Tepper et al (Rapid Response, BMJ 17 Dec 2003) and Gunnell (Rapid
Response, BMJ 19 Dec 2003) suggest that in our study of the effectiveness
of the US Air Force (USAF) suicide prevention programme(1) we did not take
account of temporal changes in the US general population suicide rate.
Although we recognize that there were declines in the US, especially in
certain subpopulations, none of the reductions in suicide rates were equal
in magnitude to that which we found in the USAF following implementation
of the programme. The national decline in suicide rates is commonly
attributed to a host of factors including a robust economy with
historically low unemployment and a decline in hard drug use, neither of
which would have been expected to be a factor underlying a reduction in
the number of suicides in the USAF. We also would like to point out that
the Centers for Disease Control, Division of Injury and Prevention
provides a cautionary note(2) for comparing fatal and non fatal injury
data from 1998 and before with data after that time:
“The external cause of injury coding for 1999 and later, based on the
ICD-10 classification system, is notably different from external cause
coding for 1998 and earlier years, based on the ICD-9 classification
system. You may not be able to compare numbers of deaths and deaths rates
computed for some external causes of injury based on 1999 and later data
to those based on data from 1998 and earlier. Consequently, use caution
when doing trend analysis of numbers of deaths and death rates across
It could be argued, therefore, that a comparison between the apparent
declines in the US general population to the sustained declines in the
USAF population is not interpretable. (The USAF definition of suicide
remained stable over the decade and was not dependent on ICD 9/ICD 10
coding). Moreover, this decline continues to be sustained seven years
after the initiation of prevention efforts; the suicide rate in the
service as of Dec 11th 2003 was 9.9/100,000. As Tepper et al suggest
following the data for longer lengths of time can help to confirm
substantial changes; from 1980-1990 there was no sustained decline in the
suicide rate in the USAF similar to the one we reported on.
One final note in reference to Tepper et al’s statement that, “the
suicide rate in the USAF was already declining in 1996 (before the
preventive program was put in place) and 1997 (before the program was
fully implemented)”. We clearly stated in the paper that the programme
began in 1996 and was fully implemented by 1997. The declines in these
years are directly in keeping with our interpretation that there is
evidence of the effectiveness of the Air Force programme. While we were
conservative in our analytic approach by not including 1996 in the post-
exposure cohort, in fact the program was well underway in several commands
during that time. (Therefore any beneficial effects are attributed to the
pre-exposure cohort, thus reducing the effect size we reported.) The fact
that we still found relative risk reductions, of the magnitude seen in
this study, does provide evidence of programmatic effectiveness.
Moreover, since the programme targeted reducing risk factors and
increasing protective factors that theoretically also should reduce
violence in general, the validity of the results is strengthened by the
findings of reductions in multiple outcomes, in addition to suicide.
1. Knox KL, Litts DA, Talcott W, Feig JC, Caine ED. Risk of suicide
and related adverse outcomes after exposure to a suicide prevention
programme in the US Air Force: cohort study. Br Med J 2003; 327:1376-8.
2. Centers for Disease Control and Prevention. Web-based Injury
Statistics Query and Reporting System (WISQARS) [Online]. (2002). National
Center for Injury Prevention and Control, Centers for Disease Control and
Prevention (producer). Available from: URL: www.cdc.gov/ncipc/wisqars.
[Accessed 2004 7 Jan].
Competing interests: No competing interests