5. Notes on statistics used in this bulletin
5.1 Source of Statistics
5.1.1 Responsibility for the Fire Service was transferred from the Department for Transport, Local Government and the Regions ( DTLR) to the Office of the Deputy Prime Minister ( ODPM) in May 2002 (the Home Office was responsible for the Fire Service until June 2001).
5.1.2 The statistics in this bulletin are compiled from reports submitted to the ODPM on fires attended by local authority fire brigades. Although a variety of Scottish fire statistics are currently published annually by the ODPM (see references below) this is the fourth time that detailed data for the eight Scottish fire brigades have been collated and reported by the Scottish Executive in the form of a dedicated Scottish fire statistics bulletin. This bulletin is intended to provide a detailed overview of Scottish fire statistics over the past decade, at a brigade and Scotland level with comparison to the other UK regions. More detailed information is collected on all fires in buildings, vehicles and outdoor structures and any fires involving casualties or rescues ( i.e. "primary" fires). Less detailed aggregated information is collected on "secondary" and chimney fires; so subsequent analysis of them is limited. Brief definitions of the fires and false alarms reported on are given below, however fuller definitions are given in section 5.10.
5.1.3 Data for 2002 and 2003 include estimates for November 2002 and January and February 2003 to account for the lack of information recorded during national industrial action. These estimates have been produced using comparable data from earlier years. See section 5.14 below.
5.2 Definitions: primary fires, secondary and chimney fires and false alarms
5.2.1 " Primary" fires include all fires in buildings, vehicles and outdoor structures or any fire involving casualties, rescues, or fires attended by five or more appliances. " Secondary" fires are the majority of outdoor fires including grassland and refuse fires unless they involve casualties or rescues, property loss or five or more appliances attend. They include fires in single derelict buildings. Chimney fires are any fire in an occupied building where the fire was confined within the chimney structure (and did not involve casualties or rescues or attendance by five or more appliances). A false alarm is defined as an event in which the fire brigade believes they are called to a reportable fire and then find there is no such incident. The term " outdoor fires" used in this Bulletin refers to primary and secondary fires in road vehicles, other outdoor property, derelict buildings and derelict vehicles and more minor refuse, grassland and intentional straw/stubble fires.
5.3 Data for primary fires
5.3.1 The source of these data on "primary fires" is the fire report form FDR1 (94) introduced from January 1994. The count specifically included two categories of fire related incident not recorded in the fire statistics from the early 1980s to 1993. These categories are described as late fire calls and heat and smoke damage only incidents:
- "late fire calls" are fires not attended as an emergency because they are known to be extinguished when the call was made, or to which no emergency call was made;
- heat or smoke damage only 'fires' - where no fire damage is reported (see definitions paragraph 5.10, for fuller explanation).
5.3.2 Analyses for 1994 and later years include both "late fire calls" and heat or smoke damage only 'fires'. Because of the way that these types of fire were processed in 1995, this involved some additional estimation for missing data.
5.4.1 A person whose death is attributed to a fire is counted as a fatality even if death occurred weeks or months later. However, it is not always the case that fire was the cause of death. The figures for fatalities are thus subject to revision, as information supplied by fire brigades needs to be cross-checked against the cause of death that appears on the death certificate exercise. This publication contains the finalised figures for 2001 and 2002 deaths following the death certificates exercise.
5.4.2 A comprehensive revision exercise was undertaken in 1998, which resulted in major revisions (mainly for fires in road vehicles) to fire related deaths occurring between 1988 and 1996. A further data refinement exercise was undertaken in June 2000 to complete the revision exercise undertaken to the 1998 deaths information. The 2000 exercise resulted in some very minor changes to the information held on non-fatal casualties, rescues and the information held on the location of some fatal casualties for the years 1989 to 1993.
5.5 Deliberate and accidental fires
5.5.1 Deliberate fires include those where deliberate ignition is merely suspected, and those recorded by the brigade as "doubtful".
5.5.2 As a result of a review of the assessment of the motive of primary fires the classification of certain types of primary fires was changed from accidental to deliberate, with effect from 1994. The effect appeared most noticeable for buildings fires, where an estimated 4,000 UK fires may have been categorised as deliberate where they were previously classed as accidental. For example, the main change for fires in dwellings was those caused by 'playing' with fire - in particular if children were involved. Prior to 1994, it was assumed that they were caused accidentally. In 1994 it was estimated that 1,500 such fires would have been recorded as accidental in 1993. In buildings other than dwellings, the main change occurred for fires that spread to them from small (secondary) fires, where an estimated 2,000 were recorded as deliberate instead of accidental. Within the deliberate category, a higher proportion of fires was recorded as "doubtful" or "unspecified", rather than "ignition of other property". These changes affect comparisons before and after 1994, but do not affect comparison between 1994 and later years.
5.5.3 Data for deliberate and accidental fires were further revised in 2000. The revised data corrected an error which resulted in a small distortion (about 3,500 UK fires) in the number of deliberate and accidental fires for the years 1996 to 1998. This error meant that the number of accidental fires were under represented by about 3,500 in these years, whilst malicious fires were over represented by 3,500. The total number of primary fires (involving property or casualties) for these years were not affected.
5.6 False alarms
5.6.1 The marked changes between 1995 and 1996 in the numbers of false alarm calls for "good intent" and "due to apparatus" are partly the result of coding changes. Guidance clarifying false alarm calls to be included in the "due to apparatus" category as applied by some fire brigades has resulted in some false alarm calls being classified into this category rather than "good intent".
5.7 Selection of samples of primary fires, since 1994
5.7.1 The databases for 1994 to 2004 contain all fires with casualties but only a sample of other fires. The samples were selected systematically for each brigade, with the following approximate sampling fractions: 1994 - 10 per cent; 1995 - 40 per cent, 1996 to 2003 - 20 per cent.
5.7.2 The data have all been weighted to agreed brigade totals. The detailed analysis of data other than for casualties, or fires involving casualties are based on the sampled data grossed to brigade totals.
5.8 Statistical sample confidence limits
5.8.1 The sample may well produce figures that differ from the ones which would have been obtained if all forms (the population) had been used. We can construct statistical confidence limits for each estimate at the 95 per cent level of significance. This means we can be confident that there is a 95 per cent chance that the limit contains the true population value of interest. The methodology for constructing confidence limits is shown below.
5.8.2 A parameter (R) from the fire incident population (N) will take the form of a proportion (P=R/N) with a particular attribute, for example the number of dwelling fires as a proportion of fires in all locations. Using the normal distribution approximations gives confidence limits for p as:
Proportion of fires with a given attribute
n / N = Sampling fraction
Known population size
Normal distribution significance level a ( i.e. a=0.05 => Z a=1.96)
However, this normal approximation requires that
(i) n << min (R,N-R)
(ii) min(nP,nQ) = 30
(see COCHRAN, William G, 'Sampling Techniques', Wiley Publications 1977, (pp57-60) for details)
5.8.3 We cannot use the approximations safely if a proportion (p value) close to zero or one is being estimated. When the normal approximation does not apply, limits for p may be found from binomial tables ( e.g. Fisher and Yates, Statistical Tables (1957)).
5.9 An example for car fires
5.9.1 In 2001, there were 228,339 primary fire incidents in the UK, of which 90,447 primary fire incident forms were entered onto the database. Of these forms, 55,811 were for fires involving casualties and other types of fire where 100 per cent were coded ( e.g. data received electronically from participating brigades) and the remaining 34,636 were other types of fire which only a sample were coded. Therefore, the 34,636 (n) sampled forms are based on a population of 228,339 - 55,811 = 172,528 forms and the overall sampling fraction (f) was 20 per cent.
5.9.2 Calculating the 95 per cent confidence limits for fires in cars (88,301) gives a result of ±1,047. Therefore, the number of UK car fires with 95 per cent confidence is between 87,254 and 89,348 .
5.10.1 From 1994, with the introduction of the fire report form FDR1 (94) some definitions were changed and clarified. The following list shows definitions applicable from 1994:
A reportable fire(no specific definition prior to 1994)
A reportable fire is an event of uncontrolled burning involving flames, heat or smoke and which the brigade attended.
Late fire call(no specific definition prior to 1994)
A fire known to be extinguished when the call was made (or which no call was made, e.g. a fire which comes to the attention of the brigade as a result of a press report or inquest) and the brigade attended.
Heat or smoke damage incidents(no specific definition prior to 1994)
These are reportable 'fires' where there is no fire damage. The damage reported may be due to any combination of heat, smoke and other which will include any water damage.
The type of premises, property or countryside in which the fire started. This is not necessarily the type of premises in which most casualties or damage occurred as a result of the fire.
Cause of fire
The defect, act or omission leading to ignition of the fire.
Includes fires where deliberate ignition is merely suspected, and recorded by the brigade as "doubtful".
Includes fires where the cause was not known or unspecified.
Source of ignition
The source of the flame, spark or heat that started the fire.
Spread of fire
The extent to which fire damage (as opposed to heat, smoke or other damage) spread, for example, beyond the room of origin.
All buildings including those under construction, but excluding derelict buildings or those under demolition. Prior to 1994 'buildings' were referred to as 'occupied buildings'.
Buildings occupied by households, excluding hotels, hostels and residential institutions. From 1988, mobile homes have been specifically included in the dwelling count. In 2000, the definition of a dwelling (for FDR1 purposes) was widened to include any non-permanent structures used solely as a dwelling, such as caravans, houseboats etc. (amounts to 0.3 per cent of the total number of UK dwelling fires). This change brings the definition of a dwelling more in line with that required under Best Value legislation. All analyses from 1994 to 1998 relating to dwellings were retrospectively revised to include the new categories of dwelling (prior to 1994 these categories were included in the dwelling count). Caravans, boats etc.not used as a permanent dwelling are shown according to the type of property (caravan, vehicle etc.).
Primary fires(no specific definition prior to 1994)
These are reportable fires (as listed below i) to vi)) or any fires involving casualties, rescues, or fires attended by five or more appliances regardless of its actual physical location. An appliance is counted if either the appliance, equipment from it or personnel riding it, were used to fight fire.
(ii) Caravans, trailers etc
(iii) Vehicles and other methods of transport (not derelict)
(iv) Outdoor storage, plant and machinery
(v) Agricultural and forestry premises and property
(vi) Other outdoor structures including post boxes, tunnels, bridges etc.
These are reportable fires that:
- were not in primary fire locations (before 1994 defined as a list of locations)
- were not chimney fires in buildings
- did not involve casualties or rescues
- were attended by four or fewer appliances. An appliance is counted if either the appliance, equipment from it or personnel riding on it, were used to fight the fire.
They are reported in less detail than other fires and consequently less information concerning them is available.
These are reportable fires in occupied buildings:
- where fire was confined within the chimney structure
- that did not involve casualties or rescues
- attended by four or fewer appliances
A person whose death is attributed to a fire is counted as a fatality even if death occurred weeks or months later - see paragraph 5. for a more detailed discussion.
Non-fatal casualties consist of persons requiring medical treatment beyond first aid given at the scene of the fire, and those sent to hospital or advised to see a doctor for a check-up or observation (whether or not they actually do). People sent to hospital or advised to see a doctor as a precaution, having no obvious injury, are recorded as "precautionary check-ups".
Refers to fatal or non-fatal casualties involving persons who are not members of the fire brigade.
Refers to fatal or non-fatal casualties involving fire brigade personnel.
A false alarm is defined as an event in which the fire brigade believes they are called to a reportable fire and then there is no such incident. False alarms are categorised as:
Malicious - the call was made with the intention of getting the brigade to attend a non-existent fire-related incident. This includes a 'deliberate' and 'suspected malicious' intentions.
Good Intent - the call was made in good faith in the belief that the brigade really would attend a fire.
Due to Apparatus - the call was initiated by fire alarm and fire fighting equipment operating (including accidental initiation of alarm apparatus by a person).
Details of the main definitions and classifications that applied to data for the period 1978 to 1993 are set out in the publication "Fire Statistics United Kingdom 1993". These are the definitions used during the period of the previous fire report form ( FDR1).
5.11 Standard Industrial Classification (Trade or business)
5.11.1 In 1994, the classification of industrial fire locations was simplified. The new codes were based on the Standard Industrial Classification (Revised 1980) ( SIC) published by the Central Statistical Office but combined or added categories where relevant to fire risk. The aim is to identify the main trade or business carried on where the fire started. Comparisons with data for the period 1988 to 1993 are affected as follows:
- "Mining and quarrying" replaces "Extraction of coal etc." and includes all mining and quarrying e.g. metal extraction.
- "Metal manufacturing" and "Transport and vehicle manufacturing" replace "Metal manufacturing and extraction" (with extraction now included in "Mining and quarrying") and "Metal goods engineering and vehicle manufacture".
- "Other manufacturing" includes office machinery and computers, electrical machinery, radio, television and communication equipment, medical, precision, optical, watch and clock manufacture and other not elsewhere specified, and replaces other manufacturing, and some "other unspecified" categories although the comparison may not be exact.
- "Recycling" is a new category which was previously included in one of the "other" categories. Outdoor recycling points are included with "outdoor storage" and not as previously in "other".
- "Retail Distribution" includes the sale and repair of goods which were previously included with sale and repair of motor vehicles. A new category for "sale and repair of motor vehicles" is shown separately in detailed tables. Also included in retail distribution is trading premises of bank and building societies which may have been previously included with financial intermediation.
- "Recreational and other cultural services" now includes "places of worship" previously in "other".
- "Children's and disabled homes" excludes some other welfare establishments now in "other miscellaneous services".
5.12.1 Symbols used in the tables are:
- Nil or less than half the final digit shown.
.. Not available/Not applicable
5.13 Publications of fire statistics
Previous statistical publications on fires include:
Fire Statistical Bulletins: UK Fire Statistics Monitor: Q1 2005
Fire Statistics Monitor: Q4 2004
Fire Statistics Monitor: Q2&3 2004
Fire Statistics Monitor: Q1 2004 (1/04)
Fire Statistics, United Kingdom, 2004
Fire Statistics, United Kingdom, 2003
Fire Statistics, United Kingdom, 2002
Fire Statistics, United Kingdom, 2001 (03 FHSD 01009)
Fire Statistics Estimates, United Kingdom, 2001 (02 FPD 00874)
Fire Statistics, United Kingdom, 2000 ( FS/ PPU/2765)
Fire Statistics, Untied Kingdom, 1999 (20/00)
Fire Statistics, United Kingdom, 1998 (15/99)
Fire Statistics, United Kingdom, 1997 (25/98)
Summary Fire Statistics, United Kingdom, 1996 (1/98)
Summary Fire Statistics, United Kingdom, 1995 (8/97)
Summary Fire Statistics, United Kingdom, 1994 (13/96)
Up to and including data for 1993, detailed information on fires, including location, cause, spread and casualties were published by the Home Office in "Fire Statistics United Kingdom". Copies of the above publications can be downloaded from the Office of the Deputy Prime Minister website at http://www.safety.odpm.gov.uk/fire.
Scottish Fire Statistical Bulletins: HM Chief Inspector of Fire Services for Scotland Report for 2004-2005 ( SE/2005/212)
Fire Statistics Scotland, 2003 (CrJ/2005/3)
HM Chief Inspector of Fire Services for Scotland Report for 2003-2004 ( SE/2004/275)
Fire Statistics Scotland, 2002 (CrJ/2004/2)
HM Chief Inspector of Fire Services for Scotland Report for 2002-2003 ( SE/2004/3)
HM Chief Inspector of Fire Services for Scotland Report for 2001-2002 ( SE/2002/335)
Fire Statistics Scotland, 2001 (CrJ/2003/4)
HM Chief Inspector of Fire Services for Scotland Report for 2000-2001 ( SE/2001/266)
Fire Statistics Scotland, 2000 (CrJ/2002/3)
HM Chief Inspector of Fire Services for Scotland Report for 1999-2000 ( SE/2001/86)
The above publications report on the number of primary fires, secondary fires and casualties at brigade area/ Scotland level and are published by the Scottish Executive. Copies of the above publications can be downloaded from the Scottish Executive's website at http://www.scotland.gov.uk/publications .
5.14 Industrial Action
5.14.1 Due to the industrial action by firefighters in November 2002 and January and February 2003 the reporting of fires for these days was disrupted. In total 15 24-hour periods were affected:
- From 18:00 on 13th to 18:00 on 15th November 2002 (2 days)
- From 09:00 on 22nd to 09:00 on 30th November 2002 (8 days)
- From 09:00 on 21st to 09:00 on 22nd January 2003 (1 days)
- From 09:00 on 28th to 09:00 on 30th January 2003 (2 days)
- From 09:00 on 1st to 09:00 on 3rd February 2003 (2 days)
5.14.2 Some information relating to this period has been received via data collected by the Ministry of Defence and also from special summary reports submitted by brigades.
5.14.3 In order to present the most accurate picture of the fires, casualties and false alarms which occurred in 2002 and 2003, in some cases estimates have been incorporated into the figures to account for missing information. These have been produced using comparable data for November 2001, January 2002 and February 2002. However, for the more complex tables the figures are as reported by the fire and rescue service and therefore exclude any estimates for strike days. All the tables are footnoted accordingly.