Céline Caserio-Schönemann

Syndromic surveillance during the Paris terrorist attacks

1 Remuzzi G, Horton R. Italy, the Land of Holy Miracles—revisited. Lancet 2016; 387: 11–12. 2 Pastoral Constitution on the Church in the Modern World. Gaudium et Spes promulgate by His Holiness, Pope Paul VI on December 7, 1965. Vatican City: The Holy See, 1965. http://www.vatican.va/arc hive/hist_councils/ ii_vatican_council/documents/vat-ii_ const_19651207_gaudium-et-spes_en.html (accessed Jan 3, 2016). 3 Russel RJ, Murphy N, Stoeger WR, eds. Scientifi c perspectives on divine action. Twenty years of challenge and progress. Vatican City State–Berkeley, CA: Vatican Observatory Publications, The Center for Theology and Natural Sciences, 2009. Pilgrims with serious disease or highrisk health conditions are participating in the event at a great risk to their own health and a substantial burden for the country. For example, pilgrims with severe cardiovascular diseases, kidney failure requiring dialysis or women in the late stages of pregnancy are allowed to attend Hajj. During the 2015 Hajj, 2200 kidney dialysis procedures, 27 open heart surgeries, and 688 cardiac catheterisation operations were done, in addition to seven birth deliveries. In view of the above, we call for a pragmatic review of the Hajj Health Requirements to extend beyond communicable disease prevention, and for a serious discussion among all stakeholders, both national and international, on whether certain pilgrims should be prevented from going to the Hajj pilgrimage on medical grounds for their own safety as well as to reduce morbidity, mortality, and associated burden. From a religious point of view, the Hajj is only required for those physically and fi nancially able to perform it.

How to use near real-time health indicators to support decision-making during a heat wave: the example of the French heat wave warning system

Introduction The French warning system for heat waves is based on meteorological forecasts. Near real-time health indicators are used to support decision-making, e.g. to extend the warning period, or to choose the most appropriate preventive measures. They must be analysed rapidly to provide decision-makers useful and in-time information. The objective of the study was to evaluate such health indicators. Methods A literature review identified a range of possible mortality and morbidity indicators. A reduced number were selected, based on several criteria including sensitivity to heat, reactivity, representativity and data quality. Two methods were proposed to identify indicator-based statistical alarms: historical limits or control charts, depending on data availability. The use of the indicators was examined using the 2006 and 2009 heat waves. Results Out of 25 possible indicators, 5 were selected: total mortality, total emergency calls, total emergency visits, emergency visits for people aged 75 and over and emergency visits for causes linked to heat. In 2006 and 2009, no clear increases were observed during the heat waves. The analyses of real-time health indicators showed there was no need to modify warning proposals based on meteorological parameters. Discussion These findings suggest that forecasted temperatures can be used to anticipate heat waves and promote preventive actions. Health indicators may not be needed to issue a heat wave alert, but daily surveillance of health indicators may be useful for decision-makers to adapt prevention measures.

User-friendly Rshiny web applications for supporting syndromic surveillance analysis

Objective The presentation describes the design and the main functionalities of two user-friendly applications developed using R-shiny to support the statistical analysis of morbidity and mortality data from the French syndromic surveillance system SurSaUD. Introduction The French syndromic surveillance system SursaUD® has been set up by Sante publique France, the national public health agency (formerly French institute for public health - InVS) in 2004. In 2016, the system is based on three main data sources: the attendances in about 650 emergency departments (ED), the consultations to 62 emergency general practitioners’ (GPs) associations SOS Medecins and the mortality data from 3,000 civil status offices [1]. Daily, about 60,000 attendances in ED (88% of the national attendances), 8,000 visits in SOS Medecins associations (95% of the national visits) and 1,200 deaths (80% of the national mortality) are recorded all over the territory and transmitted to Sante publique France. About 100 syndromic groupings of interest are constructed from the reported diagnostic codes, and monitored daily or weekly, for different age groups and geographical scales, to characterize trends, detect expected or unexpected events (outbreaks) and assess potential impact of both environmental and infectious events. All-causes mortality is also monitored in similar objectives. Two user-friendly interactive web applications have been developed using the R shiny package [2] to provide a homogeneous framework for all the epidemiologists involved in the syndromic surveillance at the national and the regional levels. Methods The first application, named MASS-SurSaUD, is dedicated to the analysis of the two morbidity data sources in Sursaud, along with data provided by a network of Sentinel GPs [3]. Based on pre-aggregated data availaible daily at 10:30 am, R programs create daily, weekly and monthly time series of the proportion of each syndromic grouping among all visits/attendances with a valid code at the national and regional levels. Twelve syndromic groupings (mainly infectious and respiratory groups, like ILI, gastroenteritis, bronchiolitis, pulmonary diseases) and 13 age groups have been chosen for this application. For ILI, 3 statistical methods (periodic regression, robust periodic regression and Hidden Markov model) have been implemented to identify outbreaks. The results of the 3 methods applied to the 3 data sources are combined with a voting algorithm to compile the influenza alarm level for each region each week: non-epidemic, pre/post epidemic or epidemic. The second application, named MASS-Euromomo, allows consulting results provided by the model developed by the European project EuroMomo for the common analysis of mortality in the European countries (www.euromomo.eu). The Euromomo model, initially developed using Stata software, has been transcripted in R. The model has been adapted to run in France both at a national, regional and other geographical administrative levels, and for 7 age groups. Results The two applications, accessible on a web-portal, are similarly designed, with: - a dropdown menu and radio buttons on the left hand side to select the data to display (e.g. filter by data source, age group, geographical levels, syndromic grouping and/or time period), - several tab panels allowing to consult data and statistical results through tables, static and dynamic charts, statistical alarm matrix, geographical maps,... (Figure 1), - a “help” tab panel, including documentations and guidelines, links, contact details. The MASS-SurSaUD application has been deployed in December 2015 and used during the 2015-2016 influenza season. MASS- Euromomo application has been deployed in July 2016 for the heat- wave surveillance period. Positive feedbacks from several users have been reported. Conclusions Business Intelligence tools are generally focused on data visualisation and are not generally tailored for providing advanced statistical analysis. Web applications built with the R-shiny package combining user-friendly visualisations and advanced statistics can be rapidly built to support timely epidemiological analyses and outbreak detection. Figure 1: screen-shots of a page of the two applications

Improving regional influenza surveillance through a combination of automated outbreak detection methods: the 2015/16 season in France

The 2014/15 influenza epidemic caused a work overload for healthcare facilities in France. The French national public health agency announced the start of the epidemic – based on indicators aggregated at the national level – too late for many hospitals to prepare. It was therefore decided to improve the influenza alert procedure through (i) the introduction of a pre-epidemic alert level to better anticipate future outbreaks, (ii) the regionalisation of surveillance so that healthcare structures can be informed of the arrival of epidemics in their region, (iii) the standardised use of data sources and statistical methods across regions. A web application was developed to deliver statistical results of three outbreak detection methods applied to three surveillance data sources: emergency departments, emergency general practitioners and sentinel general practitioners. This application was used throughout the 2015/16 influenza season by the epidemiologists of the headquarters and regional units of the French national public health agency. It allowed them to signal the first influenza epidemic alert in week 2016-W03, in Brittany, with 11 other regions in pre-epidemic alert. This application received positive feedback from users and was pivotal for coordinating surveillance across the agency’s regional units.

Early assessment of public health impacts of major floods in France in June 2016 and winter 2018

Introduction The Seine River rises at the northeast of France and flows through Paris before emptying into the English Channel. On June 2016 (week 22) and January 2018 (weeks 4 to 6), major floods occurred in the Basin of Seine River, after important rainy periods. The second period was also marked by the occurrence on the same area of a cold wave including heavy snowfall and ice conditions on week 6. Floods are known to have potential health impacts on population living in those regions, both at short-, medium- and long-term both on physical and mental health. The objective of the study is to present the results of the daily monitoring of health indicators conducted by the French public health agency (SpFrance) during the two major floods, in order to early identify potential impact of those disasters on the population. Methods Since 2004, SpFrance set up a national syndromic surveillance system SurSaUD, enabling to ensure morbidity and mortality surveillance. Morbidity data are daily collected from a network involving emergency departments (ED) and emergency general practitioners’ associations SOS Medecins. Both administrative (age, gender, date and location of consultation) and medical information (medical diagnosis using ICD10 codes in ED and specific thesauri in SOS Medecins associations) are recorded for each patient. The daily and weekly evolution of the number of all-cause ED attendances and SOS Medecins consultations during the flood periods were compared to the evolution on the two previous years. The number of hospitalisations after ED discharge was also monitored. The immediate health impact of floods was assessed by monitoring eight syndromic indicators related to flood exposure: gastroenteritis, carbon monoxide poisoning, burnt, stress, faintness, drowning, injuries and hypothermia. Analyses were performed by age group and at different geographical levels (national, Paris region and districts located in the Basin of Seine River). Results During week 22-2016, a decrease of total number of ED attendances all ages and more specifically for children aged less than 15 years old was observed in region Centre (−15% compared with weeks 21 and 23) and to a lesser extent in Paris area (−4% and −12% respectively compared with week 21 and week 23). Locally, in the most impacted cities, an increase of total ED attendances was also observed for people aged 65 years and more. At the opposite, the number of all-cause SOS Medecins consultations remained stable in all age groups. Analysis by syndromic indicator did not show unusual variations during the first flood period. In 2018, syndromic surveillance did not show any major impact on all-cause ED attendances and SOS Medecins consultations from week 4 to week 6, neither in Paris area nor in other areas along the Seine River. The numbers were comparable to the two precedent years in all age groups. During week 6 in Paris area, an increase of ED attendances was observed for injuries (+4% compared to the past weeks) and to a lesser extent for hypothermia (16 attendances compared to less than 9 for the past weeks). Conclusion For both flood episodes the rising water level was slow with foreseeable evolution, compared to other sudden flood events occurring in south of France in 2010 due to violent thunderstorms. This progressive evolution allows French authority to deploy wide specific organization in order to mitigate impact on concerned populations. That may explain the limited impact observed during the two flood disasters. The evolution of injuries during 2018 episode is attributable to the cold wave that occurred simultaneously. As the French syndromic surveillance system is implemented on the whole territory and collects emergency data routinely since several years, it constitutes a reactive tool to assess the potential public health impact of both sudden and predictable disasters. It can either contribute to adapt management action or reassure decision makers if no major impact is observed.

The influence of a major sporting event upon emergency department attendances; A retrospective cross-national European study

Major sporting events may influence attendance levels at hospital emergency departments (ED). Previous research has focussed on the impact of single games, or wins/losses for specific teams/countries, limiting wider generalisations. Here we explore the impact of the Euro 2016 football championships on ED attendances across four participating nations (England, France, Northern Ireland, Wales), using a single methodology. Match days were found to have no significant impact upon daily ED attendances levels. Focussing upon hourly attendances, ED attendances across all countries in the four hour pre-match period were statistically significantly lower than would be expected (OR 0.97, 95% CI 0.94–0.99) and further reduced during matches (OR 0.94, 95% CI 0.91–0.97). In the 4 hour post-match period there was no significant increase in attendances (OR 1.01, 95% CI 0.99–1.04). However, these impacts were highly variable between individual matches: for example in the 4 hour period following the final, involving France, the number of ED attendances in France increased significantly (OR 1.27, 95% CI 1.13–1.42). Overall our results indicate relatively small impacts of major sporting events upon ED attendances. The heterogeneity observed makes it difficult for health providers to predict how major sporting events may affect ED attendances but supports the future development of compatible systems in different countries to support cross-border public health surveillance.

Are the French SAMU data relevant for health surveillance

Objective To evaluate whether SAMU data could be relevant for health surveillance and proposed to be integrated into the French national syndromic surveillance SurSaUD® system. Introduction The syndromic surveillance SurSaUD® system developed by Sante publique France, the French National Public Health Agency collects daily data from 4 data sources: emergency departments (OSCOUR® ED network) [1], emergency general practioners (SOS Medecins network), crude mortality (civil status data) and electronic death certification including causes of death [2]. The system aims to timely identify, follow and assess the health impact of unusual or seasonal events on emergency medical activity and mortality. However some information could be missed by the system especially for non-severe (absence of ED consultation) or, in contrast, highly severe purposes (direct access to intensive care units). The French pre-hospital emergency medical service (SAMU) [3] represents a potential valuable data source to complete the SurSaUD® surveillance system, thanks to reactive pre-hospital data collection and a large geographical coverage on the whole territory. Data are still not completely standardized and computerized but a governmental project to develop a national common IT system involving all French SAMU is in progress and will be experimented in the following years. Methods A pilot study was performed in the South of France PACA region, where data from the six local SAMU structures are centralized into an interconnected database. A minimal set of variables required for health monitoring (administrative and medical items) and modalities for data extraction and transmission to Sante publique France were defined. SAMU data were transmitted daily to Sante Publique France and the PACA regional team developed a Microsoft Access® application to import decrypted data, request database and analyze indicators. Retrospective part of the study was performed over a 2-year period (2013-2014) and the prospective part during 2015 was based on daily data collection. Completeness and quality of variables were analyzed. SAMU indicators including several level of specificity were built and compared to existing SurSaUD® indicators in different situations (for detection, seasonal follow-up and health impact assessment) using Spearman coefficient correlation. Results During the pilot study, data from five of the six SAMU structures of PACA region were structured enough to be analyzed. On the study period, almost 2,400,000 files were recorded and 89% contain medical information. Data completeness was high (87%) and stable during the whole period. The annual rate of SAMU solicitation was 16 for 100 inhabitants at the regional scale. 15% of the records were opened only for medical advice. In contrast, patients were evacuated directly in intensive care unit in 9.5% of cases without ED admission. Coding quality depended on the existence and the use of official thesauri and varied widely among SAMU structures. Despite coding variations, SAMU indicators for winter epidemics were significantly correlated with ED and SOS Medecins indicators. Respectively with ED flu, bronchiolitis and gastroenteritis indicators, the strongest correlations were found for SAMU lower respiratory infection (0.74), SAMU bronchiolitis (0.72) and SAMU gastroenteritis / diarrhea / vomiting (0.81). Conclusions This pilot study demonstrated the feasibility to collect daily SAMU activity data. The key strengths of SAMU data were a large geographic coverage, the subsidiarity with SurSaUD® system data sources, the follow-up of prehospital activity and for patients directly admitted into an intensive care unit. Some limitations were highlighted related to differences in coding practices especially for medical diagnosis. The generalization of this study will require the standardization of coding practices and homogenization of thesaurus. The implementation of the national SAMU information system should allow in a very next future to widely progressing on these topics. References [1] Fouillet A, Bousquet V, Pontais I, Gallay A and Caserio-Schonemann C. The French Emergency Department OSCOUR Network:Evaluation After a 10-year Existence. Online Journal of Public Health Informatics ISSN 1947-2579-7(1):e74, 2015 [2] Caserio-Schonemann C, Bousquet V, Fouillet A, Henry V. Le systeme de surveillance syndromique SurSaUD (R). Bull Epidemiol Hebd 2014;3-4:38-44. [3] Baker, D.J.. The French prehospital emergency medicine system (SAMU): An introduction (2005) CPD Anaesthesia, 7 (1), pp. 20-25.

Impact assessment of mass gatherings using labelling procedure in ED, Nouvelle-Aquitaine, 2016

Objective To access the potential health impact on the population during mass gathering over time using labelling procedure in emergency department (ED). Introduction The massive flow of people to mass gathering events, such as festivals or sports events like EURO 2016, may increase public health risks. In the particular context of several terrorist attacks that took place in France in 2015, the French national Public Health agency has decided to strengthen the population health surveillance systems using the mandatory notification disease system and the French national syndromic surveillance SurSaUD®. The objectives in terms of health surveillance of mass gathering are: 1/ the timely detection of a health event (infectious cluster, environmental exposure, collective foodborne disease…) 2/ the health impact assessment of an unexpected event such as a terrorist attack. In collaboration with the Regional Emergency Observatory (ORU), a procedure for the labeling of emergencies has been tested to identify the ED records that could be considered as linked to the event. Methods During summer 2016, the procedure was tested on seven major festive events throughout the region. In addition to the main medical diagnosis, a specific ICD-10 code “Y3388” was chosen to be used in associated diagnosis for records that were supposed to linked to the event. Information on the labeling procedure was insured by the ORU to the emergency departments. All records with medical diagnoses or medical pattern beginning by Y33 have been analyzed. Results No significant increase in the global indicator was observed in the ED impacted by mass gathering. The ED labelling procedure identified 260 records: two thirds corresponded to young men and 17% came from abroad. Among the 250 records labeled in associated diagnosis, 39% were associated to traumatisms and 31% corresponded to alcohol intake. Conclusions This study shows that a labelling procedure allows the identification, quantification and characterization of the population ED records associated with mass gathering. Additionally, a labelling procedure to assess a potential impact of an event as mass gathering can be implemented fairly rapidly.

Mapping influenza activity in emergency departments in France using Bayesian model-based geostatistics

Maps of influenza activity are important tools to monitor influenza epidemics and inform policymakers. In France, the availability of a high‐quality data set from the Oscour® surveillance network, covering 92% of hospital emergency department (ED) visits, offers new opportunities for disease mapping. Traditional geostatistical mapping methods such as Kriging ignore underlying population sizes, are not suited to non‐Gaussian data and do not account for uncertainty in parameter estimates.

Retrospective observational study of emergency department syndromic surveillance data during air pollution episodes across London and Paris in 2014

Introduction Poor air quality (AQ) is a global public health issue and AQ events can span across countries. Using emergency department (ED) syndromic surveillance from England and France, we describe changes in human health indicators during periods of particularly poor AQ in London and Paris during 2014. Methods Using daily AQ data for 2014, we identified three periods of poor AQ affecting both London and Paris. Anonymised near real-time ED attendance syndromic surveillance data from EDs across England and France were used to monitor the health impact of poor AQ. Using the routine English syndromic surveillance detection methods, increases in selected ED syndromic indicators (asthma, difficulty breathing and myocardial ischaemia), in total and by age, were identified and compared with periods of poor AQ in each city. Retrospective Wilcoxon-Mann-Whitney tests were used to identify significant increases in ED attendance data on days with (and up to 3 days following) poor AQ. Results Almost 1.5 million ED attendances were recorded during the study period (27 February 2014 to 1 October 2014). Significant increases in ED attendances for asthma were identified around periods of poor AQ in both cities, especially in children (aged 0–14 years). Some variation was seen in Paris with a rapid increase during the first AQ period in asthma attendances among children (aged 0–14 years), whereas during the second period the increase was greater in adults. Discussion This work demonstrates the public health value of syndromic surveillance during air pollution incidents. There is potential for further cross-border harmonisation to provide Europe-wide early alerting to health impacts and improve future public health messaging to healthcare services to provide warning of increases in demand.