Scientific
publications

AIMS/HYPOTHESIS: Our objectives were to explore whether the phenomenon of HbA 'tracking' occurs in individuals with type 1 diabetes, how long after diagnosis does tracking take to stabilise, and whether there is an effect of sex and age at diagnosis on tracking. METHODS: A total of 4525 individuals diagnosed with type 1 diabetes between 1 January 1995 and 1 May 2015 were identified from The Health Improvement Network (THIN) database. Mixed models were applied to assess the variability of HbA levels over time with random effects on general practices (primary care units) and individuals within practices. RESULTS: 4525 individuals diagnosed with type 1 diabetes were identified in THIN over the study period. The greatest difference in mean HbA measurement (-7.0 [95% CI -8.0, -6.1] mmol/mol [0.6%]) was seen when comparing measurements made immediately after diagnosis (0-1 year since diagnosis) with those at 10 or more years (the reference category). The mean difference in HbA for the successive periods compared with 10 or more years after diagnosis declined and was no longer statistically significant after 5 years. In the stratified analysis using sex and age group there was considerable heterogeneity with adult onset type 1 diabetes appearing to track earlier and at a lower mean HbA . CONCLUSIONS/INTERPRETATION: In individuals with type 1 diabetes, glycaemic control measured by HbA settles onto a long-term 'track' and this occurs on average by 5 years following diagnosis. Age at diagnosis modifies both the rate at which individuals settle into their track and the absolute HbA tracking level for the next 10 years.

BACKGROUND: Cancer registry data show that survival of colorectal cancer (CRC) in the United Kingdom is poor compared with other European countries and the United States, yet these data sources lack information on patient comorbidities and medication use, which could help explain these differences. METHODS: Among individuals aged 40-89 years in The Health Improvement Network (2000-2014), we identified first ever cases of CRC and calculated incidence rates with 95% confidence intervals (CIs). For CRC cases and non-cases in two separate calendar years (2002 and 2014), we evaluated patient demographics, lifestyle factors, comorbidities and medication use and bowel screening. RESULTS: The incidence of CRC remained relatively constant across the study period; incidence rates per 10,000 person-years (95% CIs) were 9.27 (8.59-1.01) in 2000, 10.65 (10.15-11.18) in 2007 and 8.37 (7.93-8.83) in 2014. Incidence rates per 10,000 person-years were higher in men than women at 11.44 (95% CI: 10.35-12.66) vs. 7.40 (95% CI: 6.59-8.32) in 2000, and 9.39 (95% CI: 8.74-10.10) vs. 7.38 (95% CI: 6.81-8.00) in 2014. An increase was seen in the proportion of CRC cases diagnosed at age < 60 years. In 2002, 3.5% of CRC cases were diagnosed at age 40-49 compared with 5.1% in 2014 (p = 0.064). Similarly, in 2002, 12.5% were diagnosed at age 50-59 years compared with 16.2% in 2014 (p = 0.002). Between 2002 and 2014, previous bowel screening increased in both CRC cases (+ 10.6%) and non-cases (+ 9.7%)(p < 0.001 for both groups). Greater rises in the following were seen among CRC cases compared with non-cases: diabetes (+ 9.3% vs. + 3.3%; p < 0.001 for both), obesity (+ 14.5% vs. + 10.1%; p < 0.001 for both), hypertension (+ 8.3% vs. + 3.6%; p < 0.001 for both), atrial fibrillation (+ 2.6% [p < 0.01] vs. + 0.3% [p < 0.001]), and use of proton pump inhibitors (+ 11.5% vs. + 9.0%), anti-hypertensives (+ 9.9% vs. + 1.4%) and warfarin (+ 3.2% vs. + 0.4%); p < 0.001 for CRC cases and non-cases with respect to each medication. CONCLUSIONS: CRC incidence has remained relatively stable in the UK over the last decade. The increased prevalence of some comorbidities and medications among CRC cases should be considered when evaluating patterns in CRC survival.

BACKGROUND: Depression is prevalent in primary biliary cholangitis (PBC) patients. Our aims were to examine the effects of depression and antidepressants on hepatic outcomes of PBC patients. METHODS: We used the UK Health Improvement Network database to identify PBC patients between 1974 and 2007. Our primary outcome was one of three clinical events: decompensated cirrhosis, liver transplantation and death. We assessed depression and each class of antidepressant medication in adjusted multivariate Cox proportional hazards models to identify independent predictors of outcomes. In a sensitivity analysis, the study population was restricted to PBC patients using ursodeoxycholic acid (UDCA). RESULTS: We identified 1,177 PBC patients during our study period. In our cohort, 86 patients (7.3%) had a depression diagnosis prior to PBC diagnosis, while 79 patients (6.7%) had a depression diagnosis after PBC diagnosis. Ten-year incidence of mortality, decompensated cirrhosis, and liver transplantation were 13.4%, 6.6%, and 2.0%, respectively. In our adjusted models, depression status was not a predictor of poor outcomes. After studying all classes of antidepressants, using the atypical antidepressant mirtazapine after PBC diagnosis was significantly protective (Adjusted HR 0.23: 95% CI 0.07-0.72) against poor liver outcomes (decompensation, liver transplant, mortality), which remained statistically significant in patients using UCDA (HR 0.21: 95% CI 0.05-0.83). CONCLUSIONS: In our study, depression was not associated with poor clinical outcomes. However, using the antidepressant mirtazapine was associated with decreased mortality, decompensated cirrhosis and liver transplantation in PBC patients. These findings support further assessment of mirtazapine as a potential treatment for PBC patients.

Type 2 diabetes (T2D) is associated with numerous comorbidities that significantly reduce quality of life, increase mortality and complicate treatment decisions. In a recent cardiovascular outcomes trial, Empagliflozin Cardiovascular Outcome Event Trial in Type 2 Diabetes Mellitus Patients (EMPA-REG OUTCOME), the sodium-glucose cotransporter 2 (SGLT2) inhibitor empagliflozin was shown to reduce cardiovascular (CV) mortality and heart failure in high-risk patients with T2D with a previous CV event or with established CV disease (CVD). Recently published data from the Canagliflozin Cardiovascular Assessment Study (CANVAS-PROGRAM) study suggested that the cardiovascular benefits of empagliflozin are also seen with the SGLT2-inhibitor canagliflozin, indicating a class effect of SGLT2 inhibitors. Evidence for a class effect has also been shown by meta-analyses and real-world studies, including the Comparative Effectiveness of Cardiovascular Outcomes in New Users of SGLT-2 Inhibitors (CVD-REAL) and The Health Improvement Network (THIN) databases. These findings also suggest the results of EMPA-REG OUTCOME can be applied to patients with T2D with a broader CV risk profile, including people at low risk of CVD.

Relatively little is known about the risk for incident liver disease in psoriasis (PsO), psoriatic arthritis (PsA), and rheumatoid arthritis (RA). We performed a cohort study among patients with PsO, PsA, or RA and matched controls in The Health Improvement Network from 1994 to 2014. Outcomes of interest were any liver disease, nonalcoholic fatty liver disease, and cirrhosis (any etiology). Among patients with PsO (N = 197,130), PsA (N = 12,308), RA (N = 54,251), and matched controls (N = 1,279,754), the adjusted hazard ratios for any liver disease were elevated among patients with PsO (without systemic therapy [ST] 1.37; with ST 1.97), PsA (without ST 1.38; with ST 1.67), and RA without an ST (1.49) but not elevated in patients with RA prescribed an ST (0.96). Incident nonalcoholic fatty liver disease was highest in patients with PsO prescribed an ST (2.23) and PsA with an ST (2.11). The risk of cirrhosis was highest among patients with PsO with an ST (2.62) and PsA without an ST (3.15). Additionally, the prevalence of liver disease and cirrhosis increased in a stepwise fashion with increasing body surface area affected by PsO (P for trend <0.001). More so than RA, PsO and PsA are associated with liver disease, particularly nonalcoholic fatty liver disease and cirrhosis, and this was true even among patients without ST exposure.

An increase in fracture risk has been reported in patients with type 2 diabetes mellitus (T2DM) treated with canagliflozin, possibly mediated by effects induced by all members of the sodium-glucose co-transporter-2 (SGLT2) inhibitor class. It is unclear whether initiation of dapagliflozin is followed by an increase in the risk of fracture; therefore, we performed a population-based, open cohort study (from January 2013 to January 2016) using The Health Improvement Network (THIN). A total of 22 618 people with T2DM (4548 exposed to dapagliflozin and 18 070 receiving standard antidiabetic treatment, matched for age, sex, body mass index and diabetes duration) with no history of fractures at baseline were included. The primary outcome was the occurrence of any fragility fracture (hip, spine, wrist) during the observation period. Risk of any fracture served as a secondary outcome. Adjusted hazard rate ratios (HRs) with 95% confidence intervals (CIs) were calculated using Cox regression. A total of 289 fractures (132 fragility fractures) were recorded. No difference in the risk of fragility fracture was detected between participants prescribed dapagliflozin and matched control participants (crude HR 0.90, 95% CI 0.59-1.39, P = .645; adjusted HR 0.87, 95% CI 0.56-1.35, P = .531). Similarly, no difference in the risk of any fracture was detected (adjusted HR 0.89, 95% CI 0.66-1.20; P = .427). Sensitivity analyses limited to the subset of the population at high risk of fracture produced similar results; thus, there was no evidence to suggest an increase in the risk of treatment-emergent fractures in patients with T2DM who initiated treatment with dapagliflozin.

OBJECTIVE: To emulate three target trials: single treatment vs. no treatment, joint treatment vs. no treatment, and head-to-head comparison of two treatments, we explain how to estimate the observational analogs of intention-to-treat and per-protocol effects, using hazard ratios and survival curves. For per-protocol effects, we describe two methods for adherence adjustment via inverse-probability weighting. STUDY DESIGN AND SETTING: Prospective observational study using electronic medical records of individuals aged 55-84 with coronary heart disease from >500 practices in the United Kingdom between 2000 and 2010. RESULTS: The intention-to-treat mortality hazard ratio (95% confidence interval) was 0.90 (0.84, 0.97) for statins vs. no treatment, 0.88 (0.73, 1.06) for statins plus antihypertensives vs. no treatment, and 0.91 (0.77, 1.06) for atorvastatin vs. simvastatin. When censoring nonadherent person-times, the per-protocol mortality hazard ratio was 0.74 (0.64, 0.85) for statins vs. no treatment, 0.55 (0.35, 0.87) for statins plus antihypertensives vs. no treatment, and 1.13 (0.88, 1.45) for atorvastatin vs. simvastatin. We estimated per-protocol hazard ratios for a 5-year treatment using different dose-response marginal structural models and standardized survival curves for each target trial using intention-to-treat and per-protocol analyses. CONCLUSION: When randomized trials are not available or feasible, observational analyses can emulate a variety of target trials.

Essentials We estimated the cardiovascular risk of patients with idiopathic thrombocytopenic purpura (ITP). The risk of cardiovascular disease was 38% higher in ITP patients compared with controls. Among the ITP patients, splenectomy was associated with higher cardiovascular disease. Clinicians should consider cardiovascular risk when managing ITP patients. SUMMARY: Background Idiopathic thrombocytopenic purpura (ITP) is classically characterized by a transient or persistent decrease of platelet count. Mortality is higher in the ITP population than the general population, with a possible association with increased cardiovascular disease (CVD). Objectives The objective was to assess the strength of the association between ITP and CVD, with a secondary aim to assess the impact of splenectomy on CVD. Methods A population-based retrospective, open cohort study using clinical codes was performed using data from 6591 patients with ITP and 24 275 randomly matched controls (up to 1:4 ratio matched by age, sex, body mass index and smoking status). The main outcome was the risk of CVD, which included ischemic heart disease, stroke, trans-ischemic attack and heart failure. Adjusted incidence rate ratios were calculated using Poisson regression. Results During a median 6-year observation period there was a CVD diagnosis recorded in 392 (5.9%) ITP patients and 1114 (4.5%) control patients. There was an increased risk of developing CVD in the ITP cohort (incidence rate ratio [IRR], 1.38; 95% confidence interval [CI], 1.23-1.55), which remained robust even after a sensitivity analysis only including incident cases of ITP. Findings suggested that patients who had undergone splenectomy were at even further increased risk of developing CVD when compared with the ITP population who had not undergone splenectomy (adjusted IRR, 1.69; 95% CI, 1.22-2.34). Conclusion There is an increased risk of developing CVD in patients with ITP and even further increased risk for those patients with ITP who underwent splenectomy.

OBJECTIVE: To investigate the population impact of previously reported interactions between warfarin and other drugs on international normalized ratio (INR) levels. METHODS: Using The Health Improvement Network (THIN), a United Kingdom primary care database, a cohort of warfarin users between 2005 and 2013 (  = 121,962) was followed until the first qualifying prescription for the potential interacting drugs was evaluated. Sixteen sub-cohorts, one for each study drug, and a control sub-cohort of warfarin were ascertained. Short-term changes in INR levels were assessed by comparing INR values measured before and after initiation of the interacting drug with paired Student's -test. We also evaluated the proportion of patients with INR values outside the therapeutic range (INR: 2-3). RESULTS: Miconazole use was associated with the highest mean increase in INR (+3.35), followed by amiodarone (+1.28), fluconazole (+0.79), metronidazole (+0.75) and nystatin (+0.65). After subtracting the natural INR variation observed in the control sub-cohort, supra-therapeutic levels (INR > 3) were found in 53.2% (miconazole), 45.5% (amiodarone), 23.3% (metronidazole), 23.2% (fluconazole) and 17.6% (nystatin) of patients initiating treatment with these drugs. Carbamazepine use was associated with a mean INR decrease of -0.63 and infra-therapeutic levels (INR < 2) were observed in 46.2% of patients initiating carbamazepine. For all other drugs, the change was small to moderate, in absolute INR units (+0.23 to +0.55) and in the proportion of patients with INR levels out of therapeutic range (<16%). CONCLUSIONS: Clinically potentially important interactions were observed in several study drugs. The majority of them, although confirmed, had little impact after adjusting for standard INR variability in the general population of warfarin users.

BACKGROUND: Androgen excess is a defining feature of polycystic ovary syndrome (PCOS), which affects 10% of women and represents a lifelong metabolic disorder, with increased risk of type 2 diabetes, hypertension, and cardiovascular events. Previous studies have suggested an increased risk of nonalcoholic fatty liver disease (NAFLD) in individuals with PCOS and implicated androgen excess as a potential driver. METHODS AND FINDINGS: We carried out a retrospective longitudinal cohort study utilizing a large primary care database in the United Kingdom, evaluating NAFLD rates in 63,120 women with PCOS and 121,064 age-, body mass index (BMI)-, and location-matched control women registered from January 2000 to May 2016. In 2 independent cohorts, we also determined the rate of NAFLD in women with a measurement of serum testosterone (n = 71,061) and sex hormone-binding globulin (SHBG; n = 49,625). We used multivariate Cox models to estimate the hazard ratio (HR) for NAFLD and found that women with PCOS had an increased rate of NAFLD (HR = 2.23, 95% CI 1.86-2.66, p < 0.001), also after adjusting for BMI or dysglycemia. Serum testosterone >3.0 nmol/L was associated with an increase in NAFLD (HR = 2.30, 95% CI 1.16-4.53, p = 0.017 for 3-3.49 nmol/L and HR = 2.40, 95% CI 1.24-4.66, p = 0.009 for >3.5 nmol/L). Mirroring this finding, SHBG <30 nmol/L was associated with increased NAFLD hazard (HR = 4.75, 95% CI 2.44-9.25, p < 0.001 for 20-29.99 nmol/L and HR = 4.98, 95% CI 2.45-10.11, p < 0.001 for <20 nmol/L). Limitations of this study include its retrospective nature, absence of detailed information on criteria used to diagnosis PCOS and NAFLD, and absence of data on laboratory assays used to measure serum androgens. CONCLUSIONS: We found that women with PCOS have an increased rate of NAFLD. In addition to increased BMI and dysglycemia, androgen excess contributes to the development of NAFLD in women with PCOS. In women with PCOS-related androgen excess, systematic NAFLD screening should be considered.

PURPOSE: Inappropriate use of an inhaled corticosteroid (ICS) for COPD has clinical and economic disadvantages. This retrospective analysis of The UK Health Improvement Network (THIN) database identified factors influencing treatment escalation (step-up) from a long-acting muscarinic antagonist (LAMA) to triple therapy (LAMA + long-acting β-agonist-ICS). Secondary objectives included time to step up from first LAMA prescription, Global Initiative for Chronic Obstructive Lung Disease (GOLD) grouping (2011/2013, 2017), and Medical Research Council (MRC) grade prior to treatment escalation. MATERIALS AND METHODS: Data were included from 14,866 people ≥35 years old with a COPD diagnosis (June 1, 2010-May 10, 2015) and initiated on LAMA monotherapy. The most commonly used LAMA at baseline was tiotropium (92%). RESULTS: Multivariate analysis (10,492 patients) revealed that COPD exacerbations, lower forced expiratory volume in 1 second (FEV ), "asthma", MRC grade, proactive and reactive COPD primary care, elective secondary-care contact, cough, and number of short-acting bronchodilator prescriptions were positively associated with treatment escalation ( <0.05). Being older, a current/ex-smoker, or having increased sputum symptom codes were negatively associated with treatment escalation ( <0.05). Median MRC score was 2 at baseline and 3 prior to treatment escalation. Using the last MRC reading and exacerbation history in the year prior to escalation, GOLD 2017 groupings were A 27.4%, B 37.3%, C 15.3%, and D 20%. In patients with available FEV measures, exacerbations, and MRC code (n=1,064), GOLD 2011/2013 groupings were A 20.4%, B 19.2%, C 24.8%, and D 35.6%. CONCLUSION: While the presence of COPD exacerbations seems to be the main driver for treatment escalation, according to the 2017 GOLD strategy many patients appear to be overtreated, as they would not be recommended for treatment escalation. Reviewing patients' treatment in the light of the new GOLD strategy has the potential to reduce inappropriate use of triple therapy.