Search

IntroductionThe French emergency department (ED) surveillance network OSCOUR transmits data on ED visits to Santé publique France (the national public health agency). As these data are collected daily and are almost exhaustive at a national level, it would seem relevant to use them for national epidemiological surveillance of mild traumatic brain injury (mTBI). This article presents the protocol of a planned study to validate algorithms for identifying mTBI in the OSCOUR database. Algorithms to be tested will be based on International Classification of Diseases (ICD)-10 codes.Methods and analysisWe will perform a multicentre validation study of algorithms for identifying mTBI in OSCOUR. Different combinations of ICD-10 codes will be used to identify cases of mTBI in the OSCOUR database. A random sample of mTBI cases and non-cases will be selected from four EDs. Medical charts will serve as the reference standard to validate the algorithms. The sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of the different algorithms, as well as their 95% CIs, will be calculated and compared.Ethics and disseminationThe ethics committee of the French National Data Protection Authority (CNIL) approved this study (n° 921152, 1 August 2021). Results will be submitted to national and international peer-reviewed journals and presented at conferences dedicated to trauma and to methodologies for the construction and validation of algorithms.

Background The capacity to use data linkage and artificial intelligence to estimate and predict health indicators varies across European countries. However, the estimation of health indicators from linked administrative data is challenging due to several reasons such as variability in data sources and data collection methods resulting in reduced interoperability at various levels and timeliness, availability of a large number of variables, lack of skills and capacity to link and analyze big data. The main objective of this study is to develop the methodological guidelines calculating population-based health indicators to guide European countries using linked data and/or machine learning (ML) techniques with new methods. Method We have performed the following step-wise approach systematically to develop the methodological guidelines: i. Scientific literature review, ii. Identification of inspiring examples from European countries, and iii. Developing the checklist of guidelines contents. Results We have developed the methodological guidelines, which provide a systematic approach for studies using linked data and/or ML-techniques to produce population-based health indicators. These guidelines include a detailed checklist of the following items: rationale and objective of the study (i.e., research question), study design, linked data sources, study population/sample size, study outcomes, data preparation, data analysis (i.e., statistical techniques, sensitivity analysis and potential issues during data analysis) and study limitations. Conclusions This is the first study to develop the methodological guidelines for studies focused on population health using linked data and/or machine learning techniques. These guidelines would support researchers to adopt and develop a systematic approach for high-quality research methods. There is a need for high-quality research methodologies using more linked data and ML-techniques to develop a structured cross-disciplinary approach for improving the population health information and thereby the population health.

Background The use of machine learning techniques is increasing in healthcare which allows to estimate and predict health outcomes from large administrative data sets more efficiently. The main objective of this study was to develop a generic machine learning (ML) algorithm to estimate the incidence of diabetes based on the number of reimbursements over the last 2 years. Methods We selected a final data set from a population-based epidemiological cohort (i.e., CONSTANCES) linked with French National Health Database (i.e., SNDS). To develop this algorithm, we adopted a supervised ML approach. Following steps were performed: i. selection of final data set, ii. target definition, iii. Coding variables for a given window of time, iv. split final data into training and test data sets, v. variables selection, vi. training model, vii. Validation of model with test data set and viii. Selection of the model. We used the area under the receiver operating characteristic curve (AUC) to select the best algorithm. Results The final data set used to develop the algorithm included 44,659 participants from CONSTANCES. Out of 3468 variables from SNDS linked to CONSTANCES cohort were coded, 23 variables were selected to train different algorithms. The final algorithm to estimate the incidence of diabetes was a Linear Discriminant Analysis model based on number of reimbursements of selected variables related to biological tests, drugs, medical acts and hospitalization without a procedure over the last 2 years. This algorithm has a sensitivity of 62%, a specificity of 67% and an accuracy of 67% [95% CI: 0.66–0.68]. Conclusions Supervised ML is an innovative tool for the development of new methods to exploit large health administrative databases. In context of InfAct project, we have developed and applied the first time a generic ML-algorithm to estimate the incidence of diabetes for public health surveillance. The ML-algorithm we have developed, has a moderate performance. The next step is to apply this algorithm on SNDS to estimate the incidence of type 2 diabetes cases. More research is needed to apply various MLTs to estimate the incidence of various health conditions.

Background The InfAct (Information for Action) project is a European Commission Joint Action on Health Information which has promoted the potential role of burden of disease (BoD) approaches to improve the current European Union-Health Information System (EU-HIS). It has done so by raising awareness of the concept, the methods used to calculate estimates and their potential implications and uses in policymaking. The BoD approach is a systematic and scientific effort to quantify and compare the magnitude of health loss due to different diseases, injuries, and risk factors with estimates produced by demographic characteristics and geographies for specific points in time. Not all countries have the resources to undertake such work, and may therefore start with a more restricted objective, e.g., a limited number of diseases, or the use of simple measures of population health such as disease prevalence or life expectancy. The main objective to develop these recommendations was to facilitate those countries planning to start a national burden of disease study. Results These recommendations could be considered as minimum requirements for those countries planning to start a BoD study and includes following elements: (1) Define the objectives of a burden of disease study within the context of your country, (2) Identify, communicate and secure the benefits of performing national burden of disease studies, (3) Secure access to the minimum required data sources, (4) Ensure the minimum required capacity and capability is available to carry out burden of disease study, (5) Establish a clear governance structure for the burden of disease study and stakeholder engagement/involvement, (6) Choose the appropriate methodological approaches and (7) Knowledge translation. These were guided by the results from our survey performed to identify the needs of European countries for BoD studies, a narrative overview from four European countries (Belgium, Germany, The Netherlands and Scotland) and the summary of a comparative study of country health profiles with national health statistics. Conclusions These recommendations as minimum requirements would facilitate efforts by those European countries who intend to perform national BoD studies.

Background The use of artificial intelligence is increasing to estimate and predict health outcomes from large data sets. The main objectives were to develop two algorithms using machine learning techniques to identify new cases of diabetes (case study I) and to classify type 1 and type 2 (case study II) in France. Methods We selected the training data set from a cohort study linked with French national Health database (i.e., SNDS). Two final datasets were used to achieve each objective. A supervised machine learning method including eight following steps was developed: the selection of the data set, case definition, coding and standardization of variables, split data into training and test data sets, variable selection, training, validation and selection of the model. We planned to apply the trained models on the SNDS to estimate the incidence of diabetes and the prevalence of type 1/2 diabetes. Results For the case study I, 23/3468 and for case study II, 14/3481 SNDS variables were selected based on an optimal balance between variance explained and using the ReliefExp algorithm. We trained four models using different classification algorithms on the training data set. The Linear Discriminant Analysis model performed best in both case studies. The models were assessed on the test datasets and achieved a specificity of 67% and a sensitivity of 62% in case study I, and a specificity of 97 % and sensitivity of 100% in case study II. The case study II model was applied to the SNDS and estimated the prevalence of type 1 diabetes in 2016 in France of 0.3% and for type 2, 4.4%. The case study model I was not applied to the SNDS. Conclusions The case study II model to estimate the prevalence of type 1/2 diabetes has good performance and will be used in routine surveillance. The case study I model to identify new cases of diabetes showed a poor performance due to missing necessary information on determinants of diabetes and will need to be improved for further research.

As in many countries, there is an urgent need to prioritize the issues in public health care and prevention in France. The French GBD project aims to offer a unique set of combined morbidity and mortality estimates to inform health policy makers in France, at the national level. However, because of the variability in health status and risk factors across geographical areas, policy makers need comparable estimates at the regional and local levels to target specific populations and to implement actions to improve health. Besides, although the accuracy of the indicators provided by the GBD has not been evaluated in the French context, it would be likely improved by enhancing and enriching the French data used as input. The French national health data system is one of the largest healthcare claims database in Europe. Based on these data, the French National Health Insurance (Cnam) and Santé publique France (SpFrance), in collaboration with clinicians and researchers, have developed a set of algorithms to identify diseases based on health care reimbursements, which could improve GBD estimates. Cnam and SpFrance both contribute to the definition, prioritization and evaluation of health policies in France and have therefore decided to pool their strengths and mutualize their resources to build a common rationale and provide consistent French Burden of Diseases estimates. A collaboration with the Institute of Health Metrics and Evaluation (IHME) has also been initiated, to allow Cnam and SpFrance benefiting from IHME's infrastructure and expertise and together seek to improve the estimates of the indicators using comparable methods at local and national levels, and consistent with the overall Global Burden of Disease project. Finally, beyond the ongoing production of the GBD metrics, another objective is the knowledge transfer to Cnam and SpFrance as learning organizations. The project has started with a framework between the three parties and the expression of interest.

BackgroundThe availability of data generated from different sources is increasing with the possibility to link these data sources with each other. However, linked administrative data can be complex to use and may require advanced expertise and skills in statistical analysis. The main objectives of this study were to describe the current use of data linkage at the individual level and artificial intelligence (AI) in routine public health activities, to identify the related estimated health indicators (i.e., outcome and intervention indicators) and health determinants of non-communicable diseases and the obstacles to linking different data sources.MethodWe performed a survey across European countries to explore the current practices applied by national institutes of public health, health information and statistics for innovative use of data sources (i.e., the use of data linkage and/or AI).ResultsThe use of data linkage and AI at national institutes of public health, health information and statistics in Europe varies. The majority of European countries use data linkage in routine by applying a deterministic method or a combination of two types of linkages (i.e., deterministic & probabilistic) for public health surveillance and research purposes. The use of AI to estimate health indicators is not frequent at national institutes of public health, health information and statistics. Using linked data, 46 health outcome indicators, 34 health determinants and 23 health intervention indicators were estimated in routine. The complex data regulation laws, lack of human resources, skills and problems with data governance, were reported by European countries as obstacles to routine data linkage for public health surveillance and research.ConclusionsOur results highlight that the majority of European countries have integrated data linkage in their routine public health activities but only a few use AI. A sustainable national health information system and a robust data governance framework allowing to link different data sources are essential to support evidence-informed health policy development. Building analytical capacity and raising awareness of the added value of data linkage in national institutes is necessary for improving the use of linked data in order to improve the quality of public health surveillance and monitoring activities.

Background The availability of data generated from different sources is increasing with the possibility to link these data sources together. However, linked administrative data can be complex to use and may require advanced expertise and skills in statistical analysis. The main objectives of this study were to describe the current use of data linkage at the individual level and the artificial intelligence (AI) in routine public health activities, and to identify the related health outcome and intervention indicators and determinants of health for non-communicable diseases. Method We performed a survey across European countries to explore the current practices applied by national institutes of public health and health information and statistics for innovative use of data sources (i.e., the use of data linkage and/or the AI). Results The use of data linkage and the AI at national institutes of public health and health information and statistics in Europe varies. The majority of European countries use data linkage in routine by applying a deterministic method or a combination of two types of linkages (i.e., deterministic & probabilistic) for public health surveillance and research purposes. The use of AI to estimate health indicators is not frequent at national institutes of public health and health information and statistics. Using linked data, 46 health outcome indicators related to seven health conditions, 34 indicators related to determinants and 23 to health interventions were estimated in routine. Complex data regulation laws, lack of human resources, skills and problems with data governance, were reported by European countries as obstacles to link different data sources in routine for public health surveillance and research. Conclusions Our results highlight that the majority of European countries have integrated data linkage in routine public health activities but a few use the AI. A sustainable national health information system and a robust data governance framework allowing to link different data sources are essential to support evidence-informed health policy development process. Building analytical capacity and awareness of the added value of data linkage in national institutes is necessary for improving the utilization of linked data in order to improve the monitoring of public health activities.

Objectives In France, the early mortality monitoring, conducted by Sante publique France, the French National Public Health agency (SpFrance) (formerly French Institute for public health surveillance-InVS), is based on the administrative data provided by the National Institute for Statistic and Economic Studies (INSEE) and consequently does not allow analyses on medical causes of death. Since 2007, the physicians can certify deaths electronically. In this electronic system (Electronic Death Registration System; EDRS), the medical causes of death, in free-text format, are directly transmitted to SpFrance. In the future, these data could be used in a real-time surveillance system by medical causes of death. The objective of this study was to evaluate the pertinence of e-death certification using the following assessment criteria: timeliness, representativeness, and completeness of sociodemographic and medical information included in the e-death certificates. Study design This study consisted of a descriptive analysis of the information collected by e-death certificates recorded between January 1, 2012 and July 31, 2014. Methods The study quantified the temporal and geographical evolution of the deployment of the EDRS between 2012 and 2014. The timeliness of the system was estimated by calculating the delay between the dates of death and of data availability for analysis. Sociodemographic and death-related characteristics were described. The frequency of missing data was measured for each variable. The number of completed fields per certificate and the number of words per field and per certificate were calculated for the medical causes of death. Results Between January 2012 and July 2014, 77,776 e-death certificates were collected. A slight increase in the use of the e-death certification was observed during the study period, reaching 6.1% of the total number of deaths in 2014. Good national coverage was noted. Nearly 79% of e-certificates were submitted to SpFrance on the day of the death. We observed a high completeness of the e-certificates. The rate of missing data did not exceed 2.7% for sociodemographic variables. On average, 10 words, distributed in three fields, were used to describe the medical causes of death. Conclusions E-death certificates constitute a reactive source of information on medical causes of death. The deployment of EDRS is of major public health interest for the development of a real-time warning surveillance system of mortality by cause.

Background Previous studies on asthma mortality and hospitalizations in Reunion Island indicate that this French territory is particularly affected by this pathology. However, no estimates are provided on the prevalence of asthma among adults. In 2016, a cross-sectional survey was conducted to estimate the prevalence of asthma and to identify its associated factors in the adult population of Reunion Island. Methods A random sample of 2,419 individuals, aged 18-44 years, was interviewed by telephone using a standardized, nationally validated questionnaire. Information was collected on the respiratory symptoms, description of asthma attacks and triggering factors for declared asthmatics, as well as data on the indoor and outdoor home environment. “Current asthma” was defined as an individual declaring, at the time of the survey, having already suffered from asthma at some point during his/her life, whose asthma was confirmed by a doctor, and who had experienced an asthma attack in the last 12 months or had been treated for asthma in the last 12 months. “Current suspected asthma” was defined as an individual presenting, in the 12 months preceding the study, groups of symptoms suggestive of asthma consistent with the literature. Results The estimated prevalence of asthma was 5.4% [4.3-6.5]. After adjustment, women, obesity, a family member with asthma, tenure in current residence and presence of indoor home heating were associated with asthma. The prevalence of symptoms suggestive of asthma was 12.0% [10.2-13.8]. After adjustment, marital status, passive smoking, use of insecticide sprays, presence of mold in the home and external sources of atmospheric nuisance were associated with the prevalence of suspected asthma. Conclusions Preventive actions including asthma diagnosis, promotion of individual measures to reduce risk exposure as well as the development of study to improve knowledge on indoor air allergens are recommended. Key messages Individuals with suspected asthma who have not been medically diagnosed have been identified. Consequently, the development of a strategy to improve the diagnosis of asthma on Reunion is recommended. Aerating homes to reduce indoor moisture and mould growth, maintaining hygiene standards through cleaning, limiting the use of spray insecticides and household cleaning sprays at home.

Background Health information systems both at the national and international level play a key role in ensuring that timely and reliable evidence is used for operational and strategic decision making inside and outside the health sector. The availability of data generated from different sources is increasing with the possibility to link these data sources together. However, more efficient data generation processes are required to use data collected for different purposes initially, as well as advanced statistical techniques to generate comparable and timely health information. The main objective is to explore the innovative use of health information for better public health policy across the Member States. Methods As part of InfAct, we have conducted as survey among EU-MS to describe the innovative use of data sources. We are collecting inspiring examples on the innovative use of health information based on national or European data networks involved with health policy-making at national, regional or local level. We are further developing generic methods to estimate health indicators using machine learning techniques and mathematical modelling. Results These approaches will generate a roadmap on the innovative use of health information across Member States, enlarge the existing list of health indicators estimated from linked data and/or advanced statistical techniques, inform on the implications of these indicators in health policy with inspiring examples from Member States, and provide methodological guidelines for using linked data and advanced statistics to estimate health indicators, and composite outcome measures. Conclusions This work will highlight the gaps in the innovative use of data sources, and improve the comparability of health indicators and the capacity of EU-Member states to apply innovation for increased relevance and timeliness of health information for public health policy-making.

Timely mortality surveillance in France is based on the monitoring of electronic death certificates to provide information to health authorities. This study aims to analyze the performance of a rule-based and a supervised machine learning method to classify medical causes of death into 60 mortality syndromic groups (MSGs). Performance was first measured on a test set. Then we compared the trends of the monthly numbers of deaths classified into MSGs from 2012 to 2016 using both methods. Among the 60 MSGs, 31 achieved recall and precision over 0.95 for either one or the other method on the test set. On the whole dataset, the correlation coefficient of the monthly numbers of deaths obtained by the two methods were close to 1 for 21 of the 31 MSGs. This approach is useful for analyzing a large number of categories or when annotated resources are limited.

Objective The aim of this study is to present the syndromic groups that will be routinely monitored for the reactive mortality surveillance based on free-text medical causes of death. Introduction In 2004, Sante publique France, the French Public Health Agency set up a reactive all-cause mortality surveillance based on the administrative part of the death certificate, in the final objectives 1/ to detect unexpected or usual variations in mortality and 2/ to provide a first evaluation of mortality impact of events. In 2007, an Electronic Death Registration System (EDRS) was implemented, enabling electronic transmission of the medical causes of death to the agency in real-time. To date, 12% of the mortality is registered electronically. A pilot study demonstrated that these data were valuable for a reactive mortality surveillance system based on causes of death [1]. A strategy has thus been developed for the analysis in routine of the medical causes of death with the objectives of early detection of expected and unexpected outbreaks and reactive evaluation of their impact. This system will allow approaching the cause accountability when an excess death will be observed. Methods Mortality syndromic groups (MSG) were defined as clusters of medical causes of death (pathologies, syndromes or symptoms) that meet the objectives of the surveillance system. The causes of death are available reactively in free-text (words, terms, expressions) and with a delay of 6 to 24 months in ICD10 codes format. We explored multiple biomedical classifications such as the Mesh, SNOMED, UMLS or ICD10 to learn from their various ways to classify diseases. Based on ICD10, we defined MSGs by a list of ICD10 codes, each codes belonging to a unique MSGs. Each MSG definition was then discussed in working group including medical and epidemiological experts. Additionally, we used a dictionary (provided by the Epidemiology Center on Causes of Death (Inserm-CepiDc)) of each term/expression found in the death certificates since the early 2000 to enrich variety of expression of each MSG. We classified causes of death into MSGs from E-death certificates from 2012 to 2016: 1/ using the ICD10 codes assigned by Inserm-CepiDc based on rules defined by WHO in order to produce the national mortality statistics and 2/ using a linear Support Vector Machine (SVM) method to classify free-text causes of death. Then we compared the fluctuations of the weekly numbers of each MSG built by using both classification methods (ICD10 codes and the SVM classification) [2]. Results A list of a hundred MSGs was defined, divided into 20 topics (Respiratory conditions, Digestive conditions, Infectious conditions, Cardio and Cerebrovascular conditions, General symptoms…). 60 MSGs were dedicated to alert and detection of both expected seasonal epidemics (12 MSGs) and unexpected events (42 MSGs). They contain unspecific or acute pathologies and symptoms. 40 MSG included medical causes of death related to chronic diseases and medical history. The list of established MSGs was composed of: - MSGs for detection of expected seasonal events such as: “Influenza”, “Low acute respiratory infection”, “Gastroenteritis”, “Chikungunya”, “Heat related death”, “Dehydration”… - MSGs for detection of the impact of unexpected events such as: “Epilepsy”, “Choc”, “Coma”, “Unspecified fever”, “Headache”, “Suicide”, “Drugs/opioids poisoning”… - MSGs for Chronic diseases and Medical history: “Chronic digestive diseases”, “Chronic endocrine diseases”, “Genitourinary chronic diseases”, “History of diseases”… The weekly number of MSGs built using SVM classification was close and highly correlated to the weekly number of MSGs built using ICD10 codes (Figure 1). Seasonality and peaks were visible using both classifications. For instance, the increase of the MSG “Influenza” occurred during winter months which are known to be the circulating months of the influenza virus (Figure 1, left) [3]. For unusual and rare events such as death due to burns, we observed that the weekly numbers of MSG “Burns” were also similar using both methods. We observed (Figure 1, right) that the outbreak that occurred in September 2016 related to a major bus accident was found using ICD10 codes or SVM classification. Conclusions The use of free-text causes of death for reactive mortality surveillance requires the development of a strategy for the analysis of these data. Defining MSGs was essential for the implementation of automatic classification methods of the death certificates in routine. The dynamic of MSGs using ICD10 codes or SVM classification were comparable. However, the use of ICD10 codes for reactive mortality surveillance is not an option due to the delay of availability of the codes. The uses of machine learning methods, thus, enable to harness free-text causes of death for the reactive mortality surveillance with an objective of detection and early impact assessment. References 1. Lassalle M, Caserio-Schonemann C, Gallay A, Rey G, Fouillet A: Pertinence of electronic death certificates for real-time surveillance and alert, France, 2012–2014. Public Health 2017, 143:85-93.[1] 2. Baghdadi Y, Bourree A, Robert A, Rey G, Gallay A, Zweigenbaum P, Grouin C, Fouillet A: Automatic classification of medical causes from free-text death certificates for reactive mortality surveillance in France. Int J Med Inf 2018, Under review.[2] 3. Bedford T, Riley S, Barr IG, Broor S, Chadha M, Cox NJ, Daniels RS, Gunasekaran CP, Hurt AC, Kelso A et al : Global circulation patterns of seasonal influenza viruses vary with antigenic drift. Nature 2015, 523(7559):217-220.[3]

International audience; BACKGROUND: Mortality surveillance is of fundamental importance to public health surveillance. The real-time recording of death certificates, thanks to Electronic Death Registration System (EDRS), provides valuable data for reactive mortality surveillance based on medical causes of death in free-text format. Reactive mortality surveillance is based on the monitoring of mortality syndromic groups (MSGs). An MSG is a cluster of medical causes of death (pathologies, syndromes or symptoms) that meets the objectives of early detection and impact assessment of public health events. The aim of this study is to implement and measure the performance of a rule-based method and two supervised models for automatic free-text cause of death classification from death certificates in order to implement them for routine surveillance.METHOD: A rule-based method was implemented using four processing steps: standardization rules, splitting causes of death using delimiters, spelling corrections and dictionary projection. A supervised machine learning method using a linear Support Vector Machine (SVM) classifier was also implemented. Two models were produced using different features (SVM1 based solely on surface features and SVM2 combining surface features and MSGs classified by the rule-based method as feature vectors). The evaluation was conducted using an annotated subset of electronic death certificates received between 2012 and 2016. Classification performance was evaluated on seven MSGs (Influenza, Low respiratory diseases, Asphyxia/abnormal respiration, Acute respiratory disease, Sepsis, Chronic digestive diseases, and Chronic endocrine diseases).RESULTS: The rule-based method and the SVM2 model displayed a high performance with F-measures over 0.94 for all MSGs. Precision and recall were slightly higher for the rule-based method and the SVM2 model. An error-analysis shows that errors were not specific to an MSG.CONCLUSION: The high performance of the rule-based method and SVM2 model will allow us to set-up a reactive mortality surveillance system based on free-text death certificates. This surveillance will be an added-value for public health decision making.

Objective This study aims to implement and evaluate two automatic classification methods of free-text medical causes of death into Mortality Syndromic Groups (MSGs) in order to be used for reactive mortality surveillance. Introduction Mortality is an indicator of the severity of the impact of an event on the population. In France mortality surveillance is part of the syndromic surveillance system SurSaUD and is carried out by Sante publique France, the French public health agency. The set-up of an Electronic Death Registration System (EDRS) in 2007 enabled to receive in real-time medical causes of death in free-text format. This data source was considered as reactive and valuable to implement a reactive mortality surveillance system using medical causes of death (1). The reactive mortality surveillance system is based on the monitoring of Mortality Syndromic Groups (MSGs). An MSG is defined as a cluster of medical causes of death (pathologies, syndromes, symptoms) that meet the objectives of early detection and impact assessment of events (2). Since causes of death are entered in free-text format, their automatic classifications into MSGs require the use of natural language processing methods. We observe a constant increase in the use of these methods to classify medical information and for health surveillance over the last two decades (3). Methods Data consisted of the medical part of electronic death certificates received in routine by Sante publique France from 2012 to 2016. We split the dataset into training and test sets. Among each set, a subset of certificates was selected by a random sampling without replacement. Two annotators manually assigned MSGs to each death certificates in all subsets. Discordances were discussed and corrected if necessary. The agreement rate between the two annotators was 0.90 on the test set. Final annotated subsets represent the ground truth against which the methods tested were evaluated. The final evaluation was performed on the test set of 1,000 death certificates while the classifiers were trained on 3500 death certificates. Two classification methods were implemented: a rule-based method and a supervised machine learning method. The rule-based method was based on four processing steps: applying standardization rules, splitting of medical expression using delimiters, spelling correction and dictionary projection. The supervised machine learning method was set up using a linear Support Vector Machine (SVM) classifier. We trained a multi-label classifier using the one-versus-all strategy. We implemented two models: one based on surface features (SVM model) and the other, a hybrid model, combining surface features and features obtained by the rule-based method. Surface features were bags-of-word unigrams and bigrams and of character trigrams. The rule-based method and the two supervised machine learning models were evaluated using the three evaluation measures: precision (Positive Predictive Value), recall (Sensitivity) and F-measure (P/R/Fm). The study focused on the classification performance of MSGs defined for the reactive detection of outbreaks and are composed of unspecific or acute pathologies, or general symptoms (related to pain, fever, cognitive disorder…). Only the 40 MSGs mentioned at least 3 times in the test set were considered in this study, they belonged to 13 topics (Respiratory conditions, Cardio and cerebrovascular conditions, Infectious diseases, Digestive conditions…). Results With the rule-based method, among the 40 MSGs, 24 obtained a P/R/Fm over 0.90. They belonged mainly to the topics Cardio and Cerebrovascular conditions (5 MSGs), Respiratory conditions (6), and General symptoms (5). Four MSGs obtained P/R/Fm below 0.85 belonging to the topics Infectious conditions (2), Blood condition (1) and Unspecified causes of death (1). The hybrid model obtained P/R/Fm over 0.90 for 25 MSGs. Among them, 21 were the same as the rule-base method. Performance of the rule-based method and the hybrid model were over 0.95 for the same 13 MSGs. The hybrid model obtained P/R/Fm below 0.85 for 4 MSGs also belonging to the same topics as those of the rule-based method. The SVM model had lower classification performance than the two other models. Conclusions For syndromic mortality surveillance both precision and recall are important for all MSGs. Indeed to meet the objective of a reactive detection of events, high precision is needed to limit false alarms. To measure the impact of an event, the surveillance system should have high recall, to avoid an underestimation of this impact. This is especially true for rarer diseases. The results showed that the rule-based method and the hybrid model are the most effective to classify causes of death into MSGs. For some MSGs with less than 5 mentions in the test set (7%), these results must be qualified. Also, to improve classification performance for MSGs with performance below 0.90, and to confirm these results further analysis must be conducted. The results suggest the relevance of these methods to set up a reactive mortality surveillance system for detection and alert based on free-text causes of death. Such a system will provide useful information to health authorities regarding the causes of death during an event, helping them to adapt counter and prevention measures. References 1. Lassalle M, Caserio-Schonemann C, Gallay A, Rey G, Fouillet A. Pertinence of electronic death certificates for real-time surveillance and alert, France, 2012–2014. Public Health. 2017;143:85-93. (1) 2. Baghdadi Y, Gallay A, Caserio-Schonemann C, Thiam M-M, Fouillet A. Towards real-time mortality surveillance by medical causes of death: A strategy of analysis for alert. Rev Epidemiol Sante Publique. 2018;66:S402. (2) 3. Wang Y, Wang L, Rastegar-Mojarad M, Moon S, Shen F, Afzal N, et al. Clinical information extraction applications: A literature review. J Biomed Inf. 2018;77:34-49. (3)

International audience; BACKGROUND:Previous studies on asthma mortality and hospitalizations in Reunion Island indicate that this French territory is particularly affected by this pathology. Epidemiological studies conducted in schools also show higher prevalence rates in Reunion than in Mainland France. However, no estimates are provided on the prevalence of asthma among adults. In 2016, a cross-sectional survey was conducted to estimate the prevalence of asthma and to identify its associated factors in the adult population of Reunion Island.METHODS:A random sample of 2419 individuals, aged 18-44 years, was interviewed by telephone using a standardized, nationally validated questionnaire. Information was collected on the respiratory symptoms, description of asthma attacks and triggering factors for declared asthmatics, as well as data on the indoor and outdoor home environment. "Current asthma" was defined as an individual declaring, at the time of the survey, having already suffered from asthma at some point during his/her life, whose asthma was confirmed by a doctor, and who had experienced an asthma attack in the last 12 months or had been treated for asthma in the last 12 months. "Current suspected asthma" was defined as an individual presenting, in the 12 months preceding the study, groups of symptoms suggestive of asthma consistent with the literature.RESULTS:The estimated prevalence of asthma was 5.4% [4.3-6.5]. After adjustment, women, obesity, a family member with asthma, tenure in current residence and presence of indoor home heating were associated with asthma. The prevalence of symptoms suggestive of asthma was 12.0% [10.2-13.8]. After adjustment, marital status, passive smoking, use of insecticide sprays, presence of mold in the home and external sources of atmospheric nuisance were associated with the prevalence of suspected asthma.CONCLUSION:Preventive actions including asthma diagnosis, promotion of individual measures to reduce risk exposure as well as the development of study to improve knowledge on indoor air allergens are recommended.

Background In France, a mortality syndromic surveillance system was set up with objectives of early detection and reactive evaluation of the impact of expected and unexpected events to support decision makers. This study aims to describe the characteristics of the system and its usefulness for decision makers. Methods Anonymized data from the administrative part of death certificates were daily collected from 3062 computerized city halls and were transmitted to Santé publique France in routine. Coverage of the system was measured as the proportion of deaths registered by the system among the complete number of deaths and analyzed by age, month and region. Deaths were described by gender, age and geographical level using proportion. The excess periods of deaths were described based on the comparison of the weekly observed and expected numbers of deaths between 2012 and 2016. Results The system recorded 77.5% of the national mortality covering the whole territory. About 81% of deaths were aged 65 years old and more. The surveillance system identified mortality variations mainly during winter and summer, for some concomitant with influenza epidemic or heatwave period, and thus provided information for decision makers. Conclusion The ability of the system to detect and follow mortality outbreaks in routine in the whole territory has been demonstrated. It is a useful tool to provide early evaluation of the impact of threats on mortality and alert decision makers to adapt control measures. However, the absence of information on medical causes of death may limit the ability to target recommendations.

Background/Introduction The Covid19 epidemic has a considerable impact on healthcare systems in France. Firstly by a flooding of hospital resuscitation services, as well as of the call platforms for emergency services. The strict lockdown of the French population decreed on March 17, 2020 has also completely changed the conditions for seeking healthcare. Purpose The aim of our study was to give a first estimation of the impact of lockdown during the Covid19 crisis on the use of hospital emergency departments for the management of three groups of acute cardiovascular diseases: ischemic heart disease (IHD), stroke and pulmonary embolism (PE) Methods The French hospital emergency departments OSCOUR network daily collects 93 % of national ED visits, which are transmitted each day to the French public health agency. The weekly number of visits and the proportion of hospitalization following these visits between January 1, 2019 and April 12, 2020 were compared with the same weeks of 2019. Results During the first month of lockdown, the number of visits and hospitalizations after visits decreased by 35 % for IHD, 27 % for stroke, and 28 % (visits) and 18 % (hospitalizations) for PE compared to the same period in 2019. The decrease was even higher during the first three weeks of lockdown. The proportion of hospitalizations after visits was high and increased slightly for IHD (+3 %) and for stroke (+2 %), and markedly for PE (+8 %). Conclusion These preliminary data show a major impact of Covid19 lockdown on the management of acute cardiovascular diseases with a likely delay for treatment. This could lead to an increase in cardiovascular sequelae and mortality in the following months. Data on hospitalization and healthcare expenditure should complete this first analysis to have a more precise estimation of the loss of chance caused by this crisis for patients not affected by covid19.

Waterborne disease outbreaks (WBDOs) remain a public health issue in developed countries, but to date the surveillance of WBDOs in France, mainly based on the voluntary reporting of clusters of acute gastrointestinal infections (AGIs) by general practitioners to health authorities, is characterized by low sensitivity. In this context, a detection algorithm using health insurance data and based on a space&ndash, time method was developed to improve WBDO detection. The objective of the present simulation-based study was to evaluate the performance of this algorithm for WBDO detection using health insurance data. The daily baseline counts of acute gastrointestinal infections were simulated. Two thousand simulated WBDO signals were then superimposed on the baseline data. Sensitivity (Se) and positive predictive value (PPV) were both used to evaluate the detection algorithm. Multivariate regression was also performed to identify the factors associated with WBDO detection. Almost three-quarters of the simulated WBDOs were detected (Se = 73.0%). More than 9 out of 10 detected signals corresponded to a WBDO (PPV = 90.5%). The probability of detecting a WBDO increased with the outbreak size. These results underline the value of using the detection algorithm for the implementation of a national surveillance system for WBDOs in France.

SUMMARYWaterborne disease outbreaks (WBDO) of acute gastrointestinal illness (AGI) are a public health concern in France. Their occurrence is probably underestimated due to the lack of a specific surveillance system. The French health insurance database provides an interesting opportunity to improve the detection of these events. A specific algorithm to identify AGI cases from drug payment reimbursement data in the health insurance database has been previously developed. The purpose of our comparative study was to retrospectively assess the ability of the health insurance data to describe WBDO. Data from the health insurance database was compared with the data from cohort studies conducted in two WBDO in 2010 and 2012. The temporal distribution of cases, the day of the peak and the duration of the epidemic, as measured using the health insurance data, were similar to the data from one of the two cohort studies. However, health insurance data accounted for 54 cases compared to the estimated 252 cases accounted for in the cohort study. The accuracy of using health insurance data to describe WBDO depends on the medical consultation rate in the impacted population. As this is never the case, data analysis underestimates the total number of AGI cases. However this data source can be considered for the development of a detection system of a WBDO in France, given its ability to describe an epidemic signal.

Objective To describe the surveillance indicators implemented for the health impact assessment of a potential health event occurring before, during or after the UEFA Euro 2016 football matches in order to timely implement control and prevention measures. Introduction France hosted 2016 UEFA European Football Championship between June 10 and July 10. In the particular context of several terrorist attacks occurring in France in 2015 [1], the French national public health agency « Sante publique France » (formerly French Institute for Public Health Surveillance-InVS) was mandated by the Ministry of Health to reinforce health population surveillance systems during the UEFA 2016 period. Six French regions and 10 main stadiums hosted 51 matches and several official and nonofficial dedicated Fan Zones were implemented in many cities across national territory. Three types of hazard have been identified in this context: outbreak of contagious infectious disease, environmental exposure and terrorist attack. The objectives of health surveillance of this major sporting event were the same as for an exceptional event including mass gathering [2] : 1/ timely detection of a health event (infectious cluster, environmental pollution, collective foodborne disease...) to investigate and timely implement counter measures (control and prevention), 2/ health impact assessment of an unexpected event. The French national syndromic surveillance system SurSaUD® was one of the main tools for timely health impact assessment in the context of this event. Methods French national syndromic SurSaUD® system has been set up in 2004 and supervised by Sante publique France for 12 years. It allows the daily automatic collation of individual data from over 650 emergency departments (ED) involved in the OSCOUR® network and 61 emergency general practitioners’ (GPs) associations (SOS Medecins) [3]. About 60,000 attendances in ED (88% of the national attendances) and 8,000 visits in SOS Medecins associations (95% of the national visits) are daily recorded all over the territory and transmitted to Sante publique France. Medical information such as provisional medical diagnosis coded according to the International Classification of Diseases, 10th Revision (ICD-10) for EDs and specific thesaurus for SOS Medecins is routinely monitored through different syndromic indicators (SI). SI are defined by medically relevant clusters of one or several diagnoses, serving as proxies for conditions of public health interest. From June 10 to July 10, 19 SI were daily analyzed through automatic national and regional dashboards. SI were divided into 3 groups of public health surveillance interest : 1/ description of population health: injuries, faintness, myocardial infarction, alcohol, asthma, heat-related symptoms, anxious troubles ; 2/ infectious diseases/symptoms with epidemic potential or diseases/symptoms linked with an environmental exposure: fever, fever associated with cutaneous rash, meningitis, pneumonia, gastroenteritis, collective foodborne disease ; 3/ symptoms potentially linked with a CBRN-E exposure: influenza-like illness, burns, conjunctivitis, dyspnea/ difficulty breathing, neurological troubles, acute respiratory failure. Daily analysis were integrated into specific UEFA 2016 surveillance bulletins and daily sent to the Ministry of Health including week-ends. Results SI followed during the UEFA Euro 2016 period were nonspecific and potentially affected or influenced by several events appart from the championship. Between June 10 and July 10, two moderate heat-wave periods occurred on a large part of mainland France : the first one from June 22 to 25 (beginning in the West-South of France and then moving North and East of the country) and the second one from July 8 to 11 in the East-South. An increase in heat-related indicators (hyperthermia/heat stroke, dehydration, hyponatremia and burns) has been observed during both periods in five French regions including four hosting regions. Only minor increases in the other SI followed during the Euro 2016 period were observed. Conclusions Health surveillance implemented during 2016 UEFA European Football Championship through a daily analysis of non-specific SI from the French syndromic surveillance system SurSaUD® did not show any major variation associated with the sporting event. The observed variations were related with specific environmental conditions (heat-waves). Together with the health surveillance system, preventive plans were set up during the event essentially by offering flyers with information and useful tips on the main preventive attitudes and measures to adopt in a summer festive context (risks associated with alcohol and drug intake, injuries, heat and sun exposure, dehydration, unprotected sexual behaviour...).

International audience; Following of the emergence of Zika virus in Brazil in 2015, an epidemiological surveillance system was quickly implemented in the French overseas Territories of America (FTA) according to previous experience with dengue and chikungunya and has detected first cases of Zika. General practitioners and medical microbiologists were invited to report all clinically suspectedcases of Zika, laboratory investigations were systematically conducted (RT-PCR). On 18 December, the first autochthonous case of Zika virus infection was confirmed by RT-PCR on French Guiana and Martinique, indicating introduction of Zika virus inFTA. The viral circulation of Zika virus was then also confirmed on Guadeloupe and Saint-Martin. We report here early findings on 203 confirmed cases of Zika virus infection identified by RT-PCR or seroneutralisation on Martinique Island between 24 November 2015 and 20 January 2016. All cases were investigated. Common clinical signs were observed (maculopapularrash, arthralgia, fever, myalgia and conjunctival hyperaemia) among these patients, but the rash, the foundation of our case definition, may be absent in a significant proportion of patients (16%). These results are important for the implementation of a suspected case definition, the main tool for epidemiological surveillance, in territories that may be affected by ZIKV emergence, including Europe.

Presentation of the impact of two major heat waves occurring in July 2015 in France on heat-related causes in emergency health care facilities, using the French syndromic surveillance system SurSaUD®.

Zika virus (ZIKV) has recently spread widely and turned into a major international public health threat. Réunion appears to offer conditions particularly favourable to its emergence and therefore prepared to face possible introduction of the virus. We designed a scaled surveillance and response system with specific objectives, methods and measures for various epidemiological phases including a potential epidemic. Several tools were developed in order to (i) detect individual cases (including a large information campaign on the disease and suspicion criteria), (ii) monitor an outbreak through several complementary systems allowing to monitor trends in disease occurrence and geographic spread and (iii) detect severe forms of the disease in collaboration with hospital clinicians. We put the emphasis on detecting the first cases in order to contain the spread of the virus as much as possible and try to avoid progress towards an epidemic. Our two main strengths are a powerful vector control team, and a close collaboration between clinicians, virologists, epidemiologists, entomologists and public health authorities. Our planned surveillance system could be relevant to Europe and island settings threatened by Zika virus all over the world.

Background Substantial increases in syphilis have been reported since the early 2000s in northern countries, particularly among men who have sex with men (MSM). The authors aimed to identify risk factors for early syphilis in MSM in Lille, a large urban area of northern France. Methods A matched case–control study was conducted in MSM aged ≥18 years. Cases were diagnosed with primary, secondary or early latent syphilis between April 2008 and June 2010. Controls sought care in STIs clinics or were followed in an HIV clinic. Controls had no history of and no current syphilis. They were matched to cases for age and HIV status. Multivariate conditional logistic regression models were used to identify risk factors for early syphilis. Results 53 patients with early syphilis were enrolled. Average age was 37 years, and 47% were HIV-infected. For analysis, they were matched to 90 controls. Factors associated with syphilis were: low educational attainment (OR=5.38, 95% CI 1.94 to 14.94; p=0.001), receptive oral sex with casual male partners without a condom (OR=4.86, 95% CI 1.63 to 14.48; p=0.005) and anal sex toy use with casual male partners (OR=2.72, 95% CI 1.01 to 7.32; p=0.05). Seeking of sex partners online (OR=5.17, 95% CI 1.33 to 20.11), use of poppers (OR=2.2, 95% CI 1.1 to 4.3) and erectile dysfunction drugs (OR=1.9, 95% CI 1.0 to 13.2) were associated with syphilis only in the univariate analysis. Conclusions Receptive oral sex without a condom and use of anal sex toys were identified as presenting a major risk of syphilis infection. Although these practices have been shown to present low risk of HIV transmission, the general public is unaware of their impact on transmission of other STIs.

Recently, the first Neisseria gonorrhoeae strain (H041) highly resistant to the expanded-spectrum cephalosporins (ESCs) ceftriaxone and cefixime, which are the last remaining options for first-line gonorrhea treatment, was isolated in Japan. Here, we confirm and characterize a second strain (F89) with high-level cefixime and ceftriaxone resistance which was isolated in France and most likely caused a treatment failure with cefixime. F89 was examined using six species-confirmatory tests, antibiograms (33 antimicrobials), porB sequencing, N. gonorrhoeae multiantigen sequence typing (NG-MAST), multilocus sequence typing (MLST), and sequencing of known gonococcal resistance determinants ( penA , mtrR , penB , ponA , and pilQ ). F89 was assigned to MLST sequence type 1901 (ST1901) and NG-MAST ST1407, which is a successful gonococcal clone that has spread globally. F89 has high-level resistance to cefixime (MIC = 4 μg/ml) and ceftriaxone (MIC = 1 to 2 μg/ml) and resistance to most other antimicrobials examined. A novel penA mosaic allele ( penA-CI ), which was penA-XXXIV with an additional A501P alteration in penicillin-binding protein 2, was the primary determinant for high-level ESC resistance, as determined by transformation into a set of recipient strains. N. gonorrhoeae appears to be emerging as a superbug, and in certain circumstances and settings, gonorrhea may become untreatable. Investigations of the biological fitness and enhanced understanding and monitoring of the ESC-resistant clones and their international transmission are required. Enhanced disease control activities, antimicrobial resistance control and surveillance worldwide, and public health response plans for global (and national) perspectives are also crucial. Nevertheless, new treatment strategies and/or drugs and, ideally, a vaccine are essential to develop for efficacious gonorrhea management.

Background In 2004 the French Public Health Agency set up a reactive all-causes mortality surveillance-based on the administrative part of the death certificate, in the final objectives: –to detect unexpected or usual variations in mortality; –to provide a first evaluation of mortality impact of public health events. In 2007, an Electronic Death Registration System (EDRS) was implemented, enabling electronic transmission of the medical causes of death to the agency in real time. It allows approaching the cause accountability when an excess death is observed. A strategy has been developed for the analysis in routine of the medical causes of death with the objectives of early detection of expected and unexpected events and reactive evaluation of the impact of these events. The aim of this study is to present the syndromic indicators and their definitions that will be routinely followed for mortality surveillance. Method Mortality syndromic indicators were defined as a cluster of medical causes of death (pathologies, syndromes or symptoms) with the same nosological meaning and meeting the objectives of alert and impact evaluation of mortality surveillance. The causes of death are either free-text (words, terms, expressions) or ICD-10 codes. To build the pertinent syndromic indicators, we explored multiple biomedical classifications such as the Mesh, SNOMED, UMLS. We also used a dictionary [provided by the Center of Epidemiological Causes of Death (Inserm-CepiDc)] of each term/expression found in the death certificates since 2005 and the associated ICD-10 code. Each indicator and its definition were also discussed in working groups including medical and epidemiological experts. Result The list of established indicators was composed of: –indicators for detection of expected seasonal events such as: “influenza”, “pneumonia”, “gastroenteritis”, “chikungunya”, “heat related death”, “dehydration”…; –other complementary indicators for detection of the impact of unexpected events such as: “epilepsy”, “aneurysm rupture”, “pulmonary embolism”, “choc”, “coma”, “unspecified fever”, “headache”…; –indicators for the decomposition of the mortality linked to an event; they include the previous indicators. Examples of detection of winter seasonal events and heat related events will be presented as well as the detection of death with chikungunya cause in the death certificate. Conclusion The use of free-text causes of death for mortality surveillance required the development of a strategy for the analysis of these data. The building of a list of syndromic indicators for alert and impact evaluation was essential for the implementation of systematical classification methods of the death certificates in routine.

Background In France, after the 2003 heat wave, a mortality syndromic surveillance system was set up by the National Public Health Agency, with objectives of early detection and reactive evaluation of the impact of expected and unexpected events to support decision makers. This study aims to describe the characteristics of the system and the main excess mortality period detected by the system since 2011 that led to alert decision makers for actions. Methods Anonymized data from the administrative part of death certificate were daily collected from 3042 computerized city halls and were transmitted to the agency in routine. Representativeness was measured as the proportion of death registered by the system among the exhaustive number of deaths and analyzed by age, month and region. Deaths were described by gender, age group at national and regional levels using proportion. Mortality surveillance is performed routinely based on the comparison of the weekly observed and expected number of deaths estimated with the model developed by the European consortium EuroMomo between 2012 and 2016. Based on criteria, when an excess of death is observed, decision makers are alerted and an investigation is launched. Results The system recorded 77.5% of the national mortality covering the whole territory and varying from 67% to 96.8% in metropolitan regions and overseas. Three major excess deaths episodes occurred during winters 2011–12, 2012–13 and 2014–15 and were highly correlated with flu epidemics periods. Five punctual excess periods were identified during the summer months, among them 4 were concomitant with a heat wave or elevated temperature period. No particular event was found correlated to the fifth excess. The system was also able to detect events such as an increase in mortality in Guadeloupe in 2014 concomitant with chikungunya epidemics. National and regional epidemiological mortality reports were published weekly (daily if required) and transmitted to the Ministry of Health and regional health authorities with potential recommendations. Conclusion The effectiveness of the reactive mortality surveillance system to alert decision makers has been demonstrated. It is useful to adapt along the event recommendation for control measures as well as evaluation of impact of the event on mortality. However, the absence of medical causes of death may limit the ability to target recommendations. Facing the challenge of emergence of epidemics and climate changes, this reactive mortality system will be complementary for other surveillance systems for the future health crisis. The development of mortality surveillance with causes of death will ensure more information for interpretation of all-cause mortality changes.

Background. To better document the risk factors for sporadic Campylobacter infection in France, we conducted a national case-control study from September 2002 to June 2004. Methods. Cases with confirmed Campylobacter infection were sampled through the national surveillance laboratory network. Cases and controls who were matched for age, as well as attending physicians, were interviewed about foods consumed, food preparation practices, travel history, contact with cases and animals during the 8 days before the onset of infection, and any antibiotic use occurring during the 30 days before onset. Matched odds ratios [ORs] were calculated using conditional logistic regression and multiple imputation methods. Results. A total of 285 pairs of cases and matched controls were enrolled. "Ate undercooked beef" (OR, 2.86; 95% confidence interval [CI], 1.65-4.95), "ate at restaurant" (OR, 2.20; 95% CI, 1.23-3.93), and "poor utensils hygiene in the kitchen" (OR, 2.12; 95% CI, 1.33-3.37) were the main independent risk factors for infection. Cases infected with a ciprofloxacin-resistant Campylobacter jejuni strain were more likely than controls to have used antibiotics in the month before onset. Conclusion. Good hygiene practices in the kitchen remain a strong recommendation to avoid cross-contamination. However, studies are needed to explore the mechanism of contamination throughout the food chain. The use of antibiotics in humans may favor the development of a resistant infection.

Background The food-borne viruses in Europe (FBVE) network database was established in 1999 to monitor trends in outbreaks of gastroenteritisdue to noroviruses (NoVs), to identify major transmission routes of NoV infections within and between participating countries and to detectdiffuse international food-borne outbreaks.Methods We reviewed the total of 9430 NoVoutbreak reports from 13 countries with date of onset between 1 January 2002 and 1 January2007 for representativeness, completeness and timeliness against these objectives.Results Rates of reporting ranged from a yearly average of 1.8 in 2003 to 11.6 in 2006. Completeness of reporting of an agreed minimumdataset improved over the years, both for epidemiological and virological data. For the 10 countries that provided integrated (epidemiological ANDvirological) reporting over the 5-year period, the completeness of the minimum dataset rose from 15% in 2003 to 48% in 2006. Two countrieshave not been able to combine both data types due to the structure of the national surveillance system (England and Wales and Germany).Timeliness of reporting (median days between the onset of an outbreak and the date of reporting to the FBVE database) differed greatly betweencountries, but gradually improved to 47 days in 2006.Conclusion The outbreaks reported to the FBVE reflect the lack of standardization of surveillance systems across Europe, making directcomparison of data between countries difficult. However, trends in reported outbreaks per country, distribution of NoV genotypes, and detectionof diffuse international outbreaks were used as background data in acute questions about NoV illness and the changing genotype distributionduring the 5-year period, shown to be of added value. Integrated reporting is essential for these objectives, but could be limited to sentinelcountries with surveillance systems that allow this integration. For successful intervention in case of diffuse international outbreaks, completenessand timeliness of reporting would need to be improved and expanded to countries that presently do not participate.Keywords Epidemiology, Food safety, Public health

Implemented 10 years ago, the French syndromic surveillance system Oscour, based on emergency departments, has been assessed using four major evaluation criteria in syndromic surveillance: stability and regularity of data transmission, the coverage at the national level, data quality, particularly for medical information and the utility of the system for the public health surveillance. In 2014, about 40,000 daily attendances are extracted automatically from 600 ED departments located all over the territory, covering 80% of the national attendances. About 12,800 different ICD10 codes have been used in 2013, enabling a large public health surveillance.

The frequency of Campylobacter infections in human, their potential severity, and the existence of preventive measures warrant the creation of a surveillance system. A Campylobacter surveillance system based on private laboratories (PL) was set up in France in 2002 to complement the hospital laboratory-based system(HL). Since 1 January 2001 (for HL), 1 April 2002 (for PL) and 31 December 2003, the Campylobacter National reference Centre analyzed 3 698 strains. Initially thought to be 3.4/ 100,000, the overall isolation rate was grossly underestimated, as the latest figure is 14/ 100,000 among children under 5 years old. The strains were mainly C. jejuni (76.9%), followed by C. coli (17.0%) and C. fetus (5.4%). Resistance rates were 41% for ampicillin and 28% for nalidixic acid. Surveillance must be maintained to collect more data on Campylobacter epidemiology in France and to estimate its incidence in the general population., La fréquence des infections humaines à Campylobacter, leur gravité potentielle et l’existence de mesures de prévention justifient une surveillance. En France, un système de surveillance des infections à Campylobacter a été mis en place en avril 2002, à partir des laboratoires de ville (LABM), en complément de celui du réseau des laboratoires hospitaliers (LH) existant depuis 1986. Entre le 1/01/01 (pour les LH), le 1/04/02 (pour les LABM) et le 31/12/03, le Centre National de Référence des Campylobacter a expertisé 3 698 souches. Le taux global d’isolement de 3,4/100 000 était très largement sous-estimé : il était de 14/100 000 pour les enfants âgés de moins 5 ans. C. jejuni représentait 76,9% des souches, suivi de C. coli (17,0 %) et de C. fetus (5,4 %). Le taux de résistance à l’ampicilline était de 41% et à l’acide nalidixique de 28 %. Les efforts pour la surveillance doivent être poursuivis, afin de mieux connaître les caractéristiques épidémiologiques des infection à Campylobacter en France et de faire des estimations d’incidence en population générale., Gallay Anne, Prouzet-Mauleon Valérie, Labadi Leila, Desenclos Jean-Claude, Megraud Francis, De Valk Henriette, Vaillant Véronique. Les infections à Campylobacter chez l'homme en France : bilan des trois années de surveillance 2001 - 2003. In: Bulletin de l'Académie Vétérinaire de France tome 158 n°4, 2005. pp. 369-376.

The frequency of Campylobacter infections in humans, their potential severity, and the existence of preventive measures justify the implementation of a surveillance system for these infections. Before the implementation of the surveillance system, a survey of the Campylobacter diagnostic practices in the laboratories was performed. In the laboratories that responded, most investigated for Campylobacter at least once in 1999. Identification of the Campylobacter species was carried out by 86% of hospital laboratories and 37% of private laboratories. Antibiotic sensitivity tests were carried out by 75% and 32% of them respectively. Many laboratories test for Campylobacter in stool samples using comparable methods showing the feasibility of a surveillance system.