Publications
scientifiques

Arthroplasty is indicated for patients with rheumatoid arthritis (RA) who experience significant joint damage, including bone erosions, cartilage degradation and joint deformities. However, studies on its associations with all-cause mortality, cardiovascular disease (CVD), and venous thromboembolism (VTE) among patients with RA are scarce. Our aim was to evaluate the relation of knee arthroplasty or hip arthroplasty to all-cause mortality, relative risk of CVD and incident VTE among patients with RA. We included patients with RA (ages≥20 years) from a large United Kingdom primary care database (i.e., IQVIA Medical Research Database). The primary outcome was all-cause mortality (n = 4,774 for knee arthroplasty, n = 3,362 for hip arthroplasty). The secondary outcomes included incident CVD (n = 4,350 for knee arthroplasty, n = 2,390 for hip arthroplasty) and incident VTE (n = 4,574 for knee arthroplasty, n = 3,174 for hip arthroplasty). We conducted propensity score-matched cohort studies to compare the risks of each outcome between subjects with and without knee arthroplasty (n = 2,387 each) and those with and without hip arthroplasty (n = 1,681 each), respectively. We found that subjects with knee arthroplasty had a 23% lower risk of mortality than those without knee arthroplasty (HR: 0.77, 95%CI: 0.65-0.90). Similarly, a lower, albeit non-statistically significant, risk of mortality was observed among subjects with hip arthroplasty than those without arthroplasty (HR: 0.87, 95%CI: 0.73-1.04). Compared with those without arthroplasty, subjects with knee or hip arthroplasty had a lower risk of CVD. The corresponding HRs were 0.86 (95%CI: 0.73-1.01) and 0.84 (95%CI: 0.69-1.02), respectively. Both subjects with knee or hip arthroplasty showed a higher risk of VTE than their counterparts (HR for knee arthroplasty: 1.63 [95%CI: 1.23-2.17]; HR for hip arthroplasty: 2.19 [95%CI: 1.54-3.11]). The associations of arthroplasty with the risks of mortality, CVD and VTE were generally consistent across strata of age and sex, with HR ranges from 0.71-3.75 for knee arthroplasty and 0.66-3.36 for hip arthroplasty. In this large population-based cohort of patients with RA, knee arthroplasty was associated with a lower risk of all-cause mortality, while both knee and hip arthroplasty were associated with a higher risk of VTE. No significant associations were observed with CVD. These findings highlight potential long-term benefits and risks of joint replacement in RA, but given the observational design and possibility of residual confounding, the results should be interpreted as associations rather than causal effects. Further studies are warranted to confirm these observations and to better understand the mechanisms underlying these associations.

BACKGROUND: The role COVID-19 vaccination on healthcare resource utilization (HCRU) and cost remains unclear, especially during Omicron predominance and among high risk UK populations. RESEARCH DESIGN AND METHODS: A retrospective cohort study using UK The Health Improvement Network (THIN) primary care data included adults (≥18 years) with confirmed or suspected COVID-19 between September 2022 and May 2023. Three cohorts were defined: Highest risk (eligible for two seasonal doses), High Risk (eligible for one dose), and All COVID-19 patients. Long COVID was identified as ≥1 symptom or diagnostic/referral code, ≥4 weeks post COVID-19 diagnosis. Inverse probability of treatment weighting assessed associations between vaccination status (yes/no and time since vaccination) and long COVID, HCRU, and costs. RESULTS: In both Risk Cohorts, COVID-19 vaccination was not associated with long COVID incidence. However, in the High Risk (  = 1,889) and All Patients cohorts (  = 8,507) outpatient specialist referrals were significantly lower in the 3-6-month post-vaccination group versus > 6 months (rate ratio: 0.28; 95% CI: 0.10-0.79,  < 0.05 and 0.46; 95% CI: 0.27-0.79;  ≤ 0.01, respectively). CONCLUSIONS: COVID-19 vaccination was not consistently associated with incidence of long COVID or all-cause HCRU. Findings suggest potential subgroup-specific benefits, highlighting the importance of annual vaccination. Further research with larger sample size and longer-term follow-up is warranted.

BACKGROUND: C-reactive protein (CRP) is a biomarker of inflammation used in diagnosis of inflammatory diseases and to guide treatment decisions. Variation in within-individual measured CRP may affect its clinical utility but estimates of within-individual variation are based on limited data and so may not be accurate. METHODS: A retrospective cohort study was performed using data on CRP results and sociodemographic, lifestyle and comorbidity covariates extracted from the IQVIA Medical Research Database (IMRD) database using the DEXTER tool. A minimum of four measurements for each individual was the only inclusion criterion. CRP data were log-transformed for analysis. Within-individual measured variation was calculated as a coefficient of variation (CV) using a linear regression random effects model for the whole population and various subgroups. RESULTS: 472,811 participants were included in this study, making it the largest study of variation of CRP to date by a factor of approximately five. The overall coefficient of variation for CRP was 1.604 (95% CI 1.602 to 1.606). This is much higher than the median reported CV for CRP of previous studies which was 0.41. CV increased with patient median. STRENGTHS AND LIMITATIONS: The large number of participants and the real-world nature of the results are important strengths of this study. Weaknesses included the problem of accounting for confounding by indication, and the short half-life of CRP making it hard to distinguish between acute illness and physiological variation. CONCLUSIONS: Estimated within-individual variation in this analysis of real-world data is very high and is higher than previously reported. Variation increases with patient median CRP, that is with more severe disease status. This has important implications for the diagnosis, monitoring and clinical decision-making for inflammatory disease.

BACKGROUND: Drug shortages can negatively impact patient care. We aimed to estimate the incidence and prevalence of use of medicines with shortages announced by the European Medicines Agency between January, 2013, and September, 2023, and to characterise the users of these drugs. METHODS: In this multinational network cohort study, we used routinely collected data from 52 databases across 18 European countries and the USA covering primary care, secondary care, health insurance claims, and disease registries. We included all participants with a minimum of 365 days of medical history between 2010 and 2024. We estimated annual incidence rates and period prevalence of use of medicines with a reported shortage (n=16), and their key alternatives (n=41). A reduction of 33% or more in incidence or prevalence after the shortage announcement was considered confirmation of a shortage. Additionally, we analysed changes in utilisation in terms of age, sex, indication, duration, and dosage. FINDINGS: Eight drugs had a 33% or higher reduction in incidence and nine drugs had a 33% or higher reduction in prevalence. Varenicline and amoxicillin (alone or combined with clavulanate) were the medicines affected in the highest number of countries and databases. Additionally, we observed changes in the indication of alteplase (pulmonary embolism indication increased in hospitals in Finland and Germany during the shortage period) and sarilumab (rheumatoid arthritis indication decreased in databases in the UK, Spain, Finland, and Sweden); and among incident users of sarilumab, a decrease in the cumulative dose was observed in databases in the Netherlands (from 84 mg in 2020 to 28 mg in 2023) and a reduction in treatment duration was observed in databases in Finland (from 104 days in 2020 to 1 day in 2022) and Belgium (from 71 days in 2020 to 30 days in 2022). INTERPRETATION: This study highlighted changes in incidence and prevalence of use of medicines after shortage announcements, and changes observed in patient care in terms of the indication, duration, or prescribed dose of medicines. Our findings showed that some reductions in use were observed across Europe and the USA, and others differed across countries. More research is needed to reduce the effects of drug shortages globally. FUNDING: Innovative Medicines Initiative 2 Joint Undertaking.

BACKGROUND: HbA1c is a marker for diabetes mellitus that reflects average glucose concentrations over the previous eight to twelve weeks. It is used to aid in the diagnosis and management of diabetes. Variation in within-individual measured HbA1c may affect its clinical utility but estimates of this are based on limited data that is often not generalisable to real-world settings. METHODS: A retrospective cohort study was performed using data on HbA1c results and sociodemographic, lifestyle and comorbidity covariates extracted from the IQVIA Medical Research Database (IMRD) database using the DEXTER tool. A minimum of four measurements in the same individual was the only inclusion criterion. Within-individual measured variation was calculated as a coefficient of variation (CV) using a linear regression random effects model for the whole population and various subgroups. RESULTS: 587,023 participants were included in this study, making it the largest study of variation of HbA1c to date. The overall measured within-individual coefficient of variation (CVT) was 0.20 (95%CI 0.20 to 0.20). This is around three times higher than reported in a previous systematic review. CVT increased with patient mean HbA1c level. STRENGTHS AND LIMITATIONS: The large number of participants and the real-world nature of the results are important strengths of this study. Weaknesses included the problem of accounting for confounding by indication. CONCLUSIONS: Estimated within-individual variation in this analysis of real-world data is very high and is higher than previously reported. Variation increases with patient mean HbA1c, that is with more severe disease status. This has important implications for the diagnosis, monitoring and clinical decision-making for diabetes.

BACKGROUND: There is no consensus for using statins for primary prevention of cardiovascular disease (CVD) and all-cause mortality in adults with type 1 diabetes mellitus (T1DM), because no randomized controlled trial has exclusively investigated statins in this population. OBJECTIVES: In this study, the authors sought to evaluate the long-term risks and benefits of statins for primary prevention in adults with T1DM. METHODS: We performed a sequential target trial emulation comparing statin initiation vs noninitiation using UK primary care data from the IQVIA Medical Research Data database. Persons aged 25 to 84 years with a diagnosis record of T1DM with prescription of insulin from January 2005 to December 2016 were included if they had baseline low-density lipoprotein-cholesterol (LDL-C) ≥2.6 mmol/L (100 mg/dL) or non-high-density lipoprotein cholesterol ≥3.4 mmol/L (130 mg/dL). Persons with preexisting coronary artery disease, myocardial infarction, stroke, heart failure, myopathy, liver disease, rheumatic heart disease, schizophrenia or cancer were excluded. Main outcome measures were all-cause mortality, major CVD and adverse events (myopathy and liver dysfunction). We estimated 10-year absolute risk differences (RDs) for the observational analogues of the intention-to-treat (ITT) and per-protocol (PP) effects. RESULTS: We included 4,176 statin initiator (mean age of 45 years, 33.1% <40 years, 40.6% female) and 16,704 noninitiator person-trials with median follow-up of 6 years. Compared with noninitiation, statins were associated with reductions in all-cause mortality (RD : -1.66% [95% CI: -2.79% to -0.45%]; RD : -3.48% [95% CI: -4.68% to -2.07%]) and major CVD (RD : -1.63% [95% CI: -2.57% to -0.53%]; RD : -2.69% [95% CI: -4.00% to -1.22%]). Some analyses suggested a slight association with increased risk of liver dysfunction but no association with myopathy. In subgroup analyses, absolute risk reductions were generally larger in women, persons ≥40 years of age, persons with baseline LDL-C ≥3.4 mmol/L (130 mg/dL), and persons with a higher predicted cardiovascular risk. CONCLUSIONS: Among adults with T1DM, statin initiation for primary prevention was associated with reductions in all-cause mortality and major CVD with a very low risk of adverse effects. The differences in absolute risk reductions can help guide personalized statin treatment decisions in T1DM.

Low body mass index (BMI) is a common feature of severe chronic obstructive pulmonary disease (COPD) but in the general population, cigarette smoking is also associated with low body weight. Many people with COPD remain smokers after diagnosis, and it is unclear whether low BMI is because of the disease itself or its most common risk factor. We aim to assess the independent and combined effects of smoking and COPD on BMI trajectories. 27,651 patients without COPD and 25,990 with COPD from The Health Improvement Network (2005-2019) were grouped into: never-smokers, former smokers, sustained quitters, intermittent smokers, and continuous smokers (ten total COPD-smoking status groups). BMI trajectories over 10-year time horizon were modeled by these status groups using multivariable mixed-effect models adjusted for age (in continuous years), sex, Townsend score (a measure of material deprivation), alcohol consumption (yes/no), exacerbation history (yes/no, only for COPD patients) and any history of asthma, cancer, chronic kidney disease, diabetes, or cardiovascular disease (yes/no). Individuals with COPD who smoked at baseline (intermittent, sustained quitter, or continuous smokers) had a lower initial BMI (27.1 kg/m² [26.9-27.3]; 26.6 [26.4-26.9]; 26.2 [26.0-26.4], respectively) than non-COPD controls in the same smoking categories (28.0 [26.6-28.2]; 27.6 [27.2-27.9]; 26.7 [26.4-26.9]). Current smokers had lower initial BMIs than never and former smokers, regardless of COPD status. In individuals with COPD, compared to former smokers, continuous smokers lost weight faster (-0.071 kg/m²/year [-0.097 to -0.045]; p < 0.001), while quitters gained weight (0.266 [0.233 to 0.298]; p < 0.001). Non-COPD controls showed similar but less pronounced patterns when continuous smokers and quitters (-0.059 [-0.090 to -0.028] and 0.213 [0.173 to 0.254], respectively; both p < 0.001) were compared to former smokers. Those with a baseline BMI of < 30 also showed a decrease in longitudinal BMI, especially among COPD patients. COPD patients had lower baseline BMI than controls, but BMI trajectories were similar between groups, with continuous smokers losing weight faster and quitters gaining weight. These findings suggest that smoking behaviour significantly influences weight loss in COPD, emphasizing its importance in clinical evaluations and nutritional support consultations.

BACKGROUND: Pancreatic ductal adenocarcinoma (PDAC) presents an urgent challenge that necessitates improved early risk identification. We investigated the association between predictors of PDAC present at least two years before diagnosis and PDAC occurence. METHODS: This case-control study used electronic health records from The Health Improvement Network Database (UK). Cases (10,575) were matched with controls (105,750) in a 1:10 ratio by gender, age, follow-up duration, and year of inclusion in the database. Variables included clinical features, comorbidities and blood result abnormalities, reported at least 2 years before PDAC diagnosis or equivalent timeframe for controls. Conditional logistic regression model with backward estimated odds ratios (OR) and 95% confidence interval (CI). RESULTS: Electronic health records of cases reported higher prevalence of alcohol intake, cigarette smoking, dyslipidemia, increased blood pressure, and diabetes of more than four years' duration. Independent risk factors included pancreatic cysts (OR = 4.39), pancreatitis (OR = 2.16), dyslipidemia (OR = 1.50), smoking (OR = 1.09), and alcohol intake (OR = 1.09). Laboratory markers associated with higher risk included elevated alkaline phosphatase (OR = 3.21), bilirubin (OR = 2.48), alanine aminotransferase (OR = 1.76), erythrocyte sedimentation rate (OR = 1.27), and decreased albumin (OR = 1.54). CONCLUSIONS: Primary care electronic records can identify individuals at medium-term increased risk of PDAC, thus raises the opportunity to develop early detection models and address modifiable factors.

PURPOSE: Real-world data (RWD) and real-world evidence (RWE) offer significant potential for clinical development and clinical evidence generation. This review addresses the complexity surrounding the use of RWD and RWE in drug repurposing, such as antihypertensive fixed-dose combinations of known compounds from the 5 major pharmacological classes. METHODS: We performed a narrative review examining RWD and RWE use in the development and registration of antihypertensive fixed-dose combinations, covering their role as examples of value-added medicines, regulatory policies, scientific perceptions, application in hypertension research, regulatory use cases, and economic factors. FINDINGS: Well-designed RWE studies can yield clinical performance data comparable with findings of randomized controlled trials while addressing certain limitations. However, current European Union (EU) legislation regarding the usability of RWD and RWE is conservative and needs to be updated. Divergent opinions among EU member states pose uncertainty and risk for industry in marketing authorization applications involving RWD and RWE. IMPLICATIONS: Real-world data and RWE offer a transformative opportunity through data sourced beyond traditional clinical trials. This approach can expedite regulatory decisions, reduce development timelines and costs, and accelerate the delivery of valuable therapies to patients. Discussions at the centralized EU regulatory level are needed to recognize and accept RWE as valid for demonstrating clinical efficacy and safety.

BACKGROUND: Gilbert's syndrome (GS) is a common genetic disorder marked by elevated bilirubin levels due to UGT1A1 enzyme deficiency. While jaundice and some adverse drug reactions are the primary recognised clinical features, individuals with GS frequently report non-specific symptoms like fatigue, brain fog, and abdominal pain. This study investigates the symptoms and diagnostic triggers of GS using UK primary care electronic health records. METHODS: We analysed data from the IQVIA Medical Research Database, covering over 11 million active UK patients. Individuals with a recorded GS diagnosis were identified and their sociodemographic profiles described. Using a nested case-control design, we applied machine learning-based feature selection to pinpoint key clinical features recorded up to five years before diagnosis. These features were then examined longitudinally by sex to distinguish persistent symptoms from short-term diagnostic triggers. RESULTS: The estimated UK prevalence of GS was 180.4 per 100,000 (95% CI: 174.4-186.6), with diagnoses more common in men, peaking around age 35, and more frequent in areas of least social deprivation. Among 9,240 GS cases and 150,846 controls, machine learning identified key diagnostic themes including jaundice, abnormal liver function tests, abdominal pain, fatigue, bowel changes, and sleep disturbances. While most of these features appeared primarily in the year prior to diagnosis, only abdominal pain and fatigue were consistently more common in GS cases up to five years before diagnosis. CONCLUSION: Our findings highlight both expected and novel GS diagnostic triggers. While many features likely reflect known symptomology or incidental detection via routine testing, the persistent presence of fatigue and abdominal pain suggests they may be under-recognised symptoms of GS. These findings warrant further investigation, and the data-driven approach used here may help uncover early signs of other underdiagnosed genetic conditions.

OBJECTIVE: To study the COVID-19 disease course in people living with HIV (PLWH) based on meaningful case numbers, information about comorbidities, antiretroviral and COVID-19 treatment and HIV disease-related variables. METHODS: Multi-centre, observational, retrospective study of people living with HIV with COVID-19 matched to HIV-uninfected individuals with COVID-19 (HUC) and a case-control study of people living with HIV with COVID-19 matched to COVID-19 negative people living with HIV. Kaplan-Meier estimates and Cox proportional-hazards models stratified on each matched pair were used for COVID-19 outcomes, and conditional logistic regression models were used to identify risk factors for COVID-19 infection. RESULTS: Five hundred people living with HIV and COVID-19, 1106 HUC and 992 people living with HIV without COVID-19 were included. Rates for chronic kidney disease, peripheral vascular disease, dementia, prior pneumonia and liver disease in people living with HIV with COVID-19 were significantly higher by 4.1-, 2.9-, 2.6-, 2.4- and 1.6-fold, respectively, compared to HIV seronegative COVID-19 patients. Chronic kidney disease, chronic obstructive pulmonary disease (COPD), body mass index (BMI) ≥30 kg/m , Centers for Disease Control and Prevention stage B versus A and HIV viral load ≥50 copies/mL were significant risk factors for COVID-19 in people living with HIV. Critical care admission or death in people living with HIV was comparable to HUC, whilst people living with HIV stayed longer in hospital (11 vs. 9 days) and in intensive care unit (ICU) (18 vs. 7 days) and had a higher age-adjusted Charlson comorbidity index. CONCLUSIONS: Our study highlights the importance of a well-controlled HIV infection to prevent severe COVID-19-related outcomes. In people living with HIV and COVID-19, chronic kidney disease and a higher Charlson comorbidity index are risk factors that should prompt early treatment of COVID-19.

INTRODUCTION: To estimate the recording rates and impact of overweight/obesity across selected countries in Europe and Asia-Pacific. METHODS: IMPACT-O was a retrospective cohort study of electronic medical records and claims databases from France, Germany, Italy, Spain, the UK, Australia, and Japan. Recording levels and the number of adults with overweight/obesity [based on diagnostic codes and/or body mass index (BMI)] during 2018-2022 were estimated. BMI thresholds/categories and obesity-related complications (ORCs) were described for adults with ≥ 1 BMI record of ≥ 25.0 kg/m and ≥ 12 months observation before the index date. RESULTS: The number of active subjects across databases ranged from 1,203,674 to 22,413,902. BMI was recorded for 4.8-38.9% of active patients. The number of adults with overweight/obesity ranged from 3.5% in Australia to 24.9% in the UK. The percentage of individuals with BMI > 25 to < 30 kg/m ranged from 43.6% in Spain to 77.9% in Japan; the percentage with BMI ≥ 30.0 kg/m ranged from 22.1% in Japan to 54.8% in Spain. Among those with overweight/obesity, the percentage with diagnosis codes in their records ranged from 2.9% in France to 50.1% in Germany. Most (59.6-85.0%) individuals had ≥ 1 ORC; 35.9-65.8% had multimorbidity. The most frequently recorded ORCs were hypertension, dyslipidemia, depression, and type 2 diabetes. CONCLUSION: Only a small proportion of people with overweight/obesity are formally diagnosed in healthcare databases. Most present with ≥ 1 ORC, commonly hypertension, dyslipidemia, depression, and type 2 diabetes.

BACKGROUND: Stage 3 chronic kidney disease (CKD) often remains undiagnosed until more severe symptoms appear. This study assessed awareness and management of CKD among Italian general practitioners (GPs), focusing on early detection and current practices. METHODS: A nation-wide, retrospective observational study was conducted using data from The Health Improvement Network (THIN®) database. Each participant was required to have had at least one interaction with a GP for either medical or administrative purposes (considering the index date), and to have a minimum of three years of retrospective data available  from January 2021 to June 2022. The study evaluated the proportion of individuals aged ≥ 40 years who underwent a second serum creatinine test after ≥ 90 days, referrals to nephrologists, and CKD diagnosis confirmation and categorization. Multivariable Poisson regression models analyzed data to identify associations between patient characteristics and outcomes, in both the overall cohort and in the sub-group with available urine albumin-to-creatinine ratio (uACR) measurement. RESULTS: Among 347,548 adults aged ≥ 40 years, 18,002 (5.2%) had an initial estimated glomerular filtration rate (eGFR) indicating possible stage 3 CKD (30-59 mL/min/1.73 m ), and 1495 of these had a concomitant uACR assessment. Data concerning follow-up testing and specialist referrals were inconsistent, and available only for 53.0% and 9.0% of the patients, respectively. Overall, 15.3% met the criteria for KDIGO stage 3 CKD, yet CKD ICD-9-CM diagnostic codes were recorded for only 905 (5.0%) patients. Factors associated with these outcomes were analyzed, including age, comorbidities, treatments, and laboratory values. CONCLUSIONS: Substantial gaps in GP awareness and adherence to CKD management guidelines were identified, particularly in follow-up testing, referral practices, and diagnostic coding. Targeted educational interventions and standardized care protocols are needed to enhance CKD detection and management in primary care, improving patient outcomes and healthcare system performance.