J Community Health (2010) 35:543548

DOI 10.1007/s10900-010-9240-y

ORIGINAL PAPER

Evaluation of Fire-Safety Programs that use 10-Year Smoke

Alarms
Mark Jackson Jonathan Wilson Judith Akoto
Sherry Dixon David E. Jacobs Michael F. Ballesteros

Published online: 24 February 2010

Centers for Disease Control and Prevention 2010

Abstract The Centers for Disease Control and Prevention began funding a Smoke Alarm Installation and Fire
Safety Education (SAIFE) program in 1998. This program
involves the installation of lithium-powered 10-year
smoke alarms in homes at high risk for fires and injuries.
This study aimed to (1) determine among original SAIFE
homes if the lithium-powered alarms were still present and
functional 810 years after installation and (2) understand
factors related to smoke alarm presence and functionality.
Data on a total of 384 homes and 601 smoke alarms in five
states were collected and analyzed. Only one-third of
alarms were still functional; 37% of installed alarms were
missing; and 30% of alarms were present, but not functioning. Alarms were less likely to be functioning if they
were installed in the kitchen and if homes had a different
resident at follow-up. Of the 351 alarms that were present

M. Jackson (&)
M. F. Ballesteros

Division of Unintentional Injury Prevention, National Center
for Injury Prevention and Control, Centers for Disease Control
and Prevention, 4770 Buford Highway NE, Mailstop F-62,
Atlanta, GA 30341, USA
e-mail: mcj4@cdc.gov
M. F. Ballesteros
e-mail: mballesteros@cdc.gov
J. Wilson
J. Akoto
S. Dixon
D. E. Jacobs
National Center for Healthy Housing, 10320 Little Patuxent
Parkway, Suite 500, Columbia, MD 21044, USA
e-mail: jwilson@nchh.org
J. Akoto
e-mail: jakoto@nchh.org
S. Dixon
e-mail: sdixon@nchh.org
D. E. Jacobs
e-mail: djacobs@nchh.org

and had a battery at the time of the evaluation, only 21%

contained lithium-powered batteries. Of these, 78% were
still functioning. Programs that install lithium-powered
alarms should use units that have sealed-in batteries and
hush buttons. Additionally, education should be given
on smoke alarm maintenance that includes a message that
batteries in these alarms should not be replaced. Lithiumpowered smoke alarms should last up to 10 years if
maintained properly.
Keywords Smoke alarms
Fires
Injury prevention

Evaluation

Introduction
Fatal injuries from house fires continue to be an important
problem in the United States. In 2007, there were 414,000
residential fires reported in the United States resulting in
2,865 civilian deaths, 14,000 civilian injuries, and $7.5
billion in property damage [1]. A number of risk factors
have been associated with fire related deaths at home
including socioeconomic status [2], the presence of smoke
alarms in homes [3], and older housing [4, 5]. Numerous
interventions and programs have been implemented in
hopes of addressing this problem. Most of these interventions [6] take the form of home safety education via
counseling and provision of low cost or free smoke alarms.
A targeted intervention in Oklahoma involving a smokealarmgiveaway program (with installation when requested) resulted in an 80% drop in hospitalizations and deaths
related to fire [7]. However, when residents routinely fail to
install their alarms, smoke alarm give-away programs do
not reduce injuries [8]. In fact, one study found that in
programs that gave away vouchers for free alarms, almost

123

544

half of the participants did not even redeem the voucher

[9]. These facts suggest that programs that install alarms
are more effective.
The Centers for Disease Control and Preventions
(CDC) National Center for Injury Prevention and Control
(NCIPC) began funding state health departments to
implement Smoke Alarm Installation and Fire Safety
Education (SAIFE) programs in 1998 [10]. Fourteen state
health departments were awarded 3-year cooperative
agreements. Each grantee was required to install lithiumpowered smoke alarms in homes at high risk for fires and
injuries. Lithium-powered smoke alarms are designed to
last up to 10 years, which eliminates the need for owners to
replace alkaline batteries on an annual basis. The smoke
alarms installed were not sealed units, therefore allowing
access to the battery compartment. Programs targeted older
homes and homes in low-income neighborhoods. At the
time, the lithium-powered smoke alarm was new to the
market and there were concerns about whether the alarm
would last 10 years as designed.
The objectives of this study were (1) to determine
among homes enrolled in CDCs SAIFE program if the
lithium-powered smoke alarms were present and functional
810 years after installation and (2) to understand factors
related to smoke alarm presence and functionality.

Methods
Site Selection
CDC and the National Center for Healthy Housing
(NCHH) identified five state programs (Georgia, Kentucky,
Oklahoma, Virginia and Washington) that were awarded
SAIFE funds in 1998. State program managers confirmed
their interest in participating in the evaluation, that they
had local participants available to assist with the inspections, and that they had records of the original installation
sites. These records indicated the type, location, number,
and date of smoke alarm(s) installed.
To avoid selection bias in sampling, state coordinators
randomly selected 100 dwellings in each state from the
total number of houses that received smoke alarms from
1998 to 2001. In all the states, more homes than needed
were randomly selected to increase the likelihood that 100
dwellings were ultimately enrolled.

J Community Health (2010) 35:543548

observation (e.g., is there evidence of a smoker at the

dwelling?).
Inspectors conducted visual inspections of smoke alarms
that were present and recorded that information on data
forms. They checked that the smoke alarms (1) matched
the number of alarms originally installed; (2) were the
same alarms as those originally installed; and (3) were
operable. If a battery was missing and there was no physical damage to the alarm, the inspectors inserted a lithium
battery to determine if the alarm was operable. If the smoke
alarm was determined to be nonfunctional, the inspectors
replaced the old smoke alarm with a new one. Working
alarms that were installed by the original SAIFE program
were also replaced since these alarms were within a year of
their optimal life (10 years).
Data forms were reviewed for completeness and consistency, and data were entered into an Excel database and
checked for data entry accuracy.
Data Analysis
Key outcomes of interest were: (1) the alarms being
functional; and (2) a home having any original alarms
missing. We examined the bivariate relationships between
the key outcomes and potential predictors of these outcomes (i.e., smoker present, original resident, room location of alarm, state).
We analyzed data using SAS System for Windows,
version 9.1.3 (SAS Institute Inc., Cary, NC), and calculated
the odds ratios (OR).

Results
Number of Dwellings/Alarms Sampled
Local inspectors evaluated 427 dwelling units. Forty-three
units were excluded because inspectors did not have information on the number of smoke alarms installed. In the
remaining 384 units 601 installed alarms were evaluated,
which is an average of 1.6 alarms per dwelling unit. Table 1
presents dwelling and smoke alarm numbers by state.
Fifty-seven percent of the dwellings had one alarm installed
810 years ago; 35% of dwellings had two installed; 5%
had three installed and 3% had four or five installed.
Demographics of Communities/Dwellings Sampled

Data Collection
The inspectors traveled to each home for data collection.
All demographic data collected for this study came from
records from the time of installation (e.g., is resident at
time of installation still present?) or from inspector

123

Out of the 384 homes with complete smoke alarm data, 211
(55%) were in urban/metropolitan areas; 92 (24%) were in
mid-sized cities; and 81 (21%) were in rural areas. For this
evaluation, enrolling homes in rural communities was
challenging, resulting in less than 500 units being recruited

J Community Health (2010) 35:543548

Table 1 Number of dwellings
and smoke alarms evaluated

545

State

Dwellings units
evaluated

Dwellings with
complete alarm data

Installed
alarms

Georgia

152

115

141

1.2

Kentucky

40

40

82

2.1

Oklahoma

93

92

117

1.3

Virginia

100

96

154

1.6

42

41

107

2.6

427

384

601

1.6

Washington
Total

in 27% of homes, in 49% there were no smokers, and for

24% a determination could not be made. (Table 2)

Table 2 Ownership status, resident tenure and presence of smoker

Number

Percentage (%)

Smoke Alarm Status

Ownership status
Owned

249

65

Rented

109

28

26

Yes

233

61

No

121

32

30

Yes

103

27

No
NA*

188
93

49
24

384

100

NA*

Eight to ten years after the installation of the lithiumpowered smoke alarms, the inspectors found that one-third
of the alarms were still functional (Table 3). Thirty-seven
percent of the installed alarms were missing, and 30% of
the alarms were present but not functioning. Of the 180
alarms that were present but not functional, 43% had a
dead battery; 17% had no battery; 13% appeared to be
nonfunctional because of physical damage, and remaining
27% were not functioning for some other reason such as
missing parts and dust accumulation. At the time of this
evaluation, 38% of the dwellings had at least one of the
originally installed alarms still functional. Thirty (30%)
percent of the dwellings had all of the originally installed
alarms still functional. For 34% of the dwellings, all of the
originally installed alarms in the home were missing.

Original resident

NA*
Smoker present

Total

Average alarms
per dwelling

* Information not collected or unavailable

into the project and more homes being evaluated in urban

areas than in rural areas.
Sixty-five percent of the homes were owner-occupied,
28% were rented, and for 7% the occupancy status was
unknown. Higher percentages of rental units were found in
the urban/metropolitan areas. Sixty-one (61%) percent of
the homes had the same resident as when the alarms were
originally installed, 32% had a different resident, and for
8%, the residency tenure was unknown. A smoker resided

Table 3 Status of smoke

alarms

Several significant bivariate associations were seen

(Table 4). Alarms that were originally installed in the
kitchen were less likely to be functioning at follow-up (odds
ratio (OR) = 0.34, P = 0.006). Homes that had a different
resident at follow-up were less likely to have smoke alarms

Alarm status

Number

Functional

201

Missing

220

37

180

30

601

100

Present, not functional

Number

Percentage (%)

Dead battery

77

43

Missing battery

30

17

Physical damage

24

13

Other*

34

19

Unknown
Total
* Includes missing parts and
dust accumulation

Influence of Other Factors on Alarm Performance

Total

15

180

100

Percentage (%)
33

123

546

J Community Health (2010) 35:543548

Table 4 Bivariate odds ratios

Smoke alarm
functionality

Smoke alarm missing

Odds ratio

P value*

Odds ratio P value*

No

1.45

0.131

0.6

0.042

Unknown
Yesa

2.37
1.00

0.005

1.07
1.00

0.814

No

0.6

0.049

2.53

\0.001

Unknown

0.4

0.034

1.61

0.226

Yesa

1.00

1.00

Smoker present

Original resident

had non-lithium powered batteries, and the inspectors did

not report a battery-type for 109 (31%). Seventy-eight
percent (78%) of the alarms with lithium-powered batteries
were functioning at inspection compared with only 53% of
the alarms with non-lithium-powered batteries (Table 5).
Twenty percent (20%) of the alarms with non-lithiumpowered batteries (33 of 169) had dead batteries, compared
with 14% of the alarms with lithium-powered batteries (10
of 73). Thirty-one (31%) of the alarms with no report of a
battery type had a dead battery (34 of 109).

Discussion

Room location of alarm

Kitchen

0.34

0.006

Unknown

0.82

0.886

Non-kitchena

1.00

Ownership status
\0.001

Owner occupied 1.59

0.085

0.39

Unknown

0.36

0.049

0.95

0.912

Rentala

1.00

1.00

GA
KY

1.22
3.62

0.811
0.074

0.25
0.12

\0.001
\0.001

OK

7.25

0.015

0.15

\0.001

VA

1.52

0.617

0.17

\0.001

1.00

1.00

State

WA

* P value of statistical test that odds ratio equals the odds ratio of the
reference group
a

Reference group

functioning (OR = 0.60, P = 0.049) and more likely to

have alarms missing (OR = 2.53, P \ 0.001). Homes with
only non-smokers were less likely to have alarms missing
(OR = 0.60, P = 0.042). Additionally, differences among
states were also observed for both alarm functionality and
smoke alarm missing at follow-up.
Type of Battery Present
Of the 381 alarms that were present at the time of evaluation, inspectors reported that the battery was missing from
30 of the alarms (Table 3). Of the remaining 351 alarms,
73 (21%) contained lithium-powered batteries, 169 (48%)

This evaluation aimed to examine whether 10-years smoke

alarms actually provide protection for 10 years. Most
smoke alarm follow-up studies report shorter follow-up
times [11, 12], with the longest being 3 years [13]. After 8
10 years, we found that only 33% of the original alarms
were still functioning at follow-up. In a large proportion of
homes, the original alarm was physically removed. Other
homes had alarms present but not functional.
A large proportion of lithium batteries had been
replaced, which defeats the purpose of using a long-lasting
battery. At the time of the original SAIFE, smoke alarms
that were installed were not sealed units, so it was possible
to remove the batteries. Among the 77 alarms with dead
batteries, only 10 were confirmed to be lithium batteries.
Residents may have removed the original batteries to use
them in other devices; they may have followed widely
advertised recommendations that indicate batteries should
be changed each year; or it is possible that the lithium
batteries died at some point prior to our follow-up, and the
resident replaced it with a non-lithium battery. However,
our study was not able to determine the extent to which any
of these explanations occurred. Lithium-powered alarms
that are sold today often are tamper-resistant or have
sealed-in batteries, minimizing the possibility of battery
replacement. This was not the case 10 years ago. Among
the 73 alarms that had lithium batteries at the time of our
evaluation, almost 80% were still functional, indicating
that if alarms are not physically taken down and batteries
are not changed, then lithium batteries will last 810 years.
While this is encouraging, smoke alarm installation

Table 5 Type of battery present in alarms

Lithium n = 73

Non-lithium n = 169

Battery type not reported n = 109

Number

Percentage (%)

Number (%)

Percentage (%)

Number (%)

Percentage (%)

Functional

57

78

89

53

55

50

Non-functional*

16

22

80

47

54

50

* For non-functional alarms the number due to dead batteries were 10 (Lithium), 33 (Non-Lithium), and 34 (battery type not reported)

123

J Community Health (2010) 35:543548

programs, such as SAIFE, that choose to use lithium alarms,

need to provide stronger education to residents on smoke
alarm maintenance, with particular emphasis on the fact that
lithium alarms should not have their batteries replaced
annually. Current lithium smoke alarms do not have a label
on them that indicate that the batteries should not be
replaced. Programs that use these types of smoke alarms
should consider labeling the alarm to indicate the battery
should not be replaced. Most of the SAIFE grantees do this.
Our findings support guidance issued by the U.S. Consumer Product Safety Commission (CPSC) and the CDC in
2002 that 10-years or long-life smoke alarms not have
removable, replaceable batteries [14]. In response to the
CPSC report, the Underwriters Laboratory (UL) revised
standard UL217Single and Multiple Station Smoke
Alarms -to require long-life alarms to have non-removable
batteries and demonstrate a 10-years life through testing.
Some manufacturers of 10-years smoke alarms now may
provide a warranty for both the alarm mechanism and the
battery, whereas at the time that the alarms in this evaluation were installed, the battery was only warranted by its
manufacturer.
For the alarms that were present at the time of evaluation, just over half (53%) of the alarms were operational.
Only a small proportion of the alarms that were still present
were installed in kitchens. False or nuisance alarms may
occur from cooking that could have led to residents disabling alarms by removing the battery or by taking down
the entire unit (and not reinstalling later). Installing smoke
alarms in kitchens may not be optimal for this reason, and
design improvements are needed to reduce false alarms.
Smoke alarm programs should use alarms with silence or
hush buttons, and properly educate residents about using
this feature for nuisance alarms. UL217 requires that
smoke alarms with non-replaceable batteries include a hush
button to allow users to temporarily silence a nuisance
alarm without removing the smoke alarm or its batteries.
Our study found an association between missing alarms
and rental properties, and missing alarms and homes where
the resident changed since the time of alarm installation.
There are a number of possible reasons for this. Previous
occupants may have taken the alarms with them when they
moved out, landlords may have removed them during
property turnover/maintenance (e.g., painting) and not
replaced them, or new residents who were not present at
the time of the SAIFE program visit may have removed
them upon occupancy.
This study was subject to several limitations. First, we
were not able to interview residents; therefore, it is unclear
why alarms were removed or not properly maintained, and
what fire safety knowledge individual residents recalled
from the original education given through SAIFE. Knowing this information could improve fire safety programs by

547

strengthening education that addresses barriers to proper

smoke alarm maintenance.
Second, resources allowed us to examine only five
states. There were 14 original grantees. Each grantee had
the ability to tailor their program to their communities and
needs. If more grantees were able to participate in this
evaluation, it may have been possible to better understand
specific program characteristics that were associated with
smoke alarms remaining functional.
Third, inspectors in this study did not systematically
report the presence or functionality of alarms that were not
part of the original installation program. Some inspectors
documented that non-program smoke alarms were in some
dwellings; however, this information was not collected
consistently. While we were unable to quantify this, we do
know that the percentage of homes with working alarms is
higher than the 38% reported here. It is possible that there
is an indirect benefit from the original SAIFE program.
Through this program, residents may have become more
familiar with fire prevention and safety, and proactively
installed their own alarms. Future evaluation efforts might
attempt to capture this indirect effect.
Evaluation and follow-up are essential to smoke alarm
installation programs. Because alarms can become nonfunctional for reasons other than battery failure, fire
departments and organizations that do installation programs should plan to revisit homes 68 months after the
initial installation to assess alarm functionality. This is
particularly important for areas with high occupancy
turnover. Even if residents properly maintain their lithium
smoke alarms, after 810 years the batteries will ultimately
fail and if they do not fail, the current recommendation is to
replace all smoke alarm units after 10 years. Sustained long
standing programs should consider revisiting high-risk
homes after 10 years to replace older alarms with new
ones. Ongoing systematic program evaluation can be used
to monitor progress toward goals, improve service, and
demonstrate the programs effectiveness [15].
Acknowledgments The authors would like to thank Mary Borges
from the Washington Department of Health, Robert McCool from the
Kentucky Department of Health, Pam Archer from the Oklahoma
State Department of Health, Steve Davidson from the Georgia
Department of Health, and Lenny Recupero from the Virginia
Department of Health, and to all the local fire departments for their
assistance provided during this evaluation.
Disclaimer The findings and conclusions in this report are those of
the authors and do not necessarily represent the official position of the
Centers for Disease Control and Prevention.

References
1. Karter, M. J. (2008). Fire loss in the United States during 2007.
Quincy, MA: National Fire Protection Association.

123

548
2. Roberts, I., & Power, C. (1996). Does the decline in child injury
mortality vary by social class? A comparison of class specific
mortality in 1981 and 1991. BMJ, 313, 784786.
3. Ahrens, M. (2004). U.S. experience with smoke alarms and other
fire alarms. Quincy, MA: National Fire Protection Association.
4. Istre, G. R., McCoy, M. A., Osborn, L., Barnard, J. J., & Bolton,
A. (2001). Deaths and injuries from house fires. New England
Journal of Medicine, 344, 19111916.
5. Shenassa, E. D., Stubbenick, A., & Brown, M. J. (2004). Social
disparities in housing and related pediatric injury: A multilevel
study. American Journal of Public Health, 94, 633639.
6. Warda, L. J., & Ballesteros, M. F. (2007). Interventions to prevent residential fire injury. In L. S. Doll, S. E. Bonzo, J. A.
Mercy, & D. A. Sleet (Eds.), Handbook of injury and violence
prevention (pp. 97115). New York, NY: Springer.
7. Mallonee, S., Istre, G. R., Rosenberg, M., Reddish-Douglas, M.,
Jordan, F., Silverstein, P., et al. (1996). Surveillance and prevention of residential-fire injuries. New England Journal of
Medicine, 335, 2731.
8. DiGuiseppi, C., Roberts, I., Wade, A., Sculpher, M., Edwards, P.,
Godward, C., et al. (2002). Incidence of fires and related injuries
after giving out free smoke alarms: Cluster randomized controlled trail. BMJ, 325, 995.
9. Harvey, P. A., Aitken, M., Ryan, G. W., Demeter, L. A., Givens,
J., Sundararaman, R., et al. (2004). Strategies to increase smoke
alarm use in high-risk households. Journal of Community Health,
29, 375385.

123

J Community Health (2010) 35:543548

10. Ballesteros, M. F., Jackson, M. L., & Martin, M. W. (2005).
Working toward the elimination of residential fire deaths: The
Centers for Disease Control and Preventions Smoke Alarm
Installation and Fire Safety Education (SAIFE) program. Journal
of Burn Care and Rehabilitation, 26, 434439.
11. Mickalide, A., & Validzic, A. (1999). Smoke alarm maintenance
of low-income families. American Journal of Public Health, 89,
1584.
12. Thompson, C. J., Jones, A. R., Davis, M. K., & Caplan, L. S.
(2004). Do smoke alarms still function a year after installation? A
follow-up of the Get-Alarmed Campaign. Journal of Community
Health, 29, 171181.
13. Shults, R. A., Sacks, J. J., Briske, L. A., Dickey, P. H., Kinde, M.
R., Mallonee, S., et al. (1998). Evaluation of three smoke detector
promotion programs. American Journal of Preventive Medicine,
15, 165171.
14. Lee, A. (2002). Preliminary test results on lithium batteries used
in residential smoke alarms. Washington, DC: U.S. Consumer
Product Safety Commission.
15. Thompson, N. J., & McClintock, H. O. (1998). Demonstrating
your programs worth: A primer on evaluation for programs to
prevent unintentional injury. Atlanta, GA: Centers for Disease
Control and Prevention, National Center for Injury Prevention
and Control.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.