Search

AbstractObjectives This study aimed to assess the impact of on-demand versus continuous prescribing of proton pump inhibitors (PPIs) on symptom burden and health-related quality of life in patients with gastroesophageal reflux disease (GERD) presenting to primary care.Methods Thirty-six primary care centres across Europe enrolled adult GERD patients from electronic health records. Participants were randomised to on-demand or continuous PPI prescriptions and were followed for 8 weeks. PPI intake, symptom burden, and quality of life were compared between the two groups using mixed-effect regression analyses. Spearman’s correlation was used to assess the association between changes in PPI dose and patient-reported outcomes.Results A total of 488 patients (median age 51 years, 58% women) completed the initial visit, with 360 attending the follow-up visit. There was no significant difference in PPI use between the continuous and on-demand prescription groups (b=.57, 95%CI:0.40-1.53), although PPI use increased in both groups (b = 1.33, 95%CI:0.65 − 2.01). Advice on prescribing strategy did not significantly affect patient-reported outcomes. Both symptom burden (Reflux Disease Questionnaire, b=-0.61, 95%CI:-0.73 − -0.49) and quality of life (12-item Short Form Survey physical score b = 3.31, 95%CI:2.17 − 4.45) improved from baseline to follow-up in both groups. Increased PPI intake correlated with reduced reflux symptoms (n = 347, ρ=-0.12, p = 0.02) and improved quality of life (n = 217, ρ = 0.16, p = 0.02).Conclusion In real-world settings, both continuous and on-demand PPI prescriptions resulted in similar increases in PPI consumption with no difference in treatment effects. Achieving an adequate PPI dose to alleviate reflux symptom burden improves quality of life in GERD patients. EudraCT number 2014-001314-25.

Abstract Introduction Clinical decision support (CDS) systems (CDSSs) that integrate clinical guidelines need to reflect real‐world co‐morbidity. In patient‐specific clinical contexts, transparent recommendations that allow for contraindications and other conflicts arising from co‐morbidity are a requirement. In this work, we develop and evaluate a non‐proprietary, standards‐based approach to the deployment of computable guidelines with explainable argumentation, integrated with a commercial electronic health record (EHR) system in Serbia, a middle‐income country in West Balkans. Methods We used an ontological framework, the Transition‐based Medical Recommendation (TMR) model, to represent, and reason about, guideline concepts, and chose the 2017 International global initiative for chronic obstructive lung disease (GOLD) guideline and a Serbian hospital as the deployment and evaluation site, respectively. To mitigate potential guideline conflicts, we used a TMR‐based implementation of the Assumptions‐Based Argumentation framework extended with preferences and Goals (ABA+G). Remote EHR integration of computable guidelines was via a microservice architecture based on HL7 FHIR and CDS Hooks. A prototype integration was developed to manage chronic obstructive pulmonary disease (COPD) with comorbid cardiovascular or chronic kidney diseases, and a mixed‐methods evaluation was conducted with 20 simulated cases and five pulmonologists. Results Pulmonologists agreed 97% of the time with the GOLD‐based COPD symptom severity assessment assigned to each patient by the CDSS, and 98% of the time with one of the proposed COPD care plans. Comments were favourable on the principles of explainable argumentation; inclusion of additional co‐morbidities was suggested in the future along with customisation of the level of explanation with expertise. Conclusion An ontological model provided a flexible means of providing argumentation and explainable artificial intelligence for a long‐term condition. Extension to other guidelines and multiple co‐morbidities is needed to test the approach further.

Abstract Open and practical exchange, dissemination, and reuse of specimens and data have become a fundamental requirement for life sciences research. The quality of the data obtained and thus the findings and knowledge derived is thus significantly influenced by the quality of the samples, the experimental methods, and the data analysis. Therefore, a comprehensive and precise documentation of the pre‐analytical conditions, the analytical procedures, and the data processing are essential to be able to assess the validity of the research results. With the increasing importance of the exchange, reuse, and sharing of data and samples, procedures are required that enable cross‐organizational documentation, traceability, and non‐repudiation. At present, this information on the provenance of samples and data is mostly either sparse, incomplete, or incoherent. Since there is no uniform framework, this information is usually only provided within the organization and not interoperably. At the same time, the collection and sharing of biological and environmental specimens increasingly require definition and documentation of benefit sharing and compliance to regulatory requirements rather than consideration of pure scientific needs. In this publication, we present an ongoing standardization effort to provide trustworthy machine‐actionable documentation of the data lineage and specimens. We would like to invite experts from the biotechnology and biomedical fields to further contribute to the standard.

Objectives We aimed to develop and externally validate a generalisable risk prediction model for 30-day stroke mortality suitable for supporting quality improvement analytics in stroke care using large nationwide stroke registers in the UK and Sweden.Design Registry-based cohort study.Setting Stroke registries including the Sentinel Stroke National Audit Programme (SSNAP) in England, Wales and Northern Ireland (2013–2019) and the national Swedish stroke register (Riksstroke 2015–2020).Participants and methods Data from SSNAP were used for developing and temporally validating the model, and data from Riksstroke were used for external validation. Models were developed with the variables available in both registries using logistic regression (LR), LR with elastic net and interaction terms and eXtreme Gradient Boosting (XGBoost). Performances were evaluated with discrimination, calibration and decision curves.Outcome measures The primary outcome was all-cause 30-day in-hospital mortality after stroke.Results In total, 488 497 patients who had a stroke with 12.4% 30-day in-hospital mortality were used for developing and temporally validating the model in the UK. A total of 128 360 patients who had a stroke with 10.8% 30-day in-hospital mortality and 13.1% all mortality were used for external validation in Sweden. In the SSNAP temporal validation set, the final XGBoost model achieved the highest area under the receiver operating characteristic curve (AUC) (0.852 (95% CI 0.848 to 0.855)) and was well calibrated. The performances on the external validation in Riksstroke were as good and achieved AUC at 0.861 (95% CI 0.858 to 0.865) for in-hospital mortality. For Riksstroke, the models slightly overestimated the risk for in-hospital mortality, while they were better calibrated at the risk for all mortality.Conclusion The risk prediction model was accurate and externally validated using high quality registry data. This is potentially suitable to be deployed as part of quality improvement analytics in stroke care to enable the fair comparison of stroke mortality outcomes across hospitals and health systems across countries

Introduction A substantial proportion of COVID-19 survivors continue to have symptoms more than 3 months after infection, especially of those who required medical intervention. Lasting symptoms are wide-ranging, and presentation varies between individuals and fluctuates within an individual. Improved understanding of undulation in symptoms and triggers may improve efficacy of healthcare providers and enable individuals to better self-manage their Long Covid. We present a protocol where we aim to develop and examine the feasibility and usability of digital home monitoring for capturing daily fluctuation of symptoms in individuals with Long Covid and provide data to facilitate a personalised approach to the classification and management of Long Covid symptoms.Methods and analysis This study is a longitudinal prospective cohort study of adults with Long Covid accessing 10 National Health Service (NHS) rehabilitation services in the UK. We aim to recruit 400 people from participating NHS sites. At referral to study, 6 weeks and 12 weeks, participants will complete demographic data (referral to study) and clinical outcome measures, including ecological momentary assessment (EMA) using personal mobile devices. EMA items are adapted from the COVID-19 Yorkshire Rehabilitation Scale items and include self-reported activities, symptoms and psychological factors. Passive activity data will be collected through wrist-worn sensors. We will use latent class growth models to identify trajectories of experience, potential phenotypes defined by co-occurrence of symptoms and inter-relationships between stressors, symptoms and participation in daily activities. We anticipate that n=300 participants provide 80% power to detect a 20% improvement in fatigue over 12 weeks in one class of patients relative to another.Ethics and dissemination The study was approved by the Yorkshire & The Humber—Bradford Leeds Research Ethics Committee (ref: 21/YH/0276). Findings will be disseminated in peer-reviewed publications and presented at conferences.Trial registration number ISRCTN15022307.

BackgroundPalliative and end-of-life care (PEoLC) played a critical role in relieving distress and providing grief support in response to the heavy toll caused by the COVID-19 pandemic. However, little is known about public opinions concerning PEoLC during the pandemic. Given that social media have the potential to collect real-time public opinions, an analysis of this evidence is vital to guide future policy-making. ObjectiveThis study aimed to use social media data to investigate real-time public opinions regarding PEoLC during the COVID-19 crisis and explore the impact of vaccination programs on public opinions about PEoLC. MethodsThis Twitter-based study explored tweets across 3 English-speaking countries: the United States, the United Kingdom, and Canada. From October 2020 to March 2021, a total of 7951 PEoLC-related tweets with geographic tags were retrieved and identified from a large-scale COVID-19 Twitter data set through the Twitter application programming interface. Topic modeling realized through a pointwise mutual information–based co-occurrence network and Louvain modularity was used to examine latent topics across the 3 countries and across 2 time periods (pre- and postvaccination program periods). ResultsCommonalities and regional differences among PEoLC topics in the United States, the United Kingdom, and Canada were identified specifically: cancer care and care facilities were of common interest to the public across the 3 countries during the pandemic; the public expressed positive attitudes toward the COVID-19 vaccine and highlighted the protection it affords to PEoLC professionals; and although Twitter users shared their personal experiences about PEoLC in the web-based community during the pandemic, this was more prominent in the United States and Canada. The implementation of the vaccination programs raised the profile of the vaccine discussion; however, this did not influence public opinions about PEoLC. ConclusionsPublic opinions on Twitter reflected a need for enhanced PEoLC services during the COVID-19 pandemic. The insignificant impact of the vaccination program on public discussion on social media indicated that public concerns regarding PEoLC continued to persist even after the vaccination efforts. Insights gleaned from public opinions regarding PEoLC could provide some clues for policy makers on how to ensure high-quality PEoLC during public health emergencies. In this post–COVID-19 era, PEoLC professionals may wish to continue to examine social media and learn from web-based public discussion how to ease the long-lasting trauma caused by this crisis and prepare for public health emergencies in the future. Besides, our results showed social media’s potential in acting as an effective tool to reflect public opinions in the context of PEoLC.

Abstract Introduction Randomised controlled trials have shown that steroids reduce the risk of dying in patients with severe Coronavirus disease 2019 (COVID-19), whilst many real-world studies have failed to replicate this result. We aim to investigate real-world effectiveness of steroids in severe COVID-19. Methods Clinical, demographic, and viral genome data extracted from electronic patient record (EPR) was analysed from all SARS-CoV-2 RNA positive patients admitted with severe COVID-19, defined by hypoxia at presentation, between March 13th 2020 and May 27th 2021. Steroid treatment was measured by the number of prescription-days with dexamethasone, hydrocortisone, prednisolone or methylprednisolone. The association between steroid > 3 days treatment and disease outcome was explored using multivariable cox proportional hazards models with adjustment for confounders (including age, gender, ethnicity, co-morbidities and SARS-CoV-2 variant). The outcome was in-hospital mortality. Results 1100 severe COVID-19 cases were identified having crude hospital mortality of 15.3%. 793/1100 (72.1%) individuals were treated with steroids and 513/1100 (46.6%) received steroid ≤ 3 days. From the multivariate model, steroid > 3 days was associated with decreased hazard of in-hospital mortality (HR: 0.47 (95% CI: 0.31–0.72)). Conclusion The protective effect of steroid treatment for severe COVID-19 reported in randomised clinical trials was replicated in this retrospective study of a large real-world cohort.

Abstract Backgrounds We aimed to develop and validate machine learning (ML) models for 30-day stroke mortality for mortality risk stratification and as benchmarking models for quality improvement in stroke care. Methods Data from the UK Sentinel Stroke National Audit Program between 2013 to 2019 were used. Models were developed using XGBoost, Logistic Regression (LR), LR with elastic net with/without interaction terms using 80% randomly selected admissions from 2013 to 2018, validated on the 20% remaining admissions, and temporally validated on 2019 admissions. The models were developed with 30 variables. A reference model was developed using LR and 4 variables. Performances of all models was evaluated in terms of discrimination, calibration, reclassification, Brier scores and Decision-curves. Results In total, 488,497 stroke patients with a 12.3% 30-day mortality rate were included in the analysis. In 2019 temporal validation set, XGBoost model obtained the lowest Brier score (0.069 (95% CI: 0.068–0.071)) and the highest area under the ROC curve (AUC) (0.895 (95% CI: 0.891–0.900)) which outperformed LR reference model by 0.04 AUC (p 15%). The XGBoost model reclassified 1648 (8.1%) low-risk cases by the LR reference model as being moderate or high-risk and gained the most net benefit in decision curve analysis. Conclusions All models with 30 variables are potentially useful as benchmarking models in stroke-care quality improvement with ML slightly outperforming others.

Abstract Background Within-consultation recruitment to primary care trials is challenging. Ensuring procedures are efficient and self-explanatory is the key to optimising recruitment. Trial recruitment software that integrates with the electronic health record to support and partially automate procedures is becoming more common. If it works well, such software can support greater participation and more efficient trial designs. An innovative electronic trial recruitment and outcomes software was designed to support recruitment to the Runny Ear randomised controlled trial, comparing topical, oral and delayed antibiotic treatment for acute otitis media with discharge in children. A qualitative evaluation investigated the views and experiences of primary care staff using this trial software. Methods Staff were purposively sampled in relation to site, role and whether the practice successfully recruited patients. In-depth interviews were conducted using a flexible topic guide, audio recorded and transcribed. Data were analysed thematically. Results Sixteen staff were interviewed, including GPs, practice managers, information technology (IT) leads and research staff. GPs wanted trial software that automatically captures patient data. However, the experience of getting the software to work within the limited and complex IT infrastructure of primary care was frustrating and time consuming. Installation was reliant on practice level IT expertise, which varied between practices. Although most had external IT support, this rarely included supported for research IT. Arrangements for approving new software varied across practices and often, but not always, required authorisation from Clinical Commissioning Groups. Conclusions Primary care IT systems are not solely under the control of individual practices or CCGs or the National Health Service. Rather they are part of a complex system that spans all three and is influenced by semi-autonomous stakeholders operating at different levels. This led to time consuming and sometimes insurmountable barriers to installation at the practice level. These need to be addressed if software supporting efficient research in primary care is to become a reality.

BackgroundFollowing COVID-19, up to 40% of people have ongoing health problems, referred to as postacute COVID-19 or long COVID (LC). LC varies from a single persisting symptom to a complex multisystem disease. Research has flagged that this condition is underrecorded in primary care records, and seeks to better define its clinical characteristics and management. Phenotypes provide a standard method for case definition and identification from routine data and are usually machine-processable. An LC phenotype can underpin research into this condition. ObjectiveThis study aims to develop a phenotype for LC to inform the epidemiology and future research into this condition. We compared clinical symptoms in people with LC before and after their index infection, recorded from March 1, 2020, to April 1, 2021. We also compared people recorded as having acute infection with those with LC who were hospitalized and those who were not. MethodsWe used data from the Primary Care Sentinel Cohort (PCSC) of the Oxford Royal College of General Practitioners (RCGP) Research and Surveillance Centre (RSC) database. This network was recruited to be nationally representative of the English population. We developed an LC phenotype using our established 3-step ontological method: (1) ontological step (defining the reasoning process underpinning the phenotype, (2) coding step (exploring what clinical terms are available, and (3) logical extract model (testing performance). We created a version of this phenotype using Protégé in the ontology web language for BioPortal and using PhenoFlow. Next, we used the phenotype to compare people with LC (1) with regard to their symptoms in the year prior to acquiring COVID-19 and (2) with people with acute COVID-19. We also compared hospitalized people with LC with those not hospitalized. We compared sociodemographic details, comorbidities, and Office of National Statistics–defined LC symptoms between groups. We used descriptive statistics and logistic regression. ResultsThe long-COVID phenotype differentiated people hospitalized with LC from people who were not and where no index infection was identified. The PCSC (N=7.4 million) includes 428,479 patients with acute COVID-19 diagnosis confirmed by a laboratory test and 10,772 patients with clinically diagnosed COVID-19. A total of 7471 (1.74%, 95% CI 1.70-1.78) people were coded as having LC, 1009 (13.5%, 95% CI 12.7-14.3) had a hospital admission related to acute COVID-19, and 6462 (86.5%, 95% CI 85.7-87.3) were not hospitalized, of whom 2728 (42.2%) had no COVID-19 index date recorded. In addition, 1009 (13.5%, 95% CI 12.73-14.28) people with LC were hospitalized compared to 17,993 (4.5%, 95% CI 4.48-4.61; P

BackgroundMost studies of long COVID (symptoms of COVID-19 infection beyond 4 weeks) have focused on people hospitalized in their initial illness. Long COVID is thought to be underrecorded in UK primary care electronic records. ObjectiveWe sought to determine which symptoms people present to primary care after COVID-19 infection and whether presentation differs in people who were not hospitalized, as well as post–long COVID mortality rates. MethodsWe used routine data from the nationally representative primary care sentinel cohort of the Oxford–Royal College of General Practitioners Research and Surveillance Centre (N=7,396,702), applying a predefined long COVID phenotype and grouped by whether the index infection occurred in hospital or in the community. We included COVID-19 infection cases from March 1, 2020, to April 1, 2021. We conducted a before-and-after analysis of long COVID symptoms prespecified by the Office of National Statistics, comparing symptoms presented between 1 and 6 months after the index infection matched with the same months 1 year previously. We conducted logistic regression analysis, quoting odds ratios (ORs) with 95% CIs. ResultsIn total, 5.63% (416,505/7,396,702) and 1.83% (7623/416,505) of the patients had received a coded diagnosis of COVID-19 infection and diagnosis of, or referral for, long COVID, respectively. People with diagnosis or referral of long COVID had higher odds of presenting the prespecified symptoms after versus before COVID-19 infection (OR 2.66, 95% CI 2.46-2.88, for those with index community infection and OR 2.42, 95% CI 2.03-2.89, for those hospitalized). After an index community infection, patients were more likely to present with nonspecific symptoms (OR 3.44, 95% CI 3.00-3.95; P

Abstract Background Well-established electronic data capture in UK general practice means that algorithms, developed on patient data, can be used for automated clinical decision support systems (CDSSs). These can predict patient risk, help with prescribing safety, improve diagnosis and prompt clinicians to record extra data. However, there is persistent evidence of low uptake of CDSSs in the clinic. We interviewed UK General Practitioners (GPs) to understand what features of CDSSs, and the contexts of their use, facilitate or present barriers to their use. Methods We interviewed 11 practicing GPs in London and South England using a semi-structured interview schedule and discussed a hypothetical CDSS that could detect early signs of dementia. We applied thematic analysis to the anonymised interview transcripts. Results We identified three overarching themes: trust in individual CDSSs; usability of individual CDSSs; and usability of CDSSs in the broader practice context, to which nine subthemes contributed. Trust was affected by CDSS provenance, perceived threat to autonomy and clear management guidance. Usability was influenced by sensitivity to the patient context, CDSS flexibility, ease of control, and non-intrusiveness. CDSSs were more likely to be used by GPs if they did not contribute to alert proliferation and subsequent fatigue, or if GPs were provided with training in their use. Conclusions Building on these findings we make a number of recommendations for CDSS developers to consider when bringing a new CDSS into GP patient records systems. These include co-producing CDSS with GPs to improve fit within clinic workflow and wider practice systems, ensuring a high level of accuracy and a clear clinical pathway, and providing CDSS training for practice staff. These recommendations may reduce the proliferation of unhelpful alerts that can result in important decision-support being ignored.

Abstract Background The National Early Warning Score (NEWS2) is currently recommended in the UK for the risk stratification of COVID-19 patients, but little is known about its ability to detect severe cases. We aimed to evaluate NEWS2 for the prediction of severe COVID-19 outcome and identify and validate a set of blood and physiological parameters routinely collected at hospital admission to improve upon the use of NEWS2 alone for medium-term risk stratification. Methods Training cohorts comprised 1276 patients admitted to King’s College Hospital National Health Service (NHS) Foundation Trust with COVID-19 disease from 1 March to 30 April 2020. External validation cohorts included 6237 patients from five UK NHS Trusts (Guy’s and St Thomas’ Hospitals, University Hospitals Southampton, University Hospitals Bristol and Weston NHS Foundation Trust, University College London Hospitals, University Hospitals Birmingham), one hospital in Norway (Oslo University Hospital), and two hospitals in Wuhan, China (Wuhan Sixth Hospital and Taikang Tongji Hospital). The outcome was severe COVID-19 disease (transfer to intensive care unit (ICU) or death) at 14 days after hospital admission. Age, physiological measures, blood biomarkers, sex, ethnicity, and comorbidities (hypertension, diabetes, cardiovascular, respiratory and kidney diseases) measured at hospital admission were considered in the models. Results A baseline model of ‘NEWS2 + age’ had poor-to-moderate discrimination for severe COVID-19 infection at 14 days (area under receiver operating characteristic curve (AUC) in training cohort = 0.700, 95% confidence interval (CI) 0.680, 0.722; Brier score = 0.192, 95% CI 0.186, 0.197). A supplemented model adding eight routinely collected blood and physiological parameters (supplemental oxygen flow rate, urea, age, oxygen saturation, C-reactive protein, estimated glomerular filtration rate, neutrophil count, neutrophil/lymphocyte ratio) improved discrimination (AUC = 0.735; 95% CI 0.715, 0.757), and these improvements were replicated across seven UK and non-UK sites. However, there was evidence of miscalibration with the model tending to underestimate risks in most sites. Conclusions NEWS2 score had poor-to-moderate discrimination for medium-term COVID-19 outcome which raises questions about its use as a screening tool at hospital admission. Risk stratification was improved by including readily available blood and physiological parameters measured at hospital admission, but there was evidence of miscalibration in external sites. This highlights the need for a better understanding of the use of early warning scores for COVID.

Abstract Background Acute otitis media (AOM) is a common painful infection in children, with around 2.8 million cases presenting to primary care in England and Wales annually. Nearly all children who present to their general practitioner (GP) with AOM or AOM with discharge (AOMd) are treated with orally administered antibiotics. These can cause side effects; contribute to the growing problem of antimicrobial resistance, and more rarely, allergic reactions. Alternative treatments, such as an antibiotic eardrops, or ‘delayed’ orally administered antibiotics, could be at least as effective and safe as immediate orally administered antibiotics for children with AOMd. Methods/design REST is a pragmatic, three-arm, individually randomised, non-inferiority trial being conducted in 175 GP practices across the United Kingdom (UK). The study aims to recruit 399 children aged (≥ 12 months and

Background: Acute otitis media is a painful infection of the middle ear that is commonly seen in children. In some children, the eardrum spontaneously bursts, discharging visible pus (otorrhoea) into the outer ear. Objective: To compare the clinical effectiveness of immediate topical antibiotics or delayed oral antibiotics with the clinical effectiveness of immediate oral antibiotics in reducing symptom duration in children presenting to primary care with acute otitis media with discharge and the economic impact of the alternative strategies. Design: This was a pragmatic, three-arm, individually randomised (stratified by age

BackgroundMental health problems are widely recognized as a major public health challenge worldwide. This concern highlights the need to develop effective tools for detecting mental health disorders in the population. Social networks are a promising source of data wherein patients publish rich personal information that can be mined to extract valuable psychological cues; however, these data come with their own set of challenges, such as the need to disambiguate between statements about oneself and third parties. Traditionally, natural language processing techniques for social media have looked at text classifiers and user classification models separately, hence presenting a challenge for researchers who want to combine text sentiment and user sentiment analysis. ObjectiveThe objective of this study is to develop a predictive model that can detect users with depression from Twitter posts and instantly identify textual content associated with mental health topics. The model can also address the problem of anaphoric resolution and highlight anaphoric interpretations. MethodsWe retrieved the data set from Twitter by using a regular expression or stream of real-time tweets comprising 3682 users, of which 1983 self-declared their depression and 1699 declared no depression. Two multiple instance learning models were developed—one with and one without an anaphoric resolution encoder—to identify users with depression and highlight posts related to the mental health of the author. Several previously published models were applied to our data set, and their performance was compared with that of our models. ResultsThe maximum accuracy, F1 score, and area under the curve of our anaphoric resolution model were 92%, 92%, and 90%, respectively. The model outperformed alternative predictive models, which ranged from classical machine learning models to deep learning models. ConclusionsOur model with anaphoric resolution shows promising results when compared with other predictive models and provides valuable insights into textual content that is relevant to the mental health of the tweeter.

Background: The threat of antimicrobial resistance has led to intensified efforts to reduce antibiotic utilisation, but serious bacterial infections are increasing in frequency. Objectives: To estimate the risks of serious bacterial infections in association with lower antibiotic prescribing and understand stakeholder views with respect to safe antibiotic reduction. Design: Mixed-methods research was undertaken, including a qualitative interview study of patient and prescriber views that informed a cohort study and a decision-analytic model, using primary care electronic health records. These three work packages were used to design an application (app) for primary care prescribers. Data sources: The Clinical Practice Research Datalink. Setting: This took place in UK general practices. Participants: A total of 706 general practices with 66.2 million person-years of follow-up from 2002 to 2017 and antibiotic utilisation evaluated for 671,830 registered patients. The qualitative study included 31 patients and 30 health-care professionals from primary care. Main outcome measures: Sepsis and localised bacterial infections. Results: Patients were concerned about antimicrobial resistance and the side effects, as well as the benefits, of antibiotic treatment. Prescribers viewed the onset of sepsis as the most concerning potential outcome of reduced antibiotic prescribing. More than 40% of antibiotic prescriptions in primary care had no coded indication recorded across both Vision® and EMIS® practice systems. Antibiotic prescribing rates varied widely between general practices, but there was no evidence that serious bacterial infections were less frequent at higher prescribing practices (adjusted rate ratio for 20% increase in prescribing 1.03, 95% confidence interval 1.00 to 1.06; p = 0.074). The probability of sepsis was lower if an antibiotic was prescribed at an infection consultation, and the number of antibiotic prescriptions required to prevent one episode of sepsis (i.e. the number needed to treat) decreased with age. For those aged 0–4 years, the number needed to treat was 29,773 (95% uncertainty interval 18,458 to 71,091) in boys and 27,014 (95% uncertainty interval 16,739 to 65,709) in girls. For those aged > 85 years, the number needed to treat was 262 (95% uncertainty interval 236 to 293) in men and 385 (95% uncertainty interval 352 to 421) in women. Frailty was associated with a greater risk of sepsis and a smaller number needed to treat. For severely frail patients aged 55–64 years, the number needed to treat was 247 (95% uncertainty interval 156 to 459) for men and 343 (95% uncertainty interval 234 to 556) for women. At all ages, the probability of sepsis was greatest for urinary tract infection, followed by skin infection and respiratory tract infection. The numbers needed to treat were generally smaller for the period 2014–17, when sepsis was diagnosed more frequently. The results are available using an app that we developed to provide primary care prescribers with stratified risk estimates during infection consultations. Limitations: Analyses were based on non-randomised comparisons. Infection episodes and antibiotic prescribing are poorly documented in primary care. Conclusions: Antibiotic treatment is generally associated with lower risks, but the most serious bacterial infections remain infrequent even without antibiotic treatment. This research identifies risk strata in which antibiotic prescribing can be more safely reduced. Future work: The software developed from this research may be further developed and investigated for antimicrobial stewardship effect. Funding: This project was funded by the National Institute for Health Research (NIHR) Health Services and Delivery Research programme and will be published in full in Health Services and Delivery Research; Vol. 9, No. 9. See the NIHR Journals Library website for further project information.

Background: People of minority ethnic backgrounds may be disproportionately affected by severe COVID-19. Whether this relates to increased infection risk, more severe disease progression, or worse in-hospital survival is unknown. The contribution of comorbidities or socioeconomic deprivation to ethnic patterning of outcomes is also unclear. Methods: We conducted a case-control and a cohort study in an inner city primary and secondary care setting to examine whether ethnic background affects the risk of hospital admission with severe COVID-19 and/or in-hospital mortality. Inner city adult residents admitted to hospital with confirmed COVID-19 (n = 872 cases) were compared with 3,488 matched controls randomly sampled from a primary healthcare database comprising 344,083 people residing in the same region. For the cohort study, we studied 1827 adults consecutively admitted with COVID-19. The primary exposure variable was self-defined ethnicity. Analyses were adjusted for socio-demographic and clinical variables. Findings: The 872 cases comprised 48.1% Black, 33.7% White, 12.6% Mixed/Other and 5.6% Asian patients. In conditional logistic regression analyses, Black and Mixed/Other ethnicity were associated with higher admission risk than white (OR 3.12 [95% CI 2.63–3.71] and 2.97 [2.30–3.85] respectively). Adjustment for comorbidities and deprivation modestly attenuated the association (OR 2.24 [1.83–2.74] for Black, 2.70 [2.03–3.59] for Mixed/Other). Asian ethnicity was not associated with higher admission risk (adjusted OR 1.01 [0.70–1.46]). In the cohort study of 1827 patients, 455 (28.9%) died over a median (IQR) of 8 (4–16) days. Age and male sex, but not Black (adjusted HR 1.06 [0.82–1.37]) or Mixed/Other ethnicity (adjusted HR 0.72 [0.47–1.10]), were associated with in-hospital mortality. Asian ethnicity was associated with higher in-hospital mortality but with a large confidence interval (adjusted HR 1.71 [1.15–2.56]). Interpretation: Black and Mixed ethnicity are independently associated with greater admission risk with COVID-19 and may be risk factors for development of severe disease, but do not affect in-hospital mortality risk. Comorbidities and socioeconomic factors only partly account for this and additional ethnicity-related factors may play a large role. The impact of COVID-19 may be different in Asians. Funding: British Heart Foundation; the National Institute for Health Research; Health Data Research UK.

Abstract Human phenomics responds to an urgent need in the medical research community; namely, reproducibility.

ObjectiveThis study evaluated whether serious bacterial infections are more frequent at family practices with lower antibiotic prescribing rates.DesignCohort study.Setting706 UK family practices in the Clinical Practice Research Datalink from 2002 to 2017.Participants10.1 million registered patients with 69.3 million patient-years’ follow-up.ExposuresAll antibiotic prescriptions, subgroups of acute and repeat antibiotic prescriptions, and proportion of antibiotic prescriptions associated with specific-coded indications.Main outcome measuresFirst episodes of serious bacterial infections. Poisson models were fitted adjusting for age group, gender, comorbidity, deprivation, region and calendar year, with random intercepts representing family practice-specific estimates.ResultsThe age-standardised antibiotic prescribing rate per 1000 patient-years increased from 2002 (male 423; female 621) to 2012 (male 530; female 842) before declining to 2017 (male 449; female 753). The median family practice had an antibiotic prescribing rate of 648 per 1000 patient-years with 95% range for different practices of 430–1038 antibiotic prescriptions per 1000 patient-years. Specific coded indications were recorded for 58% of antibiotic prescriptions at the median family practice, the 95% range at different family practices was from 10% to 75%. There were 139 759 first episodes of serious bacterial infection. After adjusting for covariates and the proportion of coded consultations, there was no evidence that serious bacterial infections were lower at family practices with higher total antibiotic prescribing. The adjusted rate ratio for 20% higher total antibiotic prescribing was 1.03, (95% CI 1.00 to 1.06, p=0.074).ConclusionsWe did not find population-level evidence that family practices with lower total antibiotic prescribing might have more frequent occurrence of serious bacterial infections overall. Improving the recording of infection episodes has potential to inform better antimicrobial stewardship in primary care.

Background and purposeMachine learning (ML) has attracted much attention with the hope that it could make use of large, routinely collected datasets and deliver accurate personalised prognosis. The aim of this systematic review is to identify and critically appraise the reporting and developing of ML models for predicting outcomes after stroke.MethodsWe searched PubMed and Web of Science from 1990 to March 2019, using previously published search filters for stroke, ML, and prediction models. We focused on structured clinical data, excluding image and text analysis. This review was registered with PROSPERO (CRD42019127154).ResultsEighteen studies were eligible for inclusion. Most studies reported less than half of the terms in the reporting quality checklist. The most frequently predicted stroke outcomes were mortality (7 studies) and functional outcome (5 studies). The most commonly used ML methods were random forests (9 studies), support vector machines (8 studies), decision trees (6 studies), and neural networks (6 studies). The median sample size was 475 (range 70-3184), with a median of 22 predictors (range 4-152) considered. All studies evaluated discrimination with thirteen using area under the ROC curve whilst calibration was assessed in three. Two studies performed external validation. None described the final model sufficiently well to reproduce it.ConclusionsThe use of ML for predicting stroke outcomes is increasing. However, few met basic reporting standards for clinical prediction tools and none made their models available in a way which could be used or evaluated. Major improvements in ML study conduct and reporting are needed before it can meaningfully be considered for practice.

ObjectivesEffective secondary stroke prevention strategies are suboptimally used. Novel development of interventions to enable healthcare professionals and stroke survivors to manage risk factors for stroke recurrence are required. We sought to engage key stakeholders in the design and evaluation of an intervention informed by a learning health system approach, to improve risk factor management and secondary prevention for stroke survivors with multimorbidity.DesignQualitative, including focus groups, semistructured interviews and usability evaluations. Data was audio recorded, transcribed and coded thematically.ParticipantsStroke survivors, carers, health and social care professionals, commissioners, policymakers and researchers.SettingStroke survivors were recruited from the South London Stroke Register; health and social care professionals through South London general practices and King’s College London (KCL) networks; carers, commissioners, policymakers and researchers through KCL networks.Results53 stakeholders in total participated in focus groups, interviews and usability evaluations. Thirty-seven participated in focus groups and interviews, including stroke survivors and carers (n=11), health and social care professionals (n=16), commissioners and policymakers (n=6) and researchers (n=4). Sixteen participated in usability evaluations, including stroke survivors (n=8) and general practitioners (GPs; n=8). Eight themes informed the collaborative design of DOTT (Deciding On Treatments Together), a decision aid integrated with the electronic health record system, to be used in primary care during clinical consultations between the healthcare professional and stroke survivor. DOTT aims to facilitate shared decision-making on personalised treatments leading to improved treatment adherence and risk control. DOTT was found acceptable and usable among stroke survivors and GPs during a series of evaluations.ConclusionsAdopting a user-centred data-driven design approach informed an intervention that is acceptable to users and has the potential to improve patient outcomes. A future feasibility study and subsequent clinical trial will provide evidence of the effectiveness of DOTT in reducing risk of stroke recurrence.

Abstract The last 6 years have seen sustained investment in health data science in the United Kingdom and beyond, which should result in a data science community that is inclusive of all stakeholders, working together to use data to benefit society through the improvement of public health and well‐being. However, opportunities made possible through the innovative use of data are still not being fully realised, resulting in research inefficiencies and avoidable health harms. In this paper, we identify the most important barriers to achieving higher productivity in health data science. We then draw on previous research, domain expertise, and theory to outline how to go about overcoming these barriers, applying our core values of inclusivity and transparency. We believe a step change can be achieved through meaningful stakeholder involvement at every stage of research planning, design, and execution and team‐based data science, as well as harnessing novel and secure data technologies. Applying these values to health data science will safeguard a social licence for health data research and ensure transparent and secure data usage for public benefit.

Objectives Integrating Electronic Health Record (EHR) systems into the field of clinical trials still contains several challenges and obstacles. Heterogeneous standards and specifications are used to represent healthcare and clinical trial information. Therefore, this work investigates the mapping and data interoperability between healthcare and research standards: EN13606 used for the EHRs and the Clinical Data Interchange Standards Consortium Operational Data Model (CDISC ODM) used for clinical research. Methods Based on the specifications of CDISC ODM 1.3.2 and EN13606, a mapping between the structure and components of both standards has been performed. Archetype Definition Language (ADL) forms built with the EN13606 editor were transformed to ODM XML and reviewed. As a proof of concept, clinical sample data has been transformed into ODM and imported into an electronic data capture system. Reverse transformation from ODM to ADL has also been performed and finally reviewed concerning map-ability. Results The mapping between EN13606 and CDISC ODM shows the similarities and differences between the components and overall record structure of the two standards. An EN13606 archetype corresponds with a group of items within CDISC ODM. Transformations of element names, descriptions, different languages, datatypes, cardinality, optionality, units, value range and terminology codes are possible from EN13606 to CDISC ODM and vice versa. Conclusion It is feasible to map data elements between EN13606 and CDISC ODM and transformation of forms between ADL and ODM XML format is possible with only minor limitations. EN13606 can accommodate clinical information in a more structured manner with more constraints, whereas CDISC ODM is more suitable and specific for clinical trials and studies. It is feasible to transform EHR data in the EN13606 form to ODM to transfer it into research database. The attempt to use EN13606 to build a study protocol (that was already built with CDISC ODM) also suggests the possibility of using EN13606 standard in place of CDISC ODM if needed to avoid transformations.

BACKGROUND:Stroke, like many long-term conditions, tends to be managed in isolation of its associated risk factors and multimorbidity. With increasing access to clinical and research data there is the potential to combine data from a variety of sources to inform interventions to improve healthcare. A 'Learning Health System' (LHS) is an innovative model of care which transforms integrated data into knowledge to improve healthcare. The objective of this study is to develop a process of engaging stakeholders in the use of clinical and research data to co-produce potential solutions, informed by a LHS, to improve long-term care for stroke survivors with multimorbidity. METHODS:We used a stakeholder engagement study design informed by co-production principles to engage stakeholders, including service users, carers, general practitioners and other health and social care professionals, service managers, commissioners of services, policy makers, third sector representatives and researchers. Over a 10 month period we used a range of methods including stakeholder group meetings, focus groups, nominal group techniques (priority setting and consensus building) and interviews. Qualitative data were recorded, transcribed and analysed thematically. RESULTS:37 participants took part in the study. The concept of how data might drive intervention development was difficult to convey and understand. The engagement process led to four priority areas for needs for data and information being identified by stakeholders: 1) improving continuity of care; 2) improving management of mental health consequences; 3) better access to health and social care; and 4) targeting multiple risk factors. These priorities informed preliminary design interventions. The final choice of intervention was agreed by consensus, informed by consideration of the gap in evidence and local service provision, and availability of robust data. This shaped a co-produced decision support tool to improve secondary prevention after stroke for further development. CONCLUSIONS:Stakeholder engagement to identify data-driven solutions is feasible but requires resources. While a number of potential interventions were identified, the final choice rested not just on stakeholder priorities but also on data availability. Further work is required to evaluate the impact and implementation of data-driven interventions for long-term stroke survivors.

Introduction Health research using routinely collected National Health Service (NHS) data derived from electronic health records (EHRs) and health service information systems has been growing in both importance and quantity. Wide population coverage and detailed patient-level information allow this data to be applied to a variety of research questions. However, the sensitivity, complexity and scale of such data also hamper researchers from fully exploiting this potential.Objective Here, we establish the current challenges preventing researchers from making optimal use of the data sets at their disposal, on both the legislative and practical levels, and give recommendations as to how these challenges can be overcome.Method A number of projects has recently been launched in the UK to address poor research data management practices. Rapid Organisation of Healthcare Research Data (ROHRD) at Imperial College, London produced a useful prototype that provides local researchers with a one-stop index of available data sets together with relevant metadata.Findings Increased transparency of data sets’ availability and their provenance leads to better utilisation and facilitates compliance with regulatory requirements.Discussion Research data resulting from NHS data is often not utilised fully, or is handled in a haphazard manner that prevents full auditability of the research. Furthermore, lack of informatics and data management skills in research teams act as a barrier to implementing more advanced practices, such as provenance capture and detailed, regularly updated, data management strategies. Only by a concerted effort at the levels of research organisations, funding bodies and publishers, can we achieve full transparency and reproducibility of the research.

Scientists routinely analyse and share data for others to use. Successful data (re)use relies on having metadata describing the context of analysis of data. In many disciplines the creation of contextual metadata is referred to as reporting. One method of implementing analyses is with workflows. A stand-out feature of workflows is their ability to record provenance from executions. Provenance is useful when analyses are executed with changing parameters (changing contexts) and results need to be traced to respective parameters. In this paper we investigate whether provenance can be exploited to support reporting. Specifically; we outline a case-study based on a real-world workflow and set of reporting queries. We observe that provenance, as collected from workflow executions, is of limited use for reporting, as it supports queries partially. We identify that this is due to the generic nature of provenance, its lack of domain-specific contextual metadata. We observe that the required information is available in implicit form, embedded in data. We describe LabelFlow, a framework comprised of four Labelling Operators for decorating provenance with domain-specific Labels. LabelFlow can be instantiated for a domain by plugging it with domain-specific metadata extractors. We provide a tool that takes as input a workflow, and produces as output a Labelling Pipeline for that workflow, comprised of Labelling Operators. We revisit the case-study and show how Labels provide a more complete implementation of reporting queries.

BackgroundStatins are widely used as a cholesterol lowering medication, reduce cardiovascular mortality and morbidity in high risk patients; and only rarely cause serious adverse drug reactions (ADRs). UK primary care databases of morbidity and prescription data, which now cover several million people, have potential for more powerful analytical approaches to study ADRs including adjusting for confounders and examining temporal effects.MethodsCase-crossover design in detecting statin associated myopathy ADR in 93, 831 patients, using two independent primary care databases (1991-2006). We analysed risk by drug class, by disease code and cumulative year, exploring different cut-off exposure times and confounding by temporality.ResultsUsing a 12 and 26 week exposure period, large risk ratios (RR) are associated with all classes of statins and fibrates for myopathy: RR 10.6 (9.8-11.4) and 19.9 (17.6-22.6) respectively. At 26 weeks, the largest risks are with fluvastatin RR 33.3 (95% CI 16.8-66.0) and ciprofibrate (with previous statin use) RR 40.5 (95% CI 13.4-122.0). AT 12 weeks the differences between cerivastatin and atorvastatin RR for myopathy were found to be significant, RR 2.05 (95% CI 1.2-3.5), and for rosuvastatin and fluvastatin RR 3.0 (95% CI 1.6-5.7). After 12 months of statin initiation, the relative risk for myopathy for all statins and fibrates increased to 25.7 (95% CI 21.8-30.3). Furthermore, this signal was detected within 2 years of first events being recorded. Our data suggests an annual incidence of statin induced myopathy or myalgia of around 11.4 for 16, 591 patients or 689 per million per year.ConclusionThere may be differential risks associated with some classes of statin and fibrate. Myopathy related to statin or fibrate use may persist after a long exposure time (12 months or more). These methods could be applied for early detection of harmful drug side effects, using similar primary care diagnostic and prescribing data.