French National Health Insurance Database Research Articles

Trends in the Use of Eye Care Services in Adults Treated for Diabetes between 2008 and 2017 in France: A Nationwide Study

Purpose: To describe the use of eye care services among adults treated for diabetes from 2008 to 2017 in the French population. Methods: We used a random representative sampling of the French national health insurance database. Participants were adults treated for diabetes. Data regarding visits to ophthalmologists, eye examination procedures, and medical treatments were extracted using reimbursement codes. Cross-sectional analyses of eye care claims were performed each year from 2008 to 2017. Results: Less than 50% of people with diabetes underwent an annual eye examination (2008: 44.7%; 2017: 47.9%), and less than two-thirds underwent a biennial eye examination (2008–2009: 62.6%; 2016–2017: 66.4%). From 2008 to 2017, the yearly use of optical coherence tomography examinations increased steadily and markedly from 2.7% to 16.2%, while the use of fluorescein and indocyanine green angiographies decreased from 2.4% to 0.9%. In the same period, the yearly rate of intravitreal injection increased from 0.3% to 1.5%, and the use of laser photocoagulation decreased progressively from 2008 to 2014 (1.3% vs. 0.7%) but slightly increased from 2015 to 2017 (0.8% vs. 1.0%). Conclusions: Strategies to increase compliance with eye care recommendations are needed to improve early detection and management of diabetic eye complications.

Preoperative evaluation of patients receiving bariatric surgery: a comprehensive Breton study

Abstract Background With large demonstrated benefits, volume of bariatric surgery (BS) has increased almost threefold since the late 2000s in France. However, patients seeking for BS must be carefully preoperatively evaluated to reduce risks for late complications (nutritional, neurological and psychiatric). This study aimed to assess preoperative evaluation of BS according to French recommendations. Methods The preoperative evaluation (gastrointestinal, nutritional, cardiovascular and psychiatric) during the year preceding BS was assessed from the French national health insurance database. All patients with a primary BS in Brittany in 2015 were included. For each patient, a score of preoperative evaluation was carried-out (from 0 to 6) and classified as follows: “complete” (6), “incomplete” (4-5), “unsatisfactory” (2-3) or “poor” (0-1). Comparisons (chi-square tests) by sex, geographical department, type of institution and health professionals (HP) were made. Results Among the 1,446 patients finally included, 1,195 (83%) were women and the median age was 42 (quartiles: 33-51) years old. Gastric by-pass (66%) and sleeve-gastrectomy (33%) were the most performed procedure. The preoperative evaluation was classified as ’complete’ for 501 (35%) patients, ’incomplete’ for 647 (45%) patients, ’unsatisfactory’ for 238 (17%) patients, and ’poor’ for 60 (4%) patients. Better scores were found in 2 departments (Côtes d’Armor and Finistère, p < 0.001), in public institutions (p < 0.001) and in institutions performing 50 or more acts per year (p < 0.001). Heterogeneity according to the HP was also noted in private institutions (median score per HP: 4.5, min-max: 2-6). Conclusions This large regional comprehensive French study confirms that preoperative evaluation before BS is not optimal despite clear French recommendations. Collective reflection should be led to improve the pathway of care of patients seeking BS. Key messages There are disparities in the performance of BS’s preoperative evaluation in Brittany. This work suggests that preoperative evaluation for patients seeking for BS should be improved and standardized.

PIN150 ANALYSIS OF TREATMENT SEQUENCES FROM THE FRENCH NATIONAL SNIIRAM DATABASE: CASE STUDY OF INCIDENT PEOPLE LIVING WITH HIV IN 2013

The study purpose was to describe the changes and discontinuations of antiretroviral combination therapy (cART) (number of treatments, occurrence, duration) over the cohort of incident people living with HIV (PLWHIV) in 2013, included in the French national Health Insurance database (SNIIRAM). It also aims at experimenting data mining methods to study treatment sequences. From the SNIIRAM, we extracted PLWHIV affiliated to the general insurance scheme (N=96,423) through specific chronic diseases status and/or reimbursement of HIV laboratory tests and/or HIV-related hospitalizations and/or reimbursements of antiretroviral drugs in 2013. Incident patients (N=3,373) were followed for 2 years to identify their cART, categorized as single, double, triple or quadruple+ combination therapies. The treatment sequences were arranged according to a similarity criterion. It was performed by an Agglomerative Clustering algorithm configured with the Hamming distance and the Ward linkage method. The result is a graph containing one discretized timeline per patient. The lines were put in order one above the other and follow the 2-year follow-up. Finally, a smoothing was applied to the image to homogenize and highlight distinctive patterns. The combination of Agglomerative Clustering and of image processing forms an innovative methodology called TAK: Time-sequence Analysis through K-clustering. Over the 3,373 incident patients, 363 (11%) received no antivirals drugs during the follow-up, 385 (11%) were only treated after 6 months and 255 (8%) stopped cART at least one period during follow-up. 1,675 (50%) patients solely received a triple therapy over the follow-up, 484 (14%) changed at least once from a triple therapy to a single/double. In 2013, a triple therapy was the reference treatment and 30% of the patients had a 6-month period without any cART. This study validates the relevance of combining clustering methods with image processing to visualize the treatment sequences of a cohort in a clear and synthetic way.

Monitoring Fetal Alcohol Spectrum Disorder during the neonatal period in France

Abstract Background Alcohol is a known teratogenic and foetotoxic agent. At birth, only the complete foetal alcohol syndrome (FAS) and at most a suspicion for some incomplete syndromic forms can be diagnosed. Yet, other Consequences of prenatal Alcohol Exposure (CAE) can also be observed and recorded in the neonatal period. Our goal was to describe the frequency of diagnosis codes for FAS and CAE at the national and regional level. Methods Between 2006 and 2013, we identified the ICM-10 codes Q860 (FAS) and P043 (CAE) in the hospital records for stays occurring in the 28 first days of life in the French national health insurance database (SNDS). Our “potential Foetal Alcohol Spectrum Disorders group” (pFASD), included the FAS and CAE subgroups. The pFASD prevalence was estimated per 1000 live births at the national and regional levels overall and then comparing: 2006-2009 and 2010-2013. Results From 2006 to 2013, 3207 neonates were diagnosed with pFASD during the neonatal period, i.e. 0.48 cases per 1,000 live births, including 0.07‰ of FAS. Between 2006-2009 and 2010-2013, pFASD remained stable (p = 0.6). At the regional level, the proportion of pFASD was the most frequent in one of the overseas territories (La Reunion Island, 1.22‰ births) and in the north-eastern part of mainland France (0.90 ‰ births to 1.02 ‰). Conclusions This study is the first to produce a national estimate of the frequency of neonatal diagnosis of FAS. It shows a stability of the frequency of pFASD cases recorded over the 2006-2013 period which is certainly underestimated but gives a first minimal estimate of the burden of alcohol use during pregnancy in France. Key messages Alcohol use during pregnancy remains a public health issue in France. Estimating the prevalence of FAS is of undeniable priority given the public health implications of these disorders that hinder the development of children exposed, and their avoidable nature.

Long-term adverse events after sleeve gastrectomy or gastric bypass: a 7-year nationwide, observational, population-based, cohort study

Concerns are rising about the late adverse events following gastric bypass and sleeve gastrectomy. We aimed to assess, over a 7-year period, the late adverse events after gastric bypass and sleeve gastrectomy compared with matched control groups. In this nationwide, observational, population-based, cohort study, we used data extracted from the French National Health Insurance (Système National des Données de Santé) database. All patients undergoing gastric bypass or sleeve gastrectomy in France in 2009, except those who had undergone bariatric surgery in the previous 4 years before inclusion, were matched with control patients with obesity in terms of age, sex, BMI category, baseline antidiabetic therapy, and baseline insulin therapy. Exclusion criteria for the control group included cancer, pregnancy, chronic infectious disease, serious acute or chronic disease in 2008-09, or previous (2005-09) or forthcoming (2010-11) bariatric surgery. The incidence rate was calculated for each type of adverse event leading to inpatient hospital admission over a 7-year period; incidence rate ratios (with 95% CIs) were computed to compare the rate of complications among the bariatric surgery and control groups. Risks of complications during follow-up were compared using Cox proportional-hazards regression analyses. Data were analysed according to the intention-to-treat methodology. From Jan 1, 2009, to Dec 31, 2009, 8966 patients who underwent bariatric surgery (7359 [82%] women; mean age 40·4 years [SD 11·3]) and 8966 matched controls (7359 [82%] women; mean age 40·9 years [11·4]) were included in analyses 4955 (55%) off 8966 patients in the bariatric surgery group had a primary gastric bypass and 4011 (45%) patients had sleeve gastrectomy. With a mean follow-up of 6·8 years (SD 0·2), mortality was lower in the gastric bypass group than in its control group (hazard ratio 0·64 [95% CI 0·52-0·78]; p<0·0001) and in the sleeve gastrectomy group than in its control group (0·38 [0·29-0·50]; p<0·0001). The gastric bypass and sleeve gastrectomy groups had higher risk than did their control groups for invasive gastrointestinal surgery or endoscopy (incidence rate ratio 2·4 [95% CI 2·1-2·7], p<0·0001, for gastric bypass vs control and 1·5 [1·3-1·7], p<0·0001, for sleeve gastrectomy vs control); for gastrointestinal disorders not leading to invasive procedures (1·9 [1·7-2·1]), p<0·0001, for gastric bypass vs control and 1·2 [1·1-1·4], p<0·0001, for sleeve gastrectomy vs control); and for nutritional disorders (4·9 [3·8-6·4], p<0·0001, for gastric bypass vs control and 1·8 [1·3-2·5], p<0·0001, for sleeve gastrectomy vs control). For psychiatric disorders, there was no significant association (1·1 [0·9-1·4], p=0·190, for gastric bypass vs control and 1·1 [0·8-1·3], p=0·645, for sleeve gastrectomy vs control), except for gastric bypass and alcohol dependence (1·8 [1·1-2·8], p=0·0124). Despite lower 7-year mortality, patients undergoing gastric bypass or sleeve gastrectomy had higher risk of hospital admission at least once for late adverse events, except for psychiatric disorders, than did control patients, with a higher risk observed after gastric bypass than with sleeve gastrectomy. None.

Persistence of apremilast in moderate-to-severe psoriasis: a real-world analysis of 14 147 apremilast- and methotrexate-naive patients in the French National Health Insurance database.

Real-world data on the persistence of apremilast vs. methotrexate are inconclusive. To assess and compare the long-term persistence of apremilast and methotrexate in a large cohort of patients with psoriasis. All adult patients with psoriasis registered in the French national health insurance database ('Système National des Données de Santé') between 2009 and 2017 were eligible for inclusion. The study population comprised apremilast- and methotrexate-naive patients, defined as those with a first prescription of apremilast or methotrexate. Levels of persistence were compared using a Cox model with propensity-score matching that included potential confounders (notably age, sex, psoriatic arthritis, comorbidities and previous exposure to topical and systemic treatments). In this nationwide population-based cohort, 14147 adult patients with psoriasis (mean age 52·3 years, 55·2% male) were found to be naive to both apremilast and methotrexate. After propensity-score matching, two subgroups of 4805 patients with similar baseline characteristics were included, of whom 3207 apremilast-treated patients and 2736 methotrexate-treated patients discontinued their treatment. Kaplan-Meier survival propensity-score analyses revealed a discontinuation rate of 69% for apremilast and 59% for methotrexate in the first year of treatment. Apremilast-treated patients had a higher risk of discontinuation than methotrexate-treated patients when considering the study population as a whole (hazard ratio 1·28, 95% confidence interval 1·23-1·34) or in a propensity-score-matched analysis (hazard ratio 1·34, 95% confidence interval 1·27-1·41; P < 0·001). Our real-world data suggest that in the first year of treatment, the discontinuation rate was significantly higher for apremilast-treated patients than for methotrexate-treated patients, regardless of the previous therapeutic lines received. What's already known about this topic? Psoriasis is a common chronic, relapse-remitting, inflammatory skin disease associated with severe psychosocial impact. Apremilast, a phosphodiesterase 4 inhibitor, is one of the most recently commercialized psoriasis drugs. Little is known about the long-term clinical effectiveness of apremilast. What does this study add? The discontinuation rate at 1 year for apremilast was 69%, compared with 58% for methotrexate, in a nationwide population-based cohort including 14147 nonselected adult patients with psoriasis. Patients in the apremilast cohort had a higher risk of discontinuation than patients in the methotrexate cohort using propensity-score matching, including potentially relevant individual risk factors such as age, sex, comorbidities and psoriatic arthritis, and regardless of the previous therapeutic lines received. In daily practice, physicians should take these results into account when choosing between methotrexate and apremilast as a first-line systemic therapy.

Trends in sodium valproate prescriptions among children aged 0 to 14 years between 2010 and 2016: A study based on the French National Health Insurance Database

PurposeAfter a huge campaign of information on the teratogenic risk of sodium valproate (VPA) having taken place in France we aimed to evaluate the trend of its prescriptions in young epileptic girls. MethodUsing the French National Health Insurance Database we searched for patients aged 0–14 years being supplied an antiepileptic drug (AED) between 2010 and 2016. Results113,362 children received at least one AED, 61,259 boys and 52,103 girls. Compared to 2010–2014 years, VPA was less prescribed in 2016 as first AED (29% vs 37.3% respectively). The difference between the two periods was greater for girls (−41%) than for boys (−12%). ConclusionThe changing trend of VPA as first AED prescribed, particularly in girls, reflects published evidence in terms of safety.

A capture-recapture method for estimating the incidence of off-label prescriptions: the example of baclofen for alcohol use disorder in France

The local/regional incidence of off-label prescriptions can be difficult to estimate. Capture-recapture models can be used to indirectly estimate population sizes. Here, we used a capture-recapture model to estimate the number of patients treated off-label with baclofen for alcohol use disorder in northern France in 2013. Three capture sources were used: (i) the active case file at the region's largest Addiction Unit, (ii) the regional pharmacovigilance centre, and (iii) a sample of community pharmacies. After between-source overlaps had been identified, we used a log-linear model to produced eight estimates. Two models displayed the best goodness-of-fit, with estimates [95% confidence interval] of 1123 [714-2162] and 2180 [1598-2870] subjects, respectively. These two values are in line with a previous estimate of 1624 patients, based on an analysis of the French national health insurance database in 2013. Capture-recapture methods can be usefully applied to estimate the prevalence of OLPs in a specific geographical area, when direct counting is not feasible or the estimate through claim database is not possible.

French national health insurance database analysis and field study focusing on the impact of secure prescription pads on zolpidem consumption and sedative drug misuse: ZORRO study protocol

IntroductionIn recent years, data collected by the French Addictovigilance Network have shown the potential for abuse and addiction associated with zolpidem (the most sold hypnotic drug in France). Since 10...

Economic burden of chemotherapy-treated recurrent and/or metastatic squamous cell carcinoma of the head and neck in France: real-world data from the permanent sample of national health insurance beneficiaries

Aims: Overall survival (OS) of patients with recurrent or metastatic (R/M) squamous cell carcinoma of the head and neck (SCCHN) is extremely poor. New therapeutic options emerge but need to establish their economic value. The objective was to describe the direct and related costs of R/M SCCHN in France.Materials and methods: We selected all adult patients treated with chemotherapy for R/M SCCHN between 1 January 2009 and 31 December 2014 from the permanent sample of the French national health insurance database (EGB). Data were analyzed from the index date (first chemotherapy) until patients’ death or 31 December 2015. “Treatment period” and “end-of-life” (EoL) (from last chemotherapy until death) were distinguished. Costs included all hospitalizations for SCCHN and ambulatory care. Costs of hospitalized and non-hospitalized adverse events (AEs) were estimated.Results: Among 267 patients identified, 85% were men, 44% had metastases at the index date and the mean age was 62.0 years (±9.9). The most common tumor location was oropharynx (29%) but 39% of patients had multiple locations. Median OS was 9.3 (95% CI: 7.9–11.8) months for the overall population. The average total direct cost per patient was €49,954, broken down into €32,908 (95% CI: 29,525–36,290) for hospitalizations and €17,047 (14,941–19,152) for ambulatory care. Main cost drivers were drug acquisition and administration (€14,538) during the treatment period (209 days on average) and palliative care (€3,750) during the EoL period (125 days). Regarding related costs, around 12% of patients received disability pensions (€1,397 per patient [624–2,171]) and sick leave payments (€1,592 [888–2,297]). “Metabolism and nutrition disorders” and “Infections and infestations” were the most expensive hospitalized AEs (€1,513 and €1,180 per patient, respectively). Febrile neutropenia was the most expensive non-hospitalized AE (€766 per patient).Conclusions: This analysis of real-world data confirms the poor prognosis of patients with R/M SCCHN and provides cost data for future economic evaluations.

ConvSCCS: convolutional self-controlled case series model for lagged adverse event detection

With the increased availability of large electronic health records databases comes the chance of enhancing health risks screening. Most post-marketing detection of adverse drug reaction (ADR) relies on physicians' spontaneous reports, leading to under-reporting. To take up this challenge, we develop a scalable model to estimate the effect of multiple longitudinal features (drug exposures) on a rare longitudinal outcome. Our procedure is based on a conditional Poisson regression model also known as self-controlled case series (SCCS). To overcome the need of precise risk periods specification, we model the intensity of outcomes using a convolution between exposures and step functions, which are penalized using a combination of group-Lasso and total-variation. Up to our knowledge, this is the first SCCS model with flexible intensity able to handle multiple longitudinal features in a single model. We show that this approach improves the state-of-the-art in terms of mean absolute error and computation time for the estimation of relative risks on simulated data. We apply this method on an ADR detection problem, using a cohort of diabetic patients extracted from the large French national health insurance database (SNIIRAM), a claims database containing medical reimbursements of more than 53 million people. This work has been done in the context of a research partnership between Ecole Polytechnique and CNAMTS (in charge of SNIIRAM).

The value of a health insurance database to conduct pharmacoepidemiological studies in oncology

Some concerns have emerged about the evidence of benefits on survival outcomes or quality of life of new anticancer drugs. In parallel, the decreased cancer mortality leads to an increased number of patients exposed to cancer treatment-related consequences. In this context, pharmacoepidemiology is crucial to assess anticancer drug use, effectiveness and safety in real life conditions. We aimed to describe strengths, limitations and considerations associated with the use of the French national health insurance database (système national des données de santé [SNDS]) to conduct pharmacoepidemiological studies in oncology. The SNDS represents a powerful tool in pharmacoepidemiology owing to its extensive coverage, accurate description and quantification of drug exposure and individual data on patients. The main limitations of this database ensue from the administrative nature resulting in technical difficulties in its management and gaps in availability of data. Another limitation is the lack of accurate identification of diseases, comorbidities or outcomes and potential confounding with notably the lack of data regarding cancer stage, prognosis or risk factors. Finally, the accurate identification of the nature of chemotherapy received by patients is sometimes complex. To minimize these limitations, several approaches and statistical methods could be used as highlighted by national or international initiatives. First, the SNDS may be linked with cancer registry or clinical data. Then, several data sources could be combined using meta-analytical methods. The development of methodological tools and the use of standardized methods are crucial to enhance the quality of studies that can impact clinical practice and guide public decision. Pharmacoepidemiological approaches and pharmacovigilance represent an important cornerstone in oncology for signal detection or long-term follow up of cancer patients. In this context, validated methods to identify cancer patients and to describe chemotherapy regimens within these data should be promoted and remain too scarce despite international guidelines. Moreover, limits and strength of each data sources should be systematically discussed according to the research question. Optimized and framed use of claims database represents a future challenge in onco-pharmacoepidemiology.

Is there overuse of proton pump inhibitors in B-cell non-Hodgkin lymphomas? A cohort study based on the French health insurance database in the Midi-Pyrénées region.

Patients suffering from B-cell non-Hodgkin lymphomas (B-NHL) have an increased likelihood of being exposed to proton pump inhibitors (PPIs), related to several factors which have been reported in the literature. PPIs are among the drugs most likely to be prescribed inappropriately. Consequently, B-NHL patients could be particularly at risk of inappropriate PPI prescription, with potential adverse drug reactions. We aimed to evaluate the incidence of PPIs use and to identify factors associated with PPIs initiation during the active treatment phase of B-NHL. We conducted a new-user cohort study using regional data from the French national health insurance database in the Midi-Pyrénées region (southwestern France). Incident B-NHL patients were selected according to an algorithm of selection, validated with data from a cancer registry. Our study revealed that 48.9% (95% confidence interval [CI]: 45.2-52.6) of patients initiated PPIs during chemotherapy after B-NHL diagnosis. According to information available in the SNDS, recommended indications for PPI prescriptions were identified in 21.1% of cases. Median duration of treatment was 65.3days (CI: 35-112). Determinants of PPIs initiation were peptic ulcer disease, gastroprotection (appropriate or not) for medications considered at risk (NSAIDs, glucocorticoids and anticoagulants), age, nonfollicular lymphoma, polypharmacy, gastroenterologists' consultations and being hospitalized in a university hospital. Around 50% of patients initiated PPI treatment during the chemotherapy phase with only one-fifth identified as appropriate prescriptions and with long durations of treatment in most cases. Given this background, appropriate PPI prescription should be promoted in B-NHL to avoid potential inappropriate chronic use and related adverse events.

Use of oral anticoagulants in the treatment of non-valvular atrial fibrillation in France: Patient characteristics from the NAXOS cohort study

Oral anticoagulants (OAC: vitamin K antagonists [VKAs], dabigatran, rivaroxaban, and apixaban) are used for prevention of stroke and systemic embolism in patients with non-valvular atrial fibrillation (NVAF). While results from pivotal randomized trials are available, French-specific, “real-life” data pertaining to the four OAC treatments available are awaited. This analysis of the NAXOS study aimed to describe demographic and clinical characteristics of NVAF patients initiating an OAC. All patients aged ≥ 18 years with NVAF and newly initiating one of the study OACs (i.e. no dispensation of the same OAC in the 24 prior months) between 2014 and 2016 were identified in the French national health insurance database (SNIIRAM–a comprehensive in- and outpatient healthcare consumption database). Patients were allocated to 4 different treatment cohorts. To maximise the comparability of populations, only OAC-naive patients (i.e. no dispensation of any OAC during the 24 months prior) were studied. Overall, 321,501 AC-naive NVAF patients aged 18+ were identified. Of these, 112,628 (35.0%) initiated VKAs, while 87,565 (27.2%), 100,063 (31.1%), and 21,245 (6.6%) initiated apixaban, rivaroxaban and dabigatran, respectively. Median follow-up durations were 218, 213, 205 and 186 days for VKA, apixaban, rivaroxaban and dabigatran cohorts, respectively. Males represented 48.8% of patients in VKA cohort (mean age: 78.5), 51.2% in apixaban cohort (mean age: 74.7), 55.1% in rivarobaxan cohort (mean age: 72) and 54.1% in dabigatran cohort (mean age: 72.7). Median age-adjusted Charlson score was 6 in the VKA cohort, and 4 in the other three cohorts. The NAXOS study confirms the widespread use of OAC treatments in France, suggesting differences between cohorts for age, gender and comorbidities. Thanks to large cohort sizes and prolonged follow-up, longitudinal analyses will allow assessment of real-life effects of OAC therapy in NVAF.

Using Healthcare Claims Data to Analyze the Prevalence of Chronic Myeloid Leukemia in France: A Nationwide Population-Based Study

Background: In France, as in many other countries, nationwide data on prevalence are rarely available and recent prevalence estimates of Chronic Myeloid Leukemia (CML) are scarce. Improved overall survival following the introduction of tyrosine kinase inhibitors (TKIs) is expected to have increased the prevalence of CML in Western countries.Aim: We sought to estimate and analyze the prevalence of CML in France for the year 2014 using a large health care claim-based dataset.Methods: Using the French national health insurance database that covers 98.8% of the French population (66 million people) we implemented a 3-step approach. First, focusing on the 2006-2014 period, we selected: 1) all patients treated with a TKI (ie, imatinib, dasatinib, nilotinib, bosutinib or ponatinib) and/or 2) identified by the ICD-10 diagnosis code C92.1 (Chronic Myeloid Leukemia, BCR/ABL-positive) among hospital discharge diagnoses and/or 3) identified by the ICD-10 diagnosis code C92 (myeloid leukemia) for coinsurance exemption. Then, we developed a claim-based algorithm to identify CML cases. Case definition was based on 1) identifying any TKI reimbursement lasting ≥ 2 months and 2) excluding patients receiving TKIs for diseases other than CML including Phi+ Acute Lymphoblastic Leukemia, Gastrointestinal Stromal Tumor, Stromal or other connective tissue tumor, and hypereosinophilic disease.Finally, prevalent CML cases were those identified by the algorithm above and having ≥ 1 healthcare reimbursement during the year 2014 and still alive on December, 31st 2014. The internal validity of the algorithm was tested on a random sample of 100 potential CML cases fulfilling ≥ 1/3 selection criteria in step 1 by comparing the results of the algorithm with the opinion of two hematologists (gold standard). For each individual, hematologists reviewed patient demographics and the sequence of care from 2006 to 2014 including healthcare resource utilization (ie, all hospitalizations and ICD-10 diagnosis codes, all medication use, all specialist consultations with date and specialist type). In addition, we assessed the external validity of the algorithm by comparing the number of incident CML patients in 2014 as identified in the French national health insurance database with the number of incident CML cases recorded in the French cancer registries for respective departments (i.e. ~ 20% of the French territory).Results: There were 10,789 prevalent CML cases in 2014 out of 68,067 individuals from the French national health insurance database who fulfilled the selection criteria for the overall 2006-2014 period. Eighty-nine percent of the prevalent CML cases were identified by at least two out of three selection criteria (TKI, ICD-10 code C92.1 among hospital discharge diagnoses, ICD-10 code C92 for coinsurance exemption). There was a 96% concordance rate (internal validity) between the algorithm and the opinion of the hematologists. For the year 2014, 162 and 150 incident CML patients were identified by the algorithm and the French cancer registries, respectively (high external validity).Median age [Inter-Quartile Range] of the prevalent population of CML patients was 63 years [51-73], with slightly more males affected (55%). On December, 31st 2014, the crude prevalence of CML was estimated at 16.3 per 100,000 inhabitants [95% confidence interval (CI) 16.0-16.6]. The crude prevalence of CML was 18.5 per 100,000 in men (95% CI 18.0-19.0) and 14.2 per 100,000 in women (95% CI 13.8-14.6). The crude prevalence of CML was less than 1.6 per 100,000 (95% CI 1.2-2.0) before 20 years of age, progressively increasing to 19.4 per 100,000 (95% CI 18.1-20.7) among those with 50-54 and reaching a peak of 48.2 per 100,000 (95% CI 45.3-51.1) at 75-79 years. There was a male preponderance in CML prevalence in all age groups. The crude prevalence of CML varied in a ratio of one to two throughout the French territory (from 10.2 to 23.8 per 100,000 inhabitants).Conclusion: Healthcare claims data are increasingly used to estimate epidemiological parameters worldwide. This approach is particularly relevant for rare diseases and administrative databases with high population coverage. Countries without national cohorts or cancer registries could easily use our algorithm to estimate their prevalence of CML. DisclosuresCony-Makhoul:Pfizer: Consultancy; BMS: Consultancy, Speakers Bureau; Incyte: Other: Travels for attending to Congress; Novartis: Consultancy, Other: Writing support, Travels for attending to Congress. Guerci-Bresler:Pfizer: Other: Fees for symposiums and boards; Novartis: Consultancy, Other: Fees for symposiums and boards; Incyte: Other: Fees for symposiums and boards; BMS: Other: Fees for symposiums and boards; Pfizer: Other: Travel fees for Congress. Delord:Incyte: Consultancy.

Two year continuation rates of contraceptive methods in France: a cohort study from the French national health insurance database

Objective: The aim of this study was to evaluate the continuation rates of reimbursed contraceptive methods in French real-world conditions.Methods: A retrospective cohort study using a representative sample of the national health insurance database, the General Sample of Beneficiaries (Echantillon Généralistes des Bénéficiaires [EGB]), was performed between 2006 and 2012. Selected women were ≥15 years of age and had started a reimbursed contraceptive method between 2009 and 2012 without prior reimbursement for an implant or an intrauterine contraceptive method between 2006 and 2008. The outcome of interest was the continuation rates, defined as the probability of women initiating a contraceptive method and continuing to use the same method over time. Continuation rates were assessed for up to 2 years. Only the first contraceptive method used during the study period was considered in the analysis. Non-parametric Kaplan–Meier survival analysis was used to assess continuation rates.Results: A population of 42,365 women representative of the 4,109,405 French women initiating any reimbursed method between 2009 and 2012 was identified in the EGB: 74.5% of women used oral contraceptives, 12.8% the levonorgestrel-releasing intrauterine system (LNG-IUS), 9.2% the copper intrauterine device (Cu-IUD) and 3.5% the subdermal etonogestrel (ENG) implant. The 2 year continuation rates varied from 9.1% for progestin-only oral contraceptives, 27.6% for first to second generation combined oral contraceptives (COCs) and 33.4% for third generation COCs to 83.6% for the ENG implant, 88.1% for the Cu-IUD and 91.1% for the LNG-IUS.Conclusion: This study conducted in real-world conditions showed that long-acting reversible contraceptive (LARC) methods remain rarely used in France despite high continuation rates over 2 years. Increasing the use of LARC methods is therefore a public health priority.

Reply to the Letter to the Editor: Is There a Cardiotoxicity Associated With Metallic Head Hip Prostheses? A Cohort Study in the French National Health Insurance Databases.

To the Editor, We thank Drs. Matharu and Whitehouse for their interest in our research [2] and for their valuable contribution. Although we appreciate their comments, we would like to make the following clarifications. As a reminder, the main objective of our study was to assess the risk of dilated cardiomyopathy or heart failure specifically associated with metallic versus nonmetallic head THA using data from the French national health insurance databases. In their first comment, Drs. Matharu and Whitehouse argue that because THA stems contain metal, not accounting for stem material in these THAs may have caused substantial residual confounding. We agree that all of the THA metal-containing components should be considered in order to properly quantify a potential role of THA metal overall. In our study, because information on stem material was not available and therefore could not be accounted for, the magnitude of the overall association between THA metal and the risk of dilated cardiomyopathy or heart failure may have been underestimated. However, since all THA stems are made of metal (most commonly cobalt chrome or titanium), and since there is no particular reason to believe that one head material (ceramic or metal) is more likely to be used with one of those stem materials, it seems unlikely to us that stem material would cause residual confounding. Another critique relates to the lack of a control group of individuals without THA. We agree that such a control group would be of interest to assess the cardiovascular risks associated with THA. However, it turns out that osteoarthritis is itself a cardiovascular risk factor [1] independent of the presence or absence of a THA, and so estimating the proper role of THA characteristics in explaining an increased risk in patients with versus without THA would require accounting for detailed information on individual osteoarthritis diagnosis and severity. Unfortunately, such information is not available in the French national health insurance databases. In our study, we compared the risk between patients with metallic versus nonmetallic head THAs. Regarding osteoarthritis diagnosis and severity, the two groups were a priori comparable after adjusting for major other cardiovascular risk factors. Such a design was the most appropriate to adequately answer the question of a potential role of metallic versus nonmetallic head THA in increasing the risk of dilated cardiomyopathy or heart failure. Lastly, we agree with Drs. Matharu and Whitehouse’s point stating that, in addition to our study focusing on morbidity outcomes, further studies considering cause-specific mortality would be of interest to assess the cardiovascular risks of metallic head THAs in more detail.

Letter to the Editor: Is There a Cardiotoxicity Associated With Metallic Head Hip Prostheses? A Cohort Study in the French National Health Insurance Databases.

Letter to the Editor: Is There a Cardiotoxicity Associated With Metallic Head Hip Prostheses? A Cohort Study in the French National Health Insurance Databases.

Multicentre cohort study of antihypertensive and lipid-lowering therapy cessation after bariatric surgery.

Few studies have assessed changes in antihypertensive and lipid-lowering therapy after bariatric surgery. The aim of this study was to assess the 6-year rates of continuation, discontinuation or initiation of antihypertensive and lipid-lowering therapy after bariatric surgery compared with those in a matched control group of obese patients. This nationwide observational population-based cohort study used data extracted from the French national health insurance database. All patients undergoing gastric bypass or sleeve gastrectomy in France in 2009 were matched with control patients. Mixed-effect logistic regression models were used to analyse factors that influenced discontinuation or initiation of treatment over a 6-year interval. In 2009, 8199 patients underwent primary gastric bypass (55·2 per cent) or sleeve gastrectomy (44·8 per cent). After 6 years, the proportion of patients receiving antihypertensive and lipid-lowering therapy had decreased more in the bariatric group than in the control group (antihypertensives: -40·7 versus -11·7 per cent respectively; lipid-lowering therapy: -53·6 versus -20·2 per cent; both P < 0·001). Gastric bypass was the main predictive factor for discontinuation of therapy for hypertension (odds ratio (OR) 9·07, 95 per cent c.i. 7·72 to 10·65) and hyperlipidaemia (OR 11·91, 9·65 to 14·71). The proportion of patients not receiving treatment at baseline who were subsequently started on medication was lower after bariatric surgery than in controls for hypertension (5·6 versus 15·8 per cent respectively; P < 0·001) and hyperlipidaemia (2·2 versus 9·1 per cent; P < 0·001). Gastric bypass was the main protective factor for antihypertensives (OR 0·22, 0·18 to 0·26) and lipid-lowering medication (OR 0·12, 0·09 to 0·15). Bariatric surgery is associated with a good discontinuation of antihypertensive and lipid-lowering therapy, with gastric bypass being more effective than sleeve gastrectomy.

Persistence of treatment with biologics for patients with psoriasis: a real‐world analysis of 16 545 biologic‐naïve patients from the French National Health Insurance database (SNIIRAM)

Long-term clinical effectiveness of biologics in psoriasis is needed. We aimed to assess the long-term persistence of biologics used to treat psoriasis in a real-life setting. All adults with psoriasis having been registered in the French National Health Insurance database (SNIIRAM) between 2008 and 2016 were eligible for inclusion. Psoriasis was defined as the fulfilment of at least two prescriptions for topical formulations of a vitamin D derivative within a 2-year period. The study population comprised biologic-naïve patients, i.e. those with a first prescription of etanercept, infliximab, adalimumab or ustekinumab. Persistence of treatment with a biologic was defined as the time interval between initiation and discontinuation. In this nationwide population-based cohort, 16 545 out of 874 549 patients with psoriasis were biologic-naïve (mean age 48·6 years; males 57·3%, mean follow-up 3·6 years). The mean ± SD length of follow-up for biologic-naïve patients was 3·6 ± 2·4 years. There were 9988 treatment discontinuations. Kaplan-Meier survival analyses revealed a persistence rate of 61·9% for the first, 33·3% for the third and 22·6% for the fifth year. Ustekinumab had a higher persistence rate than the other biologics. This finding should be interpreted with caution, in view of differences in administration between the biologics. About 85% of patients, having discontinued their first biologic, resumed systemic treatment of some type in the following year (biologics in 85% of cases). Our data suggest that biologics are less effective than physicians have been led to believe in a real-life, nonselected population. Further, long-term disease control requires several courses of different biologics.