Loading documentation...
title: EU FIRESTAT — Final Report (July 2022) description: Canonical English text of the EU FIRESTAT pilot project final report (SI2.830108). Harmonised Tier 1/2 variables, definitions, value lists, data journey, and annexes. Pair with eu-firestat-variable-cheatsheet.md for HeatWaves mapping.
Closing data gaps and paving the way for pan-European fire safety efforts
| Field | Value |
|---|---|
| Project | SI2.830108 (European Parliament pilot project, DG GROW) |
| Completed | July 2022 (first edition) |
| ISBN | 978-92-76-51989-8 |
| DOI | 10.2873/778991 |
| Licence | CC-BY 4.0 |
Consortium (alphabetical): Efectis (lead), Bundesanstalt für Materialforschung und –prüfung (BAM), Centre for Fire Statistics of CTIF (CFS-CTIF), Danish Institute of Fire and Security Technology (DBI), Lund University, National Fire Protection Association (NFPA), School of Engineering (University of Edinburgh), Dutch Burns Foundation / EuroFSA, Vereinigung zur Förderung des Deutschen Brandschutzes (VFDB).
This document reflects the views of the project authors only. The European Commission is not liable for consequences stemming from reuse of this publication.
HeatWaves edition notice: Markdown derived from the EU FIRESTAT final report PDF (July 2022, SI2.830108). Figures and charts are not included; some figure captions and chart label text from PDF extraction may remain as prose.
For the authoritative layout and all graphics, use the official PDF.
Country-level background from the March 2021 progress report is in
progress-report-1.md(culled HeatWaves edition).
The present report is the final report of project SI2.830108, financed by the European Parliament and commissioned by DG GROW at the European Commission. There is a summary of the findings within the final report, but this Executive Summary gives an overview of the aim, methods and main findings.
The aim of this pilot project was to map the terminology used and the data collected by the EU Member States regarding fire events, and to propose a common terminology and method to collect the necessary data in each EU Member State with a view to obtain meaningful datasets (based on standardised terms and definitions). This, in turn, should allow for knowledge-based decisions regarding fire safety at the Member State and the EU level regarding building fires (i.e. houses, apartment blocks, office buildings, commercial buildings, hospitals, schools and kindergartens, retirement homes, etc.).
A review of fire statistics literature shows that systems for collecting fire data have been instrumental in reducing building fires and their associated deaths, injuries, and economic damage. But EU Member States and other countries’ systems vary considerably in the type and scope of information collected, the way that data elements are defined, the levels of specificity sought, and the types of training and resources dedicated to collection efforts. The amount and quality of information in several data collection systems also appear to be influenced by whether they include information from sources outside the fire service, such as insurers or medical authorities, through data linkage or other means. In many cases, the amount of information collected appears to depend on available resources and the level of expertise of the personnel.
In general, it appears that most countries’ administrations presume the systems for collecting fire data provide an accurate representation of the national experiences with fire incidents. However, information gathered through the initial phase of the project suggests that collection systems oftentimes fail to assess the potential for bias due to missing information, differences in the way terms are defined or interpreted, and other issues that may influence data quality. The terminology and data collection methodology adopted in current fire statistics were examined in 27 EU Member States and eight other countries (Australia, Canada, New Zealand, Norway, Russia, Switzerland, UK and USA). The eight non-EU countries were chosen based on their structured and detailed fire statistics.
The review of fire data collection measures within and outside the European Union is important for understanding the degree of commonality across the various systems, and also for identifying opportunities and challenges in any efforts to improve fire safety. Systems in the European Union fall into different tiers with respect to the amount of fire data information collected. Data collection systems usually include information on the incident time, date and location and most countries include the type of property, subdivided into residential and non-residential buildings. However, additional information on building characteristics was seldom recorded in most countries. Information about fire causes and other data related to the source of ignition, item and material ignited first, material mainly responsible for the development of the fire and fire room of origin commonly appears.
The number of fatalities and casualties are recorded in most countries, along with information about the victim’s age and gender, and type of fatality or injury. Due to the lack of official definitions and precise collection methodologies in most countries, it is clear that there is significant variation in the data currently collected across the EU Member States. The research found little mention of methodologies for dealing with missing data in the systems for collecting fire data in Europe, although some countries do acknowledge that missing data is a problem that compromises data quality. Additionally, none of the reports consulted included uncertainty estimations. Naturally, this poses a major obstacle to data comparison, and thereby, to assess effectively potential best practices and successful safety approaches.
To provide relevant information about national fire safety (number of fires, fire fatalities, fire injuries, fire losses, etc.), fire statistics will have to be improved through common terminology, common methodology, and common training and qualification of persons in charge of filling in fire reports, including uncertainty estimation methods. The research accordingly reviewed a number of critical issues involved in the design and implementation of systems for collecting fire incident data and offered recommendations to improve consistency among systems.
The review proceeds from the assumption that fire incident data can serve a number of important purposes - helping to reduce fires and losses, identifying opportunities for safety interventions and education programs, guiding the allocation of public resources to areas of greatest need and impact, and monitoring progress of safety initiatives. Data collection systems can also facilitate opportunities to share experiences and successes across regions and between countries, promoting a broader diffusion of technical and other innovations that increase fire safety. To achieve these objectives, it is important that data collection systems produce reliable data. The design of the proposed data collection method was driven by the EU Member States’ needs and capabilities.
In data collection with systematic intervention purposes, as is the case with fire incident data collection, it is important to create a sufficiently robust database that can help identify risk factors and document fire incidences with reasonable confidence. Data collection systems that rely on voluntary reporting will almost certainly fall short of a complete census, while data collected by convenience sampling methods might have selective utility but would be insufficient for capturing the broad range of fire incidents at the national level. It appears that most countries currently employ a voluntary approach to data collection, coupled with expectations of fire departments to participate in filing reports, but more efforts are necessary from national programs to encourage and evaluate compliance.
Whatever form the data collection system takes, it is important that it captures reliably the experiences of the populations it seeks to measure.
As mentioned earlier, we were able to find little discussion of missing data among the systems to collect fire incident data in the European Union, as well as most fire incident data collection systems more generally. It may be the case that missing data receives the greatest attention in the United States because its data collection system is the most extensively detailed, with the greatest potential to produce items with unknown values, and potentially to discourage submission of reports altogether. Missing data may be less problematic in reporting systems that require less detail and whose population groups may have greater uniformity with respect to fire experiences.
The impact of missing data is likely to be especially problematic if it fails to account for differences in the populations impacted by fire incidents, potentially leading to imperfect interpretations of results. Such differences might include regional differences in the built environment, differences in neighbourhood conditions, including housing quality and social conditions, or differences in age. Accordingly, assessment of missing data will be especially important in countries that are characterized by diverse regional levels of economic development and diversity of economic and social conditions. On this point, it is important to note that generally the fire data collection systems examined in this research appear to be regarded as census systems of data collection. We cannot say if this is a view held by key users of fire data in these systems.
However, there is a danger in assuming that data collection systems capture all or most fire incidents in the absence of any examination of the degree and form of unreported fires or other missing data. Any systematic failure to collect data that is not randomly distributed runs the risk of failing to identify risk factors associated with social and economic disadvantage, victim characteristics, or other factors related to fire incidence or outcomes. Accordingly, it is important that the implementation of systems for collecting fire data include plans for data quality checks and procedures for handling missing data in order to verify the validity and reliability of data findings. The financial costs associated with efforts to harmonize data collection will vary by country and be influenced by the existing state of fire data collection practices and resources.
It is important to have a realistic appraisal of the economic costs of fire incident data collection if any harmonized system is to be sustainable over time. Countries and regions with stronger national traditions of data collection in support of policy objectives will require substantially less investment in supporting a harmonized data collection system than those in which collection efforts are less mature or concentrated in specific areas. It is important to note here that substantial costs differences may depend on decisions about what data to include and how to collect it. This may be an unforeseen cost in seeking to achieve harmonized data collection in countries with decentralized and non-uniform systems, even if those systems are mature.
The cost of implementing a comprehensive data collection system will be greatest in countries that have the least experience and fewest existing resources. Countries with less established or comprehensive data collection systems may have to assume significantly greater training costs when they seek to introduce data collection responsibilities in fire departments nationwide. The cost burden will be influenced also by the availability and sophistication of computer hardware and software. An estimate of the cost of running either a census or sample survey data collection was estimated for each member state taking relative differences in cost into account. This estimate assumed everyone starting from the same level and hence did not take into account the existing systems available.
The initial cost of implementing a comprehensive data collection system will be greatest in countries that have the least experience and fewest existing resources. Countries with less established or comprehensive data collection systems may have to assume significantly greater training costs when they seek to introduce data collection responsibilities in fire departments nationwide. Our review of data collection methods and systems provides a foundation for several additional concluding observations relative to national systems for collecting fire incident data:
Data collection systems should be designed with sustainability in mind. Public funding for data collection systems can lag if they fail to generate recognition as a public good or commitment among key principals.
Overly ambitious and detailed data collection systems may tax the patience of participants and undermine data quality. To encourage compliance and build competence and interest among participants, it may be useful for the architects of data collection systems to begin with comparatively modest reporting requirements and to introduce additional details incrementally as participants gain experience.
Align data collection content with realistic policy goals and use data to promote safety interventions and practices.
Use data to chart and publicize trends, demonstrate the value of fire safety interventions, data collection and build public recognition and support.
The project team developed and dispatched a survey to stakeholders in the EU Member States, namely regulators and fire authorities, to assess their opinions about the types of data needed to support fire safety policies.
Highlights
The proposal is mainly based on the opinions of the majority of the stakeholders from the EU Member States who responded to the questionnaire, with the observation that each of the proposed variables was already being collected by the majority of the EU Member States and/or the confirmation of their relevance by the opinion of the majority of the project consortium.
The following eight variables should be collected as a first priority.
Tier 1:
Once the previous eight variables have been implemented efficiently, we propose adding the second tier, which would include five additional variables:
Tier 2:
Collecting these variables as part of harmonized European fire statistics should not prevent European countries to continue collecting other variables in parallel. A set of definitions has been developed and is detailed in Chapter 4 of this report, for prospective fire data systems that seek to ensure a common understanding within the EU. The proposed terminology includes existing standards and practices. This terminology is based on knowledge of current practices and from discussions with stakeholders. For each variable, a definition and values assigned to the variable are proposed, which shall enable better fire statistics.
Four groups of categories were established, i.e., intervention characteristics, human characteristics, building characteristics and fire characteristics and several variables are assigned to the established groups of categories. The common terminology is based on the experiences from previous tasks and by researching public datasets and literature review. The proposed terminology is inspired by the definitions found in the ISO TS 17755-21 standard but are adapted to European specifications.
Highlights
Notes to the definitions or values are also presented when necessary to prevent any ambiguities. For example, there is an established threshold to determine which fires are to be collected:
Highlights
Damage in this context is considered as injuries at the fire scene, fatalities at the fire scene, damage to property of at least 100 euros and/or environmental contamination requiring clean up. Fire in this context is considered as uncontrolled self-supporting flaming, glowing or smouldering combustion. Explosions, flashes and discharges of static electricity, attempted suicide and suicide by self-immolation are excluded unless if the event resulted in a fire after the initial event.
ISO reference: Fire safety — Statistical data collection — Part 2: Vocabulary (ISO/TS 17755-2).
Additional guidance on how to collect, interpret and report data is presented in this report (Chapters 5 to 8).
A cost-benefit analysis (CBA) method is proposed in this study to provide a structured and explicit way to create a basis for decision making regarding fire safety measures.
Cost-benefit analysis is a common methodology for performing an economic analysis of fire safety investments.
CBAs has been used in literature to study different types of fire safety measures.
In particular, the installation of various kinds of water sprinkler systems has been studied in several countries and is not seen as cost-beneficial in general.
However, for specific types of buildings or for certain risk groups, the benefits may outweigh the costs.
Another measure that has been analyzed in several countries is installation of smoke alarms and it is often found to be cost-beneficial in general, primarily due to its low cost. Other measures that have been reviewed in this work include stove guards, fire extinguishers and combustible cladding. This study gives an overview of a proposed calculation procedure for conducting a cost-benefit analysis. There is also a description of the most important input variables. Important variables can include basic fire statistics, such as the number of fires, number of fire fatalities, number of injured, cause of fire and type of building. These are some of the variables suggested in this project.
More detailed fire statistics and information is needed for information on the presence, operation, and reason for failure of different technical systems (like automatic extinguishing systems and smoke alarms), as a part of a CBA analysis of such systems. It is also important to point out that there are several other input variables needed for a CBA which cannot be obtained from fire service statistics, for example the risk reduction of installing a certain measure and the cost of installation and maintenance of that same measure.
Consequently, it is strongly recommended that a cost-benefit analysis should be complemented with a sensitivity analysis to highlight the variation in the results due to uncertainty in the inputs.
Highlights
The calculation procedure presented in this project is demonstrated by three case studies. The case studies are important for explaining and illustrating further the calculation procedure. This is because the type of data needed and its availability varies between areas of study, both in terms of type of system and country studied. The case studies cover three types of possible actions i.e. implementing technical installations, improving materials/products and prevention campaigns. Included for each case study is a cost estimate for introducing a measure and an estimate of the benefit due to risk reduction and other benefits. A detailed calculation was possible for Case study 1 on smoke alarms since there have been several studies in that field and data is available for most of the important input variables.
The case study also illustrated that the measure (a smoke alarm) is cost-effective. The results of Case studies 2 and 3 on the introduction of regulations on upholstered furniture and home visits, respectively, are considered much more uncertain and harder to interpret since the benefit-cost ratio is close to 1. Several important input variables are also associated with large uncertainties that makes it necessary to complement such a study with a sensitivity analysis. As can be concluded from the case studies, reliable fire statistics are crucial for conducting this type of analysis. Data on the number of fatalities, number of fires, item first ignited etc., have been used in the case studies. Further details are available in Chapter 10 of this report.
The review of the literature shows that fire data collection systems have been instrumental in reducing building fires and their associated deaths, injuries, and economic damage. The utility of information about these fires is apparent in the design of many fire safety interventions and policy initiatives. Data on fire incidents can inform firefighting strategies, building codes, educational and training programs, and technical innovations, to cite just a few applications. For example, with populations aging more than ever before, we might expect higher death rates among senior citizen, despite early fire detection. It is logical to assume that safety efforts can benefit from strategies that have worked in other places.
However, there is substantial agreement in the literature that differences between fire data collection systems in different countries complicate the ability to make comparisons that could be useful in evidence-based planning and prevention efforts While national fire data collection systems are likely to share certain core features and to gather some fire incident data in common, there appears to be considerable variation in the type and scope of information collected, the way that data elements are defined and levels of detail they seek, as well as the types of training and resources dedicated to collection efforts. In addition, literature suggests that fire data are influenced by differences between data collection procedures and practices.
Some data collection systems appear to provide opportunities to update information that may not be available at the time an incident record is first created, such as the cause of a fire or deaths that occur sometime after the incident. The amount and quality of information in different data collection systems also appear to be influenced by whether they include information from sources outside the fire service, such as insurers or medical authorities, through data linkage or other means. Literature suggests that the issue of how much information to collect is an important area for consideration in the design of fire data collection systems.
Data collection systems that collect too little or wrong kind of information may not produce data that are useful, while overly detailed data collection systems may overwhelm data collectors, and thereby compromise data quality, as suggested by studies from the United States. In many respects, the issue of how much information to collect appears to be driven by available resources, as well as the capacities of data collectors, who mainly are fire service personnel, to collect and record information. Concise data collection records will require less support and fewer resources than those that are more complex. To that end, recent literature on fire data collection in Canada emphasizes that such factors as funding, resources, personnel, and stakeholder acceptance are critical considerations in the design and sustainability of national fire data collection systems.
In general, it appears that the fire data collection systems in most countries are presumed to provide an accurate representation of their respective experiences with fire incidents. However, information gathered through the initial phase of research suggest that they may be unaware of important limitations of their data due to missing information, differences in the way terms are defined or interpreted, and other identified issues. We identified significant issues with fire data from Australia, Bulgaria, Canada, Denmark, France, and Germany which complicate confidence in the data, particularly for their use in inter-country comparisons. Most of the issues stem from the lack of definitions for collected terms, lack of training, dispersed data, missing information and low coverage.
USA, Italy, and the Netherlands have very different systems while having each separate advantages and drawbacks. The fire data collection system in the USA has an existing terminology, includes a large number of data fields, and has vast experience in this field, but also appears to have a significant problem with missing fire incidents. However, because the EU is in a comparable situation to the USA, there are many lessons from the experience of the USA that can be directly applied to the EU. Italy has adopted a quality control system to ensure the integrity of all data treated but is missing important fire data. The approach of the Netherlands has been to reduce the problems posed by uncertainties by focussing data collection efforts on fatal residential fires.
We estimate that Austria, Russia, Sweden and the UK (in particular England, Wales and Scotland) provide data with high confidence level due to the existing definitions, important covered areas and collected terms and existing quality safeguards. The terminology and data collection methodology adopted in current fire statistics were examined in 27 EU Member States and eight other countries (Australia, Canada, New Zealand, Norway, Russia, Switzerland, UK and USA). The eight other countries have been chosen based on their structured and detailed fire statistics. The review of fire data collection measures within and outside the European Union is critical for understanding the degree of commonality across the various systems and also identifying opportunities and challenges in any efforts to improve fire safety.
Although it was not possible to gather information on fire statistics available in a limited number of countries, it appears that fire data collection systems in the European Union fall into different tiers with respect to the amount of information collected. Usually, fire incidents are described considering the incident time, date and location. The description of the property type subdivided into residential and non-residential buildings is available in the majority of countries examined while further building characteristics are seldomly recorded. Fire causes and the other fields related to the source of ignition, item first ignited, articles responsible for the development of the fire and fire room of origin are determined less often. The data recorded more often are those related to the description of the fire incidents, fatalities and injuries. However, the variables covered by these data can be referred to as different interpretations in the various countries examined.
Therefore, it is suggested to link the considerations presented for the analysis of the definitions with the elaborations of the data recorded by the various statistics and to the information related to the collection methodologies described in the short abstract of each country. Due to the lack of official definitions and precise collection methodologies, it is clear that the current fire statistics cannot be compared from one country to another (with a few exceptions). They can only be useful to describe the global fire safety situation and trends to some extent for a group of countries, or a specific fire safety situation.
To provide relevant information regarding the national fire safety situation (number of fires, fire fatalities, fire injuries, fire losses), fire statistics will have to be internationally improved through common terminology, common methodology, and common training and qualification of persons in charge of filling in the fire report, including uncertainty estimation methods. More detailed information is available in the Task 0 and Task 1 reports (EUFireStat, 2022).
In this project, we have reviewed critical issues involved in the design and implementation of fire incident data collection systems. The review proceeds from the assumption that fire incident data can serve a number of important purposes -- helping to reduce fires and losses, identifying opportunities for safety interventions and education programs, guiding the allocation of public resources to areas of greatest need and impact, and monitoring progress of safety initiatives. Data collection systems can also facilitate opportunities to share experiences and successes across regions and between countries, promoting a broader diffusion of technical and other innovations that increase fire safety. To achieve these objectives, it is important that data collection systems produce data that is reliable. We summarize some of the key factors related to data collection design and practice here.
The selection of data collection method should be determined by fire safety needs and capabilities of data collectors. In data collection with systematic intervention purposes, as is the case with fire incident data collection, it is important to create a sufficiently robust data base that can help identify risk factors and document fire incidence with reasonable confidence. Data collection systems that rely on voluntary reporting will almost certainly fall short of a complete census, while data collected by convenience sampling methods might have selective utility but would be insufficient to capture the broad range of fire incidents at the national level.
It appears that most countries currently employ a voluntary approach to data collection, with expectations that fire departments should participate in filing reports, but with mixed efforts by national programs to encourage and evaluate the completeness of data collection by fire departments. Whatever form the data collection system takes, it is important that it reliably capture the full range of the country’s fire incidents. To this end, data collection systems should be prepared to conduct follow up with non-respondents, assess the completeness of reporting, and identify any systematic patterns of non-reporting. The potential for missing data is an issue that was addressed in all phases of this project.
We were able to find little discussion of missing data among the fire incident data collection systems in the European Union, as well as most fire incident data collection systems more generally. It may be the case that missing data receives the greatest attention in the United States because its data collection system is the most extensively detailed, with the greatest potential to produce items with unknown values, and potentially to discourage submission of reports altogether. Missing data may be less problematic in reporting systems that require less detail and in which fire incidents are fairly uniform. However, it is critical that effort be made to identify the extent of missing data and the patterns it takes if data on fire incidents is to be considered reliable.
The impact of missing data is likely to be especially problematic if it fails to account for regional differences or other factors that influence fire incidents. Such differences might include regional differences in the built environment, differences in neighbourhood conditions, including housing quality and social conditions, or differences in age demographics. Multiple studies have shown the link between elevated fire risk and socio-economic status. Assessment of missing data will accordingly be especially important in countries that are characterized by diverse regional levels of economic development and diversity of economic and social conditions. On this point, it is important to note that the fire data collection systems examined in this project appear to be generally regarded as census systems of data collection.
We cannot say if this is a view held by key users of fire data in these systems. However, there is a danger in assuming that all fire incidents are captured by national data collection systems without some effort to assess the incidence of missing data. Failure to collect data that is not representative of all regions, demographics and income levels runs the risk of failing to identify risk factors associated with social and economic disadvantage. Accordingly, it is important that the implementation of fire data collection systems include plans for data quality checks and procedures for handling missing data in order to verify the validity and reliability of data findings. Financial costs will vary by country and be influenced by existing state of fire data collection practices and resources.
It is important that there be some realistic appraisal of the economic costs of fire incident data collection if any harmonized system is to be sustainable over time. Countries and regions with stronger national traditions of data collection in support of policy objectives will require substantially less investment in supporting a harmonized fire incident data collection system than those in which data collection efforts are less mature or concentrated in specific areas.. It is clear that the cost of implementing a comprehensive data collection system will be greatest in countries that have the least experience and fewest resources. Countries with less established or comprehensive data collection systems will assume significantly greater training costs in seeking to introduce data collection in fire departments nationwide.
The cost burden will also be influenced by the availability and sophistication of computer hardware and software. Considering such differences, as well as relative differences in certain costs between Member States of the European Union, we have identified the core cost components of data collection as a starting point for assessments of financial commitment. Our review of data collection methods and systems provides a foundation for several concluding observations relative to national systems of fire incident data collection.
Data collection systems should be designed with sustainability in mind. Public funding for data collection systems can lag if they fail to generate recognition as a public good or commitment among key principals.
Overly ambitious and detailed data collection systems may tax the patience of participants and undermine data quality. To encourage compliance and build competence and interest among participants, it may be useful for the architects of data collection systems to begin with comparatively modest reporting requirements and to introduce additional details incrementally as participants gain experience.
Align data collection content with realistic policy goals and use data to promote safety interventions and practices.
Use data to chart and publicize trends, demonstrate the utility of data collection, and build public recognition and support.
The nature and format of fire data varies significantly across the EU Member States. Naturally, this poses an obstacle to data comparability and the ability to effectively assess potential best practices and successful safety approaches. The current project therefore addresses the importance of developing a common European terminology regarding fire statistics in buildings.
A survey was developed to collect the opinion of the stakeholders regarding the required data that can help decision making in fire safety policy. The proposal we developed is based on the result of the survey filled by the stakeholders of the Member States. The main goal of the questionnaire was to learn from stakeholders in the Member States and outside the EU their visions, opinions and experiences regarding the data required for forming and implementing fire safety policies. Running in parallel (and interconnected) to the development of the questionnaire, the insights from the consortium were inventoried in a process of consortium opinion stocktaking. In this way, optimal use can be made of the knowledge and experience of the partners of the consortium. These insights provided important input for the proposal.
The model of influencing factors and the principles not only form the basis of the proposal, but have also been applied in the development of the questionnaire. This model (see Figure 1), based on scientific research, describes four factors that influence fire safety (Kobes et al., 2010). These factors are human characteristics, building characteristics, fire characteristics and intervention characteristics. Working with this model makes it easier to identify the variables that may influence fire safety. This ensures that an overall picture is generated of all the variables to be collected.
Figure 1 Model of influencing factors regarding the degree of fire safety (characteristics scheme)
A mailing list for distributing the digital questionnaire was developed. The main focus was to reach regulators from all EU Member States. Additionally, the questionnaire was also sent to regulators from other countries when this was possible (such as in England, Scotland, Switzerland and New Zealand). The goal was to include a representation of stakeholders that are involved in policy and legislation. The contacts were divided into three categories, listed by order of priority:
and fire (prevention) and fire service associations). The main goal was to find one organisation per country to fill in the questionnaire, preferably on behalf of the authorities and ideally complemented with responses from the other categories. In addition to the consortium’s collective network, the Federation of the European Union Fire Officer Associations (FEU) and the EC were asked to suggest contacts from certain EU-countries that were not covered. A number of criteria were used within the process for the selection of the data needed for fire statistics:
category ‘must be included in a dataset of fire statistics.
respondents. Using this limit value, including a margin of error of ± 10 points, allows for a larger coverage of opinions, such as a near majority. By doing this, more variables were considered in the justification process.
more importance than a variable that is not yet being collected.
The results of the survey among the stakeholders were compared with the data already collected by the EU Member States, and with the opinion of the consortium. Findings from the literature were used to illustrate the importance of proposed variables. Priority was given to the variables that are already collected by the majority of the EU Member States to facilitate the implementation. The following Venn diagram shows (Figure 2) the relationship between the three different sets of findings and how they overlap.
Figure 2 Different areas of the Venn diagram
Figure 3 Data confirmed by stakeholders, the consortium, and data collected by EU-27
The selected variables were divided into three tiers. Tier 1 includes the variables already collected by the majority of the EU Member States and also covered by variables selected by the majority of the stakeholders and the consortium (4 variables), or that are also covered by the set of the stakeholders only (4 variables). The variables in Tier 2 are considered important by both the stakeholders and the consortium.
Table 1 Variables per section of the Venn diagram
Area Intersection Total Variables Tier A Intersection of Consortium & Data collected & Stakeholders 4 Number of victims (fatalities & injuries); Age of fatalities, Primary causal factor of fire; Incident date.
B Intersection of Data collected & Stakeholders 4 Type of building; Incident time; Incident location; Number of injuries; C Intersection of Consortium & Stakeholders 5 Articles contributing to fire development; Heat source; Number of floors; Area of origin; Fire safety measures present D Intersection of Consortium & Data collected 0 2 E Stakeholders, excluding the intersection with other sets 13 Operation of fire safety measures; Reason for failure of fire safety measures; Construction characteristics; Number of occupants in the building; Quantification of property damage; Fire detection time; Disability; Role; Fire brigade response time; Construction type; Effectiveness of fire safety measures in reducing the fire; Direct fire costs; Type of property damage.
F Consortium, excluding the intersection with other sets 5 Fire spread at final situation; Fire spread at fire brigade arrival; Item first ignited; Size of smoke spread; Gender; G Data collected, excluding the intersection with other sets 1 Time between fire brigade arrival and withdrawal.
Tier 1 – Eight variables; covered by all three sets, or only by the stakeholders and existing data collection
Variables in tier 1 are considered essential for data collection, those include:
For example, in the Netherlands and in the USA, elderly (age 61 and older or in some literature 65 and older) are over-represented among fatalities of residential fires, and are a risk group for serious injuries from fire (Fahy & Petrillo, 2021). When studying this specific risk group, it appears that the physical and cognitive limitations are to a large extent responsible for the fact that the elderly are overrepresented in fire fatalities. Another study about the (potential) fire risks for different groups stresses the importance of taking age into account (Fire Service Academy, 2020). The variable ‘type of building’ is frequently mentioned by the stakeholders and is currently already being collected by the majority of the EU Member States.
As the proposed focus is on all types of buildings (e.g., Residential, non-residential, etc.), it is essential to collect data on the type of building so as to ensure that a distinction can be made between the fire risk of (Home Office, 2019) different types of buildings. This distinction is important as, for example, most of the fire-related fatalities are in residential fires. Both the stakeholders and the consortium indicated the ‘primary causal factor of fire’ as an important variable regarding fire characteristics. Examples of values for this variable are human act, equipment failure, natural phenomenon, etc. Additionally, this variable is already being collected by the majority of the EU Member States.
The variables regarding fire intervention characteristics which are frequently mentioned by the stakeholders, and which are also currently already being collected by the majority of the EU Member States include ‘incident location’, ‘incident date’ and ‘incident time’. Examples of values for the variable incident location can be geographical coordinates or the building address.
Tier 2 – five variables; covered by the sets of the stakeholders and the consortium
There are five other variables chosen by the stakeholders as well as the consortium that are not currently collected by the majority of the countries. For these specific five variables, we propose to include them as a Tier 2 priority to be harmonised and implemented in a second step. Those variables are the following:
‘Area of origin’, ‘heat source’ and ‘articles contributing to fire development’ are important variables, but such insight can usually only be obtained through a fire investigation at the fire scene. The number of EU Member States already collecting data about the heat source is currently unknown, but it is confirmed that Sweden already collects this data. The implementation of these variables might be more complex than the others, hence their inclusion in the Tier 2 list.
Regarding the variable area of origin, it was confirmed that at least 30% of EU countries collect the area of origin data. With regard to fires in high rise buildings, it is conceivable that the information about the ‘number of floors’ is relevant. Data about the number of floors can give a substantial amount of information about the efficiency of fire safety and any evacuation measures that have been adopted (e.g., the evacuation strategy of a high-rise building is usually different than for single floor buildings). This can also be a strong indicator when comparing data between different countries. However, consortium experience suggests that the data on the number of floors is not always answered correctly or is unreported, resulting in missing data (along with reduction of statistical power and representativeness).
Tier 3 – Other variables covered by one set Other variables are covered by the set of the stakeholders (13), the consortium (5), or existing data collection (1). Those variables are not included in this proposal, though they may be of interest for the further development of data collection. The variables are listed below in order of the number of EU countries that selected the variable concerned or in which information is already being collected.
From the results of the questionnaire distributed to the stakeholders of the EU Member States, we propose to include thirteen variables for harmonization in European fire statistics. The choice is mainly based on the opinions of the majority of the stakeholders from the EU Member States who responded to the questionnaire, with the observation that the variable was already being collected by the majority of the EU Member States and/or the confirmation by the opinion of the majority of the consortium. As a starting point, the following eight variables should be collected.
Tier 1:
Once the previous eight variables have been implemented efficiently, we propose adding the second tier, which would include five additional variables:
Tier 2:
Collecting these variables as part of the harmonized European fire statistics should not prevent the European countries to continue collecting other variables in parallel. 3 This variable was included in Tier 2 as it will be necessary to determine other selected variables such as Heat source and
Primary causal factor
The variables listed in Tier 1 and 2 are assigned to four groups of categories of interest, i.e., intervention characteristics, human characteristics, building characteristics and fire characteristics, as presented in Table 1.
Table 1. Categories of interest and the assigned variables
The objective here is to identify the most unambiguous titles and definitions for variables which describe the categories of interest to be recorded by fire officers in the immediate aftermath of a fire incident and subsequently collected at European level, as well as appropriate values which these variables can have. This terminology would constitute a minimum dataset for collection at the local level. It would not prevent a fire department or national authority from having a more comprehensive data collection, as long as they are able to provide simplified data according to the terminology of the pan-European statistics. There are several goals when identifying appropriate values for a variable:
Age of fatalities
Building characteristics Type of building Number of floors Fire safety measures present Fire characteristics Area of origin Item first ignited Article(s) contributing to fire development Cause (Heat source and Primary Causal Factor)
The following threshold is established to determine which fires are to be collected for the project. What is a fire incident?
Damage in this context is considered as
Injuries at the fire scene.
Fatalities at the fire scene.
Damage to property of at least 100 euros.
Environmental contamination requiring clean-up. Fire in this context is considered as uncontrolled self-supporting flaming, glowing or smouldering combustion. The following will be included only if the event resulted in a fire as defined above after the initial event:
Explosions, flashes and discharges of static electricity, attempted suicide and suicide by self-immolation.
Definition:
The earliest available moment of which a fire event occurred, registered in hours (24H) and minutes at local time.
Note to definition: Note 1: The earliest available moment refers to the earliest moment that the fire is reported to an official authority/system (for example: the detection time by an automatic detection system linked to the control room or calling the emergency number).
Value:
hh/mm (24H) or undetermined + local time (for example: UTC + 01:00)
Definition:
The earliest available moment of a fire event occurrence, registered in the day, month and year at local date and time.
Note to definition: Note 1: The earliest available moment refers to the earliest moment that the fire is reported to an official authority/system (for example: the detection time by an automatic detection system linked to the control room or calling the emergency number).
Value:
dd/mm/yyyy (European notation)
Definition:
The most precise place where a fire event occurred, registered in (by availability) coordinates, name of the country, region, town, postal code and/or street name and number.
Value:
If available: coordinates, country, region, town, postal code and/or street name and number where the fire occurred or unknown (in this case only country) Note to value: Note 1: Coordinates are latitude and longitude to be collected.
Definition:
Is the number of person(s) who died as a result of injuries sustained during a fire incident.
Note to definition: Note 1: Fire-related fatalities are those that would not have occurred had there not been a fire.
Note 2: Fire fatalities include people who die within 1 year because of injuries sustained from the incident. A shorter time period is accepted, but not shorter than three months. Fire fatalities also include fatalities from natural or accidental causes sustained whilst involved in the activities of fire control, attempting rescue or escaping from the dangers of the fire, including blast and defenestration.
Note 3: Fire fatalities include all persons discovered or declared dead on the location of the fire, during their transportation to the hospital or after their admission at the hospital.
Note 4: The number of the variable should include self-intended fires / suicidal fires, but they should be marked as such.
Note 5: People who died before a fire started (natural death, victims of a violent crime) are to be excluded from the statistics as soon as a forensic medical report is available.
Value:
Numerical value [to be approximated when unknown].
Definition:
The number of persons who are injured (but not counted as deaths) as a result of a fire incident.
Note to definition: Note 1: Fire-related injuries are those that would not have occurred had there not been a fire.
Note 2: Fire injuries also include injuries from natural or accidental causes sustained whilst involved in the activities of fire control, attempting rescue or escaping from the dangers of the fire, including blast and defenestration.
Note 3: Fire injuries are those treated at the scene or taken to the hospital.
Value:
Numerical value to be approximated when unknown.
Definition:
Numerical value of age of fatalities in years, at time of the fire.
Note to definition: Note 1: If actual age is not known, it should be estimated with the closest possible estimate.
Particular care should be used in estimating the age for young adults aged 15 – 25 and older adults aged 60 – 70 as the threshold between youth and adult is often set at 18 years and between adult and senior at 65 years. For children less than 12 months old the age should be estimated to be one year.
Value:
Numerical value [to be approximated when unknown].
The following definitions and classifications are extracted from the Classification of Types of Constructions and adapted to the scope of the current project (Eurostat, n.d.-b). This classification system is used by Eurostat for European statistical purposes, such as providing indicators on the development of granted building permits in the European Union (EU) (Eurostat, n.d.-b). The classification mainly differentiates the use of buildings, according to the main use (e.g. residential, non-residential) as well as their respective sub-divisions.
Definitions:
for permanent (or semi-permanent) purposes, can be entered by persons and are suitable or intended for protecting persons, animals or objects. Buildings that are under construction are excluded from the definition of Buildings but are listed as a separated type of construction (see classification in Annex A). Buildings are subdivided into residential, non-residential and mixed-use buildings.
Residential buildings are constructions that are exclusively used for residential purposes.
Non-residential buildings are constructions that are exclusively used for non-residential purposes.
Mixed-use buildings are constructions which are used for both residential and nonresidential purpose.
In residential buildings, there is the notion of ‘Dwellings’, which is defined in the following way according to Eurostat (Eurostat, n.d.-a): Buildings that are used entirely or primarily as residences, including any associated structures, such as garages, and all permanent fixtures customarily installed in residences. Houseboats, barges, mobile homes and caravans used as principal residences of households are also included, as are historic monuments identified primarily as dwellings. The notion of ‘mixed-use buildings’ has been added by the consortium as it has a significant impact on fire safety and related policies. Consequently, mixed-used buildings are based on the main apparent use, with the addition of the flag “m” following their initial classification.
The introduction of the « m » flag has a double purpose: first to allow for attributing a primary use (residential/non residential typologies), and secondly to create a third class of buildings for which there is apparent mixed use as perceived by compilers (and without setting a minimal criteria i.e. in terms of surface). In the future this could be better codified with a separate variable, which could be combined with the basic typology of the primary use in different ways. Finally it is important to avoid that the “mixed use” category becomes an excuse for not declaring the primary/main use of the building. An additional section was also introduced for building under construction, as they cannot be considered as residential or non-residential regardless of their end use. This section does not include buildings under maintenance or renovations
Values:
Table 2 – Type of building classification (extracted from Eurostat) – modifications brought by this project are underlined
Section Division Group Class
1 BUILDINGS
11 Residential
buildings
110 1100 Unknown
111 One-dwelling buildings 1110 One-dwelling buildings 112 Two- and more dwelling buildings 1121 Two-dwelling buildings 1122 Three- and more dwelling buildings 113 Residences for communities 1130 Residences for communities Non-residential buildings
120 1200 Unknown
121 Hotels and similar buildings 1211 Hotel buildings 1212 Other short-stay accommodation buildings 122 Office buildings 1220 Office buildings 123 Wholesale and retail trade buildings 1230 Wholesale and retail trade buildings 124 Traffic and communication buildings 1241 Communication buildings, stations, terminals and associated buildings 1242 Garage buildings 125 Industrial buildings and warehouses 1251 Industrial buildings 1252 Reservoirs, silos and warehouses 126 Public entertainment, education, hospital or institutional care buildings 1261 Public entertainment buildings 1262 Museums and libraries 1263 School, university and research buildings 1264 Hospital or institutional care buildings 1265 Sports halls 127 Other non-residential buildings 1271 Non-residential farm buildings 1272 Buildings used as places of worship and for religious activities 1273 Historic or protected monuments 12 74 Other buildings not elsewhere classified
3.BUILDING UNDER CONSTRUCTION4
4 It does not include buildings under maintenance or renovations More details about the type of building can be found in Annex A.
Guidelines for classifications The classification uses a decimal system and includes:
However, as mentioned above, if the main use of the building is a school, then the building is classified to 1263m.
The definitions for the number of floors are based on a deep analysis of existing definitions in the fire statistics of various countries - see Annex C2 in Task 4 report (EUFireStat, 2022). Considering the information provided by ISO/TS 17755:2014 (ISO, 2020), the number of floors could be recorded in terms of floors or height. In the ISO/TS 17755-2:2020(E) (ISO, 2020), the height of the building is defined as the ‘distance between the floor of the ground floor used by firefighters and fire engines and the floor of the highest level used by people of the building’. However, the floor height of the building can vary in different property types, especially with public and industrial buildings. Therefore, the floor is defined as the distance between the pavement and the ceiling of one floor.
Moreover, in the statistics of the UK (England) and the USA, the number of floors (or stories) is grouped into floors above or below ground level (grade level), respectively. Based on the above investigations and after deep discussions amongst the consortium members, the number of floors has been defined and subdivided into those above and below ground level referred to as the main entrance of the building. Finally, the number of floors has been determined as an easier variable to record than the height of the building.
Definition:
The number of floors above is the numerical value to capture the number of floors above and including the ground level. The number of floors below is the numerical value to capture the number of floors below and excluding the ground level.
Note to definition: The floor is defined as the distance from the pavement to the ceiling of one floor. The ground level is referred to the level of the main entrance of the building.
Value:
Numerical value for floors above5 Numerical value for floors below6
Examples:
The building has only a basement, a ground level and 2 floors.
Number of floors above is 3
Number of floors below is 1. The building has a ground level, 10 floors and 2 floors of underground parking:
Number of floors above is 11
Number of floors below is 2.
Definition:
Fire Safety Measures are devices and systems that aim at reducing the effects of a fire. They can be detecting (smoke, fire etc.) and alarming (local, central etc.) and/or preventing fire spread (sprinklers, automatic extinguishing equipment, compartmentation etc.) or any combination of those. 5To be approximated when unknown. 6 To be approximated when unknown
Value:
Were there Fire Safety Measures present? Yes / No If yes, what kind?
The proposed definition and values are based on a deep analysis of existing definitions and values in the fire statistics of various countries. The proposed values are grouped into seven groups, following the practice in USA and Australia. The selected values take into consideration the already existing values presented in the ISO/TR 17755:2014 standard (ISO, 2014), along with the existing values in several European countries. Some of the proposed values have examples attached to it, for a better understanding.
Definition:
Area of origin is the localized area where the fire started.
Values7:
7 a coding structure for these values can be proposed in a later stage of the project
o Processing or manufacturing area
o First-aid area
o Stage/Scene
o Other (write a value)
Note 3: Examples of operating areas: Computer room, Laboratory, Machine room, Projection room.
Note 4: Examples of processing or manufacturing areas: Workshop, Painting room, Drying room.
Note 5: Examples of storage areas: Closet, Tool or supply area.
Note 6: Examples of machinery areas: Server area, Technical area.
Note 7: Examples of producing/distribution area: Power house/plant/generator, Electrical distribution, Air conditioning and ventilation room, Filter room.
Note 8: Examples of ducts: Ducts, Chimney, Ventilation duct, Water duct.
Note 9: Examples of heating areas: Boiler room, Remote heat transfer station.
Note 10: Examples of shafts: Elevator shaft, Supply and disposal shaft.
Note 11: Examples of substructure areas: Cavities in ceiling, Cavity between floors, Other cavity.
Definition:
The initial fuel of the fire – the first item that had sufficient volume or heat intensity to extend to uncontrolled and self-supporting combustion.
Note:
The values for this variable must be at a level of detail which a fire officer is able to identify, hence the use of the word “item”. It is sufficient to know the item at a general level without going into more detail about the item’s structure or the material it is made from. In items powered by electricity, an ignition can occur in the casing or heat insulation. The heat for this ignition may be internal (through an electrical fault in the equipment itself) or external. We can illustrate this with two examples of fires involving a coffee maker as the item first ignited. In the first case a fault in the coffee maker ignites the plastic casing and the coffee maker is both the heat source and item first ignited. In the second example, someone carelessly leaves the coffee maker on a hot plate of a freestanding cooker, which is then turned on by accident.
The coffee maker in the latter case is the item first ignited, but the freestanding cooker is the heat source.
Values:
Definition:
Any specific article assessed by the fire officer or fire investigator to have had a significant contribution to the development of the fire beyond the item first ignited.
Note:
This variable is only relevant if the fire spread from the item first ignited. We use the term “responsible” to indicate that we are focused on the article making the greatest contribution to fire development. In most cases, several distinct articles will have contributed significantly to the fire development as a fire progresses from the item first ignited towards full surface involvement in the area of origin and then further in the building. If so, it is virtually impossible for an untrained fire officer to identify which article was most responsible for fire development. It is often a major challenge for a trained and experienced fire investigator to inspect smoke and burn patterns in the area of origin and then make a judgement on which of all the fuel sources which had been present in the area of origin that made a significant contribution to the development of the fire.
It is therefore unreasonable to expect reliable data on this from a fire officer other than in the relatively small proportion of all building fires that spread from the item first ignited, but without extensive damage to the Area of origin, in which case the data collected will not provide a significant gain in our knowledge of fire development.
Values (multiple choices allowed):
Definition:
The source of energy that initiates combustion in the item first ignited.
Notes:
Heat source corresponds to source of ignition in the survey.
The values for this variable must be at a level of detail which a fire officer is able to identify. A specialist fire investigator may be able to examine the item in more detail to discover the component in the item which produced the heat and how this failure occurred, but it is unreasonable to demand this from a fire officer. According to the NFPA Guide for Fire and Explosion Investigations (NFPA 921), the combustion reaction taking place in a fire can be characterized by four components: the fuel, the oxidizing agent, the heat, and the uninhibited chemical chain reaction which makes the combustion self-supporting. This is commonly referred to as the “fire tetrahedron”. One of the most important goals of pan-European statistics is to describe how ignition occurs at fires attended by the fire service.
A reasonably accurate understanding in quantitative terms of how fires start is necessary for well-informed fire prevention activities. Concerning the fire tetrahedron, we do not need statistics on the oxidizing agent as in all relevant cases this can be assumed to be oxygen in the surrounding air. However, we do need information on the first fuel item with enough energy to allow self-supporting combustion and the source of the heat which ignited it. Item first ignited and heat source will allow us to understand how ignition occurred, but it is not in itself sufficient for fire prevention purposes – we also need to know why the item first ignited was exposed to the heat source for long enough for ignition to occur, as described in primary causal factor and intent.
Values:
Definition:
The general causal factor that the fire officer assesses to have been the most important in explaining why the item first ignited was exposed to the heat source in a way that led to an uncontrolled combustion.
Note:
In many cases more than one of the alternatives will have played a part in the ignition event. However, it should be possible for the fire officer to assess which of these factors was of greatest importance. It is this information that is most important for fire prevention work. The word “primary” is used to signal that it is the most important of the three factors that should be recorded. The term "causal factor" is proposed instead of “fire cause" because the direct fire cause is already clear: the item first ignited has been exposed to the heat source for long enough for ignition to occur. If the causal factor is recorded as human act or omission then it is most important to know whether the damage caused by the fire was intentional or unintentional, as there are completely different prevention strategies for these two types of fire.
Values:
The following provides a description of the steps involved in fire data collection from collection at the scene of the incident to reporting at the EU level.
All fire data collection systems researched during this study start with incident reports filled out by the attending fire department. These data are the backbone of the data collection system. Other agencies may also be present at the incident scene, including police and ambulance services, while fire investigators, police and representatives from insurance companies may collect information in the days following the incident. Each of these can provide additional information of importance to the completeness of the data set such as number of fatalities, age of fatalities, Primary causal factor, heat source and article contributing to fire spread. Fire departments may collect information in addition to the data required at the EU level. This outline is not intended to restrict the data of interest at the local or even national level.
It is recommended that data are collected and entered into the local system within 24 hours of an incident. An example of a data collection sheet is presented in annex B.
This step takes place at the local fire departments in the national data collection systems reviewed in this project. Data collected at the scene of the incident is transferred to the local data management system. The data system should allow data to be recorded as unknown but to also allow for data to be updated when additional information is obtained. The first quality control of the data should occur at this step. If data are missing or entered as unknown at the item level it is proposed to consult the following data sources to obtain the required information:
Table 3 Complementary data sources for missing or unknown variables
| Variable | Examples of complementary data sources |
|---|---|
| Incident location, incident date, incident time | Local dispatch system / records |
| Number of fatalities, number of injuries, age of fatalities | Police, ambulance and hospital records |
| Type of building, number of floors | Building registration and/or property tax records |
| Item first ignited, area of origin, heat source, primary causal factor, fire safety measures present, article contributing to fire development | Investigation reports from fire department, police and/or insurance company |
It is recommended that local data management is continuous so that incident data are reviewed and updated, when necessary, with information from other data sources as these become available.
If the local data management system is not aligned with the requirements for national data, the local data should be transposed into the format required for national data. An example of this could be that information about the fire department response (number of firefighters, engines, etc.) is included in the local dataset but not required at the national level. A qualitative estimate of uncertainty of the values reported as outlined in Section 7 should be provided including any steps taken to minimize uncertainty such as consulting other data sets as mentioned under step 2. The local dataset is then sent to the national agency responsible for collecting fire incident data. In some countries, there may be intervening layers of administrative authority which receive local data before transferring it to the national level.
In other countries data might be submitted directly to the National system in step 1 and the quality control mentioned in Step 2 is then happening at the national level. It is recommended that this reporting of data to the national system takes place at least once a year at a time schedule defined by the national agency.
It is proposed to have one national point of contact in each member state responsible for collecting the data from local fire departments, e.g. a national statistics institute. Task 1 report shows that the type of agency used nationally differ significantly among member states (EUFireStat, 2022). In some countries, this responsibility falls under a ministry and in others may be the responsibility of either the state fire service, fire protection associations or insurance companies. Only Italy appears to use a national statistics centre for fire incident data. A second round of quality control should take place at national level focusing on unit-level missing data. Fire departments may fail to report fires due to time or resource constraints, including budgetary or personnel limitations. Some types of fires may also go routinely unreported because they are not considered meaningful.
Such missing data may lead to an underreporting in the true number of fires - see Task 3 report (EUFireStat, 2022). Most member states do not seem to have a national approach to missing data at the unit level and instead appear to assume they have a full census response without apparent verification. Very different approaches were observed in those countries that seek to address the problem of missing data. Sweden identifies all incidents by having the local data collection system send a message to the national system when a report is initiated. Every month the Swedish Civil Contingencies Agency (MSB) sends feedback to fire departments on reports that have been initiated but remain uncompleted in the national dataset. France uses a weighted average to deal with unit-level missing data in the number of fire interventions reported by fire departments in different regions.
When data on fire interventions are not reported by a fire department, the Ministry of Interior calculates the weighted average number of interventions by fire departments protecting populations of similar size and applies that number to the data for the nonreporting fire department. To ensure that data can be shared between EU Member States, it is necessary to ensure that missing data at the unit level is dealt with appropriately by all reporting agencies. Another challenge when managing data at the national level is item level missing data not dealt with at the local level as well as variables coded as unknown. While various methodologies exist to deal with this, as described in Task 0 report (EUFireStat, 2022), it is recommended to maintain the unknown/missing data fields as such and not attempt to distribute them among the other data fields for the variable.
This allows data analysts to use their preferred methodology for missing data/unknowns when analysing the data set. A final step in the national data management is to calculate the totals for each of the variables as well as provide an overall qualitative estimate of the uncertainty for each of the variables recorded. Finally, we wish to emphasize that the scope of this project is confined to recommendations for collection of national data on fire incidents that occur in buildings. During the review process of this report prior to publication, it was suggested that "near miss" fires can provide important information for fire safety efforts.
While we agree that near miss incidents can be instructive, attempting to propose how data from these fires -- many or most of which almost certainly go unreported -- could be collected at the national level falls outside the mandate for this project. During the course of the current project, we encountered multiple ways in which actual fire events are defined by different data collection systems. The report also emphasizes that it can be difficult to determine the completeness of fire datasets due to unreported fires and the complications that missing data may pose for the reliability of data and the accuracy of subsequent conclusions. Those methodological issues would be compounded in any effort to collect near-miss data.
Accordingly, while we concur that the pursuit of near-miss data has valuable potential for fire safety, we believe that efforts to identify feasible methods for collecting this data are a matter for separate and future research.
If the national dataset does not correspond directly to the harmonised fire incident dataset, transformation rules will be needed. These transformation rules will depend on the national variables collected, definitions used for each of those and how they correspond to the harmonised variables. Recommendations on how to potentially transform data into the harmonized data set is discussed for each of the variables. The entire database as well as the calculated national totals are then reported to a body that is tasked with dealing with fire incident data at EU level. It is recommended that this reporting of data is done annually, and that each member state provide data from the previous calendar year. An illustration of the above-mentioned steps is presented in Figure 4.
Figure 4 Description of the steps involved in fire data collection
Number of fatalities
Collecting the number of fatalities is a quite essential part of fire statistics. The most common challenge concerning the collection of the number of fatalities seems to be they do not necessarily occur at the fire scene. With the rapid intervention of the fire services and the emergency medical services, many fatalities can occur at the hospital days after the fire incident due to severe injuries. If a body is found within the incident scene it is most likely the firefighters who found it. However, the number of fire fatalities should also include persons injured by the fire and that died beyond the fire scene, such as during transportation or at the hospital. This means that data from the medical coroner might be necessary.
For special cases where a person was already dead before the fire started, then it should not be considered as a fire fatality, but this will most likely be determined by the police investigators. We recommend that the number of fatalities entered in the incident report shall be the definite number of bodies declared dead at the incident scene. This number can always be updated if any information from other organizations is given later on. Fire fatalities as estimated by fire departments will be an underestimate for two reasons – the fire department may not know about fatalities that occur once the victim has left the scene of the fire or know about fatalities from fires that they do not attend. It is therefore appropriate that fatality and injury statistics be cross-checked with morbidity statistics, where most European countries use International Classification of Diseases (ICD10).
Although, medical practice is standardised, one must keep in mind that the longer time that passes between a fire and the fatality at the hospital, the more likely it is that burn injuries are not linked to the original fire and the more likely it is for a patient to die of complications from the original injury while under treatment. We deliberately set the time limit to consider a fire fatality to one year after the fire occurrence, which is one of the longest times identified and used in the USA. From CTIF’s experience, 99-100% of all fire fatalities are covered in the 90 days following the fire incident. Of course, this estimation can vary from country to country and with the progress of medical care, but it implies that if a country is able to collect data up to 90 days following the fire, then it would also be acceptable.
Number of fire injuries
Collecting the number of injuries is an essential part of any fire statistics. With this kind of data, it is possible to quantify the impact of a fire. One of the major challenges identified by the project is the fact, that people might not introduce themselves to the Rescue Service at the scene. As the kind of injuries can vary a lot, the threshold for counting injuries may vary significantly from country to country. It is therefore important to remain consistent. Given these challenges we recommend including the number of injured people who presented themselves at the incident scene to the Rescue Service. This decision assumes that the initial data will be collected by fire officers returning from the fire scene and then collected by the national fire authority. The communication between Rescue Service and Fire Officers at the scene is usually established.
This number can be updated if additional information by medical data is given later on.
Age of fatalities
The age of persons who are killed in fire incidents is important information for fire prevention planning and prevention. In situations where the precise age of a fire fatality is not available through records or corroboration, we recommend that a best estimate of fatality be entered into the incident report, a practice that is utilized in National Fire Incident Reporting System in the United States. An alternative approach is to utilize age ranges when precise age information is unknown. The age categories will differentiate children, adults, and the elderly, which will be sufficient for some analyses. However, we believe that providing numerical estimates is a superior approach because this number can always be reordered into age categories, but it is more difficult to translate age categories into specific age estimates and a great deal of precision will be lost in the process.
Specific age is especially critical in the case of younger and older fire fatalities, where capabilities can vary greatly between the youngest and oldest in their respective categories. As described in the data journey, age estimates can be changed at a later point following local quality control and consultation with other data sets for documentation of the age of fatalities. In countries where age data is only available in categories, the data can be transformed by selecting the mid-point of the data range (e.g., for fatalities in age category 30-59, the midpoint (45) would be selected).
Type of buildings
This variable may also be difficult to assess due to the large number of possible values and the potential for complex building types. The critical information for analysis purposes is whether the fire occurred in a residential building, a non-residential building, or a mixed-use building. As for the detailed type of building, we recommend the persons filling the report to use their best estimate. As described in the data journey, it can be changed at a later point following local quality control and consultation with other data sets. Number of floors The number of floors in buildings affected by fire incidents is an important information for the optimization of fire safety strategies and evacuation plans. It also provides additional characteristics to the property types and an estimate of the height of the building.
It would be suggested to record the number of floors rather than the height of the building to facilitate and simply the collection of this variable in the aftermath of a fire incident. The floor of the building is defined as the distance between the pavement and the ceiling of one floor. Moreover, in the statistics of the UK (England) and the USA, the number of floors (or stories) is grouped into floors above or below ground level (grade level), respectively. Therefore, it would be suggested to subdivide this variable into two: number of floors above and number of floors below the ground level recorded as numerical values. It is important to specify that the ground level is referred to the level of the main entrance of the building.
For example, if a building is composed of a basement, a ground level and 2 floors, the number of floors above should be recorded as 3 and the number of floors below as 1. However, while a numerical estimate is considered to be a superior approach that can always be reordered into floor categories, it is more difficult to translate floor categories into specific floor estimates and a great deal of precision will be lost in the process. In countries where floor data are only available in categories, the data can be transformed by selecting the mid-point of the data range (e.g., for the category 1-3 floors, the midpoint (2) would be selected). At the same time, in countries where the height of the building is recorded in meters, the value can be transformed into number of floors applying an approximate estimate of 3 meters per each floor.
The number of floors above and below the ground level are fire statistical variables that can be collected at the fire scene or during the first quality control of the data that occurred during the local data management phase. In situations where the precise number of floors is not available through records or corroboration, the best estimate is suggested to be inserted into the incident report. Other sources could also be investigated able to provide the required information such as building stock or energy consumption survey. An alternative approach is to utilize floor ranges when the precise number of floors is unknown. The provided values for the number of floors will be able to differentiate low, medium and high-rise buildings covering very important aspects that are necessary in some analyses and during the fire investigation.
Primary causal factor
Primary causal factor refers to the factor that is the most important influence in an ignition event. The terminology “causal factor” is preferred to “fire cause” because the direct cause of a fire is already clear: an item first ignited has been exposed to a heat source long enough for ignition to occur. We cite the following example for illustration: a cigarette (heat source) can ignite a paper (item first ignited) due to an unintentional human act (primary causal factor) and then spread the fire via curtains (materials contributing to fire development). Information about the primary causal factor of fires can be useful for researchers and safety authorities in guiding prevention efforts, including educational campaigns, requirements for automatic extinguishing equipment, and other potential interventions. However, information on fire cause has oftentimes been difficult to capture in practice.
In the United States, for instance, there have been long standing concerns with unknown or missing data on fire cause in the National Fire Incident Reporting System. The problem is seen to be especially prevalent for serious fires or those which involve fatalities. The values for this variable are:
Human act or omission (Intentional, Unintentional, Undetermined intent)
Equipment failure
Natural phenomenon
Undetermined If the primary causal factor is recorded as a human act or omission, it is most important to know whether the ignition was intentional or unintentional since they are associated with different prevention strategies. Research has identified several factors that help explain the failure to enter information on cause.
Causal information that is initially coded as “under investigation” is not updated after an investigation is completed.
Concerns about liability in some cases deter fire department reporting of cause. Methods to address this concern include providing an option to indicate a level of uncertainty about causal determination, as well as providing immunity from liability for persons or entities who report fire incident data in good faith and without malice.
Lack of program administrators outside the fire department who can work with fire departments and perform quality control oversight and help update incident reports. We therefore recommend that respondents provide the best information on the primary causal factor of a fire that is available at the time a report is initiated or submitted. Estimates of primary causal factor is based on experience or current knowledge, such as concluding that a fire originating in the kitchen is primarily due to a human act or omission, provides more meaningful information than no information at all. When performing quality control during local data management as described in the fire data journey, effort should be made to consult other data sets that provide information for data with high uncertainty during the initial collection. We also recognize that record keepers may be reluctant to provide information while fire cause is still under investigation. Efforts should be made to update causal information as it becomes available.
National fire data collection programs can assist fire departments with data management requirements by providing program managers in overseeing quality control efforts.
In a broad sense uncertainty relates to a situation when there is a doubt about the validity of values recorded for a particular variable. An overview of concepts associated with uncertainty as well as a discussion on the uncertainties associated with different data collection techniques are presented in the Task 3 report (EUFireStat, 2022). A total of fourteen variables have been suggested for inclusion in harmonized European fire statistics. Below we present a qualitative assessment of possible uncertainty issues connected to these variables.
Number of fatalities
As seen in studies in both Sweden (MSB, n.d.) and France (Belanger et al., 2008; Carlotti et al., 2017; Lasbeur et al., 2012; Lasbeur & Thélot, 2014), underreporting of fatalities can be as high as 20-30%. Underreporting may occur when a victim dies after transport from the fire scene or because the fire service is never called to the scene, or for some other reason. Studies of hospital records can help in determining the actual number of fire fatalities, but automated solutions can be complicated since personal information (like social security number) is seldom recorded by the fire service. Even so, the best data quality check is to regularly perform specific studies which compare hospital records of fire fatalities with the outcomes from fire statistics.
Indeed, medical data (for example those based on International Classification of Diseases (ICD10) should be cross-referenced with other sources to find an agreement. Similar initiatives for injury related mortalities have been examined across Europe (Belanger et al., 2008). In Sweden, MSB hopes to follow up all reported fatal fires by collecting supplementary information from the Police and the Board of Forensic Medicine. Number of injuries Fire injuries are even more difficult to systematically record than fatalities. It is likely that the fire service will keep track of how many people they rescue (Runefors, 2020), but people with injuries might evacuate by themselves or with the assistance of others than the fire service. Although firefighters will be able to collect some injury data on the scene, they might not be competent to evaluate injury severity, complicating data quality.
In cases where people are transported by ambulance from the scene, the data can be used to perform specific studies of the accuracy of fire service reports of injuries, as in the case of fire fatalities referenced above. However, there are likely to be situations in which injury victims may be transported by family or friends before fire service arrival. Complications in recording injury is illustrated by comparing France and Italy, which are similar with respect to populations, building methods and fatalities per 100.000 inhabitants, but which have completely different outcomes for injuries (around 1 fire injury per 100 000 inhabitants in Italy and around 20 fire injuries per 100 000 inhabitants in France), leading to a doubt about the difference in the definitions.
High level comparisons between countries could be an appropriate tool for this variable that can help identifying major discrepancies.
Primary causal factor
Primary causal factor is likely to be prone both to measurement and response errors, since confusion, ignorance, or carelessness of the reporter might result in faulty inputs. Another complicating factor is that evidence at the scene may have been destroyed by the fire. The reporter may also feel an uncertainty or unease when assigning the primary causal factor, which results in assigning it as unknown. In NFPA analyses of NFIRS data, the unknown fires are distributed in the same proportion as the fires for which the data are known (Ahrens et al., 2003). However, this might lead to model assumption errors. When detailed fire investigations occur, they can be used to update the primary causal factor first assigned, and thus improve the accuracy. Even so, the destructive nature of fires can result in it being impossible to determine the primary causal factor. Type of building This variable may also be prone to measurement errors due to confusion or ignorance.
As an example, there are different views in different countries regarding what is included in the term “residential building”. Holiday homes are considered residential in some countries but not in others. A category like “public building” might also be interpreted differently in different countries. Clear definitions and instructions to the reporter are needed on how to interpret the variable and the different categories. An additional possibility is that building information can be double-checked with real estate information or records at a municipal level. Incident location If the fire incident reporting is connected to a dispatch system where the location is recorded (address and/or coordinates), the uncertainties of incident location can be reduced.
However, there might be problems with measurement errors (faulty inputs by dispatcher or reporter at the scene) and there might also be non-responses (for example address missing in the report). Possible errors can be reduced if both address and Global Positioning System (GPS) coordinates are reported, as seen in a study where fatal fires in Sweden were connected to real estate information by utilizing both information on address and coordinate (Johansson, n.d.). In some cases, address information was lacking and data on coordinates could be used, in other cases the coordinates were wrong, and the address could be used. Incident date If incident reporting is connected to a dispatch system where the time and date for call received, unit dispatched and unit at fire scene are automatically recorded, the uncertainties regarding this variable are considered small.
If the variable is entered manually, it will potentially be prone to measurement and response errors (incidents that occur close to midnight will likely be most affected). Systematic errors may also occur but are most often likely to be random in nature. Incident time The uncertainty connected to incident time is considered to be small if the time and date is collected and recorded automatically. Errors are more likely if incident time is recorded manually, but errors are again likely to be random. In cases where incident time is recorded as a rough estimate of the time (e.g., night, morning, noon, afternoon, evening) the error will most likely be small.
Age of fatalities
There are a number of uncertainties regarding the ability of fire service to record the age of a victim at a fire scene. For example, there may be no one at the scene to attest to the age of the deceased in the event of a fatal fire. Age information will be available in other databases if the victim has been hospitalised or is deceased. Cross references to such databases can be made in order to quantify and evaluate the information in the fire service database. Number of floors The number of floors in a building should be quite straightforward to report if the variable is well defined and understood by the reporter. It must be clear for the reporter how to interpret basement floors, attic floors, mezzanine floors and ground level for uneven floors. Studies of the accuracy of this variable can be done by studying documentation and images of fire-exposed buildings.
Area of origin
The area of origin will most likely be associated with less uncertainty than fire cause. If the building is still standing or there are some cues based on eyewitness information, the area of origin should be easy to determine. Still, distinct categories are necessary to avoid systematic errors. As an example, a category labelled as “storage” could be interpreted as designated storage room or as a room used for storage in a basement. The latter can be confusing if “basement” is itself a possible category.
Heat source and Item first ignited Similar to primary causal factor of fire, it might be difficult to determine the heat source and item first ignited due to the destructive potential of the fire. The reporter might need to rely on a fire investigation or second-hand data, such as information from residents or other eyewitnesses if the fire itself has destroyed cues to the heat source and the item first ignited. The category “unknown” might cause issues with heat source for similar reasons as primary causal factor of fire (see above). Problems can also arise if the reporter is confused or unable able to distinguish between item first ignited and the heat source. Article contributing to fire development The uncertainty connected to this variable is considered similar to primary causal factor of fire and heat source.
As long as the fire is kept in the area of origin, the damage will most likely not be too severe to be able to determine the article contributing to fire development. However, there might be situations when relevant knowledge in fire development and fire dynamics is required by the reporter in order to accurately categorise this variable. Fire safety measures present There are two distinctions to be made regarding this variable. Firstly, the possibility to determine if a fire safety measure was present or not is considered to be good if the building is still standing and the fire scene can be inspected. The uncertainty in that regard is therefore considered to be low, and it can even be reduced further if fire safety documentation of the building can be studied to complement the information retrieved at the fire scene.
Secondly, if the fire safety measures present were working or not is more difficult to determine, and it also related to the type of fire safety measure. For example, it is hard to determine if a fire alarm has sounded in the initial stages of a fire if there is no eye-witness information, but it might be easier for the fire service to see if a fire door has performed as it should or not at the fire scene. Summary of uncertainties connected to the described variables Based on the description and discussion above, we can make some general estimates of the uncertainty associated with the different variable assigned in
Table 4. The estimates are rough qualitative estimates that indicate which variables can be
expected to be affected by the largest degree of uncertainty. It should also be stressed that these uncertainties can be reduced by applying different measures, like the measures discussed above.
Table 4: Estimated associated uncertainties with the selected variables
Variable Estimated associated uncertainty
Number of fatalities
Medium Number of injuries High Type of building Medium Incident location Low Incident date Low Incident time Low
Age of fatalities
High
Number of floors
Low Area of origin Low Item first ignited High Fire safety measures present Medium
Heat source
High Article contributing to fire development Medium
Primary causal factor
High
Once agreement is reached on the fire data variables and data collection methods, it is important to provide guidance for their proper interpretation. Overall, decision-making based on fire statistics should include the following steps:
Identify trends indicated by the statistical outcomes
Verify whether this trend is true (evaluate the potential for random variation, uncertainty, and other forms of interference)
Identify the reasons for the trend by connecting the variable to other relevant variables
Compare this observation with trends in other countries
Investigate possible causes of the trends
Discuss possible prevention measures (comparison with other regions or countries)
Establish appropriate actions and interventions
Here we provide examples of data interpretation and the type of deliberation involved using variables proposed for inclusion in fire data collection.
Number of fatalities
Data on fire fatalities is important for monitoring progress in fire safety. The identification of potential trends in fatality data (see Figure 4, data from the Netherlands) can be useful for planning safety interventions. However, it is difficult to establish trends due to random statistical variation that can be important. The experience of the consortium suggests that it is necessary to conduct observations over a period of eight to ten years, with no changes in methodology (definition, collection, process etc.) or outlier fire incidents, before establishing trends. It is also crucial to assess variables in relation to other variables in order to increase confidence that a trend is true. This suggestion is applicable in general terms to all guidance provided in this chapter.
Figure 5. Number of fatal residential fires (blue) and fatalities (red) in the Netherlands.
Number of injuries The estimation of fire injuries is necessary when making policies that attempt to reduce injuries and ensure adequate resources to accommodate injury victims. Examples of changing circumstances that might influence injury trends are changes in building methods and materials or changes in residential populations (such as populations with physical limitations (mobility, sight, hearing etc.). If the number of injuries is consistent over time, then the evolution of this variable should be compared with other variables, such as number of fires, number of fire fatalities, articles contributing to fire development, type of buildings, or other relevant variables.
Age of fatalities
With an aging population in Europe, data on the age of fire fatalities assumes greater importance. Physical and cognitive impairments can complicate the ability to escape a fire. The data from the Netherlands in Figure 6 shows that occupants in the age groups older than 61 years old, constitute approximately 50% of all fatalities.
Figure 6. Age of fire fatalities over the years 2008-2017 in the Netherlands.
Data on fatal residential fires has been collected in the Netherlands since 2008. Based upon the number of deaths per year and the age of these fatalities, prevention policies in the Netherlands are strengthened by means of:
Academy, 2018) The chart provides an indication regarding the age groups that are impacted by fire fatalities in a number of European countries. Compared to the size of the population (gray line), children aged 0-14 seem to be less impacted than other age groups, especially those above 65. As Europe’s aging population is expected to peak in the year 2040, it is important to begin formulating fire safety policies in anticipation of this development. This example is purely for illustrative purposes, as current fire statistics are not harmonised, and data of the countries and the average distribution relate to different years. Incident location This variable is important in order to be able to generate a mapping of incident location within a country or a region and to correlate the distribution of fire incidents in rural or urban areas.
It can also be instructive to map fire location in relation to other data, such as socio-economic or fire intervention data, such as identifying geographical areas that are not well covered due to limited resources for the fire rescue services. Information on fire location might be complicated by restrictions in providing exact address and/or GPS coordinates of fire incident if they are not well aggregated due to data protection regulations. Accordingly, the location criteria will have to be considered and possible solutions could be providing only address of the county or municipality. Time and date of the incident Data on the time and date of building fires provides basic information for planning safety interventions, assist fire department preparation and response capabilities.
Table 5 uses Spanish statistics on residential fires with fatalities to show the distribution of fire
deaths during day- and night-time hours (Fundacion MAPFRE & APTB, 2020).
Table 5 Distribution of residential fire fatalities during the day and night, extracted
from (Fundacion MAPFRE & APTB, 2020) Time of the fire Number of fatalities % Night 91 55 Day 71 43 Unknown 3 2 Total 165 100 As defined by (Fundacion MAPFRE & APTB, 2020), day fires occurred in the hours from 8:00 am to 8:00 pm while night fires occurred between 8 pm to 8 am. When looking into more details (Table 6 and Figure 8), we notice that a higher percentage of fire fatalities occur between 4 am and 8 am. Therefore, the notion of granularity of data is very important.
Table 6 Distribution of residential fire fatalities by hour ranges, extracted from (Fundacion MAPFRE & APTB, 2020)
Hour ranges Number of fatalities % 0 to 4 27 16.4 4 to 8 40 24.2 8 to 12 24 14.5 12 to 16 19 11.5 16 to 20 28 17.0 20 to 24 24 14.5 Unknown 3 1.8 Total 165 100
Figure 8 Distribution of residential fire fatalities by hour ranges, extracted from (Fundacion MAPFRE & APTB, 2020)
Statistics from Spain also (Fundacion MAPFRE & APTB, 2020) illustrate the distribution of fire fatalities over the twelve months of the year (see Figure 6). It appears that the coldest months of the year account for the highest number of fire deaths. Fundacion MAPFRE & APTB estimates that the need to generate heat in cold weather months leads to more fires and more deaths. Such observations should therefore be validated by comparing with other variables, such as primary causal factor and heat source, to determine the necessary mitigation solutions and policy measures.
Figure 9 Distribution of residential fire fatalities over the year, extracted from (Fundacion MAPFRE & APTB, 2020)
0,0 5,0 10,0 15,0 20,0 25,0 30,0 0 to 4 4 to 8 8 to 12 12 to 16 16 to 20 20 to 24 Unknown
Number of fatalities (%)
Hour ranges Number of fatalitites Month of the year Building type In this section, we present several hypothetical examples to illustrate the type of output that can be generated through data collection. Note that the trends are exaggerated for purpose to facilitate the interpretation.
Note: These statistics were inspired from existing data extracted from a EU Member State, where the labels were modified as some of the existing building classifications were different.
Figure 10 illustrates the number of fires by type of building over a one-year period. As fires
primarily occur in residential (one dwelling and residences for communities) followed by mixeduse buildings, it appears necessary to cross-reference these values with others. For instance, if we compare these data to fatality data (Figure 11) by type of building, we might notice that the deadliest fires occur in mixed residences for communities and in one dwelling residential buildings. This implies that it will be important to focus on prevention of these types of buildings and to verify the safety measures that are present as well as if the applicable regulation. Examining the time of the fire for the different type of buildings could provide indications such as if most fires occur in residential areas during the day and if they result in fewer fire deaths, while night fires could be less frequent but more fatal.
It is also important to examine mixeduse buildings in relation to the fire and items first ignited to understand if fires occurred in residential or non-residential areas or influenced by residential versus non-residential factors. It is also important to mention the utility using Eurostat typology of buildings for deriving incidence indicators, in particular for setting appropriate comparisons between type of buildings. If residential buildings are 10 or 50 times more numerous than the mixed ones, the incidence of fires in the latter could become ten times higher than in residential. Such proportions could be very informative for cost/benefit computations and to decide how/where to allocate resources for prevention measures.
Figure 10 Hypothetical example of number of fires by type of building in one year (other type of building are not shown as they
constitute less than 1% of the total number) 20% 34% 3% 10% 13% 20% One-dwelling buildings (111) Residences for communities (113) Hotels and similar buildings (121) Public entertainment, education, hospital or institutional care buildings (126) Mixed: Residences for communities (113m) Mixed: Public entertainment, education, hospital or institutional care buildings (126m)
Figure 11 Hypothetical example of number of fire fatalities by type of building in one year (other type of building are not shown
as they constitute less than 1% of the total number) Heat source and Primary causal factor Harmonized data on building fires will allow interested parties to examine how often certain kinds of fires occur. The statistics will facilitate the identification of the most common ways that ignition occurs (item first ignited and heat source), and why an item was exposed to heat long enough for ignition to occur (primary causal factor). Such analysis may help to illustrate different dominant fire scenarios for fires with different outcomes, such as fatal fires, fires with nonfatal injuries, and fires that only damage property or the environment). Analysis could also be applied to compare fires in different types of buildings, at different time periods (day, week, or month), and in different regions or countries. If data is collected over a period of several years, it may be possible to observe changes in the frequency of certain types of fires.
Harmonized data may enable the identification of trends that only become apparent at the broad European level due to the effect of random variation with smaller sample size at the national level. Statistical data from fire incidents could be used to inform prevention actions. If specific interventions are implemented in a country or region, it should be easier to determine their effectiveness since outcomes are commonly evaluated on a before/after comparison. However, it is often difficult to take account of the influence of other external factors during the period of study. European-level statistics will enable a more reliable evaluation since the outcome in the country/region can be compared to the fire scenario frequency in other European countries for the same period.
Fire safety Measures in place 34% 24% 36% 6% One-dwelling buildings (111) Residences for communities (113) Mixed: Residences for communities (113m) Mixed: Public entertainment, education, hospital or institutional care buildings (126m) Data on smoke detectors in residential buildings can provide information about the distribution and effectiveness of smoke detectors in residences and be used in considering fire regulation requirements and enforcement for smoke detectors in certain settings. For instance, smoke detectors might be missing in certain local areas even if it is required by local regulation. This can lead to better awareness and other measures to help to improve the distribution, installation, maintenance, and effectiveness of smoke detectors in residential buildings. Another striking example is the use of automatic fire alarm systems in industrial buildings.
Recent statistics from Germany show that the average amount of damage from fires in industrial facilities without automatic fire alarm systems was 850,000 euros per fire incident. Of the companies affected, 60% were no longer viable after 100 days, despite claims settlement by the insurance company. In the case of fires in industrial buildings equipped with automatic fire alarm systems, the average loss amount was 18,000 euros, and only 20 percent of companies affected were no longer viable after 100 days. Fire characteristics - A Case study from the UK
Figure 12 shows the area of fire origin in dwellings, in the UK (UK Home Office, 2021). The
most common areas of origin for fire are kitchen, followed by bedroom/bedsitting room and then the living room.
Figure 12. The area of origin - dwellings UK extracted from (UK Home Office, 2021).
4000 8000 12000 16000 20000 Primary fires dwellings Year Kitchen External fittings and structures Garage Other Roof/ Roof Space Stairs/ Under stairs (enclosed area) Bedroom/ Bedsitting Room Refuse Store Dining Room/ Utility Room/ Conservatory Bathroom/ Toilet Living Room Interestingly, Figure 13 illustrates the items first ignited in dwellings, in UK. The most common items first ignited are:
Food (cooking oil and fat; other)
Textile, upholstery and furnishing (Foam, rubber, plastic - Plastic - raw material only; Clothing/Textiles - Other textiles; Clothing/Textiles – Clothing; Clothing/Textiles – Bedding)
Structure and fittings (Structural/fixtures/fittings - Internal - Wiring insulation; Structural/Fixtures/Fittings - Internal - Internal Fittings; Structural/Fixtures/Fittings - External
External fittings)
Other material (Other; Wood - Other wooden)
Paper / cardboard (Paper/Cardboard - Household paper/Cardboard)
Not known Figure 13. Item first ignited - dwellings extracted from (UK Home Office, 2021). In comparison, Figure 14 illustrates the articles responsible for the fire development, in dwellings in the UK. The most common articles responsible for the fire development are:
Food (cooking oil and fat; other)
2000 4000 6000 8000 10000 12000 14000 Primary fires dwellings Year Food Textiles, upholstery and furnishings Paper, cardboard Structure and fittings Agricultural and forestry product Explosive Gases & Chemicals Rubbish/Waste/Recycling Other materials Not known None
The data show an increase for responses marked “None” over time, which may signify that new articles have been developed and used or that these articles should be identified and included in the code options. The data also show that:
2000 4000 6000 8000 10000 12000 Primary fires dwellings Year Food Textiles, upholstery and furnishings Paper, cardboard Structure and fittings Agricultural and forestry product Explosive Gases & Chemicals Rubbish/Waste/Recycling Other materials Not known None Unspecified
The most common article that contributed to the fire development is Food (cooking oil and fat; other) These findings illustrate the utility of fire incident data in guiding fire safety interventions. In the UK, for instance, public education campaigns have been initiated with the following messages:
Keep a clean kitchen. A clean kitchen is not only hygienic – it is a lot safer. Regularly clean your stove, range hood and filters. Built-up oil and burnt food can cause fires. Filters can be cleaned in the dishwasher.
Keep flammable objects (e.g., curtains, tea towels, oven mitts) away from the cooking area. At the same time, wear tight fitting sleeves or roll them up when you are cooking.
Turn the power or gas off (if you can) if there’s a fire on your stove.
Never throw water on a frying pan that is on fire or try to carry it outside. If you can, use a pot lid or a large flat object like a chopping board and place it over the pan to starve the fire of oxygen.
Install smoke alarms in your house and check them regularly. Work out an escape plan with your family since it could save your life. Other variables that can help better understanding the trend are the type of building and the age of injured or fatally injured victims. For instance, these additional variables can indicate if kitchen fires, or kitchen fire fatalities are the most common in housing for the elderly or other vulnerable populations. The following actions can be recommended for elderly persons:
The cooking should be assisted by a helper.
Installation of fire safe cooking devices (e.g., no open flames) Bedrooms and living rooms are the most common area of fire origin in residential buildings and the second most common items first ignited and articles contributing to the fire development include:
Textile, upholstery and furnishing (Foam, rubber, plastic - Plastic - raw material only; Clothing/Textiles - Other textiles; Clothing/Textiles – Clothing; Clothing/Textiles – Bedding)
Structure and fittings (Structural/fixtures/fittings - Internal - Wiring insulation; Structural/Fixtures/Fittings - Internal - Internal Fittings; Structural/Fixtures/Fittings - External
External fittings) Campaigns can be initiated with the following messages:
Never smoke in bed particularly when tired or on medication.
Never use candles if you are likely to fall asleep.
Keep candles away from curtains, bedding, clothes, etc.
Make sure that candles are only used in proper candle holders.
Never use candles as a night light for children.
(Use a low watt mains or battery light).
Never cover a light to make it dimmer, fit a lower watt bulb.
Be sure your bedside lamp cannot fall into the bed during the night.
Ensure that all unnecessary electrical appliances are unplugged before going to sleep.
Never place a portable heater close to a bed or other flammable materials.
Always have a flashlight available in your bedroom for emergencies.
Always have a house phone or mobile phone beside the bed in case of an emergency.
Never leave your mobile phone charging over night or when you go out.
Install smoke alarms in your house and check them regularly. Work out an escape plan with your family since it could save your life.
Check the electric plugs and systems Other variables that can help better understanding the trend are the type of building and the age of injured or the age of the victim. These additional variables can for example indicate if the bedroom and living room fires are the most common in the elderly house for example, where elderly people are injured or killed. For example, in Denmark, there was a rule that implied that the smoking in the elderly houses should be done in the common areas only.
in Europe In considering the prospect of introducing a common data collection system for fire incidents across Member States of the European Union (EU), it is worth noting that there is precedent for harmonized data collection of adverse events in the EU. Since the late 1990s, Member States of the EU have implemented a harmonized data collection system for road accidents, known as the Community database on Accidents on the Roads in Europe (CARE) (Eurostat, n.d.-c; OECD, 2018; Thomas et al., 2005). The Community database on Accidents on the Roads in Europe (CARE), supplemented in 2009 by Common Accident Data Set (CaDAS) is comprised of detailed data on individual accidents collected by the Member States of the European Union for all road accidents involving at least one moving vehicle and one injury or fatality.
At the local level, each European Union country transmits the data from its national collection to the European Commission. The data are then transferred from the European Commission to CARE database. Historically, the quality and availability of road accident data has been somewhat limited by differences in data collection form structures and the relevant data formats among the existing national databases, recorded variables and available definitions. CaDAS was subsequently introduced with the inclusion of additional variables and values with a common definition to those contained in the previous models of the CARE database. EU Member States are not obliged to adopt CaDAS and they transmit the data at the EU level choosing the level of detail.
A point of comparison between data collection in the CARE system and fire incident data collection is that data collection in CARE is limited to road accidents that result in fatality or injury, specifically excluding incidents that only involve material damage. In contrast, fire incident data collection generally includes all fires attended by fire departments, even those which do not result in fatality or injury. It seems likely that the narrower criteria for in-scope incidents in CARE data collection enhances the ability to achieve compliance with reporting requirements. In addition, preparing reports of vehicle accidents resulting in casualties is a traditional practice for police officers and likely to be an expected part of job responsibilities, but a relatively new expectation for firefighter duties.
In addition, firefighters fill out paperwork on fire incidents after rescue and extinguishment operations are completed and they return to the fire station. Firefighter fatigue may influence the accuracy of incident reporting. The differences between road accident and fire incident data practice should be carefully considered in the design and implementation of a harmonized fire data collection system. Data on road accidents are seen as an example of harmonised data collection systems involving all the EU Member States. It can be extended to fire statistics and used as a tool to promote safety initiatives by using common data measures to identify and quantify safety problems, evaluating the efficiency of existing safety interventions, and facilitating the exchange of experiences and information.
More detailed information on the CARE database is available in the Task 3 report (EUFireStat, 2022).
Fire incident data can serve a number of important purposes, i.e. helping to reduce fires and losses, identifying opportunities for safety interventions and education programs, guiding the allocation of public resources to areas of greatest need and impact, and monitoring progress of safety initiatives. Cost benefit analysis is a tool that can be used to evaluate various fire safety measures. Studies using this tool combined with fire statistics were reviewed in Task 5 report. The review of previous studies in Task 5 gives an overview of the application of cost-benefit analysis to various fire safety measures. The installation of different kinds of water sprinkler systems is a measure that has been examined in several countries.
Due to high costs water sprinkler systems are seldom seen as cost-beneficial in general; however, for specific types of buildings or for certain risk groups the benefits can out-weigh the costs. Another measure that has been analysed in several countries is the installation of smoke alarms, often seen to be cost-beneficial due to the low cost. Other measures described in the overview include stove guards, fire extinguishers and combustible cladding. Cost-benefit analysis represents a common method of socio-economic analysis. The procedure of performing such an analysis varies, but it will always include an estimate of all the costs of introducing the measure and an estimate of the benefit due to risk reduction as well as other benefits associated with the introduction of the measure.
A cost-benefit analysis is considered to provide a structured and explicit way to create basis for decision making regarding fire safety measures and it has shown to work well in several EU countries. Furthermore, an appropriate method for cost benefit assessment to be used by the Member States and/or the European Commission is proposed in Task 5. The proposal includes a general calculation procedure to conduct a cost-benefit analysis together with a description of the most important input variables. The input data in the proposed calculation procedure includes several of the statistical parameters proposed in previous tasks within the project but also other data is needed. Based on the overview of previous studies, it is evident that there can be a substantial uncertainty associated with some of the input variable values.
Consequently, it is strongly recommended that a cost-benefit analysis should be complemented with a sensitivity analysis to present the variation in the result due to uncertainty in the inputs. A sensitivity analysis provides essential background for wise decissions if a cost benefit analysis is used for making large desicisions. The sensitivity analysis comprice usually of a parameter variation of the variables with large uncertainties to study impact on the overal cost/benefit ratio.
In the Task 6 report (EUFireStat, 2022) three case studies are performed to demonstrate the proposed methodology. The topics of the case studies are:
for residential fires.
The specific procedure when performing these analysis varies somewhat between the different case studies but they are all based on the same methodology, i.e. the methodology proposed in Task 5. An estimate of the cost of introducing the measure and an estimate of the benefit due to risk reduction as well as other benefits associated with the introduction of the measure are included in all case studies. A rather detailed calculation was possible for Case study 1 since there have been several studies in the area and data is available for most of the important input variables. It was also seen that the measure (smoke alarm) is cost-effective with a benefit-cost ratio of well above 1, i.e. the benefit is much larger than the cost.
The results of Case study 2 and 3 are considered more uncertain and harder to interpret since the benefit-cost ratio is close to 1, which means that a variation in any of the input parameters can cause the benefit ratio to either grow above 1 or be reduced below 1. Several important input variables are also associated with great uncertainties which makes it especially important to include a sensitivity analysis in the cases with a cost benefit ratio close to 1. The conducted case studies underpin that good fire statistics is crucial to conduct this type of analysis. Data on the number of fatalities, number of fires, item first ignited etc. have been used in the case studies.
It is important to point out that there are several input variables needed for a cost-benefit analysis that cannot be obtained from fire service statistics, for example the risk reduction and cost of implementing and maintaining a certain measure. Accordingly, for a Member State and/or the European Commission to be able to conduct a cost-benefit analysis for a policy decision, fire statistics is a prerequisite, but it does not provide the complete dataset needed for the cost benefit analysis.
Based on what has been learned in this project, all of the variables proposed in this report are already collected by most EU countries, although they are not always formally defined. The variables collected in more than half of the EU countries are incident time, incident date, incident location, number of fatalities, number of injuries, age of fatalities, type of building and the primary cause of fire. It is advantageous that so many countries already collect this data. However, in some cases current definitions and inclusion/exclusion criteria will need to be adjusted to provide harmonized data for analysis at the European level.
The variables collected in less than half of the EU countries are the number of floors (at least 8 EU countries), the presence, type and operation of fire safety measures, the area of origin (at least 9 EU countries), item first ignited (at least 10 EU countries), article contributing to fire development (at least 5 EU countries) and the heat source (at least 9 EU countries). It is expected that adding these data to the fire statistics will require more work for implementation in countries not already collecting the variables, but on the other hand they will not have to deal with the challenge of harmonizing existing definitions. Depending on the nature of the data, some variables will require more effort to harmonize than others.
Indeed, it is expected that the variables incident time, incident date, incident location, age of victims, number of floors and fire safety measures present can be implemented and harmonized with low effort. For the implementation of variables number of fatalities and number of injuries, it is expected that the process will be difficult, especially for the countries which are currently reporting only at the fire scene. For the countries which already correct these variables after cross checking with medical records up to a certain time after the fire, it is estimated that there should be no difficulties in adapting their practices. Challenges can be anticipated in adopting proposed values for the variables type of building, area of origin, item first ignited, article contributing to fire development and cause.
We have seen in Task 1 report that each country uses its own values for each variable, hence there will be a need to adapt to the proposed new structure (EUFireStat, 2022).
We have identified 14 variables necessary to be collected as a priority in all Member States in a harmonised way in order to be able to provide recommendations and to identify areas for improvement to support fire safety and fire prevention efforts, and to enable cross-learning between different Member States, regions and local actors. The first Tier of these variables is the following:
Tier 1:
Number of fatalities
Number of injuries
Incident location
Incident date
Incident time
Age of fatalities
Primary causal factor
Type of building
Once these eight variables have been implemented efficiently, we propose adding the second tier, which would include six additional variables:
Tier 2:
Number of floors
Area of origin
Heat source
Articles contributing to fire development
Item first ignited
Fire safety measures present
For these 14 variables we have proposed precise definitions and corresponding values, and provided guidance on how to collect them and analyse them. In order to gain broad support for the pan-European fire statistics proposed, it is most important that the implementation is done in as efficient a way as possible. The other variables listed in Tier 1 (Age of fatalities, Primary causal factor and Type of building) should then be implemented in a second step, followed by the variables of Tier 2. This does not prevent Member States from collecting additional variables ahead of time.
A short survey was sent to the Member States’ regulators and persons dealing with fire statistics in all 27 EU Member States in order to survey their opinions about implementing at least five variables during the next five years as part of the pilot phase of the implementation process with the following questions:
collection at the European level?
Do you already have a national/regional/local dedicated structure (organization, department, or group) that could be responsible for managing and analysing fire data?
If yes, could you please name it? (Please provide name and contact details)
If no, would you be in favour of creating such structure nationally?
Please provide any other comments/suggestions that you think would be important to
take into consideration in the implementation of the harmonised fire statistics in Europe?
The survey response rate was 19 countries out of 27 (70%). Additionally, two countries acknowledged receiving our survey and promised to answer later.
Figure 15 Survey response analysis
It can be noted that all 19 countries that have answered the survey are in favour of providing harmonised fire statistics for collection at European level. The responses to each question of the survey are illustrated in following figure. Answered the survey Did not answer the survey Promissed to answer later Number of countries
Figure 16 Survey response analysis (questions 1 to 4)
Considering the results of this survey, we recommend the European Commission to create a core group of Member State authorities with the following role:
Standardization is necessary to provide a recognised basis for the current proposal and to facilitate its dissemination to all Members States, and potentially beyond. This can potentially be performed at two levels:
Due to the lack of common terminology or variables with a similar nomenclature covering different aspects, fire statistics and data cannot always be compared between countries. This hinders e.g., effective cross-learning about successful fire safety interventions. To develop a comprehensive evaluation, this research has identified fourteen variables that should be recorded in fire statistics as a priority.
Number of fatalities
Number of injuries
Incident location
Incident date
Incident time
Age of fatalities
Primary causal factor
Type of building
Number of floors
Area of origin
Heat source
Articles contributing to fire development
Item first ignited
Fire safety measures present
The research also focused on providing definitions and values for each variable. The proposed terminology constitutes a minimum dataset for collection at the local level and does not prevent a fire department or national authority from utilizing also more detailed data collection so as long as they can provide simplified data according to the terminology of the pan-European statistics. We described all of the necessary steps involved in fire data collection from collection at the incident to reporting at the European level as well as guidance on how to collect data. The outputs generated by this project should increase awareness about the importance of a common terminology that will generate the foundations for a harmonized fire statistics at European level.
When surveying views of fire regulators of all Member States, it was shown that at least 19 countries are in favour of providing harmonised fire statistics for collection at European level. The next step should then be to implement at least the five first proposed variables (or more) as part of an experimental phase of the implementation process. In parallel, there should be a structure at the European level which can receive national fire statistics on an annual basis, with the necessary resources to store, analyse and publish data from the various countries. Finally, standardization process seems necessary for providing a recognised basis for the proposed values and their corresponding definitions and to facilitate the dissemination to all Member States. Finally, it will be useful for a leading group of countries to implement the proposal in order to demonstrate its utility.
This will facilitate the ability of stakeholders to generate data, use them to chart and publicize trends, and build public recognition and support of fire safety policies and initiatives.
Ahrens, M., Frazier, P., & Heeschen, J. (2003). Use of fire incident data and statistics. In A. Cote (Ed.), Fire Protection Handbook (19th ed.). Fire Protection Association. Belanger, F., Ung, A. B., Jougla, E., Thélot, B., Bene, M., Bruzzone, S., & others. (2008). Analysis of injury related mortality in Europe. The ANAMORT project. Final implementation report. Carlotti, P., Parisse, D., & Risler, N. (2017). Statistical analysis of intervention reports for fires resulting in casualties deceased on the spot in Paris area. Fire Safety Journal, 92(June 2016), 77–79. https://doi.org/10.1016/j.firesaf.2017.05.017 Construction permit index overview - Statistics Explained. (n.d.). Retrieved January 21, 2022, from https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Construction_permit_index_overview Eurostat. (n.d.-a). Europa - RAMON - Nomenclature Detail View.
Retrieved January 21, 2022, from https://ec.europa.eu/eurostat/ramon/nomenclatures/index.cfm?TargetUrl=DSP_GLOSSARY_NOM_DTL_VIEW&StrNom=CODED2&StrLanguage- Code=EN&IntKey=16680085&RdoSearch=&TxtSearch=&CboTheme=&IsTer=&ter_v alid=0&IntCurrentPage=1 Eurostat. (n.d.-b). Glossary:Classification of types of construction (CC) - Statistics Explained. Retrieved January 21, 2022, from https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Glossary:Classification_of_types_of_construction_(CC) Eurostat. (n.d.-c). Road transport safety (tran_sf_road). Retrieved January 20, 2022, from https://ec.europa.eu/eurostat/cache/metadata/en/tran_sf_road_esms.htm Fahy, R. F., & Petrillo, J. T. (2021). Firefighter Fatalities in the US in 2020. https://www.nfpa.org//-/media/Files/News-and-Research/Fire-statistics-and-reports/Emergency-responders/osFFF.pdf Fire Service Academy. (2018).
Fatal residential fires in Europe. A preliminary assessment of risk profiles in nine European countries. https://www.ifv.nl/kennisplein/Documents/20181120-BA-EFA-Fatal-residential-fires-in-Europe.pdf Fundacion MAPFRE, & APTB. (2020). Victimas de incendios en España en 2019. https://www.fundacionmapfre.org/educacion-divulgacion/prevencion/semana-prevencion-incendios/estudios/ Home Office. (2019). Detailed analysis of fires attended by fire and rescue services, England, April 2018 to March 2019. Home Office National Statistics, 19(March 2019), 5. https://www.gov.uk/government/statistical-data-sets/fire-statistics-guidance ISO. (2014). ISO/TR 17755:2014 - Fire safety — Overview of national fire statistics practices. https://www.iso.org/standard/60469.html Johansson, N. (n.d.). Dödsbränder i bostäder 2012-2015: En studie av fastighetsinformation.
Kobes, M., Helsloot, I., de Vries, B., & Post, J. G. (2010). Building safety and human behaviour in fire: A literature review. Fire Safety Journal, 45(1), 1–11. https://doi.org/10.1016/j.firesaf.2009.08.005 Lasbeur, L., Barry, Y., & Thélot, B. (2012). Évolution de la mortalité par accident de la vie courante en France métropolitaine, 2000–2008. Revue d’Épidémiologie et de Santé Publique, 60. https://doi.org/10.1016/j.respe.2012.06.115 Lasbeur, L., & Thélot, B. (2014). Évolution de la mortalité par accident de la vie courante en France métropolitaine 2000–2010 : focus sur les incendies. Revue d’Épidémiologie et de Santé Publique, 62. https://doi.org/10.1016/j.respe.2014.06.267 MSB. (n.d.). Dödsbränder i Sverige Kvalitetsgranskning av MSB:s dödsbrandsdatabas. https://rib.msb.se/filer/pdf/26318.pdf OECD. (2018). Road safety annual report 2018. www.itf-oecd.org/road-safety-annual-report- 2018 Runefors, M. (2020). Measuring the Capabilities of the Swedish Fire Service to Save Lives in Residential Fires. Fire Technology, 56(2), 583–603. https://doi.org/10.1007/s10694-019- 00892-y Thomas, P., Morris, A., Yannis, G., Lejeune, P., Wesemann, P., Vallet, G., & Vanlaar, W. (2005). Designing the European Road Safety Observatory. International Journal of Injury Control and Safety Promotion, 12(4). https://doi.org/10.1080/17457300500241746
Eurostat Classification of Types of Construction (CC) — extracted and adapted for EU Firestat (project modifications underlined in the source PDF). Full class notes appear in the Eurostat CC glossary.
Recording rule: classify by main apparent use; mixed-use buildings take the primary-use code plus an m flag (e.g. 1263m). Section 3 — Buildings under construction excludes maintenance/renovation-only sites.
| Section | Division | Group | Class | Building type |
|---|---|---|---|---|
| 1 | 11 | 110 | 1100 | Residential — unknown |
| 1 | 11 | 111 | 1110 | One-dwelling buildings |
| 1 | 11 | 112 | 1121 | Two-dwelling buildings |
| 1 | 11 | 112 | 1122 | Three-and-more dwelling buildings |
| 1 | 11 | 113 | 1130 | Residences for communities |
| 1 | 12 | 120 | 1200 | Non-residential — unknown |
| 1 | 12 | 121 | 1211 | Hotel buildings |
| 1 | 12 | 121 | 1212 | Other short-stay accommodation |
| 1 | 12 | 122 | 1220 | Office buildings |
| 1 | 12 | 123 | 1230 | Wholesale and retail trade |
| 1 | 12 | 124 | 1241 | Communication buildings, stations, terminals |
| 1 | 12 | 124 | 1242 | Garage buildings |
| 1 | 12 | 125 | 1251 | Industrial buildings |
| 1 | 12 | 125 | 1252 | Reservoirs, silos and warehouses |
| 1 | 12 | 126 | 1261 | Public entertainment buildings |
| 1 | 12 | 126 | 1262 | Museums and libraries |
| 1 | 12 | 126 | 1263 | School, university and research |
| 1 | 12 | 126 | 1264 | Hospital or institutional care |
| 1 | 12 | 126 | 1265 | Sports halls |
| 1 | 12 | 127 | 1271 | Non-residential farm buildings |
| 1 | 12 | 127 | 1272 | Places of worship / religious activities |
| 1 | 12 | 127 | 1273 | Historic or protected monuments |
| 1 | 12 | 127 | 1274 | Other non-residential (e.g. prisons, barracks) |
| 3 | — | — | — | Buildings under construction (EU Firestat extension) |
Structured reference form aligned with Tier 1 + Tier 2 variables (single incident). Use [ ] as unchecked boxes.
| Field | Entry |
|---|---|
| Incident time | Hour: __ Minute: __ Local time / UTC offset |
| Incident date | Day / Month / Year |
| Incident location | Coordinates (lat/long), street, number, town, ZIP; or country only if unknown |
| Number of fatalities | |
| Number of injuries | |
| Age of fatalities | Estimate if unknown |
m flag on primary Eurostat code)Present? [ ] Yes [ ] No
| System | Present | Operated at fire? |
|---|---|---|
| Detection | [ ] | [ ] |
| Alarm | [ ] | [ ] |
| Extinguishing system | [ ] | [ ] |
| Passive fire protection (doors / compartmentation) | [ ] | [ ] |
| Smoke control systems | [ ] | [ ] |
Functional area: sleeping, bathroom/toilet, kitchen, living room, laundry, meeting, office, classroom, cafeteria/bar, sauna, stable/barn, other Area of egress: hallway/corridor, stairway, elevator, escalator, lobby, other Assembly or sales: assembly, sales, showroom, indoor swimming hall, lounge, other Technical processing: operating, processing/manufacturing, first-aid, stage/scene, other Storage: storage, parking/garage, cooling/freezer, fuel storage, trash, shipping/receiving, silo/container/barn, other Service/equipment: machinery, maintenance shop, producing/distribution, ducts, heating, shafts, other Structural: wall assembly, roof, façade, attic, balcony/terrace, substructure, awning, area under renovation, other
Food-related: cooking fat/oil, food Furnishing/clothing: armchair/sofa, curtains, bed, clothes, candle stick, table, plant pot, other furnishing Household electric appliance: freestanding cooker, hotplate/hob, oven, microwave, dishwasher, fridge/freezer, toaster, coffee maker, washing machine, tumble drier, heater, fan/ventilation, sauna heater, other Other electric: lighting, battery, charger, wiring/socket/plug, distribution board, PV panels, transformer, consumer electronics, other Building element: façade/cladding, windows, floor/wall covering, roof, masonry, inner wall, joist, other Other: renovation/maintenance items, paper/cardboard, soot/tar (chimney), wood chippings, vegetation, flammable liquid/gas, car, other vehicle, pram, rubbish, other
Household electric appliance: (same appliance list as item first ignited) Other electric appliance/tool: lamp, battery, charger, welding equipment, hot air gun, other Electric distribution: wiring/socket/plug, distribution board, transformer, other
Comments from Country A:
About the collection of geographic location data, is it possible to know which coordinate system you intend to use?
On the collection of the number of deaths and injuries: Country A collects separate data between population and firefighters. It would be useful if these data could be separated. This data will be the most difficult to collect as it involves monitoring those involved for 1 year after the fire.
Will there be a platform for collecting statistical data?
Will the data provided by each member state be public or shared among themselves?
When do you expect to start collecting statistical data? Consortium answer:
At this stage we do not separate between population and firefighter, but it can be implemented on the long term.
Ideally there should be a platform to collect statistical data
Each member state is free to publish their own data but we can also imagine a European platform that is collecting data, analysing it and making annual publications. Comments from Country B:
In this context we would like to point out that, from a quick but not approximate analysis of the minimum 5 variables to be introduced within the next 5 years (number of fire deaths, number of fire injuries, incident date, incident date time, geographical location of fire incident,) we can affirm that all the 5 variables examined are already detected and effective by us, with values corresponding to those reported by you. As regards the other 8 variables proposed, 3 of them (fire cause, type of building, source of ignition) we already detect them but the values and definitions proposed by you are not comparable to our calculations. Finally, we point out that two other variables you suggest (age of victims, room of origin), we already plan to insert them into the system.
Comments from Country C: Suggestions that would be important to take into consideration in the implementation of the harmonised fire statistics in Europe:
Comment from Country E: It is important for us to know, as we try to understand if the changes in our regulation are necessary and if there could be data protection issues.
Consortium answer:
The only variable where there might be data protection issue could be the ”incident location”. If providing exact address, or latitude and longitude could cause issues related to data protection, then perhaps providing only the County or the Municipality (or post code) of the incident location, without the detailed address or coordinates could be enough.
Comment from Country F: It is important to take in mind the different sizes of the countries and their fire and rescue services and the influence on statistics. For example: a bush fire > 10 is a “large” fire for Country F circumstances, but in another country this is kind of “daily business”. So the categories must be clear and normative and not subjective.
Consortium answer:
In this project we only focus on building fires, but we understand the message illustrated by this example. This is why we sustain form defining categories, but rather push for sharing actual data. For instance, fire fatalities are often normalised by the number of inhabitants, but we do not go into this level of details at this stage. Processing and analysing data will then be the responsibility of the Country’s analysts or at European level.
Comment from Country G: In the published final report on task 4 "Terminology" of the project, in items 3.2.1 and 3.2.2 are given definitions for "Number of deaths" and "Number of victims" (EUFireStat, 2022). The definitions and notes to them show that the "Number of deaths" includes the number of people who died because of injuries sustained from the fire within 1 year of the fire, and the "Number of victims" includes the number of people who were injured (but not accounted for as deaths) as a result of the fire within 1 year from the incident. We believe that the specified period of 1 year is too long and will create difficulties in collecting reliable data. We propose to reduce the same period to 1 month.
Consortium answer:
We deliberately set the time limit to consider a fire fatality to one year after the fire occurrence, which is one of the longest times identified. From International Association of Fire and Rescue Services (CTIF) experience, 99-100% of all fire fatalities are covered in the 90 days following the fire incident. Of course, this estimation can vary from country to country and with the progress of medical care, but it implies that if a country is able to collect data up to 90 days following the fire, then it would also be acceptable.
All written comments received during the project, with consortium responses.
| No | Body | Document | Reference | Comment | Consortium response |
|---|---|---|---|---|---|
| 1 | MSB – Sweden | Global comment on the project | — | The project mainly covers definitions and collection methods but is not proposing a common method for analysing the data. | Task 3 includes proposals for methodologies on unknowns and incomplete data. Initial analysis proposals occur in Task 3 and Task 7, with guidance on analysis and misinterpretation risks. |
| 2 | BVS – Austria | 1st progress report | Task 0 – Annex B | Efforts concerning harmonization of data collection in Austria are not reflected sufficiently. Updated diagnostic sheet and latest published fire statistics provided. | Updated diagnostic sheet and Austria information updated throughout Task 0 and Task 1 reports. |
| 3 | ANEC | 1st progress report | — | Data quality is extremely important; suggests ad hoc accuracy checks and cost/benefit analysis. Official sources may not be representative. | Data quality and error handling stated in Task 0 diagnostic sheets; importance discussed in Task 1 conclusions; cost-benefit analysis planned in Task 5. |
| 4 | ANEC | 1st progress report | — | Near misses are important (e.g. sprinklers preventing serious fires, flame-retardant furniture) but may not appear in official figures. Suggests a scheme for public/responsible-person input. | Scope is building fire incidents, not near misses (Task 1 §1.1). Near-miss collection would need household reporting or sample surveys — outside project scope but encouraged nationally/EU. |
| 5 | ANEC | 1st progress report | — | IDB-FDS injury data (mechanisms 04.14, 4.17) could complement fire statistics with age, injury type, products involved, etc. | Tasks 0–1 analyse fields recorded in fire statistics; victim/injury aspects (age, gender, cause) investigated for life-safety coverage. |
| 6 | ANEC | 1st progress report | — | Expect federation of European Fire Officer Associations to inventory national fire safety unions. | European Fire Officer Association (Steering committee) helped distribute Task 2 questionnaire contacts across Europe. |
| 7 | Fire Safe Europe | — | Task 2 | Digitalised Task 2 questionnaire circulated to members and European Fire Safety Community; 12 answers received. | Thank you; input will be analysed and fitted into project context. |
| 8 | DG ESTAT | 1st progress report | Task 1 | Questionnaires/forms and usage manuals/guidelines for data registration were not collected. | Focus is definitions and fields recorded; per-country abstracts specify collectors and systems. Form analysis beyond scope given language diversity; noted in Task 1 §1.1. |
| 9 | DG ESTAT | 1st progress report | Task 1 | Missing documentation on what is mandatory and under which legal provisions. | Added text in Task 1 §1.2 on difficulty evaluating mandatory vs optional fields across countries. |
| 10 | DG ESTAT | 1st progress report | Task 1 | No clear indication which datasource definitions refer to; fire service vs insurance may define 'accidental' or 'victim' differently. | Abstracts must be read with Appendix I/II tables; clarified in Task 1 §1.3. |
| 11 | DG ESTAT | 1st progress report | Task 1 | In country fiches, what is the difference between b and c cases? | a = fields available; b = definitions not available; c = fields unclear to authority. Explained in Task 1 §1.2; field analysis in Task 1 final report §4. |
| 12 | DG ESTAT | 1st progress report | Task 1 | Incomplete use of statistical methodology (concepts/dimensions/code lists); coding lists sometimes presented as definitions. | Tables clarify when fields use dropdown menus vs definitions; stated in Task 1 §1.2. |
| 13 | DG ESTAT | 1st progress report | Task 1 | Contradictory information in country fiches (CZ, RO, DE, NL references). | Appendix I corrected for Czech Republic, Romania, Germany; references inserted for Austria, Germany, Switzerland. |
| 14 | DG ESTAT | 1st progress report | Task 1 | What is meant by 'victims' and 'type of fatalities'? | Fatalities/victims vary by country; type of fatalities = cause of death; socioeconomic characteristics in Task 1 final report §3.8. |
| 15 | DG ESTAT | 2nd progress report | Task 2 | Expected dedicated comments on insurance companies and national statistical offices as stakeholder types. | Only 3 insurance responses (not statistically significant); no national statistics offices responded; majority from ministries and fire brigades. |
| 16 | DG ESTAT | 2nd progress report | Task 3 | Expected reflection on institutional aspects: central body, mandate, sampling framework and stratification variables. | Table 1.4 in Annex 1 covers who collects/processes/reports data; sampling design beyond Task 3 scope; may be touched in Task 7. |
| 17 | DG ESTAT | 2nd progress report | Task 3 | Timing of data collection (forms within x days, victim follow-up, reporting deadlines) affects feasibility of death/injury definitions. | Timing data not readily available; data journey steps described; local DB should be updated when new information arrives; training detail beyond Task scope. |
| 18 | DG ESTAT | 2nd progress report | Task 3 | Cost part looks weak because institutional/timing points above are absent. | Local-level costs similar across methods; national-level cost analysed from available information. |
| 19 | DG ESTAT | 2nd progress report | Task 3 | Cost analysis should compare existing costs vs additional costs for harmonised approach (e.g. periodic EU meetings). | Would require per-country research beyond feasible project effort given difficulty obtaining existing information. |
| 20 | DG ESTAT | 2nd progress report | Task 3 | Possible exploitation of insurance data as complementary source not covered. | Mentioned in chapter 2 §4; team experience is insurance microdata linking is rarely accessible; not expanded due to access barriers. |
| 21 | DG ESTAT | 2nd progress report | Task 4 | No time limit after fire for deaths/injuries may be infeasible. | Changed from 'no limit' to 1 year after fire (ISO TS 17755-2 aligned, USA practice); 90-day collection also acceptable per CTIF experience. |
| 22 | DG ESTAT | 2nd progress report | Task 4 | When reusing Eurostat building-type definition, cite source for future census updates. | Reference to Eurostat report added. |
| 23 | DG ESTAT | 2nd progress report | Task 4 | Expected per-variable paragraph on implementation difficulty vs Task 0/1 practices (e.g. already collected in 20 countries). | Added §4 discussion on implementation in EU countries (qualitative). |
| 24 | DG ESTAT | 2nd progress report | Task 4 | Expected section on key indicators derivable from variables combined with census/NUTS/urbanisation data. | Touched in Task 3 report; out of Task 4 scope; may cover in Task 7. |
| 25 | Modern Building Alliance | 2nd progress report | Task 2–4 | Rename Tier 2 terms; add 'Secondary ignition sources' variable. | 'Source of ignition' already changed to 'heat source'. Secondary ignition sources not prioritised; may overlap with articles contributing to fire development. |
| 26 | Modern Building Alliance | 2nd progress report | General | Objectives should include Eurostat integration and academic data access. | Objectives cannot change at this stage; remarks included for Task 7 discussion. |
| 27 | Ei Electronics | Final report | Task 6 – CBA | €1 replaceable battery cost too low; €3–5 more typical, affecting smoke-alarm CBA. | Agreed value is low; pricing from prior analysis to be revised in report. |
| 28 | ANEC | Final report | General | Welcomes report; near misses (small appliance/battery fires handled by owner) should be considered; hopes report revitalises EU Injury Database (EU-IDB). | Near-miss collection outside building-incident mandate (see §5.4 near-miss paragraph). EU-IDB legal basis noted as broader context. |
| 29 | Modern Building Alliance | Final report | Definitions | Distinguish material vs item/product (§4.5.3); broaden fire-safety-measures list (§4.4.3). | Changed 'material' to 'article'; added smoke control systems and passive fire protection to fire-safety-measures list. |