Scientific
publications

BACKGROUND: Fluoroquinolones (FQ) are broad-spectrum antibiotics effective against various infections. However, their use has been progressively restricted by guidelines due to safety concerns, leading to repeated reassessments of their indication. Despite a marked decline in prescriptions, misuse remains a concern. This study aimed to assess FQ prescribing indications in general practice in France and analyse their evolution between 2014 and 2023 in light of changing guidelines. METHODS: We conducted a descriptive observational study using data from the THIN® (The Health Improvement Network) database, managed by CEGEDIM-GERS DATA. This European database collects anonymized electronic health records from private practitioners in seven countries, including France since 2014. In 2023, the French panel included over 3 million patients and approximately 2000 general practitioners (GPs). Available data include administrative and medical information. Oral FQ prescriptions for patients aged≥20years were analyzed in 2014, 2019 and 2023 for compliance with five frameworks: marketing authorization (MA), SPILF 2015 (French Infectious Diseases Society), HAS 2016 (French National Authority of Health) and EMA 2018/2019 (European Medicines Agency). Results were stratified by sex, age, molecule and indication category. RESULTS: FQ prescriptions by GPs declines by 59% between 2014 and 2023. Compliance varied by reference: approximately 75% according to MA and SPILF versus less than 25% with more recent guidelines (HAS 2016 and EMA 2018/2019). Compliance was lower in women and patients aged≥75years. In 2023, the largest misuse was observed for urinary and prostate infections (from approximately 94% to 24%) due to changes affecting uncomplicated urinary tract infections in women. For respiratory or ear nose and throat (ENT) infections, compliance remained low (20-40%) across all references. CONCLUSIONS: Despite decreasing FQ use in general practice, many prescriptions remain non-compliant with updated recommendations, highlighting a persistent gap between clinical practices and evolving guidelines.

OBJECTIVE: To assess the relationship between antidepressant treatment before, during and after childbirth. DESIGN: Cohort study. SETTING: Primary care in the United Kingdom (UK). PARTICIPANTS: Women aged 15–49 with a single live birth (2006–2015), stratified by antidepressant history: Cohort A: Prescribed antidepressants ≥ 2 years before childbirth. Cohort B: Prescribed 1–2 years before childbirth. Cohort C: Prescribed within 1 year before childbirth. Cohort D: No antidepressants prescribed before childbirth. OUTCOME MEASURES: Changes in antidepressant treatment during the perinatal period. RESULTS: Of 202,303 women, 46,479 (23%) had pre-childbirth antidepressant prescriptions (Cohorts A: 37,161; B: 19,627; C: 14,363), while 155,824 (77%) did not (Cohort D). In cohort A, 76% discontinued treatment in the year before childbirth, compared to 48% in cohort B. Postpartum, 22% of those in cohort A and 33% in cohort B restarted treatment. Overall, 26,835/202,303 (13%) received antidepressants in the year after childbirth. Among women without prior treatment, 10,258/155,824 (7%) started antidepressants postpartum, compared to 16,577/46,479 (36%) of those with prior use. Cohort C had the highest postpartum prescription rate (9,232/14,363, 64%). CONCLUSIONS: Antidepressant treatment after childbirth is common, particularly among women who have received treatment before childbirth. Many women discontinue antidepressants before or during pregnancy. The study highlights the importance of considering history of antidepressant treatment when planning postnatal care and counselling women on continuing or stopping antidepressants during pregnancy. This information can aid healthcare professionals in advising women about antidepressant use before, during, and after pregnancy, considering individual circumstances and risks.

BACKGROUND: Evidence regarding the risk of new-onset myasthenia gravis (MG) following statin therapy initiation is limited. OBJECTIVES: To investigate this potential adverse effect using multinational real-world population-based data. METHODS: A self-controlled case series (SCCS) study was conducted using electronic medical records and claims databases from Hong Kong, the United Kingdom (UK) and Japan. Individuals aged ≥18 years with first diagnosis of MG and initiated statins were included. Conditional Poisson regression compared the risk of MG in different risk periods (up to 2 years after initiation) with non-exposure period, adjusted for age. Pooled results based on meta-analysis across all study sites were reported. RESULTS: In total, 2267 MG cases were analysed. Combined across all study sites, a significantly increased risk of incident MG was observed during the first year after statin initiation compared to non-exposure period, with a higher risk from Days 0-179 (pooled incidence rate ratio [IRR] [95% CI]: 2.662 [1.276-5.553]) than Days 180-364 (1.407 [1.014-1.954]). No increased risk of MG was observed more than 1 year after statin initiation (1.011 [0.848-1.206]). Moreover, the magnitude of MG risk elevation within the first 180 days after statin initiation was more pronounced with higher intensity statin regimens. CONCLUSION: In this multinational SCCS study, statin initiation may be associated with increased risk of new-onset MG during the first 6-12 months, with greater magnitude of risk elevation for higher intensity statin therapy. Consideration of the possibility of new-onset MG may be advisable within first 6-12 months after initiating statins, especially for medium-to-high-intensity statin therapy.

BACKGROUND: There is growing interest in widening the use of sodium-glucose co-transporter 2 inhibitors (SGLT2i) to all people with type 2 diabetes mellitus (T2DM). However, pivotal randomized controlled trials (RCTs) evaluated these drugs only in highly selected populations, often lacking generalizability to real-world populations. Understanding the effects of SGLT2i in populations where RCT evidence may be lacking is essential to help inform guideline development. To address this, we estimated the effect of empagliflozin in real-world users, many of whom would not have been eligible for the pivotal EMPA-REG RCT. METHODS: We designed a trial emulation in UK primary care data, based on the EMPA-REG RCT, to assess the effect of empagliflozin in a more clinically relevant population. Adults with T2DM initiating empagliflozin (intervention) or dipeptidyl peptidase-4 inhibitors (active control) between January 1, 2014 and December 31, 2022 were included. Eligibility was extended to both RCT-eligible and RCT-ineligible individuals. The effect of empagliflozin on all-cause mortality was estimated using an adjusted Cox proportional hazards model, with stratified analyses by RCT eligibility. FINDINGS: The majority of people prescribed empagliflozin would not have met the EMPA-REG RCT eligibility criteria (11,011/13,239, 83.2% RCT-ineligible). During follow-up, all-cause mortality occurred in 551 out of 13,239 (4.2%) in the empagliflozin group and 6,589 out of 49,264 (13.4%) in the active control group (adjusted HR 0.76, 95% CI 0.69 to 0.83). There was no evidence of differential treatment effect by RCT eligibility status (p-interaction=0.27). INTERPRETATION: Patients prescribed empagliflozin in real-world settings differ substantially from those enrolled in the EMPA-REG RCT. Using electronic health records, we demonstrate that the mortality benefit observed in EMPA-REG extends to a broader, more diverse real-world population, including those excluded from the original RCT. These findings provide a novel source of real-world evidence supporting the wider use of empagliflozin in routine clinical practice.

AIMS: Real-world medication use varies across clinical trial and healthcare settings; therefore, we evaluated GLP-1 receptor agonist (GLP-1RA) persistence and dose titration among adults with type 2 diabetes in UK primary care, stratified by agent, obesity status, cardiovascular disease (CVD) history, and sex assigned at birth. MATERIALS AND METHODS: Adults with type 2 diabetes initiating GLP-1RAs between 1 March 2018 and 30 June 2023, with follow-up ≥1-year, were identified using the IQVIA Medical Research Data (IMRD) incorporating data from THIN, A Cegedim Database. The absolute 1-year risk of discontinuation was estimated using the Aalen-Johansen estimator, while multivariable cause-specific Cox models assessed associations with patient characteristics. Dosing trajectories were characterised across individual prescriptions during the first therapy year. RESULTS: Among 8200 GLP-1RA initiators (46.5% dulaglutide, 38.7% semaglutide), the 1-year discontinuation risk was 41.1% (95% CI 40.1-42.2), higher among those without obesity (BMI <30 kg/m : 49.2%, 46.5-52.0) versus with BMI ≥30 kg/m (BMI 30-<35/≥35 kg/m : 40.4%/37.2%, 38.4-42.4/35.4-39.1). Discontinuation was lower with dulaglutide (36.9%, 35.4-38.4) than with subcutaneous/oral semaglutide (46.4%/55.8%, 44.4-48.4/52.3-59.4). Only 58% of subcutaneous/oral semaglutide initiators reached recommended maintenance doses after 4 weeks, while 21% remained on starting doses. At prescription 10, most common doses were 0.5/1 mg subcutaneous semaglutide (44.0%/48.6%), 7/14 mg oral semaglutide (50.2%/36.6%), and 0.75/1.5/3 mg dulaglutide (20.4%/71.5%/6.7%). Faster dose escalation occurred with higher BMI and subcutaneous semaglutide (1 mg at prescription 5: 20.8%/25.9%/33.2% with BMI <30/30-<35/≥35 kg/m ). Additional analyses (e.g., by CVD history) revealed no further heterogeneity. CONCLUSIONS: In UK primary care, GLP-1RA persistence was suboptimal and dose escalation was frequently delayed across all patient stratifications, including individuals with established CVD.

BACKGROUND: Thyrotropin levels increase with age, but standard reference intervals do not account for this, potentially leading to overdiagnosis of subclinical hypothyroidism (SCH) and overuse of levothyroxine in older adults. METHODS: Using data from The Health Improvement Network, this observational emulated target trial study assessed 10-year outcomes in adults over 50 years with SCH (thyrotropin 4.1-10.0mU/L, free thyroxine 10.0-24.0 pmol/L) who were prescribed levothyroxine versus those who were not. Subgroup analyses were limited to patients with age-specific thyrotropin levels. The primary outcome was cardiovascular events (angina, myocardial infarction, peripheral vascular disease, stent procedures, or stroke). Secondary outcomes included bone events (fragility fractures or osteoporosis) and all-cause mortality. Hazard ratios, adjusted through inverse probability of treatment weighting (IPTW) for age, sex assigned at birth, body mass index, Charlson comorbidity index, total cholesterol, hypertension, thyrotropin, hormonal medications, and smoking, were estimated. FINDINGS: Between January 1, 2006, and January 1, 2022, 22,621 patients (median age [interquartile range]: 66 [59-75] years, 76.7% female) were identified; 62% received levothyroxine and 38% did not. Levothyroxine was associated with reduced cardiovascular (IPTW-adjusted hazard ratio [aHR]: 0.82; CI: 0.74-0.91; < 0.0001) and all-cause mortality (aHR: 0.71; CI: 0.67-0.75; < 0.0001), with no adverse effects on bone (aHR: 1.04; CI: 0.93-1.17; = 0.45). Cardiovascular benefits were limited to patients with thyrotropin levels above the age-specific range and after at least five years of treatment. INTERPRETATION: Long-term levothyroxine use in older adults with SCH was associated with lower long-term cardiovascular and all-cause mortality risks, with no significant harm to bone health. Age-specific thyrotropin levels should guide treatment decisions.

AIMS: To investigate glucose-lowering treatment trajectories preceding glucagon-like peptide-1 receptor agonist (GLP-1RA) initiation in UK primary care, while assessing alignment with contemporary UK clinical guidelines, considering calendar time, cardiovascular disease (CVD) history, and obesity status. MATERIALS AND METHODS: Using the IQVIA Medical Research Data (IMRD) incorporating data from THIN, a Cegedim Database, we included adults with type 2 diabetes initiating GLP-1RAs in UK primary care between 01 January 2007 and 30 June 2023. We described treatment trajectories from the first glucose-lowering therapy to GLP-1RA initiation, stratified by calendar time (GLP-1RA initiation pre- or post-01 January 2018, when guidelines started recommending sodium-glucose co-transporter-2 inhibitors [SGLT-2is] for those at high risk or established CVD), CVD history, and body mass index (BMI) ≥35 kg/m . RESULTS: We included 29 780 GLP-1RA initiators, with 62.2% (n = 18 517) initiating pre-2018 and 37.8% (n = 11 263) post-2018. Consistent over calendar time, most GLP-1RA initiators (63.5%) received their first GLP-1RA as fourth-line (35.4%) or later-line therapy (28.1%), with fewer initiating GLP-1RAs as first- (0.8%), second- (10.5%), or third-line (25.2%) treatments. After 2018, 50.8% of individuals initiating GLP-1RA therapy used SGLT-2is concomitantly, regardless of CVD status (47.8% with established CVD vs. 51.5% without). Individuals with BMI ≥35 kg/m initiated GLP-1RA therapy earlier compared to those with BMI <35 kg/m (46.3% vs. 29.1% as first-, second-, or third-line treatment). CONCLUSIONS: GLP-1RAs were predominantly initiated following ≥3 glucose-lowering agents, consistent with contemporary UK guidance (NG28). Post-2018, most GLP-1RA initiators received SGLT-2is concomitantly. However, CVD history did not influence prescribing patterns, underscoring missed opportunities to optimise the prevention of cardiovascular events and slow the progression of CVD.

Clinical accounts and cohort studies suggest that Autistic people are disproportionately likely to be diagnosed with personality disorder. We conducted a cohort study of adults diagnosed Autistic drawn from the IQVIA Medical Research Database, with follow-up from 1 January 2000 to 16 January 2019. We included a comparison group without diagnosed autism, matched (1:10) by age, sex and primary care practice. We included 22,112 Autistic adults, of whom 6437 (29.1%) had a diagnosis of intellectual disability. Median age was 20.36 (interquartile range: 18.0-28.5), and 16,881 (76.3%) were men. The rate of new personality disorder diagnosis in Autistic people without intellectual disability was 4.8 (3.5-6.7) times higher for Autistic versus comparison men, and 4.6 (3.1-6.8) times higher for Autistic versus comparison women. For Autistic participants with intellectual disability, the rate was 2.0 (1.0-3.7) times higher for Autistic versus comparison men and 8.3 (4.0-17.2) times higher for Autistic versus comparison women. The estimated rate of new personality disorder diagnosis for Autistic people aged 20 increased from 14.67 (95% confidence interval: 10.4-20.8) per 10,000 person-years in 2009 to 22.43 (95% confidence interval: 13.9-36.3) in 2019. The findings indicate that personality disorder diagnoses are more common in Autistic people and increased overall in women from 2000 to 2019.Lay abstractSeveral research studies have suggested that Autistic people are more likely to be diagnosed with personality disorder than people who are not Autistic. We compared rates of personality disorder diagnoses between Autistic people and a comparison group of people not diagnosed Autistic using anonymised data collected by UK primary care practitioners for participants registered at a primary care (general practitioner) practice sometime between 1 January 2000 to 16 January 2019. The comparison group of people in the community who did not have an autism diagnosis were of the same age, sex and registered at the same primary care practice as their matched Autistic participant, with 10 times as many matched participants as Autistic participants. We included 22,112 Autistic adults, of whom 6437 (29.1%) had a diagnosis of intellectual disability. Median age was 20.36 years, and most, 16,881 (76.3%), were men. We included 221,120 comparison adults. New personality disorder diagnoses were more than four times as common for Autistic men and women without an intellectual disability compared to men and women in the comparison group. For Autistic participants with an intellectual disability, the rate was twice as high for Autistic versus comparison men and 8 times higher for Autistic versus comparison women. Between 2000 and 2019, there was an increase in the rate of new personality disorder diagnoses among Autistic people, and in women. The findings highlight the need for further investigation into reasons for this increase.

AIMS: 5α-Reductase inhibitors are prescribed for the treatment of benign prostatic hyperplasia (BPH) and their use is associated with increased risk of incident type 2 diabetes. This study assessed the long-term cardiovascular safety of 5α-reductase inhibitors in comparison with tamsulosin in people with co-existing BPH and type 2 diabetes. METHODS AND RESULTS: We performed a retrospective, population-based cohort study using Scottish Diabetes Research Network National Diabetes Dataset (SDRN-NDS) and IQVIA Medical Research Data (IMRD-UK). BPH patients ≥40 years with recorded type 2 diabetes mellitus and ≥2 prescriptions of 5α-reductase inhibitors or tamsulosin (2006-2021) were included. After 1:2 variable ratio propensity score matching, cause-specific Cox proportional-hazard models were used to compute the hazard ratio (HR) of incident major adverse cardiovascular events (MACE). A total of 11 969 patients were included in SDRN-NDS and 16 492 in IMRD-UK, with median follow-up durations of 3.8 (IQR: 1.7-6.8) and 4.8 (2.0-8.3) years, respectively. In SDRN-NDS, the HR of MACE in patients receiving 5α-reductase inhibitors relative to tamsulosin was 1.15 (95% CI 1.03-1.30, P = 0.007), driven by increased risk of myocardial infarction (MI) (HR 1.20, 1.03-1.40, P = 0.022). This was replicated in IMRD-UK, where HR was 1.26 (1.07-1.47, P = 0.008) for MACE and 1.33 (1.10-1.60, P = 0.005) for MI. We did not observe any increased risks in stroke, cardiovascular death, microvascular complications of diabetes, or faster progression to insulin-based therapies. CONCLUSION: Our retrospective data from two large cohorts suggest that the risk of MACE may be increased among patients with type 2 diabetes taking 5α-reductase inhibitors, potentially driven by increased risk of MI. This supports careful monitoring of macrovascular outcomes when prescribing 5α-reductase inhibitors in this population.

BACKGROUND: Depression is a heterogenous condition, with high societal and economic costs. Response to available treatments is varied, often trial-and-error, raising the need for personalization. Previous studies demonstrated success in predicting treatment outcomes for patients in psychiatric settings diagnosed with major depressive disorder, but have not done so in the general practice setting. METHODS: A retrospective cohort study analyzed electronic health records (EHRs) from the UK’s National Health Service, where depressive symptoms were recorded at two time points using the Patient Health Questionnaires (PHQ-9). Symptomatic changes between time 1 and time 2 were taken as the primary outcome of interest. Machine Learning and Causal Inference methods were used to define a data-driven antidepressant selection policy (choosing among three alternative SSRI medications) and evaluate it vis-a-vis routine clinical practice. RESULTS: In a cohort of 73,601 patients (65% female, mean age 49), evaluation of models for treatment selection suggested that: Reasonable treatment outcome prediction can be derived from EHRs (AUC 0.65), yet prediction is mostly driven by the initial PHQ-9 score (AUC 0.61 as a sole predictor). For essentially all patients with case-level depression prescribing antidepressants will reduce their symptoms by an expected 1.5 points For most patients, response to Citalopram, Fluoxetine and Sertraline is expected to be similar. Data driven selection of antidepressant type yields only a small improvement over current practice. CONCLUSIONS: In the examined setting, treatment response to antidepressants in general, and to 3 specific SSRIs can be predicted reasonably well, yet the resultant treatment selection policy does not significantly improve upon current practice. Expanding this approach to a more diverse drug selection, and augmenting it with psychological, genetic and metabolomic data, are therefore future directions for research on personalized treatment selection.

BACKGROUND: The clinical and economic burden of long COVID is poorly understood. We aim to assess all-cause healthcare resource utilization (HCRU) and costs in the primary care setting among adults with long COVID in France. METHODS: A retrospective cohort study using the electronic healthcare records (EHRs) of confirmed and/or probable COVID-19 patients from The Health Improvement Network (THIN) data between March 2020 and December 2022 was conducted. Long COVID was identified per World Health Organization (WHO) definition as suggestive symptoms present ≥3 months following acute SARS-CoV-2 infection. Patients' characteristics, HCRU, direct healthcare and indirect costs (National Health Insurance-based prices) were summarized. Costs between patients with previous SARS-CoV-2 infection who developed long COVID, patients with previous SARS-CoV-2 infection who did not develop long COVID (COVID only), and contemporaneous controls without SARS-CoV-2 infection were compared (Non-COVID). RESULTS: Long COVID developed among 30,122 (11.6%) adults; mean (SD) age was 50 (17) years, 63.6% were female and 27.5% had a Charlson Comorbidity Index score >2. During the post-infection follow-up (mean = 13 months), 97.3% of patients had general practitioner consultations (GP) and 62.4% had nursing care. Costs were highest during the first post-diagnosis year with per patient per year costs of €2,443 (total cost of €52 million), including costs for GP (€208) and specialist (€170) consultations, outpatient procedures (€413), retail pharmacy use (€595), biological testing (€147), and medical device usage (€172). Patients with long COVID had additional costs of €163 and €176 when compared to patients in the COVID only and Non-COVID cohorts, respectively. LIMITATIONS: Since the THIN database is generated from GP EHRs, there is the possibility of measurement/documentation errors and missing values which could compromise the validity and accuracy of certain results. CONCLUSION: Long COVID was associated with non-negligible HCRU, direct and indirect costs to the French healthcare system. These findings reinforce the importance of optimizing long-term resource allocation for patients infected with SARS-CoV-2.

Brief rationale: The effectiveness of bisphosphonates versus denosumab after hip fracture, especially in people with dementia or frailty, remains unclear. MAIN RESULT: Bisphosphonates were associated with a higher rate of subsequent fracture and a lower mortality rate in men. There are no differences by dementia or frailty status. Significance of the paper: Sex may influence antiresorptive treatment choice. PURPOSE: The relative effectiveness of first-line antiresorptive medications post-hip fracture in people with dementia or frailty is not understood. We investigated the risk of a subsequent fracture and death in people prescribed bisphosphonates or denosumab following hip fracture, including in people with dementia and frailty. METHODS: Parallel population-based cohort studies were conducted in Australia, Hong Kong, Taiwan, and the United Kingdom. People aged ≥ 50 prescribed or dispensed a bisphosphonate or denosumab within 60 days of discharge following their first hip fracture were included. Subgroup analyses were conducted for people with dementia, frailty, women, and men. Outcomes were second hip fracture, any subsequent fracture, and death. Inverse probability of treatment weighted Cox and competing risk models were used to estimate hazard ratios (HR) and subdistribution hazard ratios (sHRs) with 95% confidence intervals (CIs) for outcomes. Results across jurisdictions were combined using meta-analyses. RESULTS: There were 18,292 bisphosphonate users and 8560 denosumab users. Bisphosphonates versus denosumab were associated with similar rates of second hip fracture (sHR 1.13; 95% CI 0.76-1.69) and mortality (HR 0.99; 95% CI 0.94-1.04), but higher rates of any subsequent fracture (sHR 1.16; 95% CI 1.11-1.21), including in men (sHR1.27; 95% CI 1.15-1.42) but not in women (sHR 1.23; 95% CI 1.00-1.52), or in people with dementia or frailty. Men who used bisphosphonates had lower rates of mortality (HR 0.90; 95% CI 0.81-0.99) than men who used denosumab. CONCLUSION: Bisphosphonate users had higher rates of subsequent fracture than denosumab users. Mortality rates in men were lower with bisphosphonates than denosumab. There were no significant differences by dementia and frailty status.