Scientific
publications

BACKGROUND: Although cognitive impairment and dementia are common comorbidities in patients with chronic obstructive pulmonary disease (COPD), estimates of incidence following a diagnosis of COPD are inconclusive. OBJECTIVE: To determine the incidence of cognitive impairment and dementia in people with and without a COPD diagnosis. METHODS: A population-based study using UK General Practice (GP) health records from The Health Improvement Network database was conducted. Patients with confirmed COPD diagnosis, ≥40 years old, were matched to up to four subjects without a COPD diagnosis by age, sex and GP practice. Cox proportional hazards models were used to assess the incidence rates of cognitive impairment and dementia. RESULTS: Of patients with COPD (n = 62,148), 9% developed cognitive impairment, compared with 7% of subjects without COPD (n = 230,076), p < 0.001. The incidence of cognitive impairment following COPD diagnosis was greater than in subjects without COPD following index date (adjusted Hazard Ratio (aHR), 1.21; 95% CI: 1.16 ─ 1.26, p < 0.001). The coded incidence of either cognitive impairment or dementia was also greater in patients with COPD following adjustment for confounders (aHR: 1.13, 95% CI: 1.09 ─ 1.18, p < 0.001). Coded incident dementia alone was not different between patients with COPD and subjects without COPD (aHR, 0.91, 95% CI: 0.83 ─ 1.01, p = 0.053). CONCLUSION: Despite the increased incidence of cognitive impairment in patients with COPD, incidence of dementia was not as frequently recorded in patients with COPD. This raises the concern of undiagnosed dementia and emphasises the need for a systematic assessment in this population.

BACKGROUND: To manage the risk factors and to improve clinical outcomes, patients with stroke commonly receive multiple cardiovascular medications. However, there is a lack of evidence on the optimum combination of medication therapy in the primary care setting after ischemic stroke. Therefore, this study aimed to investigate the effect of multiple cardiovascular medications on long-term survival after an incident stroke event (ischemic stroke or transient ischemic attack (TIA)). METHODS: This study consisted of 52,619 patients aged 45 and above with an incident stroke event between 2007 and 2016 in The Health Improvement Network database. We estimated the risk of all-cause mortality in patients with multiple cardiovascular medications versus monotherapy using a marginal structural model. RESULTS: During an average follow-up of 3.6 years, there were 9230 deaths (7635 in multiple cardiovascular medication groups and 1595 in the monotherapy group). Compared with patients prescribed monotherapy only, the HRs of mortality were 0.82 (95% CI 0.75-0.89) for two medications, 0.65 (0.59-0.70) for three medications, 0.61 (0.56-0.67) for four medications, 0.60 (0.54-0.66) for five medications and 0.66 (0.59-0.74) for ≥ six medications. Patients with any four classes of antiplatelet agents (APAs), lipid-regulating medications (LRMs), angiotensin-converting enzyme inhibitors (ACEIs)/angiotensin receptor blockers (ARBs), beta-blockers, diuretics and calcium channel blockers (CCBs) had the lowest risk of mortality (HR 0.51, 95% CI 0.46-0.57) versus any one class. The combination containing APAs, LRMs, ACEIs/ARBs and CCBs was associated with a 61% (95% CI 53-68%) lower risk of mortality compared with APAs alone. CONCLUSION: Our results suggested that combination therapy of four or five cardiovascular medications may be optimal to improve long-term survival after incident ischemic stroke or TIA. APAs, LRMs, ACEIs/ARBs and CCBs were the optimal constituents of combination therapy in the present study.

BACKGROUND: Initiation of statins for the primary prevention of cardiovascular disease (CVD) should be based on CVD risk estimates, but their use is suboptimal. AIM: To investigate the factors influencing statin prescribing when clinicians code and do not code estimated CVD risk (QRISK2). DESIGN AND SETTING: A historical cohort of patients who had lipid tests in a database (IQVIA Medical Research Data) of UK primary care records. METHOD: The cohort comprised 686 560 entries (lipid test results) between 2012 and 2016 from 383 416 statin-naive patients without previous CVD. Coded QRISK2 scores were extracted, with variables used in calculating QRISK2 and factors that might influence statin prescribing. If a QRISK2 score was not coded, it was calculated post hoc. The outcome was initiation of a statin within 60 days of the lipid test result. RESULTS: Of the entries, 146 693 (21.4%) had a coded QRISK2 score. Statins were initiated in 6.6% (95% confidence interval [CI] = 6.4% to 6.7%) of those with coded and 4.1% (95% CI = 4.0% to 4.1%) of uncoded QRISK2 ( <0.001). Statin initiations were consistent with National Institute for Health and Care Excellence guideline recommendations in 85.0% (95% CI = 84.2% to 85.8%) of coded and 44.2% (95% CI = 43.5% to 44.9%) of uncoded QRISK2 groups ( <0.001). When coded, QRISK2 score was the main predictor of statin initiation, but total cholesterol was the main predictor when a QRISK2 score was not coded. CONCLUSION: When a QRISK2 score is coded, prescribing is more consistent with guidelines. With no QRISK2 score, prescribing is mainly based on total cholesterol. Using QRISK2 is associated with statin prescribing that is more likely to benefit patients. Promoting the routine CVD risk estimation is essential to optimise decision making.

The use of primary care electronic health records for research is abundant. The benefits gained from utilising such records lies in their size, longitudinal data collection and data quality. However, the use of such data to undertake high quality epidemiological studies, can lead to significant challenges particularly in dealing with misclassification, variation in coding and the significant effort required to pre-process the data in a meaningful format for statistical analysis. In this paper, we describe a methodology to aid with the extraction and processing of such databases, delivered by a novel software programme; the "Data extraction for epidemiological research" (DExtER). The basis of DExtER relies on principles of extract, transform and load processes. The tool initially provides the ability for the healthcare dataset to be extracted, then transformed in a format whereby data is normalised, converted and reformatted. DExtER has a user interface designed to obtain data extracts specific to each research question and observational study design. There are facilities to input the requirements for; eligible study period, definition of exposed and unexposed groups, outcome measures and important baseline covariates. To date the tool has been utilised and validated in a multitude of settings. There have been over 35 peer-reviewed publications using the tool, and DExtER has been implemented as a validated public health surveillance tool for obtaining accurate statistics on epidemiology of key morbidities. Future direction of this work will be the application of the framework to linked as well as international datasets and the development of standardised methods for conducting electronic pre-processing and extraction from datasets for research purposes.

OBJECTIVE: Effervescent, soluble, dispersible formulations contain considerable amounts of sodium. In 2013, we previously confirmed the association between sodium-containing medications and cardiovascular risks. This study aimed to determine the changes in the prescribing pattern in clinical practice following this publication. DESIGN: A longitudinal cross-sectional study. SETTING: Primary care in the UK from 2009 to 2018. PARTICIPANTS: Prescribing information in The Health Improvement Network (THIN) and Prescription Cost Analysis (PCA) databases in the UK. OUTCOME MEASUREMENTS: Prescription rates per 10 000 inhabitants were calculated using the number of prescriptions or the number of drug-using patients over the total number of inhabitants, and the prescription rates were measured at annual intervals. Prescribing trends from 2009 to 2018 were indexed with yearly data from THIN and PCA. Interrupted time series analysis (ITSA) was conducted with monthly data in THIN. RESULTS: From the THIN database, a total of 3 651 419 prescription records from 446 233 patients were included. The prescribing rate of sodium-containing medications changed from 848.3/10 000 inhabitants in 2009 to 571.6/10 000 inhabitants in 2018. The corresponding figures from PCA data were of 631.0/10 000 inhabitants in 2009 and 423.8/10 000 inhabitants in 2018. ITSA showed the prescribing trend reduced significantly during the postpublication period (prescribing rate: slope change=-0.26; 95% CI -0.45 to -0.07; p=0.009; proportion of patients: slope change=-0.22; 95% CI -0.35 to -0.09; p<0.001), but no change in postpublication level from baseline. The prescribing rates for the non-sodium-containing standard formulations were relatively stable over the study period. The reduction in the proportion of patients using sodium-containing medications was only significant in patients over 45 years old. CONCLUSIONS: The prescribing of sodium-containing medications in the UK primary care has declined significantly during the postpublication period. Changes in the prescribing trends for sodium-containing medications varied across regions of the UK and patient age groups.

IMPORTANCE: Adhesion-related complications (ARCs), including small-bowel obstruction, are common complications of intra-abdominal surgery. Statins, which have antifibrotic pleiotropic effects, inhibit adhesion formation in murine models but have not been assessed in humans. OBJECTIVE: To assess whether statin use at the time of intra-abdominal surgery is associated with a reduction in ARCs. DESIGN, SETTING, AND PARTICIPANTS: These 2 separate retrospective cohort studies (The Health Improvement Network [THIN] and Optum's Clinformatics Data Mart [Optum]) compared adults receiving statins with those not receiving statins at the time of intra-abdominal surgery. Individuals undergoing intra-abdominal surgery from January 1, 1996, to December 31, 2013, in the United Kingdom and from January 1, 2000, to December 31, 2016, in the US were included in the study. Those with obstructive events before surgery or a history of inflammatory bowel disease were excluded. Data analysis was performed from September 1, 2012, to November 24, 2020. EXPOSURE: The primary exposure was statin use at the time of surgery. MAIN OUTCOMES AND MEASURES: The primary outcome was ARCs, defined as small-bowel obstruction or need for adhesiolysis, occurring after surgery. Sensitivity analyses included statin use preceding but not concurrent with surgery, fibrate use, and angiotensin-converting enzyme inhibitor use. All analyses were adjusted for age, sex, and conditions associated with microvascular disease, such as hypertension, hyperlipidemia, obesity, and tobacco use; surgical approach and site; and diagnosis of a malignant tumor. RESULTS: A total of 148 601 individuals met the inclusion criteria for THIN (mean [SD] age, 49.6 [17.7] years; 70.1% female) and 1 188 217 for Optum (mean [SD] age, 48.2 [16.4] years; 72.6% female). A total of 2060 participants (1.4%) experienced an ARC in THIN and 54 136 (4.6%) in Optum. Statin use at the time of surgery was associated with decreased risk of ARCs (THIN: adjusted hazard ratio [HR], 0.81; 95% CI, 0.71-0.92; Optum: adjusted HR, 0.92; 95% CI, 0.90-0.95). Similar associations were appreciated between statins and small-bowel obstruction (THIN: adjusted HR, 0.80; 95% CI, 0.70-0.92; Optum: adjusted HR, 0.88; 95% CI, 0.85-0.91). CONCLUSIONS AND RELEVANCE: This study's findings suggest that, among individuals in 2 separate cohorts undergoing intra-abdominal surgery, statin use may be associated with a reduced risk of postoperative ARCs. Statins may represent an inexpensive, well-tolerated pharmacologic option for preventing ARCs.

PURPOSE: Digoxin affects several cellular pathways involved in tumorigenesis. We sought to determine the association between digoxin use and pancreatic cancer risk and survival. METHODS: A nested case-control study using The Health Improvement Network (THIN), a population-representative database from the United Kingdom (UK). Cases included all individuals with incident diagnosis of pancreatic cancer. Each case was matched to up to four controls using incidence density sampling based on age, sex, practice site, calendar time, and duration of follow-up. Exposure of interest was digoxin therapy before cancer diagnosis. Odds ratios (ORs) and 95% confidence intervals (CIs) for the association between digoxin use and pancreatic cancer risk were estimated using conditional logistic regression. We further conducted a retrospective cohort study among pancreatic cancer cases using Cox regression model in order to evaluate the association between digoxin use and overall survival. RESULTS: We identified 4,113 cases with incident pancreatic cancer and 16,072 matched controls. The adjusted OR for diagnosis of pancreatic cancer among active digoxin users was 1.41 (95% CI 1.16-1.72). The risk did not change among active users with duration of therapy of more than 1 year (adjusted OR of 1.39, 95% CI 1.11-1.76). Digoxin was not associated with change in overall survival with an adjusted hazard ratio of 0.97 (95% CI 0.81-1.18). CONCLUSIONS: Digoxin use was associated with modestly increased pancreatic cancer risk but did not affect overall survival.

OBJECTIVE: To use clinically informed machine learning to derive prediction models for early and late premature death in epilepsy. METHODS: This was a population-based primary care observational cohort study. All patients meeting a case definition for incident epilepsy in the Health Improvement Network database for inclusive years 2000-2012 were included. A modified Delphi process identified 30 potential risk factors. Outcome was early (within 4 years of epilepsy diagnosis) and late (4 years or more from diagnosis) mortality. We used regularized logistic regression, support vector machines, Gaussian naive Bayes, and random forest classifiers to predict outcomes. We assessed model calibration, discrimination, and generalizability using the Brier score, mean area under the receiver operating characteristic curve (AUC) derived from stratified fivefold cross-validation, plotted calibration curves, and extracted measures of association where possible. RESULTS: We identified 10 499 presumed incident cases from 11 194 182 patients. All models performed comparably well following stratified fivefold cross-validation, with AUCs ranging from 0.73 to 0.81 and from 0.71 to 0.79 for early and late death, respectively. In addition to comorbid disease, social habits (alcoholism odds ratio [OR] for early death = 1.54, 95% confidence interval [CI] = 1.12-2.11 and OR for late death = 2.62, 95% CI = 1.66-4.16) and treatment patterns (OR for early death when no antiseizure medication [ASM] was prescribed at baseline = 1.33, 95% CI = 1.07-1.64 and OR for late death after receipt of enzyme-inducing ASM at baseline = 1.32, 95% CI = 1.04-1.66) were significantly associated with increased risk of premature death. Baseline ASM polytherapy (OR = 0.55, 95% CI = 0.36-0.85) was associated with reduced risk of early death. SIGNIFICANCE: Clinically informed models using routine electronic medical records can be used to predict early and late mortality in epilepsy, with moderate to high accuracy and evidence of generalizability. Medical, social, and treatment-related risk factors, such as delayed ASM prescription and baseline prescription of enzyme-inducing ASMs, were important predictors.

Sodium-glucose co-transporter-2 (SGLT2) inhibitors are widely prescribed in people with type 2 diabetes. We aimed to investigate whether SGLT2 inhibitor prescription is associated with COVID-19, when compared with an active comparator. We performed a propensity-score-matched cohort study with active comparators and a negative control outcome in a large UK-based primary care dataset. Participants prescribed SGLT2 inhibitors (n = 9948) and a comparator group prescribed dipeptidyl peptidase-4 (DPP-4) inhibitors (n = 14 917) were followed up from January 30 to July 27, 2020. The primary outcome was confirmed or clinically suspected COVID-19. The incidence rate of COVID-19 was 19.7/1000 person-years among users of SGLT2 inhibitors and 24.7/1000 person-years among propensity-score-matched users of DPP-4 inhibitors. The adjusted hazard ratio was 0.92 (95% confidence interval 0.66 to 1.29), and there was no evidence of residual confounding in the negative control analysis. We did not observe an increased risk of COVID-19 in primary care amongst those prescribed SGLT2 inhibitors compared to DPP-4 inhibitors, suggesting that clinicians may safely use these agents in the everyday care of people with type 2 diabetes during the COVID-19 pandemic.

OBJECTIVES: To examine trends in the initial prescription of commonly-prescribed analgesics and patient- as well as practice-level factors related to their selection in incident OA. METHODS: Patients consulting with incident clinical OA between 2000-2016 were identified within The Health Improvement Network in the United Kingdom (UK) general practice. Excluded were patients who had history of cancer or were prescribed the analgesics of interest within 6 months before diagnosis of OA. Initial analgesic prescription included oral non-selective NSAID, oral selective cyclooxygenase-2 inhibitor, topical NSAID, paracetamol, topical salicylate or oral/transdermal opioid within 1 month after OA diagnosis. RESULTS: ∼44% of patients with incident OA (n = 125 696) were prescribed one of these analgesics. Incidence of oral NSAID prescriptions decreased whereas other analgesic prescriptions, including oral opioid prescriptions, increased (all P-for-trend < 0.001). Patients with a history of gastrointestinal disease were more likely to receive topical NSAIDs, paracetamol or oral/transdermal opioids. Only 38% of patients with history of gastrointestinal disease and 21% of patients without it had co-prescription of gastroprotective agent with oral NSAIDs. Oral/transdermal opioid prescription was higher among the elderly (≥65 years), women, obesity, current smoker, and patients with gastrointestinal, cardiovascular or chronic kidney disease. Prescription of oral opioids increased with social deprivation (P-for-trend < 0.05) and was highest in Scotland, whereas transdermal opioid prescription was highest in Northern Ireland (all P-for-homogeneity-test < 0.05). CONCLUSION: The initial prescription pattern of analgesics for OA has changed over time in the UK. Co-prescription of gastroprotective agents with oral NSAIDs remains suboptimal, even among those with prior gastrointestinal disease.

OBJECTIVE: To estimate risk of fracture in men and women with recent diagnosis of type 2 diabetes compared with individuals without diabetes. RESEARCH DESIGN AND METHODS: In this cohort study, we used routinely collected U.K. primary care data from The Health Improvement Network. In adults (>35 years) diagnosed with type 2 diabetes between 2004 and 2013, fractures sustained until 2019 were identified and compared with fractures sustained in individuals without diabetes. Multivariable models estimated time to first fracture following diagnosis of diabetes. Annual prevalence rates included at least one fracture in a given year. RESULTS: Among 174,244 individuals with incident type 2 diabetes and 747,290 without diabetes, there was no increased risk of fracture among males with diabetes (adjusted hazard ratio [aHR] 0.97 [95% CI 0.94, 1.00]) and a small reduced risk among females (aHR 0.94 [95% CI 0.92, 0.96]). In those aged ≥85 years, those in the diabetes cohort were at significantly lower risk of incident fracture (males: aHR 0.85 [95% CI 0.71, 1.00]; females: aHR 0.85 [95% CI 0.78, 0.94]). For those in the most deprived areas, aHRs were 0.90 (95% CI 0.83, 0.98) for males and 0.91 (95% CI 0.85, 0.97) for females. Annual fracture prevalence rates, by sex, were similar for those with and without type 2 diabetes. CONCLUSIONS: We found no evidence to suggest a higher risk of fracture following diagnosis of type 2 diabetes. After a diagnosis of type 2 diabetes, individuals should be encouraged to make positive lifestyle changes, including undertaking weight-bearing physical activities that improve bone health.

There is limited understanding on patterns of systemic treatment in adults with moderate-to-severe atopic dermatitis (AD) in the UK. To characterize treatment patterns in adult AD patients prescribed immunosuppressants (IMMs) in the primary care setting. Six hundred and fifty-six patients with AD (6.6%) were prescribed IMM in the analysis (mean age 52.1 years; 59.1% female; age-adjusted Charlson comorbidity index 1.4). Most prevalent (>5%) conditions at baseline were depression (10.8%), contact dermatitis (10.7%), rheumatological disease (7.9%), skin/subcutaneous tissue disorders (6.4%), upper respiratory disease (5.8%), and psoriasis (5.2%). At baseline, up to 50% of patients were prescribed ≥1 IMM. During follow-up, 42.7% of patients were prescribed oral corticosteroids (OCSs), increasing in line with IMM exposure. The most commonly prescribed IMM was methotrexate (43.3%). Ciclosporin, the only approved IMM for AD, was prescribed to 16.9% of patients. The prevalence of comorbidities and high rate of IMM prescriptions demonstrate the impact of AD on quality of life. The frequency of OCS prescribing in AD patients treated with IMMs suggests a lack of disease control with existing therapies, and an unmet need for safe and effective targeted agents for long-term disease control.

OBJECTIVES: Osteoarthritis is a leading cause of immobility and joint replacement, two strong risk factors for venous thromboembolism (VTE). We aimed to examine the relation of knee, hip and hand osteoarthritis to the risk of VTE and investigate joint replacement as a potential mediator. METHODS: We conducted three cohort studies using data from The Health Improvement Network. Up to five individuals without osteoarthritis were matched to each case of incident knee (n=20 696), hip (n=10 411) or hand (n=6329) osteoarthritis by age, sex, entry time and body mass index. We examined the relation of osteoarthritis to VTE (pulmonary embolism and deep vein thrombosis) using a multivariable Cox proportional hazard model. RESULTS: VTE developed in 327 individuals with knee osteoarthritis and 951 individuals without osteoarthritis (2.7 vs 2.0 per 1000 person-years), with multivariable-adjusted HR being 1.38 (95% CI 1.23 to 1.56). The indirect effect (HR) of knee osteoarthritis on VTE through knee replacement was 1.07 (95% CI 1.01 to 1.15), explaining 24.8% of its total effect on VTE. Risk of VTE was higher in hip osteoarthritis than non-osteoarthritis (3.3 vs 1.8 per 1000 person-years; multivariable-adjusted HR=1.83, 95% CI 1.56 to 2.13). The indirect effect through hip replacement yielded an HR of 1.14 (95% CI 1.04 to 1.25), explaining 28.1% of the total effect. No statistically significant difference in VTE risk was observed between hand osteoarthritis and non-osteoarthritis (1.5 vs 1.6 per 1000 person-years; multivariable-adjusted HR=0.88, 95% CI 0.67 to 1.16). CONCLUSION: Our large population-based cohort study provides the first evidence that knee or hip osteoarthritis, but not hand osteoarthritis, was associated with an increased risk of VTE, and such an association was partially mediated through knee or hip replacement.

PURPOSE: Treatment guidance for chronic obstructive pulmonary disease (COPD) recommends inhaled corticosteroid (ICS)+long-acting muscarinic antagonist+long-acting β -agonist (LABA) triple therapy for patients who experience recurrent exacerbations, persistent breathlessness, or exercise limitation on dual therapy. However, information is limited on pathways to triple therapy in the UK. PATIENTS AND METHODS: A retrospective cohort study was conducted using de-identified patient-level data from UK primary care electronic medical records from January 1, 2005 to May 1, 2016. Data were included from patients who had their first triple therapy regimen (index date) recorded during the study period and a minimum of 12 months' pre-index data. Treatment pathways to triple therapy were recorded, and the proportion of patients on triple therapy before their COPD diagnosis was determined. Adherence to triple therapy was estimated using the proportion of days covered (PDC). RESULTS: After applying eligibility criteria, 82,300 patients were included, with a mean age at COPD diagnosis of 64.7 years. The major treatment pathway (27.9%) was the first initiation of ICS+LABA prior to triple therapy. Following COPD diagnosis, the median time to triple therapy was approximately 3.5 years. The estimated mean adherence to triple therapy was 81.8% PDC. Multivariate analysis showed that the following groups were more likely to have received previous therapy prior to triple therapy: females (versus males), patients with asthma (versus those without asthma), severe COPD (versus those with non-severe COPD), or fewer exacerbations (versus those with more exacerbations). CONCLUSION: Treatment pathways to triple therapy in the UK are diverse, highlighting the need to better understand factors involved in clinical decision-making.

INTRODUCTION: Maintenance treatment strategies in COPD recommend inhaled corticosteroid (ICS) + long-acting muscarinic antagonist (LAMA) + long-acting β -agonist (LABA) triple therapy after initial dual therapy. Little is known about how treatment pathways to triple therapy vary across countries in clinical practice. METHODS: This multi-country, retrospective cohort study (conducted 1 January 2005-1 May 2016) included patients with a COPD diagnosis, and (UK only) evidence of smoking history, or (France, Italy, Germany, and Australia) an indicator confirming COPD diagnosis, a first instance of triple therapy recorded during the study period and ≥ 12 months of data prior to this date. Treatment pathways to triple therapy were analyzed in patients whose first instance of triple therapy was on or after the initial COPD diagnosis. The proportion of patients who initiated triple therapy prior to initial COPD diagnosis was also estimated. Meta-analyses of the main results were performed. RESULTS: In 130,729 patients across all countries, mean age (standard deviation) ranged from 63.4 (10.4) years (Germany) to 69.8 (9.9) years (Italy), and median time (interquartile range) from initial COPD diagnosis to first prescription of triple therapy ranged from 16.9 (5.7-36.2) months (Australia) to 42.5 (13.9-87.4) months (UK). ICS + LABA was the most common treatment pathway prior to triple therapy in the UK, Germany, and Italy (27.3%-31.6%); no previous maintenance therapy prior to triple therapy was the most common pathway in France and Australia (32.5% and 37.9%, respectively). Meta-analyses provided a pooled estimate of 20.4% (95% confidence interval: 13.8%-29.1%) for the proportion of patients initiating triple therapy at or before initial COPD diagnosis. CONCLUSIONS: In this retrospective cohort study, treatment pathways to triple therapy were diverse within and between countries. The differing impact of treatments may affect quality of life and disease control in patients with COPD. Further analyses should investigate factors influencing pathways to triple therapy.