Fire Statistics Review

6. Discussion

In this section, the main responses are summarised and any that differed from our proposed solution are discussed and our revised solution is explained.

6.1 Respondents

The range of respondents gave a reasonable coverage of the users of both the Fire Statistics, Scotland and Fire and Rescue Service Statistics Scotland. We are content that users' views are represented.

6.2 Format

For both publications the electronic formats (HTML and PDF) were equally used, though the HTML format is not printed out by users. Most users also either used, or would use in the future, the accompanying datasets for both publications.

6.3 Topics

Of the topics covered and the frequency of use, all topics were used, to some extent, by some respondents.

6.4 Publication presentation

For the Fire Statistics, Scotland (incident data) publication we proposed streamlining the publications so that we:

i) remove the HTML/PDF copies of tables and charts from the Appendix

ii) publish a spreadsheet containing Excel copies of all tables and charts

Similarly for the Fire and Rescue Service Statistics, Scotland we plan to reduce the level of detailed information in the commentary section, summarising key points and statistics, but maintaining the more detailed information in Excel spreadsheets.

Similar concerns were expressed on both publications which we plan to address as below:

Concern	Action
Technical difficulties with Excel spreadsheets on a website.	Monitor the published spreadsheets, provide error reporting facilities and react quickly if difficulties are identified
Single document with tables for quick reference, supported by the Excel tables for deeper analysis. Inconvenient to open the spreadsheets even for a single number.	Maintain high level data tables within the document
Need information at local authority level.	This is, and will be, provided in the supporting data

Table 11 : Comments on electronic format of the publications

We therefore plan to rationalise the publications, and acknowledge the value of including some data in the printed or on-screen report. We will include tables to an appropriate level for viewing.

6.5 Geographic level of presentation

We plan to present data for both publications at Scotland and local authority level. Most users would find this useful, though there were some requests for other geographies.

In this arena it is important to remember the quantity and therefore the value of incident and FRS data at lower geographic levels. In addition to this is the requirement to ensure that the data is non-disclosive.

We do, and will continue to, provide data to Scottish Neighbourhood Statistics^[3] (SNS) at datazone level. The data provided at datazone level is Number of all fires, Rate of accidental dwelling fires per 100,000 population and Rate of deliberate fires (excluding chimney fires) per 100,000 population. This data is carefully screened and managed to ensure it is not disclosive.

We do not plan to publish incident data at geographies smaller than local authority in the national publication, though for specific purposes, for example providing prison fire statistics, we will consider individual data requests.

We will also look into systems which can aggregate local authority data into LSO or service delivery areas if the demand for that continues.

We will publish data at Scotland and local authority level in both Fire and Rescue Service Statistics, Scotland and Fire Statistics, Scotland, and look into aggregating the data for higher levels.

We will continue to provide datazone level data for SNS.

6.6 Long-term trends

Long-term trend data is published within the additional datasets on http://www.scotland.gov.uk/Topics/Statistics/Browse/Crime-Justice/Datasets/DatasetsFire. We will continue to provide long-term trend data at Scotland level. Continuing the sub-national data may be more difficult since it was published at Fire and Rescue Service level. We will now supply the data at Scotland level and explore providing it at local authority level with some conversion method to allow comparisons of other sub-national geographies.

We will supply the long term trend data at Scotland level and explore providing it at local authority level with some conversion method to allow comparisons of sub-national data.

6.7 Publication date

Publishing later than previous years because additional time will be needed to prepare both of the publications due to the required changes is acceptable to most users. For the one request for incident data in August, this would not have been feasible even without the changes.

6.8 Publication responsibility

Handing over responsibility of Fire and Rescue Service Statistics, Scotland to the Scottish Fire and Rescue Service has the potential to cause concern to some data users who feel that there is a risk of losing data quality and consistency. These risks were considered during the decision process on the future of fire data in Scotland. In the first instance a data steering group is proposed to provide balanced oversight of the scope and content of the publication. In the longer term it is hoped that Scottish Fire and Rescue Service will become an official provider of statistics under the UK Statistics Authority badge. This would mean complying with their Code of Practice for Official Statistics, thus ensuring that the data is recognised as having "a stamp of assurance that the statistics have been produced and explained to high standards and that they serve the public good."^[4]

Scottish Government will continue to work with Scottish Fire and Rescue Service and other stakeholders to transfer responsibility for Fire and Rescue Service Statistics, Scotland to the Scottish Fire and Rescue Service. The risks raised will be noted and actions taken to mitigate them as far as possible.

6.9 Extra data requests

6.9.1 Incident data

We asked users what they would like to see published in the Fire Statistics, Scotland (incident) data . There was a considerable list of extra data requests which are considered here.

Some general points about our response to the requests are:

i) It is important to bear in mind that the question was asked about a national, annual publication with similar tables published each year. There is limited space in the publication and finite resource to produce it and so some rationalisation of requests was needed. We have therefore tried to reach an acceptable compromise which will include areas of widest interest and at a level of detail of most use to the widest audience.

ii) Data is captured using the electronic IRS data collection system, which is managed by DCLG and is currently under review. Scottish Government and Scottish Fire and Rescue Service are represented on the review group. The data collected by the system is therefore out of scope for this consultation, which focuses on how to report the data that is already collected.

iii) The IRS report is intended to be the first stage of a fire or incident investigation, collecting initial information from all incidents. The data collection needs to remain focused on that purpose, and stay within a limited scope. However, for some of the extra data requests, linking incident data with data from other sources may help to provide the information required.

iv) Within the extra data requested there was a degree of confusion over whether the data was in Fire Statistics, Scotland or in Fire and Rescue Service Statistics, Scotland. We are aware of the confusingly similar titles of these publications and have allocated the requests to the appropriate publication.

v) Several of the requests were for data that is already presented in the current publications. This suggests that it can be difficult to locate the data required. Providing more tables covering every possible aspect that is available in the data is likely to exacerbate this. However, we have noted the issue and will work to find ways of making the data easier to find and understand.

Responding to the specific respondents' points:

Premises information

There were several requests for increased information about the premises where the incident occurred, including:

• Tenure
• Structure (in particular timber framed construction)
• Age
• Heritage/listed status

While some of this information is captured in the data, not all of it is. To produce annual tables at this degree of detail depends on quality of the data returns and the level of demand for the information. Table 12 shows the detailed analysis supporting our decisions.

Tenure	This information is not recorded in IRS.
Listed/heritage buildings	Such buildings are not defined within IRS. The "Location of fire" field can be filtered to show the number of fires in typical listed buildings (e.g. castles, stately homes etc.) but such filters will not catch all listed buildings, nor would they return only listed buildings.
Age	This information is not recorded in IRS.
Construction type	IRS asks for all building fires "Was there any special method of building construction involved?" (Q8.18) with responses categorised as: None, Timber framed, Cladding, Sandwich panels, Atria, Thatch, Large single storey retail premises, Other, Not known. Tables could be produced to summarise comparisons of fires in timber-framed construction buildings to those of other construction methods. We carried out some analysis to assess the data that might be available: Findings: Over 95% of dwelling fires have special construction method "none", 1.5-2.5% are timber framed. A more detailed table combining construction method with other IRS fields (e.g. fire damage) or a special report would provide further information but would not be suitable for a national publication. Suggest: This could be a research project or the focus of separate analysis - datasets to be provided upon request to researchers.

Table 12 : Requests for premises information

We will monitor the outcomes of the IRS review and consider whether to include extra tables on the basis of that.

Specific detailed datasets can be provided on request

Cause information

There were several requests for increased information about the causes of fires and false alarms. The 2012-13 publication was extended to include a commentary section on 'Causes and source of ignition of fires' while Tables 21 to 25 of the publication all contain data on causes of ignition, sources of ignition, spread of fire and where impairment due to suspected alcohol/drugs use was a contributory factor.

In terms of annual, national reporting, we feel that there are sufficient tables already produced. While establishing the causes and contributory factors to fires is a prime example of benefitting from this data, we feel that further benefit would be derived from research rather than more tables.

We will publish datasets including the cause, source of ignition, fire spread and impairment in tables similar to those provided in the 2012-13 Fire Statistics, Scotland publication.

Detection and Suppression

There were several requests for increased information about detection and suppression systems. The 2012-13 Fire Statistics, Scotland publication contained tables showing fires and casualties by presence and operation of smoke alarms. In terms of the less common detection and suppression data, some may be suitable for regular reporting, though some contain so few instances that extracting general findings would be more appropriately achieved through research projects.

Fire safety/ suppression systems

For all primary fires in "other buildings" (non-dwellings) and some "outdoor structures", IRS provides information on the presence (Q7.11) and type of fire safety system (e.g. sprinkler, water mist, foam) present (Q7.12), if any, and whether it operated (with reason for not operating where applicable) (Q7.14).

Further details on the location (Q7.13) of the fire safety system, how many sprinkler heads operated (Q7.15) and the system's impact on the fire can also be obtained (Q7.16).

We explored the data to establish whether providing a new table was required.

Findings:

In 2012-13, of the 1,850 fires where Q7.11 (were active safety systems in place?) was asked, the answer was "yes" for 89 (~0.5%) fires. When split into the different types of active safety systems, the numbers become very small and therefore of limited analytical value. In some cases it appeared the routing rules used to determine whether this question was asked for an incident were not correctly applied e.g. the question was asked for some non-primary fires

Suggest:

There is not enough data to produce a robust analysis. We will Investigate IRS recording inconsistencies.

Fire Detection Systems

For all primary building fires, IRS asks if an alarm system was present (Q5.8). If the answer is yes, Q5.9 asks about the type of alarm system (battery operated, mains, combination etc.). Information can also be extracted on the location of the alarm (Q5.10) system, whether it operated and raised the alarm (Q5.11) and where it didn't operate, the reason for this(Q5.12).

We already provide a breakdown of fires and type of casualty by whether an alarm system was present, operated and if it raised the alarm.

Findings:

Breakdown by type (battery/mains) could be provided on request but is not suitable for national publication.

A breakdown of alarms which did not operate, by reason for alarm failure, could be provided although preliminary analysis shows that the majority were "Due to alarm not being close enough to fire or located where the fire occurred". Small numbers in each of the remaining categories may be of limited value to users.

Suggest:

Datasets to be provided on request or made available for users to access directly.

Specific detailed datasets can be provided on request.

We will explore ways of providing more of the raw data for analysis

Research projects

Several of the extra data requests were looking to examine specific hypotheses. Broadly we would propose that these are the subject of research projects, extracting specific data relevant to the hypothesis, however, some data is already published that supports the request. These requests were:

Information about seasonal variations/peaks in fire demand	Monthly trend data is published as a separate dataset on http://www.scotland.gov.uk/Topics/Statistics/Browse/Crime-Justice/Datasets/DatasetsFire/datamonthlytrend though we accept that this can be difficult to locate and we will try and make it more obvious. Primary fires and casualties by time of call are reported in Table 27 of the 2012-13 publication.
A breakdown of fires involving rubbish/refuse by type	In Table 9 and 9a on secondary and outdoor fires, Refuse fires are reported by: "Refuse - small/rubbish container (e.g. wheelie bin)" "Refuse - large/ rubbish container (e.g. skip)" "Refuse - loose/ rubbish tip"
More about the trend in the severity of dwelling fires and dwelling fire casualties	This is being explored in the current UK-wide review of IRS data. We will monitor the outcomes of the review and include them in later publications or datasets if appropriate.
Had portable extinguishers been deployed?	IRS records information on the 'Main action taken by the public prior to FRS arrival (including use of fire extinguisher)' (Q7.2), the 'Main action taken by FRS' (Q7.3) and the 'Type of firefighting equipment used by FRS' (Q7.5) (both including use of portable extinguishers). Findings: We could produce a table showing the breakdown of fires by 'Action taken by public prior to FRS arrival'. In 60-65% cases, no action is taken and, where action is taken, no information is recorded on the direct impact. It may be more useful to show this in combination with other fields though it would be difficult to show a direct link between the action taken by public and fire suppression. We could produce a table showing the breakdown of 'Main action taken by the FRS' at two levels of detail (e.g. with all portable extinguisher use grouped as one category, or broken down by type of portable extinguisher used). Given the drawbacks of data overload and the finite nature of the report, we would need to assess demand level/utility (if using in conjunction with other fields). A table showing the breakdown of 'Manual systems used by FRS' shows none to be recorded in 99% of cases. Suggest: Table showing breakdown of manual system used would not be useful as only 1% of primary fires have a manual system type recorded. Currently Tables 26 and 26a contain the number and percentage of fires by the number of appliances in attendance.
Damage to equipment and injuries to staff as a result of antisocial behaviour	Data on attacks on firefighters at an incident is extracted from IRS data, checked by Scottish Fire and Rescue Service staff and included in the Fire and Rescue Service Statistics, Scotland publication. Data on damage to equipment is not collected.

Further investigation into the utility value of providing tables of fires broken down by public/FRS action required before including in national publication. Data to be provided on request or published as dataset for users to access.

Specialist topics

Some of the data requests were for topics of particular interest to a small group of people. Data on prison fires and in kitchen grease extract systems were examples of this. This data is more suitably provided for specific research projects.

One respondent requested quality assurance data. Since IRS data undergoes multiple QA processes before and after it is received by Scottish Government, the quality of the published data has to be taken to be as good as it can be. If the request covers internal fire service processes, these are the responsibility of Scottish Fire and Rescue Service, who have to register an incident report as 'Published' in the IRS system before it is available to Scottish Government.

Specific, ad hoc data requests will be considered by Scottish Government

FRS data - not incident data

Some of the extra data requested is not collected under the IRS data collection system. The items requested are not reported as part of the initial investigation focus of the incident data, so much as being associated with related data. Examples of this are:

• Had the premises had a fire risk assessment or home fire safety visit?
• Were any notices issues or audits carried out under the Fire (Scotland) Act?

In these cases, data linkage may help to provide the information required and Fire and Rescue Service Statistics, Scotland would provide one source of the data.

6.9.2 Fire and Rescue Service Data

Fire and Rescue Service Data

As with the Fire Statistics, Scotland (incident) data, there was a considerable list of extra data requests. Table 13 shows all of the comments (i.e. if a comment is repeated by more than one respondent it will be in the table more than once) grouped.

Non-domestic fire safety	The reporting of non-domestic fire safety activity should be simplified. The existing arrangement is complex and does not readily allow benchmarking or meaningful analysis. More information with regard to enforcement visits
Data sourcing	SFRS has other data sources that would be of use (for example, the Community Safety Engagement Toolkit (CSET) which records the number and type of home fire safety visit conducted, and other community safety initiatives) Consideration should be given to transferring / incorporating summaries of the following statistical data into the Fire Statistics, Scotland publication. HFSV information and data (obtained from CSET) Recorded attacks on SFRS personnel Fire (Scotland) Act 2005 audit and inspection information and data. This would consolidate all relevant prevention and protection related data into a single reporting document.
Incident data (i.e. not related to this section of the consultation)	More detail on fire incidents and fire false alarms in prisons Detail of where fire has spread beyond a commercial kitchen via the grease extraction system and fire spread via ducting e.g. high rise social housing extract systems, factory extract systems etc. Statistics for fire incidents in Scotland's statutorily protected heritage buildings (Category A, B & CS). This will assist in the protection of this vital national resource through tailored fire safety management guidance and to inform future research.

Table 13 : Extra data requests for the Fire and Rescue Service Statistics

There were several responses here that referenced incident data. Although these are shown in Table 13, they will not be discussed as they were all already mentioned in section 6.9.1 above.

Non-domestic fire safety data

On the non-domestic fire safety data collection there was one request to simplify the data and one to extend it. The request to simplify it suggests reducing the quantity to improve the comparability of it. The current data is intended to assess the quantity of audits carried out, their targetting according to risk and the time taken to do this. To achieve these aims the data has several parameters, which are all relevant. The data definitions were discussed and agreed with the CFOA Scotland fire safety group prior to previous data collections (see Fire and Rescue Service - Data Review - Part 1^[5] ). Over 50% of respondents selected the non-domestic fire safety data as a priority topic (see Section 5.2). Overall, support for the data was felt to be strong, although the criticisms were noted.

For the current publication, we will continue to provide the non-domestic fire safety fields as they were in the previous publication. However, we note the comments and will take what quality assurance actions we can to improve the data.

Scottish Fire and Rescue Service data sources

We note that the new Scottish Fire and Rescue Service has developed and adopted nationally a number of data capture systems. Alignment with these systems is an activity that Scottish Government would be very willing to participate in, although under current plans this is the final year of Scottish Government collecting the Fire and Rescue Service data. Should this plan change, we will revisit the data. It is however important to note that the current data collections exist for a reason and continuity of data is important. It is for this reason that a data steering group is intended to have oversight of the future publication - a point underlined by a response in Section 5.3.1

When the Fire and Rescue Service Statistics are handed over to the Scottish Fire and Rescue Service, the data steering group should take into account both the continuity of existing data and alignment with Scottish Fire and Rescue Service systems.

6.10 Remove data fields

There were no requests to remove data from the Fire Statistics, Scotland publication. On the Fire and Rescue Service Statistics Scotland publication one respondent requested removing some elements of the non-domestic fire safety data collection. This request has been discussed in Section 0 above.

6.11 General comments

Some respondents discussed publication timings in the general comments of the consultation. Equal numbers stated that annual publications were suitable and that quarterly updates would be helpful. While we recognise the value of frequent data, the resources and data burden involved in providing data more frequently than once a year is prohibitive. Users also emphasised the value of timely data.

The provision of rates in addition to simple numbers is important to users so that benchmarking is possible, as is the provision of a long-term trend. Both of these are provided in the additional datasets on http://www.scotland.gov.uk/Topics/Statistics/Browse/Crime-Justice/Datasets/DatasetsFire. Because we have received several requests where respondents are asking for data that is already published, it seems that it may not be immediately clear what data exists within, and associated with, the current publications. We will consider what we can do to make it easier to know about and find all of the statistics that we publish.

Scottish Government will produce annual data publications as soon as possible after the data is collected and key statistics will continue to be provided with long-term trend data and as a rate.

We will explore ways of making published data easier to find.