The safest emergency is the one that never happens.
Most of what a fire department responds to is predictable in aggregate, even when no single incident is. Community risk reduction is the work of finding the households, buildings and populations where the next one is most likely, acting before it happens, and measuring whether the action changed anything. The technology is mostly about joining data a department already holds to data it does not, and about keeping track of work that happens nowhere near an incident.
Emergency response deals with what has already happened. Community risk reduction asks a different question: can we find out where harm is most likely to occur, and do something about it before the call comes?
It is not another name for public education, and it is not the fire prevention bureau under a newer heading. It is an all-hazards, data-informed process — identify the risks that actually exist in this community, decide which of them to work on, act in the ways most likely to change the outcome, and measure whether anything changed. Fire is part of it. So are falls, overdoses, crashes, heat, drowning and everything else a department is called to.
The part that makes it work is the last step. Prevention has always been done. What is newer is the expectation that a department can say who was helped, by how much, and how it knows.
A field guide rather than an article. Most people will want one part of it.
The whole of CRR is five steps in a loop. Everything else on this page is detail about one of them.
What risks actually exist in this community? Not the ones that make the news, and not the ones a neighboring department has — the ones the local record shows.
Which of them carry the greatest probability, consequence or vulnerability, and which offer a realistic chance of doing something about it? Not everything important is changeable.
What action is most likely to change the outcome? Usually more than one kind, aimed at a specific population in a specific place.
Did the intervention reduce risk, or only produce activity? This is the step that gets skipped, and skipping it is what makes prevention hard to fund.
Continue, change or stop, based on what the measurement showed. A program that cannot show an effect has to be allowed to end.
It is a loop, not a project. The measurement of one cycle is the identification step of the next.
A community risk assessment, or CRA, is the documented answer to the question "what is actually happening to people here, and to how?" It is the foundation the rest of CRR stands on, and it is a piece of written work rather than a piece of software.
The recognized references for how to do one are NFPA 1300, Standard on Community Risk Assessment and Community Risk Reduction Plan Development, and the material published by the U.S. Fire Administration and the National Fire Academy. Both are in the research library below, with their addresses.
The most common mistake is to treat the assessment as a mapping exercise. A heat map is a picture of where incidents happened. It is not an assessment, because it does not say who was harmed, why it kept happening, or what could realistically change it.
No department has all of this, and an assessment that waits for all of it never gets written. Start with what you hold and add to it.
Most of this exists. Very little of it is in one place, which is why the assessment is work rather than a query.
If the finished document answers these, it is an assessment. If it answers only the first two, it is a map.
Named as a population, not as a jurisdiction. "Adults over 65 living alone" is an answer. "Our residents" is not.
At the smallest level the data honestly supports — a tract, a corridor, a set of parcels. Citywide averages hide the neighborhoods that need the work.
The specific event. "Cooking fires between 5pm and 8pm" points somewhere. "Fires" does not.
The conditions behind the event. This is the question that turns a statistic into something you can act on, and the one no dataset answers on its own.
Frequency over a period long enough to be more than noise, compared against a baseline you can defend.
Injuries, deaths, displacement, loss. A rare event with severe consequences may outrank a frequent one.
Given your staffing, your budget and your partners. An intervention nobody can deliver is not a finding.
Five kinds of intervention. The point of naming them is that the first one is not the only one, and on its own it is rarely the strongest.
Teaching people what the risk is and what to do about it.
Fire service: teaching residents to test an alarm and replace the battery, or what to do if a pan catches fire.
All hazards: teaching households how to shelter or evacuate in a flood, or how to recognize heat illness.
Changing the physical conditions so the harm is less likely or less severe, whether or not anyone remembers the advice.
Fire service: installing ten-year sealed-battery alarms, or residential sprinklers in new construction.
All hazards: grab rails and improved lighting to prevent falls, or defensible space around a home in the wildland interface.
Applying the codes and rules that already exist, where they apply.
Fire service: inspecting the occupancies where a violation carries the most risk, and following through on what is found.
All hazards: property maintenance, housing and zoning codes, used with the agencies that own them.
Making the safer choice affordable, or the unsafe one costly.
Fire service: grant-funded alarms for households who cannot buy them, or a rebate for a residential sprinkler system.
All hazards: subsidized home modification for older adults, or insurance recognition for wildfire mitigation.
Improving the response to the risk that remains after the other four have done what they can.
Fire service: moving a unit, changing a dispatch protocol, or adding a capability where the need is concentrated.
All hazards: community paramedicine for the households the other four have not reached.
Almost every program that has been shown to work combined several of these. Education alone changes what people know; it does not reliably change what happens to them.
These are the operational problems departments actually describe. Each one opens onto the approaches that address it, and the products listed here that claim to.
Prevention effort is spread evenly across a jurisdiction where risk is not evenly spread, so the households most likely to have a fire are as likely to be missed as anybody else.
Home visits, alarm installs and education sessions are counted but never connected to an outcome, so the program survives or dies on budget politics rather than on evidence.
An inspection schedule built on a fixed cycle treats every occupancy as equally risky, so effort goes to buildings that were fine last year while the ones that changed hands or changed use wait their turn.
Incidents are in one system, inspections in another, EMS in a third, and the parcel and census data that would make sense of all three belongs to someone else. Every question that crosses two systems becomes a project, so most of them are never asked.
A crew notices a hoarding condition, a fall risk or a family with no working alarm, and the knowledge leaves with the shift. There is no record, no referral to the agency that could act, and no way to know whether anything came of it.
Worth being clear about
Technology does not reduce risk by itself. It helps a department identify risk, coordinate interventions, document the work and find out whether those interventions changed the outcome.
A department with a clear problem and a spreadsheet will get further than a department with a platform and no question. Buy in that order.
Four questions. It arranges your answers into the shape a CRR plan takes, so you have something to put in front of a chief, a council or a partner agency. Nothing you type is sent anywhere or saved.
Answer the questions and your plan appears here, ready to copy into a document or take to a meeting.
This is a planning and teaching aid, not a risk assessment. It calculates nothing, scores nothing and predicts nothing — every word in the plan is one you typed or chose. A real community risk assessment uses your own data and follows NFPA 1300.
Departments often assume CRR begins with buying something. It usually begins with joining together records the department already holds and has never looked at side by side.
The second half is held by other people. Most of it is public, and the parts that are not are usually obtainable with an agreement rather than a purchase.
The value is in the join. Incident history says where things happened. Census and parcel data say who lives there and in what. Neither on its own tells you who to visit on Tuesday.
Two datasets that are each unremarkable can identify a household when they are put together. Decide these things before the join, not after somebody asks.
This is a description of good practice, not legal advice. Your own counsel and your state law decide what applies to you.
Activity measures say how much work was done. Outcome measures say whether anything changed for the people the work was aimed at. Departments need both, and almost always report only the first.
What we did
We installed 2,000 smoke alarms.
What changed
Did residential fire injuries fall among the households we targeted?
What we did
We completed 500 home safety visits.
What changed
Did repeat incidents decline at the addresses we visited?
What we did
We inspected 1,200 occupancies.
What changed
Did high-risk violations decrease in that occupancy class?
What we did
We taught 3,000 schoolchildren.
What changed
Did youth-set fires fall in the areas those schools serve?
What we did
We ran a fall-prevention campaign.
What changed
Did fall-related 911 calls drop among the residents who took part?
The right-hand column is harder, slower and occasionally disappointing. It is also the only column that answers the question a council asks when the budget is tight.
The reason most prevention work cannot show an outcome is that the first step was skipped, and by then it is too late to go back for it.
Take the number before you start, for the population and area you are about to work in. Not the citywide number, and not last year's number for somewhere else.
Do the work, and record what was actually delivered, to whom, and where. A program you cannot describe precisely cannot be evaluated.
Compare against the baseline, over a period long enough for the number to mean something, and be honest about what else changed in that time.
Continue it, change it, expand it or stop it. Stopping something that did not work is a result, not a failure.
Programs whose results were written up and published, so you can read what was actually done and what was actually measured rather than a summary of it.
Ten years of door-to-door smoke alarm installation in 36 high-risk census tracts, with both the outcome and the cost measured and published in the peer-reviewed literature.
Smoke alarms · Home visits · Measurement · Fire prevention
Twelve years of a multi-part fire prevention program across 120,349 households, evaluated one initiative at a time — including the parts that did not work.
Smoke alarms · Home visits · Measurement · Fire prevention
Using incident data to find likely repeat EMS clients before the next call. Described by the U.S. Fire Administration; no measured outcome is published.
EMS · Community paramedicine · Data & analytics
Standards, government guidance and peer-reviewed research on community risk reduction. Every address here was opened and checked. Narrow the list by subject.
The federal starting point: what CRR is, the five E's, and short accounts of what named departments actually did under each one.
Why it matters
If you read one thing before writing a plan, read this. It is free, it is short, and it is the vocabulary everyone else in this list assumes you have.
The consensus standard for how a community risk assessment is conducted and how a CRR plan is developed from it.
Why it matters
This is the document a peer reviewer, an accreditation team or a sceptical council member will measure your assessment against. Read the standard before you buy anything that claims to satisfy it.
The wider USFA prevention section that the CRR material sits inside — codes, public education, data and outreach campaigns.
Why it matters
Where to go once you know which of the five E's you are working on.
NFPA's collection of CRR material — the definition used in NFPA 1300, tools, and its CRA data dashboards.
Why it matters
The bridge between the standard and something you can act on this month.
A free online self-study course on what makes up a CRR strategy and how the process works.
Why it matters
The NFA recommends taking this one before the risk assessment course. Free, self-paced, and it gives a whole department the same vocabulary.
A free online self-study course on conducting a risk assessment: the vocabulary, working with demographic and community data, and dealing with stakeholders.
Why it matters
The practical companion to NFPA 1300, and the cheapest way to find out whether you can write the assessment yourself.
How the CRA and the standards of cover fit together in CFAI accreditation: measuring fire and non-fire risk across a service area, then stating what the department can actually deliver against it.
Why it matters
If your department is accredited or heading that way, this is the frame your risk assessment has to fit. It is also the clearest published explanation of why an assessment and a deployment plan are two halves of one argument.
The IAFC's position on CRR as an evidence-based, all-hazards strategy rather than an add-on to fire prevention.
Why it matters
Useful when the argument you have to win is internal. It is the chiefs' own association saying this is the job now.
The federal replacement for NFIRS, and the incident dataset a community risk assessment will increasingly be built on.
Why it matters
Your incident history is half of any risk assessment. What NERIS collects decides what questions you will be able to ask of it later.
The NERIS programme site, run by FSRI with the USFA: documentation, onboarding and the data model itself.
Why it matters
The place to look before a vendor tells you what NERIS requires.
The outcome evaluation of Operation Installation in Dallas. Volunteers and firefighters canvassed 36 high-risk census tracts between 2001 and 2011; the study followed 8,134 programme houses against 24,346 non-programme houses in the same tracts.
Why it matters
Over a mean of 5.2 years, house fire deaths and injuries were 68% lower in programme houses (3.1 vs 9.6 per 100,000; rate ratio 0.32, 95% CI 0.10 to 0.84), and 63% lower after adjustment. It also found the effect fading after year six as alarms stopped working — which is the more useful half of the finding, and the half that gets left out of the slide.
What the same Dallas programme cost, and what it returned. Free to read in full.
Why it matters
Estimated net savings of $3.8 million over 2001-2011, a $3.21 return for every dollar spent from a societal perspective, and an incremental cost per fire injury averted of $128,800 counting only direct programme and medical costs. That last figure matters as much as the first: it is what the programme costs before you count lost productivity, and a council will ask.
Twelve years of the HomeSafe programme in Surrey, British Columbia, across 120,349 households, comparing each initiative against both the pre-intervention period and the city-wide trend.
Why it matters
It reports which parts worked and which did not. Firefighter home visits, alarm inspection and installation, and alarm checks at incidents were the three effective initiatives; door hangers and package distribution were not. Programmes that publish their failures are rarer than they should be, and more useful.
A systematic review of what has been tried, and what has been shown to work, in the neighbourhoods and communities that carry the most residential fire risk.
Why it matters
Risk is not evenly distributed, and neither is the evidence. This is the review to read before designing a programme for the part of your jurisdiction that carries most of the loss.
A synthesis of the US fire prevention programmes that had actually been evaluated, and what the evaluations found.
Why it matters
Its conclusions are still the ones that hold: canvassing and installing beats handing out, partnerships with fire departments work, using existing data systems to target and evaluate works — and alarms not working remains the recurring problem. Older than the rest of this list, and still the clearest statement of what is known.
What the discipline of injury prevention has contributed to home fire safety, and where the gaps still are.
Why it matters
The best short argument for why CRR is a public health method applied to the fire problem, rather than fire prevention with a new name.
Nothing here is paid placement. Links are chosen because they are the clearest explanation we could find, and labeled so you know who wrote them.
Not a maturity model and not a score. A list you can hold your own program against, and a reasonable standard to hold a vendor to.
CRR is not a program. It is a way of making decisions.
You do not need a data scientist, a new software platform or a fifty-page strategic plan to begin. You need one problem and the willingness to find out whether you changed it.
Pick the thing that keeps happening. Then work through these, in order, with whoever is in the room.
Then begin. The first cycle will be imperfect and will teach you more than the plan would have.
In that order. A department that arrives at a demonstration with a written problem and a measure of success gets a useful answer. A department that arrives to see what the software does gets a tour.
Nothing listed on this site is endorsed by it. The listings describe what a vendor says it does; the questions to ask are on each problem's own page.
