leveraging fine-tuning LLMs and weak supervision with virtually no domain knowledge that still achieves consistently dominant performance. Using a prompt-based approach, the LLM is used to generate weakly-labeled data for training a downstream BERT model. The weakly supervised model is then further fine-tuned on small amounts of gold standard data. We evaluate this approach using Llama2 on three different n2c2 datasets. With no more than 10 gold standard notes, our final BERT models weakly supervised by fine-tuned Llama2-13B consistently outperformed out-of-the-box PubMedBERT by 4.7-47.9% in F1 scores. With only 50 gold standard notes, our models achieved close performance to fully fine-tuned systems.

Association of patient photographs and reduced retract-and-reorder events.

Background: Wrong-patient order entry (WPOE) is a potentially dangerous medical error. It remains unknown if patient photographs reduce WPOE in the pediatric inpatient population. Materials and Methods: Order sessions from a single pediatric hospital system were examined for retract-and-reorder (RAR) events, a surrogate WPOE measure. We determined the association of patient photographs with the proportion of order sessions resulting in a RAR event, adjusted for patient, provider, and ordering context. Results: In multivariable analysis, the presence of a patient photo in the electronic health record was associated with 40% lower odds of a RAR event (aOR: 0.60, 95% CI: 0.48-0.75), while cardiac and ICU contexts had higher RAR frequency (aOR: 2.12, 95% CI: 1.69-2.67 and 2.05, 95% CI: 1.71-2.45, respectively). Discussion and Conclusion: Patient photos were associated with lower odds of RAR events in the pediatric inpatient setting, while high acuity locations may be at higher risk. Patient photographs may reduce WPOE without interruptions.

Treatment drop-in in a contemporary cohort used to derive cardiovascular risk prediction equations.

BACKGROUND: No routinely recommended cardiovascular disease (CVD) risk prediction equations have adjusted for CVD preventive medications initiated during follow-up (treatment drop-in) in their derivation cohorts. This will lead to underestimation of risk when equations are applied in clinical practice if treatment drop-in is common. We aimed to quantify the treatment drop-in in a large contemporary national cohort to determine whether equations are likely to require adjustment. METHODS: Eight de-identified individual-level national health administrative datasets in Aotearoa New Zealand were linked to establish a cohort of almost all New Zealanders without CVD and aged 30-74 years in 2006. Individuals dispensing blood-pressure-lowering and/or lipid-lowering medications between 1 July 2006 and 31 December 2006 (baseline dispensing), and in each 6-month period during 12 years' follow-up to 31 December 2018 (follow-up dispensing), were identified. Person-years of treatment drop-in were determined. RESULTS: A total of 1 399 348 (80%) out of the 1 746 695 individuals in the cohort were not dispensed CVD medications at baseline. Blood-pressure-lowering and/or lipid-lowering treatment drop-in accounted for 14% of follow-up time in the group untreated at baseline and increased significantly with increasing predicted baseline 5-year CVD risk (12%, 31%, 34% and 37% in <5%, 5-9%, 10-14% and ≥15% risk groups, respectively) and with increasing age (8% in 30-44 year-olds to 30% in 60-74 year-olds). CONCLUSIONS: CVD preventive treatment drop-in accounted for approximately one-third of follow-up time among participants typically eligible for preventive treatment (≥5% 5-year predicted risk). Equations derived from cohorts with long-term follow-up that do not adjust for treatment drop-in effect will underestimate CVD risk in higher risk individuals and lead to undertreatment. Future CVD risk prediction studies need to address this potential flaw.

Treatment of hypertension during pregnancy: a cohort of pregnancy episodes from the SIDIAP database, Catalonia, Spain.

Introduction: Hypertension during pregnancy is one of the most frequent causes of maternal and fetal morbimortality. Perinatal and maternal death and disability rates have decreased by 30%, but hypertension during pregnancy has increased by approximately 10% in the last 30 years. This research aimed to describe the pharmacological treatment and pregnancy outcomes of pregnancies with hypertension. Methods: We carried out an observational cohort study from the Information System for the Development of Research in Primary Care (SIDIAP) database. Pregnancy episodes with hypertension (ICD-10 codes for hypertension, I10-I15 and O10-O16) were identified. Antihypertensives were classified according to the ATC WHO classification: ß-blocking agents (BBs), calcium channel blockers (CCBs), agents acting on the renin-angiotensin system (RAS agents), diuretics, and antiadrenergic agents. Exposure was defined for hypertension in pregnancies with ≥2 prescriptions during the pregnancy episode. Descriptive statistics for diagnoses and treatments were calculated. Results: In total, 4,839 pregnancies with hypertension diagnosis formed the study cohort. There were 1,944 (40.2%) pregnancies exposed to an antihypertensive medication. There were differences in mother's age, BMI, and alcohol intake between pregnancies exposed to antihypertensive medications and those not exposed. BBs were the most used (n = 1,160 pregnancy episodes; 59.7%), followed by RAS agents (n = 825, 42.4%), and CCBs were the least used (n = 347, 17.8%). Discussion: Pregnancies involving hypertension were exposed to antihypertensive medications, mostly BBs. We conduct a study focused on RAS agent use during pregnancy and its outcomes in the offspring.

Using Routinely Collected Electronic Healthcare Record Data to Investigate Fibrotic Multimorbidity in England.

Background: Electronic healthcare records (EHRs) are used to document diagnoses, symptoms, tests, and prescriptions. Though not primarily collected for research purposes, owing to the size of the data as well as the depth of information collected, they have been used extensively to conduct epidemiological research. The Clinical Practice Research Datalink (CPRD) is an EHR database containing representative data of the UK population with regard to age, sex, race, and social deprivation measures. Fibrotic conditions are characterised by excessive scarring, contributing towards organ dysfunction and eventual organ failure. Fibrosis is associated with ageing as well as many other factors, it is hypothesised that fibrotic conditions are caused by the same underlying pathological mechanism. We calculated the prevalence of fibrotic conditions (as defined in a previous Delphi survey of clinicians) as well as the prevalence of fibrotic multimorbidity (the proportion of people with multiple fibrotic conditions). Methods: We included a random sample of 993,370 UK adults, alive, and enrolled at a UK general practice, providing data to the CPRD Aurum database as of 1st of January 2015. Individuals had to be eligible for linkage to hospital episode statistics (HES) and ONS death registration. We calculated the point prevalence of fibrotic conditions and multi-morbid fibrosis on the 1st of January 2015. Using death records of those who died in 2015, we investigated the prevalence of fibrosis associated death. We explored the most commonly co-occurring fibrotic conditions and determined the settings in which diagnoses were commonly made (primary care, secondary care or after death). Results: The point prevalence of any fibrotic condition was 21.46%. In total, 6.00% of people had fibrotic multimorbidity. Of the people who died in 2015, 34.82% had a recording of a fibrotic condition listed on their death certificate. Conclusion: The key finding was that fibrotic multimorbidity affects approximately 1 in 16 people.

Fibrotic conditions are scarring conditions which impact the way an organ functions and eventually lead to organ failure. We studied routinely collected health data from GPs, hospitals, and death certificates to estimate the percentage of UK adults who had fibrotic diseases. We found that 1 in 5 people had at least one fibrotic disease, and we also found that 1 in 16 people had more than one fibrotic disease.

Leveraging GPT-4 for identifying cancer phenotypes in electronic health records: a performance comparison between GPT-4, GPT-3.5-turbo, Flan-T5, Llama-3-8B, and spaCy's rule-based and machine learning-based methods.

Objective: Accurately identifying clinical phenotypes from Electronic Health Records (EHRs) provides additional insights into patients' health, especially when such information is unavailable in structured data. This study evaluates the application of OpenAI's Generative Pre-trained Transformer (GPT)-4 model to identify clinical phenotypes from EHR text in non-small cell lung cancer (NSCLC) patients. The goal was to identify disease stages, treatments and progression utilizing GPT-4, and compare its performance against GPT-3.5-turbo, Flan-T5-xl, Flan-T5-xxl, Llama-3-8B, and 2 rule-based and machine learning-based methods, namely, scispaCy and medspaCy. Materials and Methods: Phenotypes such as initial cancer stage, initial treatment, evidence of cancer recurrence, and affected organs during recurrence were identified from 13 646 clinical notes for 63 NSCLC patients from Washington University in St. Louis, Missouri. The performance of the GPT-4 model is evaluated against GPT-3.5-turbo, Flan-T5-xxl, Flan-T5-xl, Llama-3-8B, medspaCy, and scispaCy by comparing precision, recall, and micro-F1 scores. Results: GPT-4 achieved higher F1 score, precision, and recall compared to Flan-T5-xl, Flan-T5-xxl, Llama-3-8B, medspaCy, and scispaCy's models. GPT-3.5-turbo performed similarly to that of GPT-4. GPT, Flan-T5, and Llama models were not constrained by explicit rule requirements for contextual pattern recognition. spaCy models relied on predefined patterns, leading to their suboptimal performance. Discussion and Conclusion: GPT-4 improves clinical phenotype identification due to its robust pre-training and remarkable pattern recognition capability on the embedded tokens. It demonstrates data-driven effectiveness even with limited context in the input. While rule-based models remain useful for some tasks, GPT models offer improved contextual understanding of the text, and robust clinical phenotype extraction.

Retrospective analysis of hospital electronic health records reveals unseen cases of acute hepatitis with unknown aetiology in adults in Oxfordshire.

BACKGROUND: An outbreak of acute severe hepatitis of unknown aetiology (AS-Hep-UA) in children during 2022 was subsequently linked to infections with adenovirus-associated virus 2 and other 'helper viruses', including human adenovirus. It is possible that evidence of such an outbreak could be identified at a population level based on routine data captured by electronic health records (EHR). METHODS: We used anonymised EHR to collate retrospective data for all emergency presentations to Oxford University Hospitals NHS Foundation Trust in the UK, between 2016-2022, for all ages from 18 months and older. We investigated clinical characteristics and temporal distribution of presentations of acute hepatitis and of adenovirus infections based on laboratory data and clinical coding. We relaxed the stringent case definition adopted during the AS-Hep-UA to identify all cases of acute hepatitis with unknown aetiology (termed AHUA). We compared events within the outbreak period (defined as 1st Oct 2021-31 Aug 2022) to the rest of our study period. RESULTS: Over the study period, there were 903,433 acute presentations overall, of which 391 (0.04%) were classified as AHUA. AHUA episodes had significantly higher critical care admission rates (p < 0.0001, OR = 41.7, 95% CI:26.3-65.0) and longer inpatient admissions (p < 0.0001) compared with the rest of the patient population. During the outbreak period, significantly more adults (≥ 16 years) were diagnosed with AHUA (p < 0.0001, OR = 3.01, 95% CI: 2.20-4.12), and there were significantly more human adenovirus (HadV) infections in children (p < 0.001, OR = 1.78, 95% CI:1.27-2.47). There were also more HAdV tests performed during the outbreak (p < 0.0001, OR = 1.27, 95% CI:1.17-1.37). Among 3,707 individuals who were tested for HAdV, 179 (4.8%) were positive. However, there was no evidence of more acute hepatitis or increased severity of illness in HadV-positive compared to negative cases. CONCLUSIONS: Our results highlight an increase in AHUA in adults coinciding with the period of the outbreak in children, but not linked to documented HAdV infection. Tracking changes in routinely collected clinical data through EHR could be used to support outbreak surveillance.