Since smoke detectors first appeared on the shelves of stores in the 1970s, they have been heralded as the single most effective means of reducing deaths from fire. It has been close to a quarter of a century since low-cost smoke detectors have been introduced, and we are now looking at the question, Do smoke detectors really work? The National Fire Protection Association (NFPA) says they do. In fact, the NFPA claims smoke detectors are as much as 40 to 50 percent effective in preventing the loss of life by fire. But, how are these claims possible when from 1977 to 1997 there was a 45.3 percent reduction in the number of home fires but only a 42.7 percent reduction of fatalities in home fires? This article addresses the apparent discrepancy and other issues pertaining to the relationship of smoke detectors to life safety.
THE PROBLEM
NFPA data show a decline in the number of “home” fires-from 723,500 in 1977 to 395,500 in 1997, representing a decline of 45.3 percent over 21 years. During the same time frame, fatalities in home fires fell from 5,856 to 3,360, accounting for only a 42.7 percent decline. At best, this means that the life loss from home fires is directly proportional to the reduction in home fires. It could be that the only people who didn’t die in fires over the 21-year period were those who didn’t have fires in the first place. This opinion is actually borne out when we look at the fatality rate for each year in question. In 1977, there were 8.11 fatalities for every l,000 fires; in 1997, there were 8.5 fatalities for every 1,000 fires. With data like these, how can we claim smoke detectors are helping?
NFPA1 data show that of all home fires that occur in the United States, 56.1 percent occur in homes with smoke detectors and 43.9 percent in homes without detectors. From these data, the NFPA further determined that fatalities occur in homes without smoke detectors about twice as frequently as in homes with smoke detectors. This is where the 40 percent to 50 percent improvement in life safety figure comes from. But, there is a major flaw!
What isn’t taken into account is the fact that the most recent NFPA estimates indicate that as many as 93 percent of all homes currently have at least one smoke detector. This means that the 56.1 percent of all home fires that occur where smoke detectors are pres-ent is disproportionally spread among 93 percent of the population, for a relative risk of 0.6. Even more important is the fact that the seven percent of the population that has no smoke detectors suffers an alarmingly disproportionate 43.9 percent of home fires, for a relative risk of 6.22.2 To put it another way, the relative risk for people in homes without smoke detectors is 10.4 times that for people in homes with smoke detectors.
IS FIRE SAFETY CONSCIOUSNESS A FACTOR?
The question then becomes, if the 93 percent of the population that has smoke detectors accounts for just over half of the home fires, could there be a positive relationship between the installation of smoke detectors and fire safety awareness? After all, how else can we account for the fact that 93 percent of the population accounts for only just over half of all home fires? NFPA data fail to account for any possibility that homeowners who install and maintain smoke detectors are more fire safety conscious and thus are less likely to have a fatal fire in the first place. Instead, all the credit for the lower fatality rates in the 93 percent of homes with smoke detectors is in effect given to the smoke detectors.
THE STUDY
This study is aimed at answering two primary questions: (1) Do smoke detectors actually improve the chances of surviving a fire? and (2) Are people who install and maintain smoke detectors more fire safety conscious? While arriving at the answers to these questions, I also uncovered some unexpected but interesting information, which is presented near the end of this report.
Statistics from the National Fire Data Center (NFDC) in Emmitsburg, Maryland, were primarily used for this study. Unfortunately, the Center’s reliable database goes back only a few years. However, I was able to make some interesting analyses based on the information provided. References to national trends are based on NFPA data.
 |
I prepared a spreadsheet (Tables 1a, 1b, and 1c) on which the data was entered, by year, under eight categories:
 |
Fires under the following conditions:
 |
- not reported,
- detectors were in the room of origin and operated,
- detectors were not in the room of origin and operated,
- detectors were in the room of origin and did not operate,
- detectors were not in the room of origin and did not operate,
- fire too small to set off detector,
- no detectors present, and
- performance not classified.
Each category has subcategories of number of incidents, number of injuries, injury rate (number of injuries per 1,000 incidents), number of fatalities, and fatality rate. The data were then sorted into three groups: one- and two-family dwellings (yellow), multifamily dwellings (blue), and the total of these two categories (orange). Unless otherwise noted, the data discussed here were taken from the “total” category. In Table 2, “Smoke Detector Performance,” are totals for smoke detector data from the four categories of smoke detector performance presented in two composite categories: (1) where smoke detectors operated and (2) where they were present but did not operate. These two categories summarize smoke detector performance.
DO SMOKE DETECTORS WORK?
To determine if smoke detectors really work, let’s start with the data in the orange-colored group (the combined total of the one- and two-family and the multifamily dwelling categories).
 |
Table 1c shows that the fatality rate (number of fatalities per 1,000 fires) under the Annual Totals for 1996 column is 8.23. Although this is an average rate and doesn’t give specific data, it provides a point of reference. For example, where there were no smoke detectors, the fatality rate was 10.9-24 percent higher than the average fatality rate for 1996. We now know that where smoke detectors were not present, fatality rates in 1996 were significantly higher than the average. More importantly, we now have a figure against which to compare smoke detector performance.
 |
Next, note from Table 2 that where smoke detectors are present but do not work, the fatality rate is 7.28, 11.5 percent below the average. That is a reduction of 33.2 percent from cases where smoke detectors aren’t present. Looking at the spreadsheet (Tables 1a, 1b, and 1c), we see that this general trend is consistent. Is this the evidence needed to prove that people who install smoke detectors are inherently more fire safety conscious? If people with smoke detectors weren’t more fire safe, we should see the same fatality rate in homes with nonworking smoke detectors as in those where smoke detectors are not even present.
 |
From the category (Table 2) where smoke detectors were present and did work, we see that the fatality rate in 1996 was a low 4.34-a full 40 percent lower than in those cases where smoke detectors were present but not operating and an astounding 60 percent lower than where smoke detectors were not present. From these data, we can now verify that smoke detectors are effective. However, it is probably more accurate to say that in homes in which smoke detectors are installed and maintained the fatality rate is only 4.34 vs. a rate of 10.9 where no smoke detectors are present.
 |
Illustration 1, “Smoke Detectors and Fatality Rates,” indicates the direction the fatality rates are taking in each of the three categories mentioned above (no smoke detector, smoke detectors present and worked, and smoke detectors present but did not work).
The top line (red) indicates the fatality rate for homes without smoke detectors. The straight line through the graph line is a linear trend line. It gives an idea of the degree of increase or decrease of the data in the graph. Where there were no smoke detectors, the trend line is almost level. This indicates that even though the rates were gyrating wildly over the past 10 years, the fatality rate is essentially the same today as it was in 1987. But remember, this is the rate for both one- and two-family and multifamily dwellings.
Line 2 (blue) represents homes in which smoke detectors were present but did not work when the fire occurred. The slope of the trend line indicates a slight increase in the overall rate from 1987 to 1996. The reason for this increase is not known. It is also possible that this increase is so slight as to be insignificant.
The bottom (yellow) indicates home fires in which smoke detectors alerted residents to the fire. Note the overwhelming decrease in the line’s slope. Something positive is happening here! In fact, the graph of the data for this category is almost a straight line. How is this possible? Why are smoke detectors more effective today than in 1987?
This is easy to explain. Over the years since some jurisdictions have made smoke detectors mandatory, the codes have become increasingly strict. At first, only one detector was required-in the sleeping area, and it could be battery-powered. Later, the detector had to be hard-wired. Then, detectors were required on each floor level; they had to be interconnected and ultimately were required to have a battery backup. Some recent codes now require that a detector be in every bedroom as well as on each floor level; detectors are also required to be interconnected and to have battery backup. Whatever the reason, smoke detectors are a success story, at least to a degree. But to be most effective, they must be properly maintained.
As recently as 1996, only 26 percent of all home fires occurred in homes with working smoke detectors, according to NFDC data. This gives further support to the theory that people who install and maintain smoke detectors are much less likely to have a fire than people who do not install smoke detectors.
Looking closely at the data in the spreadsheet (Tables 1a, 1b, and 1c), we see that the data break down smoke detector performance into four categories-two in which the detectors work and two in which they don’t work. In the first category, which represents working smoke detectors, the detectors are actually in the room of the fire’s origin (Table 1a). For 1996, the fatality rate was a low 3.84 deaths per 1,000 fires. Where the detectors were working but not in the room of origin, the rate rose to 4.98. This means that even in a home with working smoke detectors, there is a 29.6 percent greater chance of dying in a fire if the fire does not start in a room that has a working smoke detector. In fact, if you compare the rates for these two categories over the 10 years of data provided, you’ll see a significant improvement in life safety where the detector is in the room of fire origin. How’s that for a good argument for having smoke detectors in every room? It’s the most convincing direct evidence I have found that smoke detectors are effective. In fact, of the eight categories of data, the only one that has a fatality rate lower than when the detector is in the room of origin is that where the fire was determined to be too small to trigger the smoke detector in the first place.
ONE-AND TWO-FAMILY VS. MUTIFAMILY DWELLINGS
Let’s look at the data that compare one- and two-family vs. multifamily dwellings (Illustrations 2 and 3, respectively). The data depicted in these graphs are the same as those in the spreadsheet. Looking at Illustration 2 (one- and two-family dwellings), we see that where no smoke detectors are present (blue), there is evidence of an increase (the slope of the trend line goes up) in fatality rates over the 10 years of data. A similar increase is seen where smoke detectors were present but didn’t work (yellow). The only category in which an improvement is evident (the slope of the line goes down) is where the smoke detectors actually worked (red). This means that in one- and two-family dwellings, except where smoke detectors actually work, the situation appears to be getting worse. And, on top of this, the categories in which the fatality rate is getting worse represent 76.7 percent of all one- and two-family dwelling fires.
When we look at the same data over the same time period on the graph for multifamily (Illustration 3), we see a completely different outcome. Every category shows an improvement (a decline in rates). The slightest improvement, but still an improvement, is in the category of smoke detectors present but did not work (yellow). In the other two categories, the reduction in fatality rates is dramatic.
Why have fatality rates significantly gone down in multifamily residences that have no smoke detectors? This same category shows an increase for one- and two-family residential occupancies. The answer that I can determine, from another study I have done recently, is that it is probably attributable to fire prevention efforts in the forms of enforcement of building/fire codes, installation of 13-R sprinkler systems, enforcement of strong fire prevention code, and implementation of a vigorous fire inspection program. In short, fire prevention works!
To prove this point, I will briefly discuss the results of the other study to which I referred. I looked at an increase in suppression runs over a seven-year period in a jurisdiction of just under one million people. The increase in suppression runs was compared in relation to the increase in population of the jurisdiction. From 1992 to 1998, the jurisdiction saw a 10.2 percent increase in population and a 15.4 percent increase in suppression runs. (Suppression runs include fires, reports of fires, fires out on arrival, and so on.) I then broke the suppression runs for buildings only down into two subcategories: buildings inspected and those not inspected. In the inspected group were all the commercial, industrial, mercantile, multifamily residential, and so on). The group not inspected consisted essentially of one- and two-family dwellings. The group that underwent some type of inspection on a regular basis had a 14.7 percent increase in suppression runs. The group not inspected (one- and two-family dwellings) had an increase of 110.1 percent. No, it’s not a typo! In the seven-year period, suppression runs to one- and two-family dwellings increased more than seven times the rate of suppression runs overall and 10 times the rate of the increase in the population.
Before moving on, a warning flag needs to be raised. When discussing the average fatality rates above, I mentioned that the fatality rate where no smoke detectors were present was essentially unchanged since 1987. But now that we have broken that rate down into one- and two-family dwellings and multifamily dwellings, it becomes evident that the fatality rate is actually increasing in one- and two-family dwellings. On top of the increased fatality rate where smoke detectors aren’t present, there is also an increase in the fatality rate where smoke detectors are present but don’t work. In fact, where smoke detectors are present and don’t go off, the fatality rate is increasing even faster than where smoke detectors aren’t present at all. So, even when we recognize the potential of smoke detectors, we must also note these specific categories in which things are actually getting worse.
INJURY RATE
By now, you have probably noticed that I have not made any mention of the injury rate. This is for a good reason. Good arguments can be made that we have had success in the reduction of life loss from fire even when the fatality rates show an increase simply because overall numbers have decreased. But when we discuss injuries, no such claim can be easily made. Earlier, I had pointed out that from 1977 to 1997 there was a 45.3 percent reduction in the number of home fires. In 1977, there were 21,640 injuries in home fires. By 1997, there were only 17,300 injuries in home fires. That is only a 20 percent reduction, less than half the reduction in fatalities. If we compare the number of injuries per 1,000 fires to make an injury rate, the real problem begins to appear. In 1977, the injury rate for home fires was 29.91; by 1997, it had increased to 43.74-a 46.8 percent increase. This is a problem that is currently being ignored by the U.S. fire service.
What is more to the point, since this report is about smoke detectors and life safety, is what part do smoke detectors play in the reduction of injuries? When we combine both categories where smoke detectors work, we have an average injury rate of 91.42 injuries per 1,000 home fires in 1996. What probably comes as a surprise to most is that this is the highest injury rate for any category. We can see from Table 2 that in 1996 homes with smoke detectors that didn’t work have the second highest injury rate-87.7 percent. Where smoke detectors are not present, the injury rate is only 74.71 percent (Table 1c). Once again, it is hard not to perceive a behavioral difference between people who install and maintain smoke detectors and those who don’t, even though where injuries are concerned, it is working against them.
Is it possible that people who install and maintain smoke detectors believe they can fight the fire themselves? After all, they were fire safety conscious enough to install the smoke detectors. This, more than anything else, could expose residents to the fire and cause injuries. But, it can’t be explained simply by saying this phenomenon is attributed to smoke detectors’ alerting residents to the fire in its earliest stages, thus giving them a better chance of fighting the fire themselves. After all, even where smoke detectors are present but don’t go off, the injury rate is 17 percent higher than where smoke detectors are not present.
It might be possible to make an argument that a higher injury rate is evident where smoke detectors work because without the detectors the injuries would be fatalities. But this doesn’t work when we compare the injury rate with fatality rates where smoke detectors are present but don’t go off. As shown above, there is a 33.2 percent better of chance of surviving a fire if a smoke detector is present, even when it doesn’t work. But the injury rate goes up where smoke detectors are present but don’t work compared to where no smoke detectors are present.
THE NEED FOR BETTER DATA
 |
As mentioned earlier, the primary data for this report were provided by the National Fire Data Center in Emmitsburg, Maryland. If you look at the data closely, you can see trends in various areas. As an example, if looking at the spreadsheet (Table 1c-data from the NFDC), fatality rates given under the Annual Totals column appear to be headed down. However, fatality rates using NFPA data (Table 3 and Illustration 4) appear to be increasing. So, how is it possible that two very credible sources of information are actually in opposition?
The information supplied by the NFDC is an actual count. It also has the major flaw of being an incomplete count. In fact, only about 50 percent of all fire departments in the country choose to report data to the NFDC. At this point in the history of the U.S. fire service, it is discouraging that so many departments choose to exclude themselves from a national database. There is no more accurate way of determining how many fires of what category occurred and their related particulars than to have an actual accounting.
On the other hand, the NFPA figures, which are used exclusively when national data are needed, are an estimate. They are, however, the best estimates we have concerning the U.S. fire problem. They are arrived at by combining the information from the NFDC with data from some 3,000 survey forms sent out by the NFPA. But since neither method results in a complete, accurate count, the NFPA must make predictions from the available data. The final figures are a statistically generated best guess. They may not be an actual count; but, unfortunately, they are the best statistics we have. This conflict really points out the need for a complete, national count. The NFDC already has a system in place; all we have to do is use it.
CONCLUSIONS
The answer to the original question with regard to the effectiveness of smoke detectors is that they are effective. But just how effective is not exactly known. Remember that evidence presented here supports the theory that people who install and maintain smoke detectors are less likely to have a fire than people who don’t have smoke detectors installed. This answers the second question posed at the beginning of this article. Since the fatality rate is increasing in one- and two-family dwellings, which accounts for about 74 percent of all home fires, it would provide an answer to the question of why fatality numbers have not fallen as fast as the number of fires-in short, the tremendous benefit of smoke detectors is more than offset by the ever-increasing fatality rate in one- and two-family dwellings where smoke detectors aren’t present or don’t work.
Figuring out the reason for higher injury rates in homes where smoke detectors are present is more complicated. The first answer reached is usually that the increased injury rate represents lives saved. But this doesn’t “compute” when we realize that a difference in both fatality and injury rates is evident even when smoke detectors are present but don’t work. In fact, if we look at the injury rates for 1996, where smoke detectors were present but did not work, the number of injuries per 1,000 fires goes up by 12.9 injuries compared with homes without smoke detectors. But the injury rate only goes up by another 3.7 injuries per 1,000 fires when a smoke detector works. Could this be the smoking gun that proves the injury rate difference is in large part behavioral and not all the result of the smoke detectors’ saving the lives? Remember, where smoke detectors are present but don’t work, we can’t credit the smoke detector with alerting the residents, but we still see notable and credible differences in fatality rates.
Overall, the effectiveness of smoke detectors is evident in saving lives, but this benefit is accompanied by an increased injury rate. Although it may not be possible to quantify an exact benefit figure for smoke detectors, it is clear we need to get them in all residences. States and jurisdictions that currently don’t have mandatory smoke detector legislation need to adopt laws mandating smoke detectors in every room except kitchens and utility rooms. And while we are at it, how about a true national (meaning everyone has to participate) reporting system?
Endnotes
- Hall, John R., Jr., U.S. Experience with Smoke Detectors and Other Fire Detectors, National Fire Protection Association, Aug. 1996.
- Relative risk is the percentage of total fires in homes with (without) smoke detectors divided by percentage of homes with (without) smoke detectors-e.g., 56.1/93 = 0.6.
WILLIAM F. CRAPO, whose fire service career spans more than 30 years, is a retired captain from the Washington, D.C. Fire Department. He was a volunteer firefighter with the Brentwood (MD) Fire Department, where he served as chief and training officer. In 1979-80, Crapo was involved in developing the first Disaster Planning Guide for Fire Chiefs for the Federal Emergency Management Agency; the project was spearheaded by the International Association of Fire Chiefs. He has a bachelor’s degree in fire science from the University of Maryland and associate’s degrees in management and fire science and is a member of the adjunct faculty at Lord Fairfax Community College in Virginia, where he teaches fire science.
