Skip to main content
SpringerLink
Log in

Fire Department Attended and Unattended Fires: Estimates from the 2004–2005 National Sample Survey and Comparison with Previous Surveys

  • Published:
Fire Technology Aims and scope Submit manuscript

Abstract

Previous national surveys in 1974 and 1984 have shown that although attended and unattended fires differed substantially in severity and fire losses, there were between 10 and 29 unwanted residential fires for every fire reported to, or attended by, U.S. fire departments. The study objective was to obtain new estimates of fires not attended by fire departments. Interest in unattended fires derives from the understanding that most fires begin small, then unless controlled, grow until fire department assistance is needed. To update these analyses, a national telephone survey was conducted during 2004 and 2005. The survey had 916 respondents who reported one or more residential fires during the previous 90-day period. The principal methodological issues in analyzing the survey data included: (1) determining the optimum recall period to balance sampling variance and bias, and (2) imputing incompletely specified fire dates. The resulting estimates were 7.2 million unattended residential fires per year, a 69% decrease from the 1984 survey estimate of 22.9 million fires. During the same time period, fire department attended residential fires decreased by 36%. The greater decrease in unattended fires is at variance with the conjecture in the 1984 survey that increasing availability of smoke alarms would result in more fires detected at an earlier stage when they could be controlled by residents; a conjecture that would predict a greater decrease in attended rather than unattended fires.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Figure 1

Similar content being viewed by others

Notes

  1. The full survey report is in [1]. A brief report on the methods for analyzing the survey data was presented at the 2009 Joint Statistical Meetings [2]. Both reports are available from the contributing author.

  2. A review of the literature has not resulted in discovery of any other national household probability samples of unreported fires.

  3. The Current Population Survey is conducted by the Bureau of the Census for the Bureau of Labor Statistics. See http://www.census.gov/cps/.

  4. In 1984, there were an estimated 124,100 residential structure fires involving cooking equipment reported to U.S. fire departments [9] and 605,500 residential structure fires [5]. Estimates of unreported cooking/kitchen appliance or equipment fires are found in [4, p. 37].

  5. This is described in [4, p. 11]. They estimated that there were 13 million fires of which 11,830,000 were unreported and unattended by fire departments. Subtraction results in 1,170,000 fire department attended incidents.

  6. The survey questions are found in [4, Appendix B, p. 2]. This is the actual wording. After asking the same question as the 1974 survey, “… We are interested in asking about any fires—large or small that you have had in or around your home, vacation home, or on your property,” the interviewer continued with a definition by reading the following to the respondent: “By ‘fire,’ I mean any incident—large or small—that resulted in flames or smoke, and could have caused damage to life or property if left unchecked.” If the respondent did not acknowledge a fire incident, then the interviewer prompted with various types of incidents similar to the 1974 survey.

  7. In [4, pp. 20 and 22]. The report presented a 95% confidence interval for only total residential fires, a somewhat larger quantity that included all unattended and attended home and motor vehicle fires. This was an estimated total of 25.2 million fires, with a coefficient of variation of 3.54%. We applied that coefficient of variation results to the 22.9 million unattended fires. Based on a slightly smaller sample size, the coefficient of variation and associated confidence intervals are likely to be a little larger than the values used here.

  8. The estimated number of households was reported in [4]. More recent estimates from the Bureau of the Census [10] were 85.4 million households in 1984, which would lower the per-household rate to 27.8 fires per 100 households per year.

  9. Response rate calculations were made by dividing the number of responses by the sum of responses and refusals. RR2 and RR4 differ in handling households with unknown eligibility in the calculation. A household’s eligibility for this survey was unknown if the phone was always busy, there was no answer after repeated dialings, there was always an answering machine or the respondent broke off the interview before eligibility can be established. RR2 considers all respondents with unknown eligibility to be eligible and to have refused to answer the survey. In contrast, RR4 considers a proportion of the unknown elgibles to be eligible refusals. The proportion is the known eligibles divided by the sum of the known eligibiles and known ineligibles. With a smaller denominator, RR4 is usually larger than RR2. Note that the large difference between these two rates is a result of about 29% of the numbers dialed determined to be of unknown eligibility. The largest two categories were no answer (11.3%) and where the screening interview could not be completed (12.7%). For more information on the AAPOR response rate calculations see [11, pp. 36–37].

  10. For example, if the recall period was 1 week, the estimates would be multiplied by 52; 2 weeks by 26; 3 weeks by 17, etc.

  11. The first week was disqualified as the reference value because it was “… estimated to be affected the most by the possible discrepancy between the recorded interview date and the date the respondent completed the injury section…” [18].

  12. These 12 missing fire dates were about 3.8% of the missing dates and about 1.2% of the total dates. There were more complicated imputation approaches available for imputation of these dates, but they did not seem warranted because of the few cases involved.

  13. R is a freely available language and environment for statistics and statistical computing at http://www.r-project.org/.

  14. A spline is a mathematical function (graphically, a curve), whose shape is somewhere between the regression line (a straight line with least squares properties) and a line that connects all the points (an interpolating function). The spline is calculated by imposing a weight penalty on the lack of smoothness in the interpolating function that is traded off against the sum of the squared deviations in the regression line. Large weight functions move the curve toward the regression line, while lower weight functions move it toward the interpolating function. Typically, splines are between the regression line and the interpolating function. See Hastie and Tibshirani [22] for a complete description.

  15. The NFPA estimate is a “midyear” estimate obtained by averaging the 2004 estimate of 395,500 residential structure fires and the 2005 estimate of 381,000 fires. These appeared in Karter [6, 7]. Ninety-five percent confidence intervals for these estimates are (383,500, 407,500) and (371,000, 391,000), respectively. The confidence intervals were provided separately by Michael J. Karter, Jr. of the National Fire Protection Association.

  16. Direct comparisons between the two surveys can be difficult because the 1974 survey report is based on the full year recall period not the 3 month recall period used for estimating fire incidence, and the 1984 survey report is based on the 3 month recall period, not the 1 month period used for incidence estimates. Table 3 and 4 in the 1974 study [3, p. 23] shows 1.855 million incidents (40.8%) involving cooking, grease and food of a total of 4.5 million incidents where the source of the fire or the item first ignited was known. In the 1984 study, 77.7% of incidents were shown to have involved cooking/kitchen appliances or equipment [4, p. 37]. Note that these estimates are from the full 1 year recall period for the 1974 survey and the 3 month recall period in the 1984 survey.

  17. Estimated unattended fires per 100 households as follows: 1984, 22.9 million/(85.4 million/100) = 26.8 fires per 100 households; 2004–2005 7.2 million/(113.3 million/100) = 6.3 fires per 100 households. Estimated attended fires per 100 households using the NFPA survey estimates as follows: 1984, 605,500/(85.4 million/100) = 0.71 fires per 100 households; 2004–2005 388,250/(113.3 million/100) = 0.34 fires per 100 households.

  18. There is an emerging literature on this topic, but it is not always clear that the declining response rates introduce bias into the results. See for example reference [23] on estimating the prevalence of smoking and reference [24] on political polling.

References

  1. Greene MA, Andres C (2009) 2004–2005 national sample survey of unreported residential fires. U.S. Consumer Product Safety Commission, Washington, DC

  2. Greene MA, Andres C (2009) Methods for analysis of recall of fires in a retrospective survey. In: Proceedings of the joint statistical meetings section on survey research methods, Washington, DC

  3. U.S. Consumer Product Safety Commission (1978) Special report: results of national household fire survey. U.S. Consumer Product Safety Commission, Washington, DC

  4. Audits and Surveys (1985) 1984 national sample survey of unreported, residential fires. Audits and Surveys, Princeton

  5. Ahrens M (2010) Home structure fires. National Fire Protection Association, Quincy, MA

    Google Scholar 

  6. Karter MJ (2005) Annual fire loss in the U.S. report 2004. NFPA J 99: 46-51

    Google Scholar 

  7. Karter M J Jr (2006) Annual fire loss in the U.S. report 2005. NFPA J 100:44-49

    Google Scholar 

  8. Ahrens M (2008) Home smoke alarms—the data as context for decision. Fire Technol, 44: 313-327

    Article  Google Scholar 

  9. Ahrens M (2009) Home fires involving cooking equipment. National Fire Protection Association, Quincy, MA

    Google Scholar 

  10. U.S. Bureau of the Census, Table HH-1 Households, by Type: 1940 to Present, Current Population Survey, March and Annual Social and Economic Supplements, 2009 and Earlier. www.census.gov./population/socdemo/hh-fam/hh1.xls

  11. American Association for Public Opinion Research (2000) Standard definitions: final dispositions of case codes and outcome rates for surveys.” AAPOR, Ann Arbor, MI

    Google Scholar 

  12. Harel Y, Overpeck MD, Jones DH, Scheidt PC, Bijur PE, Trumble AC, Anderson J (1994) The effects of recall on estimating annual nonfatal injury rates for children and adolescents. Am J Public Health, 84,4, 599-605

    Article  Google Scholar 

  13. Massey JT, Gonzalez JF (1976) Optimum recall periods for estimating accidental injuries in the national health interview survey. In: Proceedings of the American Statistical Association, Social Statistics Section, Boston

  14. Cummings P, Rivara FP, Thompson RS, Reid RJ (2005) Ability of parents to recall the injuries of their young children. Inj Prev, 11, pp. 43-47

    Article  Google Scholar 

  15. Landen DD, Hendricks S (1995) Effect of recall on reporting of at-work injuries. Public Health Rep, 110:3, 350-354

    Google Scholar 

  16. Mock C, Acheampong F, Adei S, Koepsell T (1999) “The effect of recall on estimation of incidence rates for injury in Ghana.” Int J Epidemiol, 28, 4, pp. 750-755.

    Article  Google Scholar 

  17. Moshiro C, Heuch I, Astrom AN, Setel P, Kvale G (2005) Effect of recall on estimation of non-fatal injury rates: a community based study in Tanzania. Inj Prev 11:48–52

    Google Scholar 

  18. Warner M, Schenker N, Heinen MA, Fingerhut LA (2005) “The effects of recall on reporting injury and poisoning episodes in the National Health Interview Survey.” Inj Prev, 11:282-287.

    Article  Google Scholar 

  19. Schaefer JL (1997) Analysis of incomplete multivariate data. Chapman and Hall, New York

    Book  Google Scholar 

  20. SAS Institute Inc (2004) SAS/STAT® 9.1 users guide. SAS Institute, Cary, NC

    Google Scholar 

  21. Little RJA, Rubin DB (1987) Statistical analysis with missing data. John Wiley and Sons, NY

    MATH  Google Scholar 

  22. Hastie TJ, Tibshirani RJ (1990) Generalized additive models. Chapman and Hall, NY

    MATH  Google Scholar 

  23. Biener L, Garrett CA, Gilpin EA, Roman AM, Currivan DB (2004) Consequences of declining survey response rates for smoking prevalence estimates. Am J Prev Med 27(3): 254-257

    Article  Google Scholar 

  24. Keeter S, Kennedy C, Dimock M, Best J, Craighill P (2006) Gauging the impact of growing nonresponse on estimates from a national RDD telephone survey. Pub Opin Q 70(5): 759-779

    Article  Google Scholar 

  25. Blumberg SJ, Luke JV (2010) Wireless substitution: early release of estimates from the National Health Interview Survey July—December 2009. National Center for Health Statistics. http://www.cdc.gov/nchs/nhis.htm

  26. Wilson J, Akoto J, Dixon S, Jacobs D (2008) Evaluation of the “10 year” smoke alarm project. Report prepared for the Centers for Disease Control and Prevention by the National Center for Healthy Housing

  27. Smith CL (1994) Smoke detector operability survey report on findings. U.S. Consumer Product Safety Commission, Washington, DC

    Google Scholar 

  28. Ballesteros MF, Kresnow MJ (2007) Prevalence of residential smoke alarms and fire escape plans in the U.S.: results from the second injury control and risk survey (ICARIS-2). Public Health Rep, 122, 224-231

    Google Scholar 

  29. Peek-Asa C, Allaredy V, Yang J, Taylor C, Lundell J, Zwerling C (2005) “When one is not enough: prevalence and characteristics of homes not adequately protected by smoke alarms.” Inj Prev, 11: 364-368

    Article  Google Scholar 

Download references

Acknowledgments

Members of the Consumer Product Safety Commission (CPSC) staff study team who contributed to this project included Linda E. Smith (retired), and William W. Zamula. Kathleen A. Stralka, Russell H. Roegner, Erlinda M. Edwards, Gregory B. Rodgers and Eileen J. Williams read drafts of this manuscript and contributed valuable suggestions. The authors also appreciate the comments of the editor and two anonymous referees. The telephone survey was conducted by Synovate, Inc. Alan Roshwalb designed the sampling plan, the sample weighting, and prepared the SAS® dataset used for analysis. Tim Amsbury and John Lavin were instrumental along with CPSC staff in the design of the questionnaire and supervised the data collection. The project was supervised by Corporate Vice President, W. Burleigh “Leigh” Seaver. In addition to funding from the Consumer Product Safety Commission, funding was also provided by the Division of Unintentional Injury Prevention of the National Center for Injury Prevention and Control in the Centers for Disease Control and Prevention, Department of Health and Human Services and the United States Fire Administration, Department of Homeland Security.

Author information

Authors and Affiliations

  1. U.S. Consumer Product Safety Commission, 4340 East West Highway, Bethesda, MD, 20814, USA

    Michael A. Greene

  2. U.S. Army Aberdeen Test Center, Aberdeen Proving Grounds, 400 Colleran Road, Aberdeen, MD, 21005-5059, USA

    Craig D. Andres

Authors
  1. Michael A. Greene
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Craig D. Andres
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Michael A. Greene.

Additional information

This report was prepared by the staff at the U.S. Consumer Product Safety Commission (CPSC) and has not been reviewed or approved by, and may not necessarily reflect the views of, the Commission. Because this report was prepared in the author’s official capacity, it is in the public domain and may be copied freely.

Michael A. Greene retired from CPSC.

Rights and permissions

Reprints and permissions

About this article

Cite this article

Greene, M.A., Andres, C.D. Fire Department Attended and Unattended Fires: Estimates from the 2004–2005 National Sample Survey and Comparison with Previous Surveys. Fire Technol 48, 269–289 (2012). https://doi.org/10.1007/s10694-011-0215-z

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s10694-011-0215-z

Keywords

Search

Navigation