Illinois Cooking Fire Risk Analysis

Cooking Fire Definition and Data Quality

A qualifying cooking fire event is defined as NFIRS incident type 113 (confined cooking fire) or incident type 111 (structure fire) where the recorded area of origin is coded 24 (kitchen). This definition is built on area of origin rather than on item first ignited or heat source, and that choice reflects a known, documented weakness in NFIRS causal data rather than a judgment call made for convenience.

Item first ignited, heat source, and cause of ignition are the fields NFIRS relies on to establish that a fire was specifically cooking related, as opposed to merely located in a kitchen. But these are also the fields with the highest rates of undetermined or missing entries anywhere in the NFIRS dataset. The U.S. Fire Administration's own 2017 review of NFIRS data quality identifies the causal factor fields, including heat source and item first ignited, as the chronic weak point of the system, a problem serious enough that the National Association of State Fire Marshals commissioned dedicated research on it. Restricting this analysis to incidents with a fully coded ignition chain would exclude a substantial share of otherwise valid records and bias the ranking toward departments with better records management rather than toward departments with genuinely higher cooking fire risk. Area of origin, a structural field with a much lower undetermined rate, is the more stable basis for a comparative ranking across hundreds of departments and five years of data, even though it is a broader net than ignition cause coding alone.

That broader net cuts in the department's favor more often than not. Peer reviewed NFPA research on NFIRS incident typing, Kinsey and Ahrens, 2016, found that a substantial share of incidents fire crews describe on scene as a fire that was out on arrival, the pattern typical of a stovetop or pan fire extinguished before the department arrived, are coded not as a fire at all but as a false alarm, a good intent call, or a service call. That same research documents burned food and cooking related incidents scattered across incident type codes well outside 111 and 113. The direction of that error is undercounting, not overcounting: a meaningful number of true cooking fires never appear in this analysis's incident type filter to begin with, because the incident was never coded as a fire in the source data.

Taken together, these two data quality issues push in opposite directions. An area of origin based definition can include a small number of kitchen area fires that were not, strictly, cooking activity fires, an electrical fire at a kitchen outlet, for instance. At the same time, the same reporting system very likely omits a meaningful number of true cooking fires that were resolved before arrival and coded as non fire incidents. This report and dashboard are built to identify departments and locations carrying elevated cooking fire risk for the purpose of directing prevention resources, not to serve as a certified count of cooking fire caused loss. The rankings and rates presented here should be read as a reliable and defensible indicator of relative risk, not as a precise or exhaustive incident count.

Departments were ranked by five year qualifying event count divided by 2024 American Community Survey five year county population estimates, expressed as a rate per 10,000 residents. Departments with fewer than 15 qualifying events over the five year period were excluded to avoid statistically unstable rates.

Sources: U.S. Fire Administration, National Fire Data Center, "Review and Assessment of Data Quality in the National Fire Incident Reporting System" (May 2017). Kinsey, K. and Ahrens, M., "NFIRS Incident Types: Why Aren't They Telling a Clearer Story?", National Fire Incident Reporting Community/NFPA (January 2016). U.S. Fire Administration, "Cooking Fires in Residential Buildings (2017-2019)," Topical Fire Report Series, Vol. 21, Issue 5 (July 2021).