A multidisciplinary and structured investigation of three suspected clusters of transverse upper limb reduction defects in France.

INTRODUCTION: Between 2019-2021, facing public concern, a scientific expert committee (SEC) reanalysed suspected clusters of transverse upper limb reduction defects (TULRD) in three administrative areas in France, where initial investigations had not identified any risk exposure. We share here the national approach we developed for managing suspicious clusters of the same group of congenital anomalies occurring in several areas. METHODS: The SEC analysed the medical records of TURLD suspected cases and performed spatiotemporal analyses on confirmed cases. If the cluster was statistically significant and included at least three cases, the SEC reviewed exposures obtained from questionnaires, environmental databases, and a survey among farmers living near to cases' homes concerning their plant product use. RESULTS: After case re-ascertainment, no statistically significant cluster was observed in the first administrative areas. In the second area, a cluster of four children born in two nearby towns over two years was confirmed, but as with the initial investigations, no exposure to a known risk factor explaining the number of cases in excess was identified. In the third area, a cluster including just two cases born the same year in the same town was confirmed. DISCUSSION: Our experience highlights that in the event of suspicious clusters occurring in different areas of a country, a coordinated and standardised approach should be preferred.

Prevalence of long COVID in the general adult population according to different definitions and sociodemographic and infection characteristics. A nationwide random sampling survey in France in autumn 2022.

OBJECTIVES: Long COVID has been recognized since early 2020, but its definition is not unanimous, which complicates epidemiological assessments. This study estimated the prevalence of long COVID based on several definitions and severity thresholds in the adult population of mainland France and examined variations according to sociodemographic and infection characteristics. METHODS: A cross-sectional survey using random sampling was conducted in August-November 2022. Participants declaring SARS-CoV-2 infection were assessed for infection dates and context, post-COVID symptoms (from a list of 31, with onset time, daily functioning impact, and alternative diagnosis), and perceived long COVID. Long COVID prevalence was estimated according to the WHO, National Institute for Health and Care Excellence, United States National Centre for Health Statistics, and United Kingdom Office for National Statistics definitions. RESULTS: Of 10 615 participants, 5781 (54.5%) reported SARS-CoV-2 infection, with 123-759 (1.2-13.4%) having long COVID, depending on the definition. The prevalence of WHO post-COVID condition (PCC) was 4.0% (95% CI: 3.6-4.5) in the overall population and 8.0% (95% CI: 7.0-8.9) among infected individuals. Among the latter, the prevalence varied from 5.3% (men) to 14.9% (unemployed) and 18.6% (history of hospitalization for COVID-19). WHO-PCC overlapped poorly with other definitions (kappa ranging from 0.18 to 0.59) and perceived long COVID (reported in only 43% of WHO-PCC). DISCUSSION: Regardless of its definition, long COVID remains a significant burden in the French adult population that deserves surveillance, notably for forms that strongly impact daily activities. More standardized definitions will improve integrated surveillance of, and better research on, long COVID.

Complex association between post-COVID-19 condition and anxiety and depression symptoms.

BACKGROUND: To assess the associations between anxiety and depressive symptoms and post-COVID-19 condition (PCC) by exploring the direction of these associations and their relevance in the definition of PCC. METHODS: Nationwide survey among French adults, recruited between March and April, 2022, using a quota method to capture a representative sample of the general population with regard to sex, age, socioeconomic status, size of the place of residence, and region. We included all participants who met the World Health Organization (WHO) definition of PCC in addition to a random sample of participants infected with SARS-COV-2 for at least 3 months but without PCC. Self-reported anxiety and depressive symptoms, chronic anxiety and depression (for more than 3 years), and anxiety and depression were measured using the GAD-2 and PHQ-2 questionnaires, respectively. RESULTS: In a sample of 1,095 participants with PCC and 1,021 participants infected with SARS-COV-2 without PCC, 21% had self-reported anxiety and 18% self-reported depression, whereas 33% and 20% had current measured symptoms of anxiety and depression, respectively. The high prevalence of these symptoms cannot only be explained by the characterization of PCC, as only 13.4% of anxiety symptoms and 7.6% of depressive symptoms met the WHO criteria for PCC. Only one participant met the WHO criteria based on self-reported anxiety or depressive symptoms alone, as these were always combined with other symptoms in patients with PCC. Chronic symptoms were associated with PCC (aOR 1.27; 95% CI: 1.00-1.61). In addition, measured anxiety was associated with PCC (aOR = 1.29; 95% CI: 1.02-1.62). CONCLUSIONS: Pre-COVID-19 chronic anxiety and depression may play a role in the development of PCC or share vulnerability factors with it. Our results challenge the inclusion of anxiety and depression in the definition of PCC.

Direct impact of COVID-19 by estimating disability-adjusted life years at national level in France in 2020.

BACKGROUND: The World Health Organization declared a pandemic of coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), on March 11, 2020. The standardized approach of disability-adjusted life years (DALYs) allows for quantifying the combined impact of morbidity and mortality of diseases and injuries. The main objective of this study was to estimate the direct impact of COVID-19 in France in 2020, using DALYs to combine the population health impact of infection fatalities, acute symptomatic infections and their post-acute consequences, in 28 days (baseline) up to 140 days, following the initial infection. METHODS: National mortality, COVID-19 screening, and hospital admission data were used to calculate DALYs based on the European Burden of Disease Network consensus disease model. Scenario analyses were performed by varying the number of symptomatic cases and duration of symptoms up to a maximum of 140 days, defining COVID-19 deaths using the underlying, and associated, cause of death. RESULTS: In 2020, the estimated DALYs due to COVID-19 in France were 990 710 (1472 per 100 000), with 99% of burden due to mortality (982 531 years of life lost, YLL) and 1% due to morbidity (8179 years lived with disability, YLD), following the initial infection. The contribution of YLD reached 375%, assuming the duration of 140 days of post-acute consequences of COVID-19. Post-acute consequences contributed to 49% of the total morbidity burden. The contribution of YLD due to acute symptomatic infections among people younger than 70 years was higher (67%) than among people aged 70 years and above (33%). YLL among people aged 70 years and above, contributed to 74% of the total YLL. CONCLUSIONS: COVID-19 had a substantial impact on population health in France in 2020. The majority of population health loss was due to mortality. Men had higher population health loss due to COVID-19 than women. Post-acute consequences of COVID-19 had a large contribution to the YLD component of the disease burden, even when we assume the shortest duration of 28 days, long COVID burden is large. Further research is recommended to assess the impact of health inequalities associated with these estimates.

Validity of algorithms for identifying mild traumatic brain injury in the French national emergency department database OSCOUR: a retrospective multicentre validation study protocol.

INTRODUCTION: The 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 ANALYSIS: We 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 DISSEMINATION: The 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.

The burden of injury in Central, Eastern, and Western European sub-region: a systematic analysis from the Global Burden of Disease 2019 Study.

BACKGROUND: Injury remains a major concern to public health in the European region. Previous iterations of the Global Burden of Disease (GBD) study showed wide variation in injury death and disability adjusted life year (DALY) rates across Europe, indicating injury inequality gaps between sub-regions and countries. The objectives of this study were to: 1) compare GBD 2019 estimates on injury mortality and DALYs across European sub-regions and countries by cause-of-injury category and sex; 2) examine changes in injury DALY rates over a 20 year-period by cause-of-injury category, sub-region and country; and 3) assess inequalities in injury mortality and DALY rates across the countries. METHODS: We performed a secondary database descriptive study using the GBD 2019 results on injuries in 44 European countries from 2000 to 2019. Inequality in DALY rates between these countries was assessed by calculating the DALY rate ratio between the highest-ranking country and lowest-ranking country in each year. RESULTS: In 2019, in Eastern Europe 80 [95% uncertainty interval (UI): 71 to 89] people per 100,000 died from injuries; twice as high compared to Central Europe (38 injury deaths per 100,000; 95% UI 34 to 42) and three times as high compared to Western Europe (27 injury deaths per 100,000; 95%UI 25 to 28). The injury DALY rates showed less pronounced differences between Eastern (5129 DALYs per 100,000; 95% UI: 4547 to 5864), Central (2940 DALYs per 100,000; 95% UI: 2452 to 3546) and Western Europe (1782 DALYs per 100,000; 95% UI: 1523 to 2115). Injury DALY rate was lowest in Italy (1489 DALYs per 100,000) and highest in Ukraine (5553 DALYs per 100,000). The difference in injury DALY rates by country was larger for males compared to females. The DALY rate ratio was highest in 2005, with DALY rate in the lowest-ranking country (Russian Federation) 6.0 times higher compared to the highest-ranking country (Malta). After 2005, the DALY rate ratio between the lowest- and the highest-ranking country gradually decreased to 3.7 in 2019. CONCLUSIONS: Injury mortality and DALY rates were highest in Eastern Europe and lowest in Western Europe, although differences in injury DALY rates declined rapidly, particularly in the past decade. The injury DALY rate ratio of highest- and lowest-ranking country declined from 2005 onwards, indicating declining inequalities in injuries between European countries.

Ability of municipality-level deprivation indices to capture social inequalities in perinatal health in France: A nationwide study using preterm birth and small for gestational age to illustrate their relevance.

BACKGROUND: Evidence-based policy-making to reduce perinatal health inequalities requires an accurate measure of social disparities. We aimed to evaluate the relevance of two municipality-level deprivation indices (DIs), the French-Deprivation-Index (FDep) and the French-European-Deprivation-Index (FEDI) in perinatal health through two key perinatal outcomes: preterm birth (PTB) and small-for-gestational-age (SGA). METHODS: We used two data sources: The French National Perinatal Surveys (NPS) and the French national health data system (SNDS). Using the former, we compared the gradients of the associations between individual socioeconomic characteristics (educational level and income) and "PTB and SGA" and associations between municipality-level DIs (Q1:least deprived; Q5:most deprived) and "PTB and SGA". Using the SNDS, we then studied the association between each component of the two DIs (census data, 2015) and "PTB and SGA". Adjusted odds ratios (aOR) were estimated using multilevel logistic regression with random intercept at the municipality level. RESULTS: In the NPS (N = 26,238), PTB and SGA were associated with two individual socioeconomic characteristics: maternal educational level (≤ lower secondary school vs. ≥ Bachelor's degree or equivalent, PTB: aOR = 1.43 [1.22-1.68], SGA: (1.31 [1.61-1.49]) and household income (< 1000 € vs. ≥ 3000 €, PTB: 1.55 [1.25-1.92], SGA: 1.69 [1.45-1.98]). For both FDep and FEDI, PTB and SGA were more frequent in deprived municipalities (Q5: 7.8% vs. Q1: 6.3% and 9.0% vs. 5.9% for PTB, respectively, and 12.0% vs. 10.3% and 11.9% vs. 10.2% for SGA, respectively). However, after adjustment, neither FDep nor FEDI showed a significant gradient with PTB or SGA. In the SNDS (N = 726,497), no FDep component, and only three FEDI components were significantly associated (specifically, the % of the population with ≤ lower secondary level of education with both outcomes (PTB: 1.5 [1.15-1.96]); SGA: 1.25 [1.03-1.51]), the % of overcrowded (i.e., > 1 person per room) houses (1.63 [1.15-2.32]) with PTB only, and unskilled farm workers with SGA only (1.52 [1.29-1.79]). CONCLUSION: Some components of FDep and FEDI were less relevant than others for capturing ecological inequalities in PTB and SGA. Results varied for each DI and perinatal outcome studied. These findings highlight the importance of testing DI relevance prior to examining perinatal health inequalities, and suggest the need to develop DIs that are suitable for pregnant women. .

The InfAct proposal for a sustainable European health information infrastructure on population health: the Distributed Infrastructure on Population Health (DIPoH).

BACKGROUND: In Europe, data on population health is fragmented, difficult to access, project-based and prone to health information inequalities in terms of availability, accessibility and especially in quality between and within countries. This situation is further exacerbated and exposed by the recent COVID-19 pandemic. The Joint Action on Health Information (InfAct) that builds on previous works of the BRIDGE Health project, carried out collaborative action to set up a sustainable infrastructure for health information in the European Union (EU). The aim of this paper is to present InfAct's proposal for a sustainable research infrastructure, the Distributed Infrastructure on Population Health (DIPoH), which includes the setup of a Health Information Portal on population health to be maintained beyond InfAct's time span. METHODS: The strategy for the proposal was based on three components: scientific initiatives and proposals to improve Health Information Systems (HIS), exploration of technical acceptability and feasibility, and finally obtaining high-level political support.. The technical exploration (Technical Dialogues-TD) was assumed by technical experts proposed by the countries, and political guidance was provided by the Assembly of Members (AoM), which gathered representatives from Ministries of Health and Science of EU/EEA countries. The results from the AoM and the TD were integrated in the sustainability plan compiling all the major outputs of InfAct. RESULTS: The InfAct sustainability plan was organized in three main sections: a proposal of a new research infrastructure on population health (the DIPoH), new health information tools and innovative proposals for HIS, and a comprehensive capacity building programme. These activities were carried out in InfAct and are being further developed in the Population Health Information Research Infrastructure (PHIRI). PHIRI is a practical rollout of DIPoH facilitating and generating the best available evidence for research on health and wellbeing of populations as impacted by COVID-19. CONCLUSIONS: The sustainability plan received wide support from Member States and was recognized to have an added value at EU level. Nevertheless, there were several aspects which still need to be considered for the near future such as: (i) a commitment of stable financial and political support by Member States (MSs), (ii) the availability of resources at regional, national and European level to deal with innovations, and (iii) a more direct involvement from EU and international institutions such as the European Centre for Disease Prevention and Control (ECDC), the World Health Organization (WHO) and the Organisation for Economic Cooperation and Development OECD for providing support and sustainable contributions.

Methodological guidelines to estimate population-based health indicators using linked data and/or machine learning techniques.

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.

Integrating technical and political views for a sustainable European Distributed Infrastructure on Population Health.

BACKGROUND: Non-Communicable diseases (NCD) are the main contributors to mortality and burden of disease. There is no infrastructure in Europe that could provide health information (HI) on Public Health monitoring and Health Systems Performance (HSP) for research and evidence-informed decision-making. Moreover, there was no EU and European Economic Area Member States (EU/EEA MSs) general consensus, on developing this initiative and guarantee its sustainability. The aim of this study is to analyze the integration of technical and political views made by the Joint Action on Health Information (InfAct; Information for Action) and the results obtained from those activities, in terms of advice and national and institutional support to develop an integrated and sustainable European Distributed Infrastructure on Population Health (DIPoH) for research and evidence-informed policy-making. METHODS: InfAct established two main boards, the Technical Dialogues (TDs) and the Assembly of Members (AoM), to provide a platform for discussion with EU/EEA MSs to establish a sustainable infrastructure for HI: 1) The TDs were composed by national technical experts (NTE) with the aim to discuss and provide feedback about scientific aspects, feasibility and EU-added value of the infrastructure proposed by InfAct. 2) The AoM gathered country representatives from Ministries of Health and Research at the highest political level, with the aim of providing policy-oriented advice for the future political acceptance, support, implementation, and development of InfAct's outcomes including DIPoH. The documentation provided for the meetings consisted in Fact-Sheets, where the main results, new methods and proposals were clearly exposed for discussion and assessment; altogether with more extended information of the DIPoH. The documentation was provided to national representatives within one more before each TD and AoM meeting. The Agenda and methodological approaches for each TD and AoM meeting consisted in the presentations of the InfAct outcomes extending the information provided in the Fact-Sheets; followed by a non-structured interaction, exchange of information, discussion and suggestions by the MSs representatives. The outcomes of the non-structured discussions were collected in Minutes of the TD and AoM meetings, and the final version was obtained with the consensus of all participants. Additionally, structured letters of political support were provided to the AoM representatives, for them to consider providing their MS written support for DIPoH. RESULTS: NTE, within the TDs, considered that DIPoH was useful for technical mutual learning and cooperation among and within countries; although they considered that the technical feasibility to uptake InfAct deliverables at the national and EU level was complex. The AoM focused on political support, resources, and expected MSs returns. The AoM representatives agreed in the interest of setting up an integrated and sustainable HI infrastructure and they considered DIPoH to be well-articulated and defined; although, some of them, expressed some barriers for providing DIPoH political support. The AoM representatives stated that the AoM is the most suitable way to inform EU MSs/ACs about future advances of DIPoH. Both boards provided valuable feedback to develop this infrastructure. Eleven countries and sixteen institutions supported the proposal, either by letters of political support or by signing the Memorandum of Understandings (MoU) and three countries, additionally, provided expression of financial commitment, for DIPoH to be added to the ESFRI 2021 roadmap. CONCLUSIONS: TDs and AoM were key forums to develop, advise, advocate and provide support for a sustainable European research infrastructure for Population Health.

Use of artificial intelligence for public health surveillance: a case study to develop a machine Learning-algorithm to estimate the incidence of diabetes mellitus in France.

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.

Recommendations to plan a national burden of disease study.

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.

Innovative use of data sources: a cross-sectional study of data linkage and artificial intelligence practices across European countries.

BACKGROUND: The 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. METHOD: We 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). RESULTS: The 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. CONCLUSIONS: Our 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.

Automatic classification of free-text medical causes from death certificates for reactive mortality surveillance in France.

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.

A New Approach to Compare the Performance of Two Classification Methods of Causes of Death for Timely Surveillance in France.

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.

Evaluation of the French reactive mortality surveillance system supporting decision making.

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.

Waterborne Disease Outbreak Detection: A Simulation-Based Study.

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⁻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.

Réunion Island prepared for possible Zika virus emergence, 2016.

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.