Clinical Practice Research Datalink Research Articles

Association between warfarin and direct oral anticoagulants users and risk of mortality: cohort studies in England and Hong Kong

Abstract Background The evidence of the association between warfarin and risk of mortality compared with direct oral anticoagulants (DOACs) has been mixed in current literature. Purpose To investigate the association between all-cause mortality and warfarin, compared with DOACs. Methods Population-based cohort studies were conducted using the data from the Hong Kong Clinical Data Analysis and Reporting System (CDARS) and United Kingdom Clinical Practice Research Datalink (CPRD) Aurum. People with atrial fibrillation aged ≥18 years who initiated warfarin or a DOAC during the study period (1st Jan 2014 - 31st Dec 2019) were identified. Follow-up started from the first date of oral anticoagulant prescription and ended at earliest of the date of death, the day before drug switching (between warfarin and DOACs), transferring out of the practice (CPRD Aurum only), last data collection date for the practice (CPRD Aurum only) or the end of study period. We estimated propensity score (PS) based on the list of potential confounders including demographics, co-morbidities, co-medications and polypharmacy. Cox proportional hazards regression model with PS-weighting was used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs). Results Among 227,190 new oral anticoagulant users, 54,367 (23.93%) died. The median follow-up was 2.96 and 2.45 years in CPRD and CDARS, respectively. Compared with DOAC use, a lower hazard of all-cause mortality was found in warfarin use (PS-weighted hazard ratio [HR]=0.94, 95% confidence interval [CI]=0.89-0.99) in CPRD; while a higher hazard was observed in warfarin users (PS-weighted HR=1.31, 95% CI=1.24-1.39) in CDARS, versus DOAC users. Consistent results were seen in both databases when stratified warfarin users by Time in Therapeutic Range (TTR). We observed a lower hazard of all-cause mortality in warfarin users with TTR≥65% (PS-weighted HR=0.90, 95% CI=0.85-0.94 in CPRD; 0.87, 0.79-0.96 in CDARS) and increased hazard in warfarin users with TTR<65% (PS-weighted HR=1.37, 95% CI=1.25-1.49 in CPRD; 1.57, 1.48-1.66 in CDARS), versus DOAC users. Conclusions The hazard of all-cause mortality associated with warfarin compared with DOACs were found to be conflicting across databases. However, both databases consistently showed that an increased hazard of all-cause mortality was only found in warfarin users with TTR<65% but not in those with TTR≥65%, suggesting that the differences in hazard of all-cause mortality associated with warfarin compared with DOAC, in part may depend on anticoagulation control in warfarin users. We recommend that warfarin users with poor INR control warrant closer clinical attention and could be considered for a switch to DOACs.

ESC Heart Failure Association age-adjusted natriuretic peptide thresholds for a new diagnosis of heart failure: diagnostic accuracy study

Abstract Background Natriuretic peptide (NP) testing is an important part of the heart failure (HF) diagnostic pathway. The European Society of Cardiology (ESC) HF guidelines currently recommend a NT-proBNP cut-off <125pg/mL to rule out HF in the community.1 However, NP level increases with age and an NT-proBNP above 125pg/mL is common in older people in the general population. The ESC Heart Failure Association (HFA) recently published a practical algorithm for early diagnosis of HF recommending age-specific rule-in NT-proBNP thresholds: ≥125pg/ml for patients aged under 50 years, ≥250pg/ml for patients aged 50-75 years, and ≥500pg/ml for patients over 75 years (Figure 1).2 Purpose Our aim was to assess NT-proBNP test performance for HF diagnosis at the ESC HFA age-specific thresholds. Method We analysed our existing diagnostic accuracy study dataset containing primary care data from the Clinical Practice Research Datalink (CPRD) linked to inpatient Hospital Episode Statistics (HES) inpatient data between 2004 and 2018. NT-proBNP results were taken from the GP record (index test) and the primary outcome (reference standard) was a HF diagnostic code entered in CPRD or HES within six months of the test. NT-proBNP accuracy for HF was assessed by calculating sensitivity, specificity, positive predictive values (PPV), and negative predictive value (NPV). Receiver operating characteristic (ROC) curves were plotted to assess overall test performance. Results In total, 229,580 patients had an NP test and 21,102 (9.2%) were diagnosed with HF. The current ESC NT-proBNP threshold ≥125pg/mL had sensitivity 94.6% (95% confidence intervals [CI] 94.2 to 95.0), specificity 50.0% (95%CI 49.7 to 50.3), PPV 16.4% (16.1 to 16.6), NPV 98.9% (98.8 to 99.0)). Area under the ROC curve was 0.874 (0.871 to 0.877). Age-specific NT-proBNP thresholds performance: Below 50 years (≥125pg/mL): 32,882 people tested, HF prevalence 3.8%. Sensitivity 83.5 % (81.3 to 85.5), specificity 77.6% (77.2 to 78.1), PPV 13.0% (12.2 to 13.7), NPV 99.2% (99.0 to 99.3). 50 to 75 years ≥250pg/mL: 105,771 people tested, HF prevalence 9.8%. Sensitivity 88.5% (87.9 to 89.1), specificity 67.8% (67.5 to 68.0), PPV 23.1% (22.6 to 23.5), NPV 98.2% (98.1 to 98.3). Above 75 years ≥500pg/mL: 16,694 people tested, HF prevalence 17.5%. Sensitivity 84.4% (83.0 to 85.7), specificity 63.5% (62.7 to 64.3), PPV 32.8% (31.7 to 33.9), NPV 95.0% (94.6 to 95.5). Conclusion The age-specific NT-proBNP thresholds in the HFA practical algorithm had higher specificity so fewer people without HF would be referred, reducing demand for echocardiography and cardiology assessment overall meaning people with HF could be seen earlier. However, lower sensitivity means some cases of HF would initially be missed. NT-proBNP was a reliable ‘rule-out’ test at population level. The optimal NT-proBNP threshold will depend on the priorities and capacity of the national healthcare system.

Patient characteristics and incidence of cardiac amyloidosis: an observational study in the UK clinical setting using electronic health records

Abstract Background Cardiac amyloidosis (CA) is associated with a high mortality rate due to sudden or progressive heart failure. It is often misdiagnosed due to non-specific symptoms, and its true incidence and prevalence in the United Kingdom (UK) remain uncertain. Purpose To describe patient characteristics and incidence of CA in the UK clinical setting using electronic health records. Methods In this retrospective observational study, anonymised, linked primary and secondary care data (Clinical Practice Research Datalink, CPRD and Hospital Episode Statistics, HES) were used to describe individuals diagnosed with CA. The CPRD database contains de-identified patient data sourced from a sample of general practitioner (GP) practices in the UK. Patients aged 18 years and older with a recorded Read or SNOMED-CT diagnosis code for CA or record of treatment with tafamidis (a licensed but not reimbursed treatment for transthyretin amyloid cardiomyopathy, ATTR-CM in the UK) in the period between January 2004 to December 2021 were included. The index date was defined as the earliest recorded date of CA diagnosis or tafamidis treatment. Demographic and clinical data were summarised using descriptive statistics. Incidence rates were reported per 100,000 person-years (PY). Results There were a total of 3,788 patients with CA (mean age: 67 years; 61% males) of which only 277 were coded as having wild-type transthyretin (ATTRwt) CA in mostly older patients (mean age: 79 years). The low patient counts of ATTRwt amyloidosis could be attributed to the lack of specificity in CA types and the low utilisation of ATTR amyloidosis codes. Heart Failure, essential hypertension and dyspnoea were the most common clinical presentations recorded in the 12 months to first diagnosis code or tafamidis prescription for CA (Table). The incidence of CA varied from 0.75 in 2004 to a peak of 1.96 in 2021, per 100,000 PY; a 2.9-fold increase in newly coded diagnoses in CA. There was a slight decline in the rates in 2020 and 2021 (Figure). Incidence rates in males and females increased over the 17-year period, with males consistently having higher rates than females. The incidence rate in females started from a lower base at 0.62/100,000 PY in 2004, increased by a factor of 1.9 to 1.1/100,000 PY, whereas in males it started at 0.88/100,000 PY and rose by a factor of 2.9 to 2.5 /100,000 PY. Incidence rates in subgroup populations have also been analysed. Conclusions In the UK clinical setting, the incidence rate of newly coded CA increased overall between 2004 and 2021, as well as in both male and female subgroups. The data suggest CA is becoming more common, or that awareness has improved, and there is now an urgent need for enhanced methods of detection, characterisation, and the need for thorough data capture to shape future pathway advancements.

Characterization of individuals with atherosclerotic cardiovascular disease, with and without chronic kidney disease: results from a UK electronic health record database

Abstract Introduction Elevated C-reactive protein (CRP) levels are often present in individuals with atherosclerotic cardiovascular disease (ASCVD), including in those who also have chronic kidney disease (CKD), and are associated with an increased risk of major adverse cardiovascular events (MACE). To understand unmet needs in the identification and management of at-risk individuals, we characterized patients with ASCVD in a UK database, according to CKD stage and CRP levels. Methods This study comprised separate retrospective cross-sectional and longitudinal cohort analyses of patients in the Clinical Practice Research Datalink (CPRD) Aurum database, a representative UK database of anonymized primary care records. In the cross-sectional analysis, CPRD Aurum was linked with Hospital Episode Statistics (HES) and Index of Multiple Deprivation data to assess patient characteristics on 1 Jan 2021 (defined as index date). In the cohort analysis, CPRD Aurum was linked with HES and Office for National Statistics mortality data to determine the incidence of 3-point MACE (3P-MACE) over approximately 5 years of follow-up (index date: 1 Jan 2016). Both analyses included individuals ≥ 18 years old with an ASCVD diagnosis in both primary and secondary care in the 5 years pre-index. Data were analysed using descriptive statistics. Patient characteristics were assessed by CKD stage; for individuals with CKD stage 3–4, characteristics were also assessed by CRP levels. Results In total, 225,767 and 230,055 individuals were included in the cross-sectional and cohort analyses, respectively. The prevalence of CKD stage 3–4 among individuals with ASCVD was 24.3% (Table 1). The proportion of individuals with elevated CRP (CRP ≥ 2 mg/L) and anaemia increased for successively higher CKD severity groups; in addition, glycated haemoglobin (HbA1c) ≥ 7% and elevated blood pressure were generally more frequently observed in individuals with CKD than in those without (Table 1). Individuals with ASCVD, CKD stage 3–4 and elevated CRP had higher HbA1c and total cholesterol levels and a greater proportion had obesity, anaemia and elevated blood pressure, compared with individuals with CRP < 2 mg/L (Table 2). Over 5 years’ follow-up, the incidence of 3P-MACE increased with CKD severity (Table 1) and with increasing CRP levels in individuals with ASCVD and CKD stage 3–4 (Table 2). However, only 28% of individuals with ASCVD and CKD stage 3–4 had CRP measurements. Conclusion Adults with ASCVD, CKD and elevated CRP identified from primary and secondary UK care records had a higher disease burden than individuals with CRP levels < 2 mg/L. In the UK, testing CRP levels is not routinely recommended, resulting in a high proportion of missing CRP measurements in this study. Routine testing of CRP may help in identifying patients with residual cardiovascular risk and has the potential to support risk assessment and management of patients with ASCVD and CKD.

Achieving integrated care through clusters of co-occurring diseases for people with multimorbidity

Abstract Background People with multimorbidity are often seen in many different specialist health services, resulting in fragmented care. Conventional services are designed around specialties based on anatomical systems, rather than diseases that occur together. We examined whether organising services around clusters of co-occurring diseases would lead to fewer interactions with unique services and so improve integrated care. Methods We used the Clinical Practice Research Datalink, a nationally representative sample of primary care data in England, including all patients registered on 1st January 2015 with multimorbidity (two or more of 212 long-term conditions (LTCs)). We used the assignment of 212 LTCs to 15 clusters of co-occurring diseases, derived from our earlier work. For each patient, we calculated the number of interactions with different specialists, under both the existing specialty-based model, and the hypothetical cluster-based model, under the assumption that a patient required review for each LTC. Results Of 6,200,973 patients with multimorbidity, the mean age was 53 years, with a median (interquartile range) of 8 (5 - 11) LTCs per person. There was considerable variation in the assignment of diseases comparing 15 clusters versus 15 specialties. Under the existing model, 409,708 (6.6%) patients interacted with only one specialty, and under the hypothetical cluster model, 584,446 (9.4%) patients interacted with only one cluster. On average, patients interacted with 3.66 specialties versus 3.30 hypothetical clusters (p < 0.001 for difference). Conclusions Health services designed around clusters of co-occurring diseases might lead patients to interact with fewer different services and so experience less fragmented care. Further work is needed to understand which specialties collaborating, and how, would have the greatest impact on improving integrated care. Key messages • Health services designed around clusters of co-occurring diseases led to a small reduction in the number of interactions with different services per year. • Designing services around clusters of co-occurring diseases might reduce fragmented care for people with multimorbidity.

The association between drugs and vaccines commonly prescribed to older people and bullous pemphigoid: a case-control study.

Bullous pemphigoid (BP) is an autoimmune skin disease that affects mainly older people. Numerous drugs have been previously associated with BP based on case series and small hospital-based studies. More reliable and precise estimates of associations between a broad selection of drugs/vaccines and BP will enable greater awareness of potential increased risk of BP following certain medicines and help identify clinical, histological and genomic characteristics of drug-induced BP for different types of culprit drugs. Greater awareness could lead to earlier recognition or suspicion of BP and referral to a dermatologist for diagnosis. Earlier diagnosis may lead to less aggressive treatment and improved well-being. To determine the association between drugs/vaccines commonly prescribed to older people and BP risk. We conducted a population-based nested case-control study between 1998-2021 using electronic primary care records from the Clinical Practice Research Datalink. We matched BP cases with up to 5 controls. Exposures were drugs/vaccines commonly prescribed to older people. We used multivariable conditional logistic regression adjusting for multiple drug use. For antibiotics, we considered prescriptions may be prescribed for undiagnosed symptoms of BP which resemble skin infection (protopathic bias) in a sensitivity analysis. Antibiotics were the therapeutic group associated with the highest risk of BP (OR: 4.60; 95%CI 4.40-4.80). However, after adjusting for protopathic bias, the OR reduced to 2.08 (95%CI 1.99-2.17). After adjusting for protopathic bias, of all antibiotic classes and subclasses, penicillins and penicillinase-resistant penicillins had the strongest associations with BP risk (OR: 3.44; 95%CI 3.29-3.60; sensitivity analysis OR; 1.74; 1.66-1.84 and OR: 7.56; 95%CI 7.15-8.00; sensitivity analysis OR: 2.64; 95%CI 2.45-2.85, respectively). Other drugs strongly associated with increased risk were gliptins (OR: 2.77; 95% CI 2.37-3.23), and second-generation antipsychotics (OR: 2.58; 95%CI 2.20-3.03). Healthcare professionals need to be aware of BP risk in older people particularly when prescribing penicillinase-resistant penicillins, gliptins, and second-generation antipsychotic drugs to recognise and manage BP early. Due to the low prevalence of disease, we do not suggest avoidance of drugs/vaccines to prevent BP. Further research should consider recency, dosage, and duration of antibiotic treatments.

Associations between general practice and outpatient care continuity in people with multimorbidity

Abstract Background Relational continuity in primary and secondary care is linked to better health outcomes for patients, but it is unclear whether metrics of continuity in each setting are associated. Our study examined the association between relational continuity in general practice (GP) and continuity of hospital outpatient specialties in people with clusters of multimorbidity. Methods We used GP records from the Clinical Practice Research Datalink, linked to Hospital Episode Statistics outpatient records. Included were patients ≥18 years, with two or more of 212 Long-Term Conditions (LTCs), registered throughout 2019 and with at least three GP and three outpatient appointments. The Continuity of Care Index (COCI) was calculated separately for visits to the same i) GP and ii) hospital specialty, where 0 represents no continuity and 1 represents all visits to the same GP or specialty. We assessed the association between the indices using univariable and multivariable fractional logit regression. We calculated expected COCI scores for each LTC and assessed variation in the COCI between LTCs assigned to one of 15 clusters of co-occurring diseases derived previously. A sensitivity analysis repeated analysis using the Sequential Continuity (SECON) score, which captures longitudinal continuity. Results Of 1,135,903 patients, 56.2% were ≥60 years. The median (IQR) COCI in GP was 0.22 (0.11-0.40) and in outpatients was 0.33 (0.20-0.52). Although statistically significant (p < 0.001) the relationship between the two measures was clinically insignificant in both univariable and multivariable models. No strong relationships were found between COCI scores and LTC clusters. Similar patterns were found using the SECON score. Conclusions We found no strong association between continuity of care in GP and outpatient settings, and no clear patterns with LTC clusters. This suggests that lack of continuity in secondary care is not compensated for by increased continuity of care in primary care. Key messages • We found no meaningful association between measures of continuity of care in general practice and hospital outpatient settings. • Our findings suggest that lack of continuity of outpatient care is not compensated for by increased continuity in primary care.

A novel dynamic risk score to predict venous thromboembolism (VTE) recurrences after 3 months of anticoagulation in patients with incident VTE and without active cancer

Abstract Background Predicting the risk of recurrent venous thromboembolism (rVTE) during and following anticoagulation is complex. Current models lack the ability to adapt to changing patient conditions over time and fail to account for the direct impact of anticoagulation in real-world clinical settings. Purpose To develop a dynamic risk prediction model for rVTE to help clinicians decide on the duration of anticoagulant therapy in conjunction with a model for bleeding (also submitted to the conference). Methods UK Clinical Practice Research Datalink data (2001-2020) was used to generate a retrospective cohort with first VTE who had received 3 months of anticoagulation. Patient episodes of rVTE were evaluated. Covariates were collected at baseline and subsequently as time-varying data to capture changes post-VTE. Hazards with 95% confidence intervals (CI) were estimated and covariates included in a Fine & Gray model used as predictors of rVTE. An additive (logarithmic) scoring scheme was developed from subdistribution hazard ratios, discrimination (expressed by the C-statistic) estimated from 10-fold cross-validation. Results 51,465 patients with a first VTE were included; 4041 rVTE were identified in 200,698 person-years of observation. Incidence rates for rVTE were 2.01 per 100 person-years. Nineteen independent predictors of rVTE (recorded before or after the VTE diagnosis) were included in the model, 13 associated with increased risk and 6 with decreased risk, Table 1. Patients recognised as lower-risk, medium and higher-risk (10.5%, 37% and 52.5% of the population, respectively, at 90 days after first VTE, assuming no anticoagulation treatment) had an annualised rVTE incidence rate of 2.20, 3.84, and 7.44 per 100 person-years. The C-statistic for rVTE was 0.65 (95%CI, 0.64-0.66). The points scored can be added or subtracted to predict risk (Table 2). Conclusions Our dynamic risk score effectively identifies patients at risk of rVTE three months post their initial VTE diagnosis. It allows for continuous monitoring of risk, and modelling of treatment impact, providing clinicians with a pragmatic tool to help determine treatment duration, and adapt their strategies in response to evolving patient conditions.Table 1:Predictors for VTE recurrenceTable 2:VTE recurrence risk score

Comorbidity neural network model for personalised recommendations in cardiometabolic risk factor management: towards early precision medicine in routine clinical practice

Abstract Background and Aims Hypertension and hypercholesterolaemia often develop in tandem to confer a risk of cardiovascular disease that may be more than additive. The present study explored early comorbid therapeutic intervention, i.e., beyond individual disease silos, while minimising data bias to provide clinical decision support tools that are portable across clinics. Methods A neural network was created to reproduce UK national clinical guidelines with an accuracy higher than 95%. Relying on real-world evidence from the UK primary care Clinical Practice Research Datalink electronic health record resource, a transfer learning methodology was then applied to achieve recommended risk factor goal attainment after two years. Shapley values derived from game theory were used to define similar digital twin cohorts, and to evaluate individualised recommendations that deviated from national clinical guidelines by rejecting a no-benefit hypothesis retrospectively in a classical proportion test. The methodology was ported to an independent hold-out dataset with anonymized records from six primary care clinics in the South of England. Results The adult patient population (n=469,496, mean+SD age 61+11yr; M 56% vs F 44%; primary 84% vs secondary prevention 16%) included 101,255 diagnosed for hypertension and 112,180 for cholesterol management. While a majority of the 52,504 therapeutic decisions in a test set agree with single morbidity guidelines, for 4639 deviations from the guidance, a significant minority was identified for whom alternative therapy was recommended. Rejecting a no-clinical benefit hypothesis at the 95% confidence level in a retrospective analysis, 1367 decisions showed statistically significant better outcomes over guidelines, for example by escalating anti-hypertensive therapies for patients receiving statin therapy (1085) or by complementing anti-hypertensive therapy with a statin (158). Similar findings were obtained when tested in a hold-out dataset with 13,725 decisions for 6531 patients. Conclusions The results support the contention that machine learning can identify patient subgroups for whom comorbid intervention deviating from clinical guidelines may be appropriate. A novel aspect of the study is the demonstration of clinically portable and actionable application of machine learning in the context of comorbid cardiometabolic diseases. The results represent practical steps towards personalised precision medicine for cardiovascular comorbidity in routine clinical practice.

A novel dynamic risk score to predict Clinically Significant Bleeding (CSB) after 3 months of anticoagulation in patients with incident Venous ThromboEmbolism (VTE) and without active cancer

Abstract Background Predicting the risk of clinically significant bleeding (CSB), bleeding requiring hospitalisation or fatal bleeding, during and following anticoagulation is complex. Current models lack the ability to adapt to changing patient conditions over time and fail to account for the direct impact of anticoagulation in real-world clinical settings. Purpose To develop a dynamic risk prediction model for CSB to help clinicians decide on the duration of anticoagulant therapy in conjunction with a model for recurrences of VTE (also submitted to the conference). Methods UK Clinical Practice Research Datalink data (2001-2020) was used to generate a retrospective cohort with first VTE who had received 3 months of anticoagulation. Patient episodes of CSB and recurrent VTE were evaluated. Covariates were collected at baseline and subsequently as time-varying data to capture changes post-VTE. Hazards with 95% confidence intervals (CI) were estimated and covariates included in a Fine & Gray model used as predictors of CSB. An additive (logarithmic) scoring scheme was developed from subdistribution hazard ratios, discrimination (expressed by the C-statistic) estimated from 10-fold cross-validation. Results 51,621 patients with a first VTE were included; 3307 CSB were identified in 206,292 person-years of observation. Incidence rates for CSB were 1.60 per 100 person-years. 18 independent predictors of CSB (recorded before or after the VTE diagnosis) were included in the model, Table 1. Patients recognised as lower-risk, medium and higher-risk (67.5 %, 15.9 % and 16.7 % of the population, respectively, at 90 days after first VTE, assuming no anticoagulation treatment) had an annualised incidence rate of 0.84, 2.94, and 9.45 per 100 person-years. The C-statistic for CSB was 0.78 (95%CI, 0.77-0.79). The points scored can be added to predict risk (Table 2). Conclusions Our dynamic score effectively identifies patients at risk of CSB three months post their initial VTE diagnosis. It allows for continuous monitoring of risk, and modelling of treatment impact, providing clinicians with a pragmatic tool to help determine treatment duration, and adapt their strategies in response to evolving patient conditions.Table 1:Predictors for CSBTable 2:CSB risk score

A real-world data analysis investigating the cardioprotective effects of glucagon-like peptide-1 receptor agonists among people with type 2 diabetes in England

Abstract Background Cardiovascular disease (CVD) is a common and major complication of type 2 diabetes (T2D). Anti-diabetic therapy aims to lower blood glucose, with primary clinical trial endpoints for new antidiabetic entrants focused on HbA1c. Research suggests some glucose-lowering drugs, including glucagon-like peptide-1 receptor agonists (GLP-1 RAs), can have a cardioprotective effect in people with T2D.1,2 Purpose To compare the real-world risk of major adverse cardiovascular event (MACE) in people with T2D treated with GLP-1 RAs compared to dipeptidyl peptidase-4 (DPP-4) inhibitors or basal insulin. Methods We conducted a retrospective cohort study using linked primary care data from the Clinical Practice Research Datalink Aurum, secondary care Hospital Episode Statistics and Office for National Statistics death registrations. We included patients aged ≥18 years with T2D at high-risk or very-high-risk of CVD (as defined by the European Society of Cardiology) who had ≥2 prescriptions of either GLP-1 RA (dulaglutide, liraglutide, and semaglutide), DPP-4 inhibitor (as part of line 4 therapy) or basal insulin from 01/01/2014-31/12/2018. We excluded those pregnant at therapy initiation or with type 1 diabetes. Follow-up started at second prescription of relevant therapy and ended at the earliest of two years later, pregnancy, death, end of primary care record, or 31/12/2019. We used data from patients’ medical histories to define baseline characteristics. We calculated the incidence of MACE per 100,000 person-years (PY), defined as non-fatal myocardial infarction, non-fatal stroke, and cardiovascular death. We used Cox proportional hazards regression, using LASSO to select confounding clinical variables including BMI, prior MACE and intercurrent antidiabetic treatments, to estimate risk difference of MACE among those treated with 1) GLP1-RA and DPP-4 inhibitor and 2) GLP1-RA and basal insulin. Results We included 63,237 patients, among whom the patients treated with GLP1-RA were younger and a higher proportion were obese or severely obese than those treated with DPP-4 inhibitor or basal insulin (Table 1). The incidence of MACE was highest in those treated with DPP-4 inhibitor (high-risk 1,503.0/100,000 PY; very-high-risk 6,000.7/100,000 PY) followed by basal insulin (high-risk 1,396.9/100,000 PY; very-high-risk 5,707.1/100,000 PY), and lowest in those treated with GLP1-RA (high-risk 377.1/100,000 PY; very-high-risk 1,946.1/100,000 PY). After adjustment, the risk of MACE among patients treated with GLP1-RA was still significantly lower than those treated with either DPP-4 inhibitor (high-risk aHR: 0.42, 95% CI 0.21-0.86 and very-high-risk aHR: 0.65, 95% CI 0.58-0.73) or those treated with basal insulin (high-risk aHR: 0.36, 95% CI 0.18-0.72 and very-high-risk aHR: 0.72, 95% CI 0.64-0.81). Conclusion(s) Our study supports clinical trial and other observational research studies which find a cardioprotective effect of GLP-1 RA use in people with T2D.

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.

Temporal trends and practice variation of paediatric diagnostic tests in primary care: retrospective analysis of 14 million tests

ObjectiveThe primary objective was to investigate temporal trends and between-practice variability of paediatric test use in primary care.Methods and analysisThis was a descriptive study of population-based data from Clinical Practice...

Bayesian blockwise inference for joint models of longitudinal and multistate data with application to longitudinal multimorbidity analysis.

Multistate models provide a useful framework for modelling complex event history data in clinical settings and have recently been extended to the joint modelling framework to appropriately handle endogenous longitudinal covariates, such as repeatedly measured biomarkers, which are informative about health status and disease progression. However, the practical application of such joint models faces considerable computational challenges. Motivated by a longitudinal multimorbidity analysis of large-scale UK health records, we introduce novel Bayesian inference approaches for these models that are capable of handling complex multistate processes and large datasets with straightforward implementation. These approaches decompose the original estimation task into smaller inference blocks, leveraging parallel computing and facilitating flexible model specification and comparison. Using extensive simulation studies, we show that the proposed approaches achieve satisfactory estimation accuracy, with notable gains in computational efficiency compared to the standard Bayesian estimation strategy. We illustrate our approaches by analysing the coevolution of routinely measured systolic blood pressure and the progression of three important chronic conditions, using a large dataset from the Clinical Practice Research Datalink Aurum database. Our analysis reveals distinct and previously lesser-known association structures between systolic blood pressure and different disease transitions.

Trajectories and management of vascular risk following the diagnosis of multiple sclerosis: A population-based matched cohort study between 1987 and 2018 in England.

People with multiple sclerosis (PwMS) have an increased cardiovascular and cerebrovascular disease burden, but this could be mitigated by vascular risk factor management. We compared the trajectories of vascular risk factors, vascular comorbidities and clinical management in PwMS against the general population post-MS diagnosis while controlling for frailty. Retrospective longitudinal analysis using English data from the Clinical Practice Research Datalink between 1987 and 2018 comprising PwMS matched with up to six controls without MS by age, sex and general practice. We compared 12,251 PwMS with 72,572 matched controls; 3.8% of PwMS had mild-moderate frailty, 1.2% more than matched controls. Compared to controls, PwMS had an elevated incidence of Type 2 diabetes (HR 1.18, 95% CI (1.04, 1.34)), and starting antihypertensive medications (HR 1.40, 95% CI (1.33, 1.47)). Among those with hypertension at baseline, blood pressure trajectories did not differ between PwMS and controls. PwMS had increased rates of meeting targets for hypertension management (HR 1.25, 95% CI (1.12, 1.41)). The observation that PwMS with hypertension are more likely to meet treatment targets than matched controls is encouraging, but the elevated rates of vascular comorbidities suggest that tighter vascular management may be needed in this population.

Prioritising primary care patients with unexpected weight loss for cancer investigation: diagnostic accuracy study (update)

ObjectiveTo quantify the predictive value of unexpected weight loss for cancer according to patient’s age, sex, smoking status, and concurrent clinical features (symptoms, signs, and abnormal blood test results).DesignDiagnostic accuracy...

Association between socioeconomic deprivation, ethnicity and health outcomes in preschool children with recurrent wheeze in England: a retrospective cohort study

BackgroundPreschool-aged children have among the highest burden of acute wheeze. We investigated differences in healthcare use, treatment and outcomes for recurrent wheeze/asthma in preschoolers from different ethno-socioeconomic backgrounds.MethodsRetrospective cohort study...

Use of cholesterol-lowering medications in relation to risk of primary liver cancer in the Clinical Practice Research Datalink.

Although the relation between statin use and liver cancer risk has been extensively examined, few studies have examined other cholesterol-lowering medications in relation to liver cancer risk. The authors examined five classes of nonstatin medications and liver cancer risk. A nested case-control including 3719 cases and 14,876 matched controls was conducted within the Clinical Practice Research Datalink. Additional matches on type 2 diabetes and chronic liver disease were also implemented. The medications examined included cholesterol absorption inhibitors, bile acid sequestrants, fibrates, niacin, and omega-3 fatty acids. Conditional logistic regression estimated odds ratios and 95% confidence intervals. Cholesterol absorption inhibitor use was associated with reduced liver cancer risk in the overall analysis (odds ratio, 0.69; 95% confidence interval, 0.50-0.96) and in analyses based on type 2 diabetes and chronic liver disease status. Although bile acid sequestrant use was associated with increased liver cancer risk in the overall analysis (odds ratio, 5.31; 95% confidence interval, 3.53-7.97), the results of the analyses based on type 2 diabetes and chronic liver disease status were inconsistent. [Correction added on 19 August 2024, after first online publication: In the preceding sentence, the value '3.534' has been changed to '3.54'.]. No associations were observed for the other medications. Cholesterol absorption inhibitors may be associated with reduced liver cancer risk. Whether bile acid sequestrant use was associated with increased risk was only partially supported in the current study.

Opportunities and challenges for identifying undiagnosed Rare Disease patients through analysis of primary care records: long QT syndrome as a test case.

Patients with rare genetic diseases frequently experience significant diagnostic delays. Routinely collected data in the electronic health record (EHR) may be used to help identify patients at risk of undiagnosed conditions. Long QT syndrome (LQTS) is a rare inherited cardiac condition associated with significant morbidity and premature mortality. In this study, we examine LQTS as an exemplar disease to assess if clinical features recorded in the primary care EHR can be used to develop and validate a predictive model to aid earlier detection. 1495 patients with an LQTS diagnostic code and 7475 propensity-score matched controls were identified from 10.5million patients' electronic primary care records in the UK's Clinical Practice Research Datalink (CPRD). Associated clinical features recorded before diagnosis (with p < 0.05) were incorporated into a multivariable logistic regression model, the final model was determined by backwards regression and validated by bootstrapping to determine model optimism. The mean age at LQTS diagnosis was 58.4 (SD 19.41). 18 features were included in the final model. Discriminative accuracy, assessed by area under the curve (AUC), was 0.74, (95% CI 0.73, 0.75) (optimism 6%). Features occurring at significantly greater frequency before diagnosis included: epilepsy, palpitations, syncope, collapse, mitral valve disease and irritable bowel syndrome. This study demonstrates the potential to develop primary care prediction models for rare conditions, like LQTS, in routine primary care records and highlights key considerations including disease suitability, finding an appropriate linked dataset, the need for accurate case ascertainment and utilising an approach to modelling suitable for rare events.

Treatment for osteoporosis and risk of osteonecrosis of the jaw among female patients in the United Kingdom Clinical Practice Research Datalink.

Osteonecrosis of the jaw (ONJ) is a rare adverse effect of antiresorptive drug use; however, the magnitude of risk in osteoporosis patients has not been clearly described. We conducted a cohort study among cancer-free female patients aged 40-89 with, or at risk for, osteoporosis in United Kingdom Clinical Practice Research Datalink (CPRD) Aurum. We followed patients from first osteoporosis treatment until first of osteonecrosis diagnosis, age 90, record end, or other prespecified censoring event, and accumulated person-time by osteoporosis treatment. ONJ cases were selected from CPRD Aurum and linked Hospital Episode Statistics data using an algorithm and manual review. We estimated incidence rates (IR) of ONJ by current treatment type and post discontinuation. We conducted a nested case-control analysis to further describe risk by cumulative dose and duration of antiresorptive therapies. Among 467,654 eligible patients, there were 208 ONJ cases. IR among patients currently treated with antiresorptives (primarily alendronate) was 1.2 (95% confidence interval [CI] 1.0-1.4) per 10,000 person-years. Compared with past use of antiresorptives, odds ratios of ONJ were 3.0 (95% CI 1.5-5.7) after 2-3 years of treatment and 8.1 (95% CI 4.4-15) after 10 years. However, absolute risks remained low (~ 0.05% after 5 years and ~ 0.18% after 10 years) and elevated risks diminished to near zero within 6 to 9 months of discontinuation. Risk of ONJ increased after 2-3 years of treatment with antiresorptives; however, the absolute risk was low and returned to baseline shortly after treatment discontinuation.