UK Primary Care Electronic Health Research Articles

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.

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.

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.

Non-osteoporotic fractures are associated with increased risk of subsequent major osteoporotic fractures

SummaryWe studied the association between non-osteoporotic fractures and future major osteoporotic fractures, using UK health records. Non-osteoporotic fractures were found to increase the risk of major osteoporotic fractures, although to a lesser extent than osteoporotic fractures. This highlights the importance of considering all previous fractures in assessing future fracture risk.PurposePrevious studies demonstrated that osteoporotic fractures—minor and major—increase the risk for future major osteoporotic fractures; we test whether non-osteoporotic fractures are also associated with such increased risk.MethodsThe study is a retrospective cohort study using UK primary care electronic health records. Exposure groups were defined according to fracture location prior to the year 2011 (index date): major, minor, and non-osteoporotic. The outcome of incident major osteoporotic fractures following the index date was compared between the exposure groups and the general population.ResultsThe general study population included 1,951,388 patients. The exposure groups included 39,931 patients with a prior major osteoporotic fracture, 19,397 with a prior minor osteoporotic fracture, and 50,115 patients with a prior non-osteoporotic fracture. The standardized Incidence Rate Ratio for future major osteoporotic fractures was 2.73 (95% confidence interval: 2.64–2.82), 2.43 (2.32–2.54), and 1.83 (1.74–1.92), respectively.ConclusionNon-osteoporotic fractures are significantly associated with increased risk for future major osteoporotic fractures relative to the general population, yet to a lesser extent compared to major and minor osteoporotic fractures.

P051 Trends in consultations for musculoskeletal pain in children and young people presenting to United Kingdom primary care: an electronic health record study

Abstract Background/Aims Musculoskeletal pain in childhood increases the risk of such pain in adulthood but there is an absence of management guidance for this group, not helped by a lack of evidence on the extent and nature of pain in children. The aim was to determine the extent and patterns of consultations for musculoskeletal pain in children and young people (CYP) presenting to UK primary care. Methods Data were obtained from an anonymised nationally representative UK primary care electronic health record database (Clinical Practice Research Datalink, CPRD). The study population was patients aged 8-18 years between 2005-2021 with consultations for musculoskeletal pain. Consultation prevalence was calculated per 10,000 registered population for each calendar year, and by age (8-11, 12-15 and 16-18 years), gender and body site (prevalence presented for 2019 prior to COVID-19 pandemic). Results 1.12 million children and young people (48% female; median age 14 years) were recorded with musculoskeletal pain, with an annual consultation prevalence of over 9% (932/10,000, 95% CI 928-937). Consultation prevalence increased until 2015, stabilised until 2019, and was lower during 2020 due to the COVID-19 pandemic, but climbed slightly in 2021. Consultation prevalence in 2019 was highest for those aged 12-15 years (1105/10,000 vs 8-11 years 736/10,000 and 16-18 years 984/10,000) and for females. Age and gender differences were observed by body site with foot/ankle the most common site of pain in younger children (ages 8-11: females 277/10,000, males 229/10,000,), back the most common site in older females (ages 16- 18: 247/10,000) and knee the most common site in older males (ages 16-18: 178/10,000). Conclusion Musculoskeletal pain in CYP is common with almost one in ten consulting primary care annually, but pain sites vary by age and gender. Understanding the extent and nature of musculoskeletal pain in CYP presenting to primary care can help inform targeted and effective care, and underpin much needed intervention studies, potentially reducing risk of continued pain into adulthood. Disclosure K.J. Mason: None. K.P. Jordan: None. J. Bailey: None. A. Jöud: None. S. Jowett: None. J. Protheroe: None. F. Mughal: None. K.I.E. Snell: None. M. Thomas: None. K.M. Dunn: None.

The epidemiology of psoriatic arthritis in the UK: a health intelligence analysis of UK Primary Care Electronic Health Records 1991-2020.

Epidemiological estimates of psoriatic arthritis (PsA) underpin the provision of healthcare, research, and the work of government, charities and patient organizations. Methodological problems impacting prior estimates include small sample sizes, incomplete case ascertainment, and representativeness. We developed a statistical modelling strategy to provide contemporary prevalence and incidence estimates of PsA from 1991 to 2020 in the UK. Data from Clinical Practice Research Datalink (CPRD) were used to identify cases of PsA between 1st January 1991 and 31st December 2020. To optimize ascertainment, we identified cases of Definite PsA (≥1 Read code for PsA) and Probable PsA (satisfied a bespoke algorithm). Standardized annual rates were calculated using Bayesian multilevel regression with post-stratification to account for systematic differences between CPRD data and the UK population, based on age, sex, socioeconomic status and region of residence. A total of 26293 recorded PsA cases (all definitions) were identified within the study window (77.9% Definite PsA). Between 1991 and 2020 the standardized prevalence of PsA increased twelve-fold from 0.03 to 0.37. The standardized incidence of PsA per 100,000 person years increased from 8.97 in 1991 to 15.08 in 2020, an almost 2-fold increase. Over time, rates were similar between the sexes, and across socioeconomic status. Rates were strongly associated with age, and consistently highest in Northern Ireland. The prevalence and incidence of PsA recorded in primary care has increased over the last three decades. The modelling strategy presented can be used to provide contemporary prevalence estimates for musculoskeletal disease using routinely collected primary care data.

Phenotyping hypophosphatasia using UK primary care electronic health records

Phenotyping hypophosphatasia using UK primary care electronic health records

The chicken or the egg of lymphopenia in atopic eczema.

Lymphopenia has been documented in atopic eczema (AE) patients mostly in the setting of an active infection.1, 2 However, the question is whether lymphopenia is a primary problem in AE leading to infection or it is a secondary problem due to infection. In the study by Hollestein et al.,3 the authors conducted a retrospective review of a large electronic database (UK primary care electronic health record from CPRD GOLD). They included 71,731 adults with AE and 126,349 adults without AE. Using the definition of lymphopenia as having two lymphocyte counts of <1 × 109/L in 3 months, they showed that 4.1% (n = 2909) of AE patients had lymphopenia as compared to 3.7% (n = 4700) of controls without AE. The adjusted OR for lymphopenia in AE patients as compared to those without AE was 1.16 and this increased to 1.89 in patients with severe AE. Their data were appropriately adjusted for age, sex, ethnicity, smoking, co-morbidities and immunosuppressive drug use. In their secondary analysis, they confirmed an increased risk of skin and systemic infections including cellulitis, varicella zoster, gastroenteritis and urinary tract infection in AE patients as compared to controls. The authors also replicated their findings of lymphopenia in AE patients based in a US database, NHANES. Their data support the hypothesis that lymphopenia may predispose to increased infections in AE patients. A possible mechanism of lymphopenia in AE patients is due to the homing of systemic lymphocytes to inflamed skin. Their finding of a higher risk of lymphopenia in more severe AE is also consistent with this mechanism. While Hollestein et al. have provided a robust and well-controlled study that lymphopenia may be a primary problem in AE patients, the findings can only be inferred as an association, rather than cause-and-effect, due to the retrospective nature of the study. In a real-life clinical situation, complete blood cell (CBC) profile is normally sent when a patient is symptomatic, that is, during an illness or an infection. Acute lymphopenia is common with viral respiratory infections.4 In AE patients, CBC may also be sent together with viral or bacterial identification testing during a skin infection, which may be associated with lymphopenia.1, 2 AE patients are not known to have a primary immunodeficiency that predisposes them to have increased infection. However, skin and soft tissue infections are more common in AE patients due skin barrier defects and suppression of cutaneous innate immunity by type 2 cytokines.5 The results of Hollestein et al. are nevertheless interesting, and will need to be confirmed with prospective studies. None to declare. P. Ong reports no conflict of interest relevant to this article. Data sharing not applicable.

Association of Recent Fatherhood With Antidepressant Treatment Initiation Among Men in the United Kingdom

There is some evidence that men may be at higher risk of depression directly following the birth of their child and that approximately 1 in 10 men will experience depression in the year after birth, but less is known about men's antidepressant treatment during this time. Likewise, few direct comparisons have been made with antidepressant treatment in men who have not recently become fathers. To determine whether recently having a child was associated with increased odds of antidepressant treatment in men. This cohort study used UK primary care electronic health records from the IQVIA Medical Research Database. Participants included men aged 15 to 55 years who had had a child in the previous year, from January 2007 to December 2016, and compared with up to 5 men who did not have a child in the same calendar year. Data were analyzed from January 2022 to March 2023. A record of having a child in the previous year was identified through linked primary care records using a family identification number. The main outcome was antidepressant initiation in the year after childbirth or in the year after this index date for men who did not have a child. Random-effects Poisson regression was used to determine associations of cohort, age group, social deprivation, history of antidepressant treatment, and calendar year with having an antidepressant prescription in the year after index date using prevalence risk rates (PPRs). Analysis included 90 736 men who had had a child in the previous year and 453 632 men in the comparison cohort. Most men in the study (463 879 men [85.2%]) were aged between 25 and 44 years, and there were more men living in the least deprived areas (130 277 men [23.9%]) than the most deprived areas (72 268 men [13.3%]). Overall, 4439 men (4.9%) had at least 1 antidepressant prescription in the year after they had a child, compared with 26 646 men (5.9%) who did not have a child in the same year. However, after adjustment there was no difference in antidepressant treatment between groups (adjusted PRR [aPRR], 1.01; 95% CI, 0.98-1.04). In fathers, those who had recently received antidepressant treatment were much more likely to receive antidepressant treatment after childbirth compared with fathers with no history of antidepressant treatment (aPRR, 32.31; 95% CI, 30.37-34.38). Fathers living in the most deprived areas were 18% more likely to have an antidepressant prescription compared with fathers living in the least deprived areas (aPRR, 1.18; 95% CI, 1.07-1.30). These findings suggest that recently having a child was not associated with an increase in antidepressant treatment among men, but previous antidepressant treatment in fathers was strongly associated with treatment after childbirth. Further research is needed to determine whether antidepressant treatment or experiencing depression can be a barrier to fatherhood and whether fatherhood is a barrier to receiving antidepressant treatment.

Valsartan, Losartan and Irbesartan use in the USA, UK, Canada and Denmark after the nitrosamine recalls: a descriptive cohort study

ObjectivesTo examine valsartan, losartan and irbesartan usage and switching patterns in the USA, UK, Canada and Denmark before and after July 2018, when the first Angiotensin-Receptor-Blocker (ARB) (valsartan) was recalled.DesignRetrospective...

Determining cardiovascular risk in patients with unattributed chest pain in UK primary care: an electronic health record study.

Most adults presenting in primary care with chest pain symptoms will not receive a diagnosis ('unattributed' chest pain) but are at increased risk of cardiovascular events. To assess within patients with unattributed chest pain, risk factors for cardiovascular events and whether those at greatest risk of cardiovascular disease can be ascertained by an existing general population risk prediction model or by development of a new model. The study used UK primary care electronic health records from the Clinical Practice Research Datalink linked to admitted hospitalizations. Study population was patients aged 18 plus with recorded unattributed chest pain 2002-2018. Cardiovascular risk prediction models were developed with external validation and comparison of performance to QRISK3, a general population risk prediction model. There were 374 917 patients with unattributed chest pain in the development data set. The strongest risk factors for cardiovascular disease included diabetes, atrial fibrillation, and hypertension. Risk was increased in males, patients of Asian ethnicity, those in more deprived areas, obese patients, and smokers. The final developed model had good predictive performance (external validation c-statistic 0.81, calibration slope 1.02). A model using a subset of key risk factors for cardiovascular disease gave nearly identical performance. QRISK3 underestimated cardiovascular risk. Patients presenting with unattributed chest pain are at increased risk of cardiovascular events. It is feasible to accurately estimate individual risk using routinely recorded information in the primary care record, focusing on a small number of risk factors. Patients at highest risk could be targeted for preventative measures.

The association between atopic eczema and lymphopenia: Results from a UK cohort study with replication in US survey data.

Lymphocyte skin homing in atopic eczema (AE) may induce lymphopenia. To determine if AE is associated with lymphopenia. We used UK primary care electronic health records (Clinical Practice Research Datalink GOLD) for a matched cohort study in adults (18 years+) (1997-2015) with at least one recorded lymphocyte count. We matched people with AE to up to five people without. We used multivariable logistic regression to estimate the association between AE and lymphopenia (two low lymphocyte counts within 3months) and linear mixed effects regression to estimate the association with absolute lymphocyte counts using all available counts. Cox proportional hazard models were used to investigate the effect of lymphopenia on common infections. We replicated the study using US survey data (National Health and Nutrition Examination Survey [NHANES]). Among 71,731 adults with AE and 126,349 adults without AE, we found an adjusted odds ratio (OR) for lymphopenia of 1.16 (95% CI: 1.09-1.23); the strength of association increased with increasing eczema severity. When comparing all recorded lymphocyte counts from adults with AE (n=1,497,306) to those of people without AE (n=4,035,870) we saw a lower mean lymphocyte (adjusted mean difference -0.047 × 109 /L [95% CI: -0.051 to -0.043]) in those with AE. The difference was larger for men, with increasing age, and with increasing AE severity and was present among people with AE not treated with immunosuppressive drugs. In NHANES (n=22,624), the adjusted OR for lymphopenia in adults with AE was 1.30 (95% CI: 0.80-2.11), and the adjusted mean lymphocyte count difference was -0.03 × 109 /L (95% CI: -0.07 to 0.02). Despite having a lower lymphocyte count, adjusting for time with lymphopenia, did not alter risk estimates of infections. Atopic eczema, including increasing AE severity, is associated with a decreased lymphocyte count, regardless of immunosuppressive drug use. Whether the lower lymphocyte count has wider health implications for people with severe eczema warrants further investigation.

Cardiovascular and renal outcomes among patients with type 2 diabetes using SGLT2 inhibitors added to metformin: a population-based cohort study from the UK

IntroductionLarge numbers of patients with type 2 diabetes receive treatment with a sodium-glucose co-transporter-2 inhibitor (SGLT2i). We investigated whether the cardiorenal preventative effects found in clinical trials are also seen...

Machine Learning for Risk Prediction of Oesophago-Gastric Cancer in Primary Care: Comparison with Existing Risk-Assessment Tools.

Simple SummaryOesophago-gastric cancer is one of the commonest cancers worldwide, yet it can be particularly difficult to diagnose given that initial symptoms are often non-specific and routine screening is not available. Cancer risk-assessment tools, which calculate cancer risk based on symptoms and other risk factors present in the primary care record, can aid decisions on referrals for cancer investigations, facilitating earlier diagnosis. Diagnosing common cancers earlier could help improve survival rates. Using UK primary care electronic health record data, we compared five different machine learning techniques for probabilistic classification of cancer patients against a current widely used UK primary care cancer risk-assessment tool. The machine learning algorithms outperformed the current risk-assessment tool, with a higher overall accuracy and an ability to reasonably identify 11–25% more cancer patients. We conclude that machine-learning-based risk-assessment tools could help better identify suitable patients for further investigation and support earlier diagnosis.Oesophago-gastric cancer is difficult to diagnose in the early stages given its typical non-specific initial manifestation. We hypothesise that machine learning can improve upon the diagnostic performance of current primary care risk-assessment tools by using advanced analytical techniques to exploit the wealth of evidence available in the electronic health record. We used a primary care electronic health record dataset derived from the UK General Practice Research Database (7471 cases; 32,877 controls) and developed five probabilistic machine learning classifiers: Support Vector Machine, Random Forest, Logistic Regression, Naïve Bayes, and Extreme Gradient Boosted Decision Trees. Features included basic demographics, symptoms, and lab test results. The Logistic Regression, Support Vector Machine, and Extreme Gradient Boosted Decision Tree models achieved the highest performance in terms of accuracy and AUROC (0.89 accuracy, 0.87 AUROC), outperforming a current UK oesophago-gastric cancer risk-assessment tool (ogRAT). Machine learning also identified more cancer patients than the ogRAT: 11.0% more with little to no effect on false positives, or up to 25.0% more with a slight increase in false positives (for Logistic Regression, results threshold-dependent). Feature contribution estimates and individual prediction explanations indicated clinical relevance. We conclude that machine learning could improve primary care cancer risk-assessment tools, potentially helping clinicians to identify additional cancer cases earlier. This could, in turn, improve survival outcomes.

Continuing specialist care into adulthood in young people with juvenile idiopathic arthritis: a retrospective cohort study using electronic health records in England.

This study aimed to measure (1) the proportion of children who continue to receive specialist care (rheumatology/ophthalmology) as adults, (2) the characteristics associated with continuing specialist care, and (3) the frequency of specialist care appointments in both paediatric and adult services. A retrospective cohort of young people with JIA was identified from UK primary care electronic health records (Clinical Practice Research Datalink) between 1 April 2003 and 31 December 2018. To be included in the study, cases needed to have at least 1 year of registration at their general practice beyond age 18 and linkage to Hospital Episodes Statistics data for secondary care information. All specialist care outpatient visits were identified from Hospital Episodes Statistics outpatient data. There were 666 young people included in the study. Of these, 427 (64%) received specialist care beyond age 18, 90 (13%) had their last recorded contact at 16-17 years and 149 (22%) did not continue after 16 years. Older age at diagnosis, female gender, less deprivation and a childhood diagnosis of uveitis were associated with continuing specialist care beyond age 18. Of those continuing beyond 18, 35% (n = 153) were subsequently discharged by the study end date. Of all those discharged, 32% had a missed appointment recorded after the last attended visit, suggesting failure to attend. Two-thirds of young people with JIA continue to receive specialist care beyond age 18. This is useful information for children and young people with JIA and their families planning for their future, and for clinicians planning health-care services.

Incremental value of risk factor variability for cardiovascular risk prediction in individuals with type 2 diabetes: results from UK primary care electronic health records.

Cardiovascular disease (CVD) risk prediction models for individuals with type 2 diabetes are important tools to guide intensification of interventions for CVD prevention. We aimed to assess the added value of incorporating risk factors variability in CVD risk prediction for people with type 2 diabetes. We used electronic health records (EHRs) data from 83 910 adults with type 2 diabetes but without pre-existing CVD from the UK Clinical Practice Research Datalink for 2004-2017. Using a landmark-modelling approach, we developed and validated sex-specific Cox models, incorporating conventional predictors and trajectories plus variability of systolic blood pressure (SBP), total and high-density lipoprotein (HDL) cholesterol, and glycated haemoglobin (HbA1c). Such models were compared against simpler models using single last observed values or means. The standard deviations (SDs) of SBP, HDL cholesterol and HbA1c were associated with higher CVD risk (P < 0.05). Models incorporating trajectories and variability of continuous predictors demonstrated improvement in risk discrimination (C-index = 0.659, 95% CI: 0.654-0.663) as compared with using last observed values (C-index = 0.651, 95% CI: 0.646-0.656) or means (C-index = 0.650, 95% CI: 0.645-0.655). Inclusion of SDs of SBP yielded the greatest improvement in discrimination (C-index increase = 0.005, 95% CI: 0.004-0.007) in comparison to incorporating SDs of total cholesterol (C-index increase = 0.002, 95% CI: 0.000-0.003), HbA1c (C-index increase = 0.002, 95% CI: 0.000-0.003) or HDL cholesterol (C-index increase= 0.003, 95% CI: 0.002-0.005). Incorporating variability of predictors from EHRs provides a modest improvement in CVD risk discrimination for individuals with type 2 diabetes. Given that repeat measures are readily available in EHRs especially for regularly monitored patients with diabetes, this improvement could easily be achieved.

Maternal Postnatal Depression and Completion of Infant Immunizations: A UK Cohort Study of 196,329 Mother-Infant Pairs, 2006-2015.

Objective: To examine the relationship between maternal postnatal depression and completion of infant vaccinations.Methods: We conducted a cohort study using data from The Health Improvement Network (THIN), a large UK primary care electronic health record database. We identified 196,329 mother-infant pairs in which the infant was born between 2006 and 2015. Postnatal depression was identified through antidepressant prescriptions or diagnoses or symptoms of depression in first year after childbirth. Primary outcome was completion of three 5-in-1 vaccination doses in infants before 1 year of age; this vaccine protects against diphtheria, tetanus, whooping cough, polio, and Haemophilus influenzae type b. We used Poisson regression models to compare likelihood of infant 5-in-1 vaccine uptake among children of women with a record of postnatal depression to likelihood among those without.Results: Of the 196,329 women, 20,802 (10.6%) had a record of postnatal depression and/or antidepressant prescription. There was no difference in infants' 5-in-1 vaccination completion between those of mothers with a record and those of mothers' without (adjusted incidence rate ratio [IRR] = 1.01; 95% CI, 0.99-1.02). Those from more socially deprived areas were less likely to complete infant vaccinations compared to those from the least deprived areas (IRR = 0.92; 95% CI, 0.90-0.93). Likelihood of completing infant vaccination decreased over time, comparing 2014-2015 to 2006-2007 (IRR = 0.90; 95% CI, 0.89-0.92).Conclusions: Among mothers who engage with primary care, maternal postnatal depression is not associated with lower rates of infant vaccination, though more research is needed to conclude if either more severe depression or unrecognized depression is associated with lower completion rates.

Development of an algorithm to classify primary care electronic health records of alcohol consumption: experience using data linkage from UK Biobank and primary care electronic health data sources

ObjectivesDevelop a novel algorithm to categorise alcohol consumption using primary care electronic health records (EHRs) and asses its reliability by comparing this classification with self-reported alcohol consumption data obtained from...

PRO14 Healthcare Resource Utilisation and Costs Saved with Earlier Diagnosis of Behcets Disease: Using UK Primary Care Electronic Health Records

To quantify the direct medical cost of achieving an earlier diagnosis of Behcet’s disease (BD) with a digital health tool, from a payer perspective. 501,188 primary care electronic health records (EHRs) were scanned and 44 patients identified with the diagnostic code for BD (SNOMED CT code: 310701003). Using a digital health tool, six BD patients could be identified on average 3.7 (2- 7) years earlier. We analysed the EHRs of the 44 BD patients with at least 3 years of EHR entries before the first diagnosis (n=32). We examined the codes related to laboratory tests, clinical visits, referrals, and primary care interventions(medications & procedures) possibly attributed to BD. Cost data from the NHS National tariff 2020/21 and private care tariff were applied to the identified codes to calculate the direct medical costs incurred in the 3 years before diagnosis. Total healthcare resource utilisation associated with BD for the 32 study patients in the 3 years before diagnosis was 452 laboratory tests, 128 clinical visits, 66 referrals and 4 interventions. This equates to a total direct medical cost of £17,125 in primary care and an average cost of £535 (£0-£2,242) per patient. The average costs per patient in the year before diagnosis was £245 (£0-£1,077), and in the second and third years before diagnosis was £140 (£0-£570) and £150 (£0- £673) respectively. This study demonstrates the use of EHRs as an approach to measure the potential economic burden of diagnostic delay in diseases with low prevalence in the UK. Digital health tools can result in potential cost-savings in primary care through early diagnosis of BD.

Healthcare use by people who use illicit opioids (HUPIO): development of a cohort based on electronic primary care records in England.

Background: People who use illicit opioids such as heroin have substantial health needs, but there are few longitudinal studies of general health and healthcare in this population. Most research to date has focused on a narrow set of outcomes, including overdoses and HIV or hepatitis infections. We developed and validated a cohort using UK primary care electronic health records (Clinical Practice Research Datalink GOLD and AURUM databases) to facilitate research into healthcare use by people who use illicit opioid use (HUPIO). Methods: Participants are patients in England with primary care records indicating a history of illicit opioid use. We identified codes including prescriptions of opioid agonist therapies (methadone and buprenorphine) and clinical observations such as 'heroin dependence'. We constructed a cohort of patients with at least one of these codes and aged 18-64 at cohort entry, with follow-up between January 1997 and March 2020. We validated the cohort by comparing patient characteristics and mortality rates to other cohorts of people who use illicit opioids, with different recruitment methods. Results: Up to March 2020, the HUPIO cohort included 138,761 patients with a history of illicit opioid use. Demographic characteristics and all-cause mortality were similar to existing cohorts: 69% were male; the median age at index for patients in CPRD AURUM (the database with more included participants) was 35.3 (interquartile range 29.1-42.6); the average age of new cohort entrants increased over time; 76% had records indicating current tobacco smoking; patients disproportionately lived in deprived neighbourhoods; and all-cause mortality risk was 6.6 (95% CI 6.5-6.7) times the general population of England. Conclusions: Primary care data offer new opportunities to study holistic health outcomes and healthcare of this population. The large sample enables investigation of rare outcomes, whilst the availability of linkage to external datasets allows investigation of hospital use, cancer treatment, and mortality.