Fire chiefs, politicians and citizens often want to know how well their fire services measure up. Those of us responsible for managing fire departments use several methods to make this analysis. Often we compare ourselves to similar agencies, a process known as benchmarking, or evaluate if we are in compliance with national standards.

For career fire departments, NFPA 1710, Organization and Deployment of Fire Suppression Operations, Emergency Medical Operations, and Special Operations to the Public by Career Fire Departments, is the one standard every fire chief has examined and asked, “How do we measure up?” For those departments attempting to comply with NFPA 1710, one of the hurdles is response time performance. If you have already examined your current response time performance against NFPA 1710, you may have several questions. How do we compare to other agencies? What can we do to improve? Before we get to the answers, let's look at how this study explored these questions.

Performance measure

The most chronic and unresolved problem in measuring performance is the difficulty of comparing apples to apples, sometimes described as definitional ambiguity. Different agencies use different definitions or have different computer-aided dispatch systems, making the comparison of data difficult. To overcome the problem, this research examined 17 metropolitan fire departments that all use the same analytical software, Deccan's CAD Analyst.

Conducting the study through a state university, each of the fire departments involved gave permission for their original CAD data, formatted into a large database, to be analyzed. From a full year's worth of data — more than 1 million incidents from the 17 agencies — this review focused on reported structure fires, a total of 43,733 calls. The results are enlightening and can provide a road map for other fire departments looking to improve their performance.

Call processing

The first component of fire department response starts with call-processing time. Typically, this is defined as the time from when the 911 call is received until emergency units are dispatched. However, technical limitations or conscious policy decisions mean there are other points of time in the dispatch center when the response time clock could actually start. Understanding this important fact allows you to make meaningful comparisons between your agency and others. While many agencies control their own dispatch center, others rely on local law enforcement or perhaps a regional communications center.

Some dispatch centers start the clock when the 911 call is first answered in the primary public safety answering point. Others wait until the call-taker has obtained all the information, input that data into a CAD system and forwarded the incident to a fire dispatcher for alerting of emergency units. This research shows that the difference between these two common methods was 47 seconds. The trick is to know how you define response time and how others define it. A word of caution! I have seen large urban dispatch centers be confused on how their own system captures response time. Until someone actually verifies your process and what each time stamp represents, be careful about making comparisons.

Six of the 17 agencies studied here captured the first time stamp when the initial 911 call was answered at the primary PSAP. The average time from 911 pickup until dispatch of emergency units was 1.4 minutes. However, one agency averaged a call-processing time of 2.1 minutes, while another was much quicker at 41 seconds. For all calls handled by these six agencies, the 60-second benchmark called for in NFPA 1221, Installation, Maintenance, and Use of Emergency Services Communications Systems, was met only 45.8% of the time.

As noted, some agencies wait until the intake operator has input all the emergency information and sent the call to a separate dispatcher. When starting the clock at this later time, which ignores call intake/data entry, the call-processing time averaged only 39 seconds. Several of the best-performing agencies, using this more liberal definition, had an average call-processing time of 18 seconds. A poorly performing agency, using the same definition, averaged as high as 1.3 minutes.

Because different agencies used different definitions, a strict apples-to-apples comparison is impossible. To provide an overall picture of these 17 agencies, the table on page 89 uses the first time stamp captured by each agency, regardless of when they started the clock. Using this approach, call-processing averaged 1.3 minutes with only 52.7% of all calls being handled within the 60-second benchmark. The 95th percentile using this approach was calculated at 3.2 minutes, over three times the recommended benchmark.

Dispatch to en route

Often referred to as turnout time, the average time from dispatch of emergency units until the units reported they were responding was 1.3 minutes. The best-performing agency maintained a 53-second average, while the worst agency averaged 2.2 minutes for turnout. Using the 60-second benchmark, the overall compliance rate among all agencies was 37%. To meet the standard 90% of the time, you would have to set the goal at 2.1 minutes — more than twice the current benchmark.

En route to unit arrival

Travel time should be at or less than four minutes for the first responding engine. The total travel time for the first-alarm assignment should be eight minutes. Overall travel time for the first engine averaged 3.4 minutes, with 67% of the overall responses meeting the four-minute benchmark. Two of the 17 agencies were meeting the four-minute goal 97% of the time, while the worst performer only met the goal 25% of the time. Unfortunately, except for new stations or more fire units, most agencies are limited in what they can do to improve travel time. Studies from the Rand Institute in the 1970s demonstrated that the distance you must actually travel mostly determines travel time. For reported structure fires studied here, the 90th percentile was 5.9 minutes for the first engine.

First-alarm assignment

NFPA 1710 also calls for the entire first-alarm assignment to arrive within eight minutes after going en route. While the exact configuration, staffing or number of units will vary among agencies, this research used the definition provided by the agency itself. In other words, what do they consider their first-alarm assignment? Based on the agency's own definition of a first-alarm assignment, the average time from en route to on scene was 5.9 minutes, or 2.5 minutes beyond the first arriving engine. The 90th percentile was 11.2 minutes.

Total time

So what does this information tell us? The chart on page 90 summarizes the total time from the first time-stamp in dispatch, however each agency captures it, until arrival on scene. The bottom bar shows the average time in minutes for each component outlined above. The top bar shows the percentage (or fractile) time — the point at which 90% (or 95% for call processing) of incidents are handled, as called for in NFPA standards.

For reported structure fires the total time for call processing, turnout and travel time averaged 5.95 minutes for the first engine. For the first-alarm assignment, the average was 8.45 minutes. However if you want to meet the standard 90% or 95% of the time, the total times would have to be 11.2 minutes for the first engine and 15.6 minutes for the first alarm.

Most important factor

A closer examination of the table provides some clues that can help improve your agency's performance. When looking at call-processing times, the four best-performing agencies averaged 30 seconds to process their calls, while the worst four spent almost four times that amount, an average of 1.9 minutes. Even the worst performer was able to meet the standard almost 13% of the time. Therefore, it's possible to meet the call-processing benchmark, but there's a lack of consistency in doing so.

When talking to high-performing agencies that have strong call-processing times, one factor seems to be most important: a focus on getting the call dispatched. These agencies are routinely meeting the benchmark because they insist call-intake operators obtain the location of the incident and the problem.

The call-taker then immediately sends the call to the dispatcher or directly alerts emergency responders. Additional information is obtained and any pre-arrival instructions given after units are dispatched. They perform consistently because these agencies pay attention to their performance, often on a shift-by-shift or daily basis. Problems are identified quickly and resolved immediately.

The same is true if you look at turnout time. The four best-performing agencies averaged 1.1 minutes for turnout while the worst four averaged 2.2 minutes. It's understood that nighttime turnout may be slower, and vehicle maintenance or training may also keep personnel from getting out the door. But again, high performers focus within the organization on how rapidly personnel are making it to fire units and getting on the road. When was the last time your supervisors questioned the time it took personnel to turn out for a reported structure fire?

Students of industrial engineering would argue that more attention needs to be given to monitoring compliance with goals and standards. Where problems are identified, you must first determine if the goal is reasonable and attainable. Then work with fire personnel to identify the issues that prevent them from meeting the goals. Many agencies publish performance information sorted by shift and station. Often, this technique of simply providing feedback will improve performance.

Where to improve

While most agencies are not strictly meeting the response time criteria of NFPA 1710, several areas were identified that can lead to improved performance.

First, agencies should attempt to further reduce call-processing times by defining, and then strictly monitoring, call-processing procedures. By ensuring that 911 operators are getting the location and type of incident — then immediately dispatching emergency units — call-processing times can be improved significantly. After the initial dispatch is made, supplemental information can be obtained and relayed to emergency responders. Second, firefighters and their supervisors need to pay attention to turnout time. The data clearly shows that achieving a 30- or 45-second turnout time is routinely possible.

Chief officers should not try to improve performance by simply issuing new regulations or intimidating orders. Instead, the first action should be to compare your agency's performance to some of the information shown here. Then, take the time to have knowledgeable staff visit both dispatch and fire stations to observe directly the call-processing and turnout actions of your personnel. Finally, meet and discuss what actions, training, or policies may be needed to improve performance. By making all interested stakeholders aware of what is occurring, and why, you will find greater support for the needed improvements.

The fire service needs to continue research and evaluation of its performance objectives. We can then base our standards on more meaningful data.

---

Bruce Moeller, Ph.D., CFO, MIFireE, is the fire chief in Sunrise, Fla. A member of the IAFC's Professional Development Committee and the Institution of Fire Engineers, Moeller previously served with Broward County, Fla., and Naperville, Ill. He has a Ph.D. in public administration from Florida Atlantic University and has taught undergraduate and graduate courses for several accredited institutions in Florida.

Surveyed Agency	Processing (min)	% < 1 min	Turnout (min)	% < 1 min	1st Engine (min)	% < 4 min	1st Alarm (min)	% < 8 min
A	0.4 (±0.4)	93.8%	2.2 (±1.0)	7.3%	5.7 (±3.1)	32.0%	7.5 (±3.5)	64.1%
B	1.3 (±1.2)	50.9%	1.2 (±0.6)	31.6%	2.9 (±2.4)	83.5%	4.2 (±4.8)	85.5%
C	1.4 (±1.1)	43.5%	1.3 (±1.1)	39.4%	2.5 (±1.5)	89.4%	5.6 (±3.6)	84.7%
D	1.9 (±3.9)	51.2%	1.5 (±0.7)	18.6%	3.6 (±2.0)	65.6%	5.6 (±3.0)	83.1%
E	0.6 (±0.6)	90.7%	1.4 (±0.5)	21.8%	3.2 (±1.3)	77.6%	5.9 (±0.6)	100.0%
F	1.3 (±0.8)	43.1%	1.3 (±0.7)	47.6%	2.1 (±1.0)	96.7%	5.0 (±2.5)	88.1%
G	1.0 (±1.3)	66.6%	1.8 (±0.6)	9.0%	3.1 (±1.6)	78.2%	7.9 (±4.0)	62.6%
H	0.4 (±0.3)	93.6%	1.8 (±1.1)	13.9%	6.0 (±3.2)	25.0%	8.4 (±3.9)	59.7%
I	1.6 (±2.5)	39.8%	1.0 (±1.0)	63.3%	3.4 (±2.0)	71.4%	6.8 (±3.4)	73.0%
J	0.6 (±0.8)	88.8%	2.1 (±1.6)	21.1%	4.8 (±2.9)	45.8%	5.5 (±3.3)	87.9%
K	2.1 (±2.1)	29.1%	1.4 (±0.8)	30.2%	2.3 (±1.6)	89.8%	7.4 (±4.0)	64.2%
L	1.8 (±1.0)	18.0%	2.5 (±1.5)	6.3%	5.0 (±2.6)	37.3%	8.6 (±4.1)	43.2%
M	1.2 (±0.6)	41.6%	1.1 (±0.6)	45.0%	2.1 (±1.0)	96.6%	4.2 (±2.2)	95.1%
N	1.1 (±1.8)	65.6%	0.9 (±0.6)	69.4%	3.9 (±2.2)	62.1%	6.8 (±3.0)	74.6%
O	1.6 (±2.2)	35.4%	2.0 (±1.0)	3.0%	3.5 (±2.6)	70.7%	8.9 (±4.1)	52.6%
P	1.9 (±1.2)	12.8%	1.2 (±0.9)	52.0%	4.1 (±2.3)	54.1%	11.9 (±2.1)	0.0%
Q	1.0 (±1.3)	56.1%	1.7 (±0.7)	9.2%	2.7 (±1.5)	86.3%	7.0 (±3.3)	72.9%
Average/Total	1.3 (±1.6)	52.7%	1.3 (±0.8)	37.0%	3.4 (±2.3)	67.4%	5.9 (±3.7)	80.6%