The “Santé Post Incendie 76” scientific research program aims to assess the overall short-, medium-, and long-term health impacts on individuals exposed to this large-scale industrial accident. It consists of four components:

1. A population-based health survey, titled “A Study Listening to Your Health”

2. Regular monitoring, over several years, of population health indicators using medical and administrative data from the National Health Data System (SNDS)

3. An assessment of the health status of workers exposed to the fire and its emissions

4. An analysis of the relevance and feasibility of detecting and measuring, in the bodies of exposed individuals, chemical substances dispersed by the fire, in light of the results of the environmental pollution assessment related to the fire

Its purpose is to enable public authorities to take health measures, if necessary, and to optimize the management of future industrial accidents.

• Learn more about the SPI76 program

Conducting a thorough assessment takes time.
The COVID-19 pandemic has complicated and slowed down the implementation of the analysis, as the teams at Santé publique France have been particularly focused on managing this health crisis, as have several members of the GAST, who are clinicians.

We received some data late, in part due to the COVID-19 crisis. Medical and
biological monitoring was implemented by the occupational health services of the relevant companies and public institutions. This included follow-ups at 1 month and 6 months. These follow-ups were delayed due to COVID-19.

Information obtained by asking people about their perception of their health is reliable when it is collected and analyzed using rigorous methods, as is the case in our study, which uses, in particular, a standardized and validated questionnaire: the SF-12.

The international scientific community recognizes the validity of measuring perceived health through questionnaires, as its correlation with mortality and morbidity indicators derived from hospital records and data recorded in medical-administrative databases has been demonstrated. Measuring health as “reported” or “perceived” by the individuals concerned therefore provides relevant and scientifically valid information on a population’s health status.

Furthermore, the data collected by asking people about the symptoms they experienced during the fire are of the same nature as the information obtained by a healthcare professional when interviewing patients during a consultation.

Yes, because the health and quality-of-life survey cannot answer all the questions regarding the potential health impact of the fire at the Lubrizol and NL Logistique warehouses. That is why Santé publique France is launching several complementary studies as part of the Santé Post Incendie 76 initiative.

This scientific study program includes medium- and long-term epidemiological surveillance—lasting about ten years or more—of the “cohort” comprising the 340,000 people living in the 122 municipalities of Seine-Maritime affected by the fire. To do this, we will utilize medical and administrative data from the National Health Data System (SNDS), which includes all data on healthcare reimbursements by the Health Insurance system (medications and diagnostic tests, consultations with doctors and other healthcare professionals), hospital data from the Medical Information System Program (PMSI), and data from the death registry (CepiDC).

All this data is continuously recorded and is already stored in the SNDS database. Its analysis, even if conducted after the date of the industrial accident, will not result in any loss of information regarding the population’s health status before and after the fire.

It will enable the analysis and monitoring of trends over time in health indicators related to pregnancy and perinatal care, the incidence of certain cancers and chronic diseases, hospitalizations, causes of death, etc.

The population of Seine-Maritime potentially exposed to the accident numbers approximately 340,000 people: this is too large a group to survey in its entirety as part of the “A Study Listening to Your Health” survey.

That is why we are collecting data from a representative sample of this population; we are proceeding in the same way for the control population. In epidemiological terms, this is a population-based survey.

Population samples are selected at random from an INSEE database (Fidéli) to ensure they are statistically representative of the populations from which they are drawn. Thus, the results observed in the study can be generalized to the entire population of the study area and the control area, which is one of the major advantages of this epidemiological survey.

The monitoring of health indicators over several years, which will be conducted using medical and administrative data from the National Health Data System (SNDS), will cover all 340,000 residents of the Seine-Maritime study area.

Yes, and this was in line with strong public expectations. Some questions specifically addressed pregnancy and breastfeeding.

However, as might be expected, the number of people affected turned out to be small because the survey was conducted among the general population. Thus, only 39 women in the sample from the exposed area reported that they were breastfeeding when the accident occurred, which did not allow for reliable estimates regarding them. A survey conducted on a sample of the general population is not methodologically appropriate for describing a specific and small group within the population.

An epidemiological follow-up of pregnancy outcomes and other perinatal indicators, using data from the National Health Data System (SNDS), will be organized to identify any potential impact of the industrial fire on this group of people. This work is part of the Santé Post Incendie 76 initiative, which focuses on the medium- and long-term monitoring of the health status of the exposed population.

Yes, we included in the study the Travellers who were residing at the Petit-Quevilly site at the time of the fire, as well as all residents living within 700 meters of the site of the fire. The administration of the questionnaire—with one adult and one child selected at random per household—was made possible thanks to the collaboration of the Relais-Accueil des Gens du voyage association.

We cannot release the results for the Travellers. The study data are anonymized before being aggregated and statistically processed by Santé publique France. Furthermore, rules regarding the protection of individuals’ anonymity do not permit the publication of results involving a small number of people, as is the case here.

This study was developed through a collaborative process, grounded in a scientific approach, and encompassing several key elements: consultation, coordination, and co-creation. The objective is twofold:

• to adhere to sound epidemiological practices,

• address the concerns and expectations of the affected population, in a spirit of openness to civil society.

A “Health Group” was formed by Santé publique France at the local level in January 2020, comprising members of the Transparency and Dialogue Committee established in Rouen by the Prefect of Seine-Maritime. It includes representatives of private and hospital-based healthcare professionals, elected officials and members of parliament, a representative of the Metropolitan Area, and associations (Rouen Respire, France Asso Santé Normandie, UFC Que Choisir).

Since its creation, the Health Group has been involved in every stage of the study’s development: drafting the questionnaire, defining the study area, selecting the control area, preparing communication materials for the public and key contacts, and interpreting the results. Several sections of the survey questionnaire were co-developed with members of the Health Group to ensure it accurately reflected what occurred during the fire and in the days that followed, and to address local expectations and concerns.

In practice, implementing this participatory approach involved a monthly meeting of the Health Group with the Santé publique France team (including during the lockdown period), numerous email exchanges, the review of many documents, and so on.

Santé publique France’s commitment to involving the public and incorporating their main areas of interest into the questionnaire was also reflected in the conduct of a qualitative pre-survey in February 2020 among mayors, healthcare professionals, and residents of four municipalities affected by the fire.

Two key guiding principles meant that the development of the study required several months of work by the teams at Santé publique France:

• the pursuit of the highest scientific, statistical, and epidemiological standards to ensure the quality and validity of our results,

• the implementation of a participatory approach, involving representatives of the affected population, to address local concerns and expectations.

Furthermore, the context of the COVID-19 pandemic complicated and slowed the implementation of the study, as the teams at Santé publique France were particularly focused on managing this health crisis, as were several of the partners involved in conducting “A Study Listening to Your Health.”

This preparatory work helped ensure that the survey questionnaire addressed the public’s main concerns and expectations.

The preliminary survey was conducted in February 2020 in four municipalities affected by the fire, representing a range of situations in terms of geographic location, exposure to the accident, and socioeconomic status: Petit-Quevilly, Bois-Guillaume, Préaux, and Buchy.

Individual interviews were conducted with mayors and healthcare professionals (general practitioners, community pharmacists, midwives, private nurses, etc.), as well as group interviews with residents of the four municipalities. The objective was to gather information on residents’ experiences of the accident and its aftermath, the symptoms they experienced, their concerns regarding the impact on health, and their expectations regarding this type of event.

The individuals selected at random by INSEE received a letter and documents explaining the study’s objectives and methodology, as well as how to participate; each was provided with a unique code to access and complete the online questionnaire. A dedicated website and toll-free number were set up for their use.

Data were collected via a questionnaire from the randomly selected individuals who agreed to participate in the study. Data collection took place in September and October 2020 for the main section of the questionnaire and in November and December 2020 for the supplementary section on mental health. The task was entrusted to Ipsos, a company specializing in population surveys, which carried it out in accordance with the Santé publique France study protocol. The survey questionnaire could be completed online or by phone with an Ipsos interviewer.

The data collected was transmitted, recorded, and analyzed by Santé publique France in a completely anonymous manner. In accordance with the law, respondents have the right to access, object to, correct, delete, and restrict the processing of data concerning themselves or their child.

Yes, the full study questionnaire is available on the Santé publique France website. The study protocol is also available online.

You can get more information by:

• by calling the toll-free number: 0 800 942 590

• on the study’s dedicated website: www.sante-post-incendie-76.fr

• on the Sante publique France website: www.santepubliquefrance.fr

Questions can also be submitted via the contact form on the study’s dedicated website: contact@sante-post-incendie-76.fr

The data analysis consisted of two main stages:

• The first stage was descriptive: it provided the characteristics of the sample (age, gender, occupation, etc.) and verify that it was indeed representative of the study population, calculate the percentages of people who reported exposure to each hazard caused by the fire, and determine the proportions of people who experienced one or more health problems during or following the fire (distinguishing between each symptom and any care received), and finally to describe the population’s perceived health, in both its physical and mental dimensions, as measured one year after this industrial accident.

• The second stage was analytical: statistical models were developed to examine the relationship between perceived exposures to the fire and, on the one hand, symptoms experienced during the industrial accident, and, on the other hand, physical and mental health as perceived by individuals one year later. These multivariate analyses were adjusted for factors known to influence health status (age, sex, body mass index, educational level, income level, tobacco use, alcohol consumption, etc.), including the COVID-19 pandemic, which was incorporated into the statistical models.

Completing this analytical step took several months because it is methodologically complex and we had a large amount of data collected from a large number of people. All the study data were analyzed, except for those from the supplementary module on mental health, which will be processed in the coming months.

“A Study Focused on Your Health” helped rigorously assess the short- and medium-term impact of the fire on public health: perceived exposure to the accident’s hazards and pollution, symptoms and health issues experienced during the fire, and health status one year after the industrial accident. From a scientific standpoint, this epidemiological study has several strengths:

• a high-quality statistical method that reflects the diversity of characteristics among those affected by the fire and produces results generalizable to the entire population of the 122 municipalities in Seine-Maritime affected by the accident, which totals 340,000 people;

• standardized questionnaires, particularly to measure individuals’ perceived health status one year after the accident (the SF-12), whose scientific validity is recognized;

• the collection of health data not available in any other health information source: its results thus complement the information provided by the immediate post-accident surveillance implemented by Santé publique France during the month following the fire, based on data from hospital emergency departments and SOS Médecins;

• the study was conducted one year after the industrial accident: this time lag made it possible to measure health events that may have persisted or occurred long after the event, particularly changes in mental health, and to assess the medium-term impact of the fire;

• the study was conducted in a control population to facilitate the interpretation of the results obtained by the study in order to assess the health impact of the accident one year after it occurred.

Scientific literature and discussions with the Health Group indicate that people retain vivid memories of such a significant event for a long time. We observed, during the questionnaire test conducted in July 2020 with Ipsos, that fewer than 10% of respondents (3 out of 40) answered “yes” to the question “Did you have difficulty answering certain questions because they concerned an event that occurred too long ago?”

However, there is a potential memory bias that may have led to an underestimation—or, conversely, an overestimation—of the accident’s impact on the health of those exposed. This bias primarily concerns the measurement of the relationship between perceived exposure and symptoms experienced during the accident, as perceived health one year after the fire was measured using a tool (the SF-12 questionnaire) that asks about the four weeks preceding the survey.

Yes. Despite the delay in its implementation, this epidemiological study provides unique insights into the nature of perceived exposure to the nuisances and pollution generated by the fire and its immediate health effects, as well as into the health status and quality of life of those exposed one year after the event.

The information it provides:

• supplements the data produced in the immediate post-accident period by the regional unit of Santé publique France and by other organizations: the Regional Health Agency, Rouen Emergency Services, Poison Control Centers, Atmo Normandie, etc.,

• highlights the importance of establishing epidemiological monitoring of the population’s health status in the medium and long term and guides the selection of health indicators to be monitored.

The study’s results enable the formulation of initial proposals for actions aimed at improving the management of the accident’s health consequences and preparing the response to future industrial accidents.

Yes. Although the impact of the COVID-19 pandemic and the lockdown on both physical and mental health is undeniable, it was essential to accurately assess the health impact of the industrial accident on September 26, 2019, and to address the concerns and questions of the affected population.

Discussions with the Health Group and media coverage of the event show that this industrial accident and its impact on the health of the exposed population remain topical issues that continue to fuel public debate.

No, because the study focuses on the short- and medium-term health effects of the fire.

However, its findings indicate a medium-term effect on overall health, linked to a decline in mental health, which could persist. They justify the need for medium- and long-term epidemiological surveillance, which will be implemented by Santé publique France, and help guide the selection of certain health indicators that will be monitored.

Despite its heavy involvement in the COVID-19 pandemic, Santé publique France considered the assessment of the health impact of the industrial accident in Rouen to be a priority, justifying the deployment of some of its staff.

The team leading the study consists of several statisticians and epidemiologists (specializing in mental health, environmental health, and post-disaster health), two public health physicians, and a specialist in population surveys. These staff members are based at the Normandy Regional Unit and at the Santé publique France headquarters.

It receives support and advice from individuals outside Santé publique France: members of the Health Group, experts from the Thematic Support Committee, and, on an ad hoc basis, specialists from INSEE and DRES.

Santé publique France sprang into action on the very day of the industrial accident in Rouen, September 26, 2019. Epidemiological surveillance of acute effects that led to visits to emergency rooms or calls to SOS Médecins was initiated on the day of the accident and continued for one month (learn more about the immediate response).

• October 2019: Santé publique France was tasked by the Directorate General of Health with conducting health monitoring following the fire. Since then, work has been underway on the overall epidemiological response to be implemented, which took shape as the "Santé Post Incendie 76" initiative.

• December 2019: Work began on developing the study in close consultation with the Health Group starting in January 2020, and with scientific support from the Thematic Support Committee beginning in March.

• First half of 2020: drafting of the study protocol, conducting a preliminary survey in four municipalities, designing and testing the questionnaire, obtaining approval from the Public Statistics Label Committee, logistical preparation for the survey, and developing informational materials about the study. This work was carried out in collaboration with the Rouen Health Group and, at the national level, with the Thematic Support Committee.

• Second half of 2020: collection of study data, preparation and post-processing of data.

• First half of 2021: statistical analysis of the data, interpretation of the results with the Health Group and the Thematic Support Committee, drafting of the main study report, presentation of the results to the Transparency and Dialogue Committee on July 5 in Rouen, publication of the report and an infographic of the main results on the Santé publique France website.

The three other components of the “Santé Post Incendie 76” initiative have also been the subject of reports and summaries, which are published on the Santé publique France website.

The fire at the Lubrizol and NL Logistique warehouses and the subsequent cleanup efforts caused odors to spread around the site due to the release into the air of substances that had been stored there and combustion byproducts. According to Atmo Normandie, the sulfur compounds identified in analyses of certain ambient air samples (such as butyl mercaptan, sulfides, and disulfides) likely contributed to the hydrocarbon- and sulfur-like odors. The “alkyl/aromatic” odors that were detected are believed to be largely due to the chemicals used in an attempt to mask the emitted odors.

The issue of health effects caused by the perception of unpleasant odors must be distinguished from the issue of the potential toxicity of the chemicals responsible for the olfactory stimulus. Indeed, a bothersome olfactory perception can induce adverse health effects through pathological mechanisms (psychological stress and somatization, intolerance linked to past events) that differ from those triggered by the chemical toxicity of the molecule on the body.

The 6,124 reports received by the Atmo Normandie association through September 2020 made it possible to describe the odors generated by the fire and to document the health effects reported as being associated with their perception: these were primarily headaches, nausea, and tingling/irritation of the throat and nostrils.

The survey “A Study Listening to Your Health” showed that a very large proportion (86%) of residents in the 122 municipalities of Seine-Maritime included in the study detected odors linked to the fire. This olfactory perception lasted several weeks or months for 20 to 30% of respondents and persisted at the time of data collection, one year after the accident. Statistical analyses revealed an association between the perception of odors and the risk of experiencing at least one symptom during the fire and its aftermath, with this statistical link being particularly pronounced for otolaryngological symptoms (tingling or burning in the nostrils, throat, or tongue) and headaches. The longer the perception of odors lasted, the greater the likelihood of experiencing a symptom. Furthermore, having experienced fire odors for a long time is linked to a decline in overall health and mental health, as measured one year later. Data from the complementary component of this study, focused on mental health, will allow for an analysis of the link between odor perception and anxiety or depression.

More than 90% of the study population, comprising 340,000 people, reported having experienced at least one form of exposure to the industrial accident on September 26, 2019. The perception of odors was the most commonly reported form of exposure (86%) and was often described as bothersome or very bothersome.

At the time of the fire and in its aftermath, two-thirds of adults reported experiencing at least one symptom or health issue they attributed to the industrial accident. These were primarily psychological symptoms (nervousness, stress, anxiety, distress, panic), otolaryngological (tingling in the nostrils or throat, nasal discharge, and nasal congestion), general (headache, malaise, fatigue), ocular (watery eyes, redness of the conjunctiva), or respiratory (cough, shortness of breath, and, more rarely, asthma attacks). Nearly 20% of people who reported having a chronic condition mentioned that it had worsened as a result of the accident. The frequency of reported symptoms was higher among people living within 1,500 meters of the accident site than among those living farther away. About 17% of symptomatic adults reported having sought medical care, primarily by consulting a general practitioner, and 19% reported self-medicating. The likelihood of reporting at least one symptom increased steadily with the number of perceived nuisances and forms of pollution (odors, plumes of black smoke, soot deposits, being awakened, or hearing the sounds of the fire). The perception of odors was the factor most strongly associated with experiencing at least one symptom.

According to the parents included in the study, 60% of the children developed or experienced a worsening of at least one health condition during or after the accident, which they attributed to the accident. The most common symptoms were related to the ears, nose, throat, and eyes. Psychological symptoms were reported for one-third of the children, and sleep disturbances for one in ten children. Parents reported that 13% of symptomatic children received a medical consultation, and in 9% of cases were self-medicated.

One year after the accident, the perceived health of those exposed to the accident is poorer than that of unexposed individuals, primarily due to a decline in their mental health scores. This decline in perceived health is observed among those who experienced multiple accidental nuisances and pollution incidents and among those who perceived the odors for an extended period.
Further results from the study will be presented before the end of the year. They will focus on the mental health of those exposed, and more specifically on anxiety, depression, and post-traumatic stress.

The study’s findings and results provide a basis for initial recommendations aimed at improving the preparedness of the public health response to future industrial accidents. Among these are

• the preemptive establishment of a surveillance and information system to be activated in the event of an accident, involving healthcare professionals (private practitioners, pharmacists, and other stakeholders in the healthcare and medico-social sectors) as well as the general public, would enable real-time assessment of the event’s health impact and the collection of concerns, needs, and expectations of residents in the affected area.

• Training healthcare professionals (both hospital-based and private practitioners) on potential exposures in the event of an industrial accident, the expected effects on the physical and mental health of the affected population, and the appropriate management protocols.

The experience gained from this industrial accident should enable the development of strategies to facilitate the conduct of a population health study shortly after an industrial accident (standard questionnaire, simplified protocol for random population sampling, etc.), thereby limiting recall bias and reducing delays in producing data on its health impact.

Key Findings

This is the first study conducted among the general population following the industrial accident. It was carried out using a questionnaire administered to a representative sample of the exposed population, which was selected through a rigorous process. The study was developed in collaboration with local stakeholders, brought together in the Health Group, to better align it with local perceptions and address specific questions from the public regarding the health consequences of the fire.
The study provides a detailed description of perceived exposures, experienced symptoms, healthcare utilization, and behaviors adopted during and after the accident. It also assessed the population’s health one year after the event using a psychometric measurement tool validated by the international scientific community (the SF-12).

The statistical models used to analyze the relationships between perceived exposure and health were adjusted for numerous factors, including the effects of the COVID-19 pandemic.

Limitations

This study is a cross-sectional survey, conducted at a single point in time, which does not allow for controlling the temporal sequence between the events under consideration (exposure and health) nor for ruling out the possibility that the observed relationships are bidirectional—that is, they may be linked to exposure but also, in some cases, induced by the respondent’s initial health status.

The one-year interval between the industrial accident and the conduct of the study (in September and October 2020) could introduce a recall bias that might lead to an underestimation or overestimation of the relationship between perceived exposures and symptoms experienced during the fire. This potential bias does not apply to the analysis of perceived health measured one year after the fire, which was conducted using the SF-12 questionnaire, as this health measurement tool focuses on health status in the four weeks preceding participation in the study.

According to the World Health Organization (WHO), mental health is defined as “a state of well-being in which an individual can realize his or her potential, cope with the normal stresses of life, work productively, and contribute to the life of the community.” European countries (the European Commission of the European Union and the WHO European Ministerial Conference) highlight two aspects of mental health. On the one hand, “positive” mental health, which relates to autonomy, well-being, and personal fulfillment; and on the other hand, “negative” mental health, which concerns reactive psychological distress and psychiatric disorders.

In our study, participants’ mental health one year after the industrial accident was assessed using the MCS (Mental Component Summary) score from the 12-item version of the Medical Outcomes Study Short Form Health Survey (MOS SF-12 or SF-12). This is a scientifically validated psychometric measurement tool, translated into French. A supplementary section of the study questionnaire is dedicated to mental health. Focusing on generalized anxiety, depression, and post-traumatic stress, its results will be presented at a later date.

Given the documented effects of the COVID-19 pandemic on mental health, three questions related to the viral outbreak were included in the questionnaire. This allowed for the adjustment of statistical models to account for these factors. In addition, our study included a control group that did not experience the industrial accident, thereby allowing us to assess the impact of the fire on the mental health of the exposed population while accounting for the potential effect of COVID-19.

This decision is up to the patients’ primary care physician. If symptoms persist, a doctor should be consulted; he or she will be able to determine the most appropriate course of action based on the patients’ health condition.

Many of the symptoms experienced by those exposed during and after the fire are of an irritative nature (tingling in the nostrils, red and watery eyes, etc.) and appear, at first glance, to be benign. However, some of the health issues reported, such as an asthma attack or the worsening of a chronic condition, can be potentially serious.

Furthermore, our results show a certain persistence of the reported sleep disturbances and psychological symptoms, as well as a decline in mental health measured one year after the accident. These observations may foreshadow a longer-term impact on the health of exposed individuals and warrant medium- and long-term epidemiological surveillance to track changes in health indicators over time.

Our study did not involve blood samples because this was not in line with its objectives. It is an epidemiological study in which data were collected via a questionnaire, designed to assess individuals’ perceived exposure and health following the industrial fire.

Our study did not collect data on the nature of the chronic conditions mentioned by respondents. It focused on their personal assessment of whether their chronic condition had worsened as a result of the industrial fire.

However, the surveillance system established by Santé publique France, based on data from the National Health Data System (SNDS), should make it possible to assess at a later date the health status of individuals with certain chronic diseases and how their conditions have evolved following the industrial accident.

These are samples of human biological materials—such as blood, urine, hair, breast milk, etc.—that are typically collected immediately following an incident, even before the exact substances released are known, and which provide the biological material needed to analyze human exposure. The collection of these types of samples for workers falls under the responsibility of occupational physicians and employers.

Biomonitoring studies are population-based studies that are complex to implement, particularly in emergency situations following an accident, due to:

• the large number of potential exposure substances or biomarkers to be screened for, the choice of biological matrix in which to measure them,

• the laboratory techniques that can be used: sample collection and storage, analytical measurement equipment and methods, and the availability of laboratories capable of performing them,

• the interpretation of biological results, especially in the absence of reference values, which in some cases is complex and requires additional data (as biases are possible) and often repeated measurements over time,

• ethical considerations: invasive sampling, advice to be given to individuals.

Fully aware of the public’s high expectations in this area, Santé publique France has been tasked by the Directorate General of Health and is considering the feasibility and necessary preparations to carry out, in the event of future industrial accidents, a campaign of precautionary biological sampling in the immediate post-accident period, particularly in situations where, immediately following an accident, the full range of substances to which workers and the general public may be exposed is unknown.

Santé publique France responded by establishing the Normandy Occupational Health Alert Group (GAST) on October 29, 2019, at the request of the Regional Directorate for Enterprises, Competition, Consumer Affairs, Labor, and Employment (DIRECCTE).

This group, operating under Santé publique France, brings together specialists from several disciplines: physicians from the Centers for Occupational and Environmental Pathology Consultation (CCPPE), a physician from the Poison Control and Toxicovigilance Center (CAP-TV), a regional labor inspector (Direccte), and epidemiologists from Santé publique France.

The Gast’s objective was to assess the extent of worker exposure, medical monitoring, and acute health effects or events observed following the fire, in order to propose a surveillance strategy for these populations to evaluate the long-term health consequences—including psychological ones—of the event.