Primary Care Electronic Health Research Articles

Patient-oriented unsupervised learning to uncover the patterns of multimorbidity associated with stroke using primary care electronic health records

BackgroundWe aimed to identify and characterise the longitudinal patterns of multimorbidity associated with stroke.MethodsWe used an unsupervised patient-oriented clustering approach to analyse primary care electronic health records (EHR) of 30 common long-term conditions (LTC) in patients with stroke aged over 18, registered in 41 general practices in south London between 2005 and 2021.ResultsOf 849,968 registered patients, 9,847 (1.16%) had a record of stroke and 46.5% were female. The median age at record of stroke was 65.0 year (IQR: 51.5-77.0) and the median number of LTCs in addition to stroke was 3 (IQR: 2-5). We identified eight clusters of multimorbidity with contrasted socio-demographic characteristics (age, gender, and ethnicity) and risk factors. Beside a core of 3 clusters associated with conventional stroke risk-factors, minor clusters exhibited less common combinations of LTCs including mental health conditions, asthma, osteoarthritis and sickle cell anaemia. Importantly, complex profiles combining mental health conditions, infectious diseases and substance dependency emerged.ConclusionThis novel longitudinal and patient-oriented perspective on multimorbidity addresses existing gaps in mapping the patterns of stroke-associated multimorbidity not only in terms of LTCs, but also socio-demographic characteristics, and suggests potential for more efficient and patient-oriented healthcare models.

Obstructive sleep apnoea syndrome and future risk of dementia among individuals managed in UK general practice

BackgroundObstructive sleep apnoea syndrome (OSAS) has been recognised as a potential risk factor for cognitive decline, yet its precise relationship with dementia remains uncertain. This study aimed to determine the...

Study protocol: Comparison of different risk prediction modelling approaches for COVID-19 related death using the OpenSAFELY platform

On March 11th 2020, the World Health Organization characterised COVID-19 as a pandemic. Responses to containing the spread of the virus have relied heavily on policies involving restricting contact between people. Evolving policies regarding shielding and individual choices about restricting social contact will rely heavily on perceived risk of poor outcomes from COVID-19. In order to make informed decisions, both individual and collective, good predictive models are required. For outcomes related to an infectious disease, the performance of any risk prediction model will depend heavily on the underlying prevalence of infection in the population of interest. Incorporating measures of how this changes over time may result in important improvements in prediction model performance. This protocol reports details of a planned study to explore the extent to which incorporating time-varying measures of infection burden over time improves the quality of risk prediction models for COVID-19 death in a large population of adult patients in England. To achieve this aim, we will compare the performance of different modelling approaches to risk prediction, including static cohort approaches typically used in chronic disease settings and landmarking approaches incorporating time-varying measures of infection prevalence and policy change, using COVID-19 related deaths data linked to longitudinal primary care electronic health records data within the OpenSAFELY secure analytics platform.

Involving People With Lived Experience in Electronic Health Record Database Studies Reflections and Learning From the CHOOSE Study.

Patient and public involvement and engagement (PPIE) is integral to health research. Reporting of PPIE methods and impact is becoming increasingly common in health research. However, reporting on PPIE in studies using large, routinely collected electronic health record data sets is less common. Anecdotal evidence suggests that involvement in this research context is more challenging and offers fewer opportunities for meaningful influence on the research process. This paper reports the involvement approach for a Clinical Practice Research Datalink (CPRD) study and critically reflects on the process and impact of involving young people, parents and carers in research using this UK primary care electronic health record data set. The CHOOSE study investigated mental health diagnoses of children and young people (1-24 years) during the COVID-19 pandemic using the CPRD. The study was informed by a Lived Experience Advisory Panel (LEAP) which consisted of 13 members including 8 young people (13-25 years) with lived experience of mental health difficulties and 5 parents/carers, with involvement activities facilitated by project partners, mental health research charity, The McPinR Foundation. We reflect on this process in this manuscript. Key benefits of involving people with lived experience in this research included making sense of and contextualising findings and ensuring that they were focused on making a difference to young people's lives. Challenges included the fixed nature of the CPRD data, which did not capture all the information people with lived experience perceived to be important. Researchers expressed limited time for PPIE activities although that was compensated by McPin colleagues who organised and facilitated online meetings, and supported the young people, parents and carers during and between meetings. This paper describes an approach to patient and public involvement in an electronic health record database study. Working collaboratively with young people, carers and other stakeholders requires sufficient time and adequate resources. We also highlight the importance of appropriate training and support and being transparent about the limitations of PPIE involvement. Three members of the CHOOSE LEAP have been involved in conceptualising and writing this paper.

Living in the vicinity of pesticide-treated crop fields: Exploring associated perceptions and psychological aspects in relation to self-reported and registry-based health symptoms

BackgroundExposure to pesticides in the living environment can be associated with the prevalence of health symptoms. This study investigates associations between health symptoms among residents in areas with fruit crop fields where pesticides are applied, and psychological perceptions and attitudes about environmental aspects and exposures.MethodsA cross-sectional survey combined with routine primary care electronic health records (EHR) data was conducted in 2017 in rural areas of the Netherlands with high concentration of fruit crops (n = 3,321, aged ≥ 16 years). Individual exposure to pesticides was estimated using geocoded data on fruit crops around the home. Validated instruments were used to assess symptom report and psychological perceptions and attitudes. Annual prevalence of various health symptoms was derived from EHRs. Multilevel regression models were used to analyze associations between health symptoms (outcome), fruit crops, and multiple psychological perceptions and attitudes (confounders).ResultsLiving in the vicinity of fruit crop fields was generally not associated with self-reported symptom duration and general practitioner (GP) registered symptoms. For self-reported symptoms, symptom prevalence decreased when crop density within 250 m and 500 m from the home increased. No associations were found at other distances. Furthermore, higher levels of environmental worries, perceived exposure, and perceived sensitivity to pesticides and attribution of symptoms to environmental exposures were generally associated with a higher number of self-reported symptoms, and longer symptom duration. Symptoms reported to GPs were not associated with psychological perceptions and attitudes, except for perceived sensitivity to pesticides.ConclusionPsychological perceptions and attitudes appear to be related to self-reported symptoms, but not to GP-registered symptoms, independent of the actual levels of exposure as measured by the size of the area of crop fields. Perceptions about environmental factors should be taken into account in environmental health risk assessment research when studying health symptoms.

Acceptability and feasibility of a chronic breathlessness diagnostic clinical algorithm in Australian primary care.

Chronic breathlessness is a frequent diagnostic challenge in primary care. Our aim is to evaluate the feasibility of a stepwise breathlessness diagnostic algorithm for primary care. This mixed-methods study included: (1) a general practitioner (GP) nominal group technique study; (2) focus groups on GPs' views on the algorithm; and (3) analysis of algorithm alignment against patterns of diagnostic referrals and diagnoses of breathlessness presentations (2014-19) from the MedicineInsight primary care electronic health record (EHR) dataset of 1,961,264 patients (405 general practice sites). All the tests in our algorithm, except for echocardiography, were ranked in the top 10 tests used by most GPs for patients presenting with chronic breathlessness. Themes from the focus group include similarity with current practice and test accessibility. Analysis of EHR diagnostic referrals revealed that all tests in the algorithm are regularly utilised and covered the major tests needed for breathlessness diagnoses recorded. The results of the three studies support the acceptability and feasibility of the clinical algorithm in primary care.

Linking national primary care electronic health records to individual records from the U.S. Census Bureau's American Community Survey: evaluating the likelihood of linkage based on patient health.

To evaluate the likelihood of linking electronic health records (EHRs) to restricted individual-level American Community Survey (ACS) data based on patient health condition. Electronic health records (2019-2021) are derived from a primary care registry collected by the American Board of Family Medicine. These data were assigned anonymized person-level identifiers (Protected Identification Keys [PIKs]) at the U.S. Census Bureau. These records were then linked to restricted individual-level data from the ACS (2005-2022). We used logistic regressions to evaluate match rates for patients with health conditions across a range of severity: hypertension, diabetes, and chronic kidney disease. Among more than 2.8 million patients, 99.2% were assigned person-level identifiers (PIKs). There were some differences in the odds of receiving an identifier in adjusted models for patients with hypertension (OR = 1.70, 95% CI: 1.63, 1.77) and diabetes (OR = 1.17, 95% CI: 1.13, 1.22), relative to those without. There were only small differences in the odds of matching to ACS in adjusted models for patients with hypertension (OR = 1.03, 95% CI: 1.03, 1.04), diabetes (OR = 1.02, 95% CI: 1.01, 1.03), and chronic kidney disease (OR = 1.05, 95% CI: 1.03, 1.06), relative to those without. Our work supports evidence-building across government consistent with the Foundations for Evidence-Based Policymaking Act of 2018 and the goal of leveraging data as a strategic asset. Given the high PIK and ACS match rates, with small differences based on health condition, our findings suggest the feasibility of enhancing the utility of EHR data for research focused on health.

Antidepressant prescribing inequalities in people with comorbid depression and type 2 diabetes: A UK primary care electronic health record study.

To compare the likelihood of being prescribed an antidepressant in depressed individuals with and without type 2 diabetes. We performed a matched cohort study using primary care record data from the UK Clinical Practice Research Datalink. We used multivariable logistic regression to compare antidepressant prescribing during the first five years of starting oral antidiabetic medication to a comparison group without type 2 diabetes, matched based on GP practice, age and sex. We performed subgroup analyses stratified by sex, age and ethnicity. People with type 2 diabetes and depression were 75% less likely to be prescribed an antidepressant compared to people with depression alone (odds ratio (OR) 0.25, 95% confidence interval (CI) 0.25 to 0.26). This difference was greater in males (OR 0.23, 95% CI, 0.22 to 0.24), people older than 56 years (OR 0.23, 95% CI, 0.22 to 0.24), or from a minoritised ethnic background (Asian OR 0.14, 95% CI 0.12-0.14; Black OR 0.12, 95% CI 0.09-0.14). There may be inequalities in access to antidepressant treatment for people with type 2 diabetes, particularly those who are male, older or from minoritised ethnic backgrounds.

Comparative risk of mortality in new users of prescription opioids for noncancer pain: results from the International Pharmacosurveillance Study.

Although opioids continue to be used internationally for noncancer pain, evidence to date on the comparative safety of different opioids is sparse and conflicting. The aim of this study was to examine the comparative risk of all-cause mortality in patients newly initiated on opioids for noncancer pain, across 3 jurisdictions in the United Kingdom (UK), United States, and Canada. A multicentre retrospective, population-based cohort study was conducted. Data sources included UK national primary care electronic health records (Clinical Practice Research Datalink), The Partners HealthCare Research Patient Data in Boston (US), and The Montreal Population Health Record data (Canada). New users of opioids aged ≥18 years without cancer were included. Patients with a diagnosis of a pain condition and with known back pain were analysed separately. Fully adjusted hazard ratios (HRs) were calculated using Cox-proportional models and adjusted for confounders. In total, 1,066,216 patients were included (UK: n = 993,294; Boston, US: n = 43,243; Montreal, Canada: n = 26,116). Compared with codeine, patients using morphine had a significantly higher adjusted risk in the UK {HR: 12.58 [95% confidence interval (CI), 11.87-13.32]}, US (HR: 8.62 [95% CI, 3.34-22.27]), and Canadian cohorts (HR: 6.69; [95% CI, 1.35-32.22]). In addition, other factors associated with higher mortality were being on combination opioids, fentanyl, buprenorphine, and oxycodone. Compared with those on <50 morphine milligram equivalents/day, patients on higher-doses experience an incremental increase in risk. In new users of opioids, compared with codeine, strong opioids, including morphine, fentanyl, buprenorphine, oxycodone, and combination opioids, and those on ≥50 morphine milligram equivalent/day were associated with a higher subsequent risk of all-cause mortality.

Trajectory of change in body mass index in Parkinson's disease

IntroductionGastrointestinal (GI) symptoms are some of the most common non-motor symptoms in Parkinson's. Weight is a nutritional metric and can be affected by dysfunction of the gastrointestinal (GI) tract. This study aims to explore the change in trajectory of body mass index (BMI) in individuals with Parkinson's over the course of the disease including the prodromal and post-diagnostic periods. MethodsThis was a retrospective longitudinal study of data from participants from the PRIME Parkinson UK cross-sectional study. Participants were included if they had had one or more weights and height recorded in the primary care electronic health record. Results287 patients were initially included but only 234 could be included in the analysis of BMI trajectory. Using a piecewise linear mixed model, we determined that there was a ‘change point’ in BMI trajectory. This occurred on average 3.73 years after diagnosis, when the mean BMI was 26.4 kg/m2. Prior to this change point, the estimated mean rate of change in BMI was −0.09 kg/m2 (95 % credible interval −0.20,0.00 kg/m2) per year. However, after the change point, we observed a more accelerated decline in BMI, with an estimated mean rate of change of −0.34 kg/m2 (95 % credible interval −0.70,-0.07 kg/m2) per year. ConclusionThere was a modest weight loss trajectory in the pre-diagnostic period consistent with clinically stable weight. However, after several years, post-diagnosis BMI loss became more marked. In clinical practice interventions could be targeted at this time point to optimize and maintain nutritional intake.

Health symptoms in individuals with and without post-covid in the Dutch general population

Abstract Background Health symptoms, often non-specific in nature such as headache and sleep problems, are more prevalent in the aftermath of disasters. Relatively little research focused on non-specific symptoms in the context of the COVID-19 pandemic, including those with post-covid conditions. The aim of this study was to assess the symptomatic profile of people with post-covid conditions, and to assess differences in health symptoms between non-infected individuals, infected individuals, and individuals with post-covid. Methods A nation-wide cross-sectional health survey performed in 2022 was combined with routine primary care electronic health records (EHRs). Prevalence of individual symptoms was analyzed, as well as symptom scores for number, duration, and severity of symptoms. The total survey sample consisted of 315,586 individuals. The EHR sample consisted of 29,797 individuals with matched survey data. Both samples were divided into three groups: non-infected, infected, and post-covid. Results Survey participants with post-covid experienced significantly (CI 99%) more (incidence risk ratio (IRR) 1.55), longer lasting (IRR 1.87), and more severe self-reported symptoms (IRR 1.95) compared to non-infected individuals. For infected compared to non-infected survey participants, there were hardly differences in symptom scores. The most distinct differences in individual symptoms were observed for loss of smell or taste, fatigue, and respiratory problems. Based on EHR data, the post-covid group is at significantly higher risk of experiencing fatigue, shortness of breath, headache, sleep problems, and dizziness. Conclusions Individuals suffering from post-covid are at the highest risk of experiencing health symptoms, both based on self-reported symptoms as well as based on EHRs. The infected group generally experienced more self-reported symptoms than the non-infected, but on average, they experienced them at a shorter duration, and they also reported them to be less severe. Key messages • Health symptom prevalence is higher in individuals with post-covid in the Netherlands. • This effect is confirmed in self-reported health surveys as well as in primary care registries.

Clinical indications associated with new opioid use for pain management in the United Kingdom: using national primary care data.

Prescription opioids for noncancer pain in the United Kingdom have increased over the past 2 decades, alongside associated harms. Policies addressing opioid prescribing must be tailored to individual patient needs with specific disease systems. The aim of this study was to evaluate clinical conditions associated with new opioid initiation in noncancer pain using nationally representative UK data. Primary care electronic health records from January 1, 2006, to September 31, 2021, were used from the Clinical Research Practice Datalink to identify incident opioid prescriptions. Patient histories were reviewed using code lists for opioid-related conditions with a 5-year look-back for chronic conditions and a 1-year look-back for surgical indications before opioid initiation. In total, 3,030,077 new opioid use episodes in 2,027,402 patients were identified, with 61% being women, 77% aged 45 years and older, and 48% from the highest deprivation quintile. Ten systems associated with opioid initiation were identified, which were not mutually exclusive, as patients could have opioids prescribed for multiple indications. The most common were musculoskeletal (80.8%), respiratory (57.6%), infections (30.4%), trauma/injury (20.4%), neurology (19.9%), and postsurgical indications (5.5%). Osteoarthritis (60.7%) and low back pain (41.0%) were the most frequent musculoskeletal conditions. Orthopedic surgeries accounted for 41.2% of all postsurgical indications. This is the first study in the United Kingdom evaluating large-scale national data to assess indications associated with opioid initiation. Nearly 3 quarters of new opioid prescriptions for noncancer pain were in patients with musculoskeletal conditions, often for conditions with limited evidence for opioid efficacy. These findings could inform targeted interventions and future policies to support nonpharmacological interventions in the most common conditions where opioid harms outweigh benefits.

A machine learning algorithm for the detection of paroxysmal nocturnal haemoglobinuria (PNH) in UK primary care electronic health records

BackgroundParoxysmal Nocturnal Haemoglobinuria (PNH) is an ultra-rare, acquired disorder that is challenging to diagnose due to varied symptoms, heterogeneous patient presentations, and lack of awareness among healthcare professionals. This leads to frequent misdiagnosis and delays in diagnosis. This study evaluated the feasibility of a machine learning model to identify undiagnosed PNH patients using structured electronic health records.MethodsThe study used data from the Optimum Patient Care Research Database, which contains electronic health records from general practitioner (GP) practices across the United Kingdom. PNH patients were identified by the presence, and control patients by the absence of a PNH diagnosis code in their records. Clinical features (symptoms, diagnoses, healthcare utilisation) from 131 patients in the PNH group and 593,838 patients in the control group, were inputted to a tree-based XGBoost machine learning model to classify patients as either “positive” or “negative” for PNH suspicion. The algorithm was finalised after additional exclusions and inclusions applied. Performance was assessed using positive predictive value (PPV), recall and specificity. As the sample used to develop the algorithm was not representative of the true population prevalence, PPV was additionally adjusted to reflect performance in the wider population.ResultsOf all the patients in the PNH group, 27% were classified as positive (recall). 99.99% of the control group were classified as negative (specificity). Of all the patients classified as positive, 60.4% had a diagnosis of PNH in their record (PPV). The PPV adjusted for the population prevalence of PNH was 19.59 suggesting nearly 1 in 5 patients flagged may warrant further PNH investigation. The key clinical features in the model were aplastic anaemia, pancytopenia, haemolytic anaemia, myelodysplastic syndrome, and Budd-Chiari syndrome.ConclusionThis is the first study to combine clinical understanding of PNH with machine learning, demonstrating the ability to discriminate between PNH and control patients in retrospective electronic health records. With further investigation and validation, this algorithm could be deployed on live health data, potentially leading to earlier diagnosis for patients who currently experience long diagnostic delays or remain undiagnosed.

Utilising primary care electronic health records to deliver the ALABAMA randomised controlled trial of penicillin allergy assessment

BackgroundUse of electronic health records (EHR) to provide real-world data for research is established, but using EHR to deliver randomised controlled trials (RCTs) more efficiently is less developed. The Allergy AntiBiotics And Microbial resistAnce (ALABAMA) RCT evaluated a penicillin allergy assessment pathway versus usual clinical care in a UK primary care setting. The aim of this paper is to describe how EHRs were used to facilitate efficient delivery of a large-scale randomised trial of a complex intervention embracing efficient participant identification, supporting minimising GP workload, providing accurate post-intervention EHR updates of allergy status, and facilitating participant follow up and outcome data collection. The generalisability of the EHR approach and health economic implications of EHR in clinical trials will be reported in the main ALABAMA trial cost-effectiveness analysis.MethodsA descriptive account of the adaptation of functionality within SystmOne used to deliver/facilitate multiple trial processes from participant identification to outcome data collection.ResultsAn ALABAMA organisation group within SystmOne was established which allowed sharing of trial functions/materials developed centrally by the research team. The ‘ALABAMA unit’ within SystmOne was also created and provided a secure efficient environment to access participants’ EHR data. Processes of referring consented participants, allocating them to a trial arm, and assigning specific functions to the intervention arm were developed by adapting tools such as templates, reports, and protocols which were already available in SystmOne as well as pathways to facilitate allergy de-labelling processes and data retrieval for trial outcome analysis.ConclusionsALABAMA is one of the first RCTs to utilise SystmOne EHR functionality and data across the RCT delivery, demonstrating feasibility and applicability to other primary care RCTs.Trial registrationClinicalTrials.gov: NCT04108637, registered 05/03/2019. ISRCTN: ISRCTN20579216.

Impacts of the COVID-19 pandemic on deprivation-level differences in cardiovascular hospitalisations: a comparison of England and Denmark using the OpenSAFELY platform and National Registry Data

ObjectivesTo examine the impact of the COVID-19 pandemic on deprivation-related inequalities in hospitalisations for cardiovascular disease (CVD) conditions in Denmark and England between March 2018 and December 2021.DesignTime-series studies in...

Phenotypic characterization of people at risk of atrial fibrillation: protocol for the FIND-AF longitudinal cohort study.

The Future Innovations in Novel Detection of Atrial Fibrillation (FIND-AF) longitudinal cohort study is a multi-centre prospective cohort study of patients identified at risk of atrial fibrillation (AF). The aim of the FIND-AF longitudinal cohort study is to provide multi-modal phenotypic characterization of these patients. A total of 1955 participants identified as at risk of AF by the FIND-AF algorithm from primary care electronic health record (EHR) data, aged 30 years and above and eligible for oral anticoagulation, will be recruited between October 2023 and November 2024 to receive home-based intermittent electrocardiogram monitoring. About 500 participants without diagnosed AF will then undergo cross-sectional phenotypic characterization including physical examination, symptoms assessment, serum blood biomarkers and echocardiography, and non-stress cardiac magnetic resonance imaging. Longitudinal information about cardio-renal-metabolic-pulmonary outcomes will be ascertained from linkages to EHR data. The study is funded by the British Heart Foundation (CC/22/250026). The study has ethical approval (North West-Greater Manchester South Research Ethics Committee reference 23/NW/0180). Findings will be announced at relevant conferences and published in peer-reviewed journals in line with the funder's open-access policy. The FIND-AF multi-centre prospective longitudinal cohort study aims to (i) provide evidence for the impact of comorbidities on AF genesis, (ii) uncover actionable targets to prevent AF, and (iii) act as a platform for cohort randomized clinical trials that investigate enhanced detection and prevention of AF.

Excess health care use is significantly and persistently reduced following diagnosis of late-onset epilepsy.

The incidence of late-onset epilepsy (LOE) is rising, and these patients may use an excess of health care resources. This study aimed to measure pre-/post-diagnostic health care use (HCU) for patients with LOE compared to controls. This was an observational open cohort study covering years 1998-2019 using UK population-based linked primary care (Clinical Practice Research Datalink [CPRD]) and hospital (HES) electronic health records. The participants included patients with incident LOE enrolled in CPRD and 1:10 age-, sex-, and general practice-matched controls. The exposure was incident LOE (diagnosed at age ≥65) using a 5-year washout. The main outcome was all HCU (primary care [PC], accident and emergency [A&E], admitted patient and outpatient care) using inverse proportional weighting to PC use and HCU by setting. An interrupted time-series analysis was used to examine pre-/post-diagnostic HCU between patients with LOE and controls over 4 years either side of diagnosis/matching date. An adjusted mixed-effects negative binomial regression was used for post-diagnosis HCU interactions. Of 2 569 874 people ≥65 years of age, 1048 (4%) developed incident LOE. Mean weighted total HCU increased by 32 visits per patient-year (95% confidence interval [95% CI]: 13-50, p = .003) until LOE diagnosis, and then dropped by a mean of 60 visits per patient-year (95% CI: -81 to -40). There was an acute rise and fall over the 1-2 years immediately pre-/post-diagnosis. Incident HCU remained higher for LOE compared to controls post-diagnosis (adjusted incidence rate ratio: 1.72; 95% CI: 1.65-1.70; p < .001), including A&E, outpatient, and admitted care. Health care use demonstrates an acute on chronic rise over the 4 years before diagnosis of LOE. To what extent the partial reversal of the acute pre-diagnosis rise, and the mediators of the accelerated increase compared to controls are attributed to epilepsy, comorbid and bidirectional disease states, or a combination of both warrants further exploration.

Optimal risk-assessment scheduling for primary prevention of cardiovascular disease

Abstract In this work, we introduce a personalized and age-specific net benefit function, composed of benefits and costs, to recommend optimal timing of risk assessments for cardiovascular disease (CVD) prevention. We extend the 2-stage landmarking model to estimate patient-specific CVD risk profiles, adjusting for time-varying covariates. We apply our model to data from the Clinical Practice Research Datalink, comprising primary care electronic health records from the UK. We find that people at lower risk could be recommended an optimal risk-assessment interval of 5 years or more. Time-varying risk factors are required to discriminate between more frequent schedules for high-risk people.

Foundational model aided automatic high-throughput drug screening using self-controlled cohort study.

Developing medicine from scratch to governmental authorization and detecting adverse drug reactions (ADR) have barely been economical, expeditious, and risk-averse investments. The availability of large-scale observational healthcare databases and the popularity of large language models offer an unparalleled opportunity to enable automatic high-throughput drug screening for both repurposing and pharmacovigilance. To demonstrate a general workflow for automatic high-throughput drug screening with the following advantages: (i) the association of various exposure on diseases can be estimated; (ii) both repurposing and pharmacovigilance are integrated; (iii) accurate exposure length for each prescription is parsed from clinical texts; (iv) intrinsic relationship between drugs and diseases are removed jointly by bioinformatic mapping and large language model - ChatGPT; (v) causal-wise interpretations for incidence rate contrasts are provided. Using a self-controlled cohort study design where subjects serve as their own control group, we tested the intention-to-treat association between medications on the incidence of diseases. Exposure length for each prescription is determined by parsing common dosages in English free text into a structured format. Exposure period starts from initial prescription to treatment discontinuation. A same exposure length preceding initial treatment is the control period. Clinical outcomes and categories are identified using existing phenotyping algorithms. Incident rate ratios (IRR) are tested using uniformly most powerful (UMP) unbiased tests. We assessed 3,444 medications on 276 diseases on 6,613,198 patients from the Clinical Practice Research Datalink (CPRD), an UK primary care electronic health records (EHR) spanning from 1987 to 2018. Due to the built-in selection bias of self-controlled cohort studies, ingredients-disease pairs confounded by deterministic medical relationships are removed by existing map from RxNorm and nonexistent maps by calling ChatGPT. A total of 16,901 drug-disease pairs reveals significant risk reduction, which can be considered as candidates for repurposing, while a total of 11,089 pairs showed significant risk increase, where drug safety might be of a concern instead. This work developed a data-driven, nonparametric, hypothesis generating, and automatic high-throughput workflow, which reveals the potential of natural language processing in pharmacoepidemiology. We demonstrate the paradigm to a large observational health dataset to help discover potential novel therapies and adverse drug effects. The framework of this study can be extended to other observational medical databases.

Residential mobility and receipt of measles, mumps and rubella vaccination: analysis of linked primary care electronic health records in a disadvantaged London region

ObjectiveCoverage for the measles, mumps and rubella (MMR) vaccination in north-east London (NEL) is the lowest in the United Kingdom. It has been hypothesised that children who move home frequently are less likely to receive vaccinations. We examined the association between number of addresses in the first two years of life, and receipt of MMR vaccination by age two. ApproachWe derived households by pseudonymously linking the addresses of 178,028 children registered with a NEL general practitioner (GP) on their second birthday (2015-2021). We included 135,221 (76.0%) children (51.0% male) after excluding 42,807 with invalid household characteristics at age 24 months. The outcome was MMR vaccination by 24 months. Residential mobility was categorised by number of addresses (1, 2, ≥3) between earliest GP-registration and vaccination date/second birthday. We calculated the percentage receiving MMR by residential mobility and estimated adjusted odds (aOR; 95% confidence intervals [CI]) for MMR receipt. Results105,392 (77.9%) children received MMR by age two, and 26,652 (19.7%) had more than one address. Receipt of MMR decreased with increasing residential mobility: 80.9% (95% CI: 80.6,81.1), 66.5% (65.9,67.1), and 58.6% (56.2,60.9) for children with one, two, or three or more addresses respectively. Children with multiple addresses were less likely to receive MMR by age two: aOR: 0.53 (0.51,0.55); 0.36 (0.32,0.40) for 2 and ≥3 addresses respectively. Conclusions Uptake of MMR vaccine by two years of age is lower in children with multiple addresses. Further research is needed to identify actionable opportunities to improve MMR uptake in mobile populations.