Psychological Medicine, 2004, 34, 1419–1430.

f 2004 Cambridge University Press

DOI : 10.1017/S0033291704002818 Printed in the United Kingdom

Psychological health of Australian veterans of

the 1991 Gulf War: an assessment using the SF-12,
GHQ-12 and PCL-S
D. P. M C K E N Z I E*, J. F. I K I N, A. C. M C F A R L A N E, M. C R E A M E R, A. B. F O R B E S,
H. L. K E L S A L L, D. C. G L A S S, P. I T T A K A N D M. R. S I M
Department of Epidemiology and Preventive Medicine, Monash University ; Australian Centre for
Posttraumatic Mental Health, University of Melbourne, Melbourne, Victoria ; Department of Psychiatry,
University of Adelaide, Adelaide, South Australia, Australia

ABSTRACT
Background. Elevated rates of psychological morbidity and symptomatology have been widely
reported in 1991 Gulf War veterans. The present study used brief self-report instruments to com-
pare the psychological health of Australian Gulf War veterans with that of a randomly sampled
military comparison group.
Method. The 12-item Short Form Health Survey (SF-12), 12-item General Health Questionnaire
(GHQ-12), Posttraumatic Stress Disorder Checklist – Specific (PCL-S) and Military Service Ex-
perience (MSE) questionnaire were administered to 1424 male Australian Gulf War veterans and
1548 male Australian Defence Force members who were operational at the time of the Gulf War
conflict, but were not deployed there.
Results. The Gulf War veterans exhibited poorer psychological health, as measured by the above
three instruments, than the comparison group members. For Gulf War veterans, the number of
stressful experiences, as measured by the MSE questionnaire, was correlated with scores on the
three instruments. SF-12 mental health component summary scores and PCL-S caseness, but not
GHQ-12 caseness, differed significantly between Gulf War veterans and comparison group mem-
bers who had been on at least one active deployment.
Conclusions. More than a decade after the 1991 Gulf War, Australian Gulf War veterans are exhi-
biting higher levels of current (past month) psychological ill-health, as measured using the GHQ-12
and PCL-S, as well as lower mental health status, as measured by the SF-12, than the comparison
group. Although not a replacement for formal psychiatric diagnosis, instruments such as those
above may aid in the assessment of veterans’ psychological health.

INTRODUCTION remain. In order to best care for those forces

returning from that region, and to improve
There has recently been a new war in Iraq,
preparation for future deployments, the long-
involving military forces from the United States,
term effects of the 1991 Gulf War on the health
United Kingdom, Australia and other countries.
of veterans must be better understood.
Unfortunately, tensions in the Gulf (sometimes
known as the Persian Gulf) region are likely to Many studies have examined the psychologi-
cal and physical health of Gulf War veterans.
* Address for correspondence : Mr Dean P. McKenzie, Monash Indeed, in a recent editorial appearing in this
University, Department of Epidemiology and Preventive Medicine, journal, Hyams & Scott (2002) suggested that
Alfred Hospital, Commercial Road, Melbourne, Victoria, Australia
3004. these veterans form one of the most highly
(Email : [Link]@[Link]) studied adult populations. Increased numbers
1419
1420 D. P. McKenzie et al.

of psychiatric diagnoses, and or symptoms instruments to compare the psychological

of substance misuse, anxiety, depression and health of the Australian 1991 Gulf War veterans
post-traumatic stress disorder (PTSD), and and the randomly sampled comparison group.
non-specific psychological symptoms such as These instruments consisted of the 12-item ver-
agitation and concentration difficulties, have sion of the Short Form Health Survey (SF-12 ;
been found in Gulf War veterans from Canada Ware et al. 1998), the 12-item version of the
(Goss Gilroy Inc., 1998), Denmark (Ishoy et al. General Health Questionnaire (GHQ-12;
1999), the UK (Unwin et al. 1999) and the USA Goldberg & Williams, 1988), and the Posttrau-
(Iowa Persian Gulf Study Group, 1997 ; Joseph matic Stress Disorder Checklist – Specific (PCL-
et al. 1997; Kang et al. 2003). A systematic S ; Weathers et al. 1993).
review of published studies has been provided Stressful events, including fear of death and
by Stimpson et al. (2003). feelings of helplessness, are associated with
The psychological and physical health of increased rates of PTSD (Adler et al. 1996).
Australian Gulf War veterans has not pre- The current study, therefore, used the Military
viously been objectively and comprehensively Service Experience (MSE) questionnaire,
analysed. As part of the recent Australian Gulf specifically developed by Ikin et al. (2004), to
War Veteran’s Health Study, Ikin et al. (2004) examine the psychological stressors experienced
compared DSM-IV (APA, 1994) diagnoses in by Australian Gulf War veterans. Finally, in
Australian Gulf War veterans with those of a order to determine whether increased psycho-
randomly sampled military comparison group. logical ill-health in Gulf War veterans is a result
Diagnoses were made using the computerized of deployment to war itself rather than some-
version of the Composite International Diag- thing specific to a particular war, we compared
nostic Interview (CIDI) (Robins et al. 1988 ; the GHQ and PCL-S caseness of Gulf War
WHO Collaborating Centre for Mental Health veterans with that of comparison group mem-
and Substance Abuse, 1997). bers who had been on at least one active de-
Ikin et al. (2004) found that the two groups ployment.
exhibited similar diagnostic patterns for the Three major questions will be investigated in
period prior to the Gulf War. The Gulf War the present paper: (1) do Australian Gulf War
veterans, however, were more likely than the veterans have lower scores (indicating poorer
comparison group to have met criteria for psy- health) on the SF-12 and exhibit more GHQ-
chiatric diagnosis at some point since the Gulf 12 and PCL-S caseness than the comparison
War, as well as during the 12 months prior to group ; (2) is poor psychological health in Gulf
CIDI administration (study conducted during War veterans related to the number of psycho-
2000–2002). Commonly occurring diagnoses logical stressors experienced during their Gulf
included PTSD, major depression and sub- War deployment, as measured using the MSE
stance abuse. questionnaire, and (3) will differences in case-
Diagnostic systems such as the CIDI require ness also be apparent between Gulf War vet-
administration by trained personnel and may erans and those comparison group members
be time-consuming when employed with large who have been actively deployed ?
groups, with the average time to complete the
computerized version reported to be around 75
METHOD
minutes (WHO Collaborating Centre for
Mental Health and Substance Abuse, 1997). Subjects
The administration of such comprehensive sys- The Australian Gulf War veteran population
tems might, therefore, be usefully supplemented consisted of 1833 males and 38 females who
by brief, self-report instruments. In addition, served in the Gulf region from 2 August 1990,
more detailed knowledge about the perform- the date of the invasion of Kuwait by Iraqi
ance of such instruments in military populations forces, to 4 September 1991, the end date of
is important to ensure their adequacy in the the Australian Department of Veterans’ Affairs
assessment of returning veterans. Nominal Roll for the Gulf War. Most (1579,
The present study seeks to extend the findings 84.4%) of the 1871 personnel that deployed to
of Ikin et al. (2004) by using brief self-report the Gulf were in the Navy. The comparison
 Psychological health of Australian Gulf War veterans 1421

group of 2850 males and 74 females was ran- Summary’ (MCS-12), an index of mental health
domly selected from 26411 Australian Defence and well-being (Ware et al. 1998). Both
Force personnel who were in operational units summary scales have the same items but are
during the above time period, but did not deploy weighted differently. In both cases lower scores
to the Gulf War conflict. The veteran and com- represent poorer health.
parison groups were matched on sex, branch of
service (Navy, Army, Air Force) and 3-year age Twelve-item General Health Questionnaire
band. (GHQ-12)
Due to the small number of female Gulf The GHQ, available in various lengths, is one
War veterans, analyses were limited to males. of the most widely used self-report screening
The participating Gulf War veteran study tests for non-psychotic psychological illness
group consisted of 1424 males [1232 (86.5%) in (Goldberg & Williams, 1988). The GHQ-12 has
the Royal Australian Navy, 87 (6.1 %) in the been shown to have high validity (Clarke et al.
Australian Army and 105 (7.4 %) in the Royal 1993; Goldberg et al. 1997 ; Donath, 2001) and
Australian Air Force]. The participating com- has been used in a variety of general (Australian
parison study group consisted of 1548 males. Bureau of Statistics, 1998 ; Pevalin & Goldberg,
The combined male and female participation 2003), and military (Schei, 1994 ; Goyne, 2001)
rates, for the Gulf War veterans and the com- applications.
parison group members (80.5 % and 56.8 % We employed the standard, or binary, method
respectively, not including persons deceased, or of scoring. In regard to choice of caseness
overseas during the study) were comparable to threshold or cut-off score, previous Gulf War
those reported in the review by Stimpson et al. studies (Unwin et al. 1999) using the GHQ-12
(2003). The average age at time of study, of the have solely involved UK veterans and employed
Gulf War veteran and comparison groups a fixed cut-off of three or more symptoms. There
respectively, was 38.1 years (S.D.=6.4) and 39.3 is strong evidence, however, that optimal GHQ
years (S.D.=6.4). Further details of the recruit- cut-off scores vary geographically (Goldberg
ment, and demographic characteristics of the et al. 1997, 1998). Donath (2001) found the
groups, are provided by Ikin et al. (2004). optimal cut-off for the general Australian popu-
lation to be one or more symptoms. As Goyne
Instruments (2001) has suggested that the optimal cut-off
for Australian Defence Force members may be
Subjects completed the SF-12, GHQ-12, PCL-S different to that of the general population we
and MSE as part of a comprehensive postal determined the optimal cut-off empirically, as
questionnaire. described below.
Short-Form-12 Health Survey (SF-12) Posttraumatic Stress Disorder
The SF-12 (Ware et al. 1998) is a subset of, and Checklist – Specific (PCL-S)
has comparable validity to, the SF-36 (Ware The PCL (Weathers et al. 1993) is a self-report
et al. 1993), a widely employed measure of rating scale for assessing the 17 DSM-IV symp-
health status and health-related quality of life. toms of PTSD.
Although no previous Gulf War studies appear Validation of the PCL has been carried out
to have employed the SF-12, several studies (e.g. by Blanchard et al. (1996), Forbes et al. (2001)
Iowa Persian Gulf Study Group, 1997 ; Voelker and Weathers et al. (1993). The instrument
et al. 2002) have used the SF-36. has been used in general psychiatric (Sampson
The SF-12 has been validated internationally et al. 2003), as well as military (Weathers et al.
(Gandek et al. 1998; Sanderson & Andrews, 1993; Barrett et al. 2002 ; Sutker et al. 2002)
2002) and has been used in a variety of studies research.
(e.g. Australian Bureau of Statistics, 1998; We made a slight modification to the specific
Herrman et al. 2002). The instrument has two event version of the PCL (PCL-S). The stem
summary scales, the ‘Physical Component question, originally reading : ‘Please consider the
Summary ’ (PCS-12), an index of physical health event that you found most stressful or upset-
and well-being, and the ‘Mental Component ting … ’, was changed to ‘Please consider the
1422 D. P. McKenzie et al.

event or group of events, military or non-military, non-supervisory, other rank – supervisory (at or
in your life that you found most stressful or above the rank of Leading Seaman in the Navy
upsetting … ’. This change was implemented or Corporal in the Army and Air Force) and
because PTSD may have been precipitated by officer. Rank categories are comparable to the
a group of related events. As our study had categories employed by Ismail et al. (2000).
a military focus, participants may otherwise If the number of subjects exhibiting caseness,
have felt obliged to nominate stressful military, in either study group, was small (arbitrarily but
rather than non-military, events. conventionally defined as being five or less),
A PCL cut-off of 50 was originally rec- exact logistic regression (Mehta & Patel, 1995)
ommended by Weathers et al. (1993), and was was performed using LogXact 4 (Cytel Software
recently employed with US Gulf War veterans Corporation, 2000).
by Barrett et al. (2002). In a study of Australian Differences in mean SF-12 summary scores
Vietnam War veterans, Forbes et al. (2001) were obtained using multiple regression, before
found little difference in screening and diag- and after adjusting for possible confounding
nostic performance between three cut-offs (45, factors. In order to determine the optimal
50, 55) and suggested that the originally rec- GHQ-12 threshold for our data, non-parametric
ommended cut-off was suitable for use with (Hanley & McNeil, 1982) Receiver Operating
Australian veterans. This threshold has, there- Characteristic (ROC) analysis (Kraemer, 1992)
fore, been adopted in the present study. was applied to the Gulf War veterans and
comparison group combined. The criterion
Military Service Experience (MSE) diagnosis was defined as the presence of any
questionnaire DSM-IV diagnosis, excluding current substance
use disorder, alcohol use disorder and specific
Psychological stressors were assessed using the
phobia, within the previous 4 weeks. These
MSE questionnaire (Ikin et al. 2004). This three diagnoses were excluded because it would
questionnaire consists of 44 items, each de-
not be expected that they would be detected by
scribing a potentially stressful experience for
the GHQ and so should not be included in an
Australian Gulf War veterans, such as boarding
assessment of its specificity and sensitivity.
hostile ships at sea, fear of entrapment below the
Substance abuse (including alcohol) in the CIDI
waterline as a result of missile attack or collision
is diagnosed on the basis of the quantity and
with sea-mine, or threat of chemical or biologi-
regularity of consumption, no specific symp-
cal attack. The instrument was derived from
toms of distress being required to satisfy
various sources, including the Combat Ex-
the diagnosis. Simple phobia is defined by the
posure Scale (Keane et al. 1989), and findings
avoidance of a specific object or situation, with
from an Australian Gulf War veteran focus
no anxiety or distress present if the individual is
group. The MSE questionnaire was scored by
not confronted by the phobic stimulus. Again,
summing the 44 binary-coded items.
this avoidance is not screened by the GHQ.
All diagnoses were made using the computerized
Statistical analyses CIDI administered by psychologists, as de-
For GHQ-12 and PCL-S caseness, odds ratios scribed by Ikin et al. (2004). Confidence intervals
were first calculated using study group status (CI) for sensitivity, specificity and diagnostic
only (crude odds ratios), and again after efficiency (Kraemer, 1992) were calculated
adjusting for possible confounding factors (ad- using the procedure described by McKenzie
justed odds ratios), using logistic regression. et al. (1997), and implemented by Mackinnon
The possible confounding factors consisted of (2000).
age as at 2 August, 1990 (<20, 20–24, 25–34, The values of the crude, and the adjusted,
o35 years), branch of service, highest education odds ratios and differences between means
level (f10, 11 or 12 years of schooling, cer- were found to be highly similar, and so only
tificate or diploma, tertiary), marital status the adjusted results are reported. Unless speci-
(married/de facto; separated, divorced or fied otherwise, all statistical analyses were car-
widowed ; single/never married) and military ried out using the Stata 7 package (StataCorp,
rank. Rank was categorized as other rank – 2001).
 Psychological health of Australian Gulf War veterans 1423

Table 1. Mean SF-12 mental component (MCS-12) and physical component (PCS-12)
summary scores : the effects of study group across subgroups of age and rank
Gulf War Comparison
veterans group
(n=1374) (n=1513) Adj. difference*
Mean (S.D.) Mean (S.D.) (95 % CI) p value

MCS-12
All subjects 47.4 (11.2) 50.9 (9.5) x3.4 (x4.2 to x2.6) <0.001
Age (years)
9
<20 45.0 (11.5) 51.9 (8.7) x7.1 (x9.5 to x4.8)>
=
20–24 47.9 (11.5) 51.1 (9.3) x3.9 (x5.4 to x2.5) 0
001#
25–35 48.0 (10.8) 50.8 (9.4) x2.7 (x3.8 to x1.7)>;
35+ 48.8 (11.3) 50.2 (10.3) x1.9 (x3.9 to 0.1)
Rank 9
Officer 49.3 (9.8) 51.7 (8.9) x2.5 (x4.1 to x0.9)=
Other rank – supervisory 47.8 (11.2) 50.3 (9.8) x2.6 (x3.7 to x1.5) 0
009#
;
Other rank – non-supervisory 45.9 (11.8) 51.1 (9.5) x5.3 (x6.6 to x3.9)
PCS-12
All subjects 49.2 (9.2) 49.9 (9.0) x0.9 (x1.6 to x0.2) 0.008
Age (years)
9
<20 49.7 (9.8) 51.1 (8.1) x1.4 (x3.5 to 0.6)>
=
20–24 49.8 (8.7) 51.0 (9.1) x1.0 (x2.3 to 0.2) 0
739#
25–35 49.3 (8.9) 49.7 (8.7) x0.4 (x1.4 to 0.5)>;
35+ 46.9 (10.6) 48.5 (9.8) x1.8 (x3.6 to 0.0)
Rank 9
Officer 50.3 (8.8) 52.1 (8.0) x1.6 (x3.0 to x0.2)=
Other rank – supervisory 49.1 (8.9) 48.7 (9.4) . . .
x0 1 (x1 0 to 0 9) 0
745#
Other rank – non-supervisory 48.8 (9.8) 50.2 (8.9) x1.7 (x2.9 to x0.5);

* Differences between means are adjusted for service branch, rank, age category, education and marital status.
# p value for interaction.

RESULTS expected mean MCS-12 score decreased by 0.77

with each increase in age category, 95% CI
SF-12 x0.149 to x0.05, p=0.037).
A total of 1374 (96.5 %) of the 1424 Gulf War There was a statistically significant (p=0.009)
veterans in the study group, and 1513 (97.7%) interaction between rank (assuming linear
of the 1548 comparison study group members trend), and study group. In the case of the Gulf
fully completed the SF-12. War veterans there was a statistically significant
Table 1 shows the mean scores of the MCS-12 positive linear relationship between rank and
and the PCS-12 for the total study population mean MCS-12 score (slope=1.84, 95 % CI
and subgroups categorized by age and rank (due 0.90–2.77, p<0.001), with the lower ranks
to the low numbers of Army and Air Force having lower scores. This relationship was not
personnel, service branch subgroups were not observed in the comparison group (p=0.334).
compared). The Gulf War veterans had lower With regard to self-reported physical health,
mean MCS-12 scores than the comparison Gulf War veterans had a lower mean PCS-12
group (p<0.001), indicating poorer psycho- score than the comparison group (p=0.008).
logical health. There were no statistically significant interac-
There was a statistically significant interac- tions between study group and age category
tion (p=0.001) between age group (linear trend) or rank. For both study groups, there was a
and study group. For the Gulf War veterans, negative linear relationship between age and
MCS-12 means increased with age (slope=0.81, mean PCS-12 scores (slope=x1.60, 95% CI
indicating that the expected mean MCS-12 x2.10 to x1.09, p<0.001), with the higher age
score increased by 0.81 with each increase in categories reporting poorer physical health.
age category, 95 % CI 0.07–1.54, p=0.032), There was a positive linear relationship between
representing better mental health. The opposite rank and mean PCS-12 scores (slope=1.51,
relationship was observed for the compari- 95% CI 0.86–2.16, p<0.001), with the lower
son group (slope=x0.77, indicating that the ranks reporting poorer physical health.
1424 D. P. McKenzie et al.

Table 2. Twelve-item General Health Questionnaire (GHQ-12) and Posttraumatic Stress Disorder
Checklist – Specific (PCL-S) caseness : the effects of study group across subgroups of age and rank
Gulf War Comparison
veterans group
(n=1422) (n=1544)
Adj.
n (%)* n (%)* OR# 95 % CI p value

GHQ-12
All subjects 564 (39.6) 502 (32.5) 1.4 1.2–1.6 <0.001
Age (years)
9
<20 74 (42.8) 29 (22.8) 2.6 1.5–4.3>
=
20–24 164 (40.6) 123 (31.3) 1.5 1.1–2.0 0
011$
25–34 267 (39.7) 266 (34.1) 1.3 1.0–1.6>;
35+ 59 (34.1) 84 (34.4) 1.1 0.7–1.6
Rank 9
Officer 94 (35.1) 117 (30.0) 1.3 0.9–1.8=
Other rank – supervisory 267 (39.0) 260 (35.2) 1.2 1.0–1.5 0
088$
;
Other rank – non-supervisory 202 (43.2) 125 (30.0) 1.8 1.4–2.4
PCL-S
All subjects 105 (7.9) 66 (4.6) 2.0 1.5–2.9 <0.001
Age (years)
9
<20 13 (7.9) 3 (2.6) 3.1· 0.8–17.7>
=
20–24 31 (8.2) 14 (3.8) 2.2 1.1–4.3 0
121$
25–34 40 (6.4) 28 (3.8) 1.9 1.2–3.2 >;
35+ 21 (12.5) 21 (9.0) 1.7 0.9–3.4
Rank 9
Officer 9 (3.6) 11 (3.0) 1.5 0.6–3.7=
Other rank – supervisory 49 (7.6) 37 (5.3) 1.8 1.1–2.9 0
163$
Other rank – non-supervisory 47 (10.6) 18 (4.7) 2.9 1.6–5.2;

* Percentage of subjects within each subgroup of age or rank.

# Odds ratios are adjusted for service branch, rank, age category, education and marital status.
$ p value for interaction.
· This odds ratio is adjusted for service branch and rank only. Confidence interval (CI) for this adjusted odds ratio was obtained using exact
logistic regression.

GHQ-12 There was a statistically significant interac-

A total of 1422 (99.9%) of the Gulf War vet- tion ( p=0.011) between age and study group.
erans in the study group, and 1544 (99.7%) of The prevalence of GHQ caseness increased with
the comparison study group members com- age for the comparison group (p=0.034), but
pleted the GHQ-12. decreased with age for the Gulf War veterans,
In determining the optimal cut-off score the latter result not being statistically significant
for the GHQ-12, the prevalence of the criterion (p=0.338). The interaction between study group
diagnosis defined earlier was observed to be and rank (assuming linear trend) narrowly failed
10 %, for both study groups combined. ROC to achieve statistical significance (p=0.088).
analysis found the optimal cut-off to be two or
more symptoms, with a sensitivity of 72 % (95 % PCL-S
CI 66–77), specificity of 68 % (95 % CI 66–69), A total of 1339 (94.0 %) Gulf War veterans in
overall diagnostic efficiency of 68 % (95 % CI the study group and 1452 (93.8 %) comparison
66–70), and area under the ROC curve of 0.77 study group subjects completed the PCL-S.
(95 % CI 0.74–0.80). Table 2 presents the prevalence of PCL-S
Table 2 presents the prevalence of GHQ-12 cases for the study groups overall, and within
cases, as defined above, for the study groups subgroups of age and rank. Gulf War veterans
overall, and within subgroups of age and rank. were more likely than comparison group sub-
Gulf War veterans were more likely than com- jects to be suffering symptoms indicative of
parison group subjects to be suffering psycho- PTSD (p<0.001).
logical distress or ill-health, as measured by the Although there were no statistically signifi-
GHQ-12 (p<0.001). cant interactions, power is lower than for the
 Psychological health of Australian Gulf War veterans 1425

Table 3. The relationship between stressful Gulf War Military Service Experience (MSE)
questionnaire score and SF-12 summary scores in Gulf War veterans
Adj.
Mean (S.D.) diff.* 95 % CI p value

SF-12 mental component (MCS-12)

MSE questionnaire score
0–4 52.1 (8.4) 0.0 —
5–8 49.8 (10.2) x2.2 x3.7 to x0.6
9–12 46.3 (10.8) x5.5 x7.2 to x3.8
>12 41.5 (12.0) x10.0 x11.6 to x8.4
Dose–response# — — x0.7 x0.8 to x0.6 <0.001
SF-12 physical component (PCS-12)
MSE questionnaire score
0–4 51.9 (6.6) 0.0 —
5–8 50.3 (8.3) x1.8 x3.2 to x0.5
9–12 49.4 (9.8) x2.7 x4 1 to x1.3
.
>12 45.6 (10.5) x6.5 x7.9 to x5.1
Dose–response# — — x0.5 x0.54 to x0.37 <0.001

* Differences in means are adjusted for service branch, rank, age category, education and marital status.
# Dose–response slope is the expected increase in mean MCS-12 or PCS-12 score per unit increase in the MSE questionnaire score.

Table 4. The relationship between Gulf War Military Service Experience (MSE)
questionnaire score and GHQ-12 and PCL-S caseness in Gulf War veterans
Adj.
n (%) OR* 95 % CI p value

GHQ-12 caseness
MSE questionnaire score
0–4 (n=320) 66 (21) 1.0 —
5–8 (n=415) 128 (31) 1.7 1.2–2.5
9–12 (n=316) 144 (46) 3.2 2 3–4.7
.
>12 (n=369) 226 (61) 6.1 4.2–8.7
Dose–response# — — 1.13 1.11–1.16 <0.001
PCL-S caseness
MSE questionnaire score
0–4 (n=320) 5 (2) 1.0$ —
5–8 (n=415) 12 (3) 2.0$ 0.6–7.3
9–12 (n=316) 21 (7) 4.2$ 1.5–14.7
>12 (n=369) 67 (19) 13.8$ 5.3–45.7
Dose–response# — — 1.22 1.17–1.27 <0.001

* Odds ratios are adjusted for service branch, rank, age category, education and marital status using logistic regression.
# Dose–response slope is the expected proportionate increase in the odds ratio per unit increase in the MSE questionnaire score.
$ These odds ratios are adjusted for service branch, rank and age (<25 v. o25 years) only. Confidence interval (CI) values for these
adjusted odds ratios were obtained using exact logistic regression.

previous analyses due to the smaller number of Gulf War service-related MSE questionnaire
observed cases. Tests for trend show that there and SF-12 summary scales. Table 4 shows the
were statistically significant relationships be- relationship between MSE questionnaire scores
tween age and PCL-S caseness [odds ratio (OR) and GHQ-12 and PCL-S caseness.
1.81, 95% CI 1.41–2.33, p<0.001], and rank A lower MCS-12 score was associated with an
and PCL-S caseness (OR 0.46, 95% CI 0.32– increasing MSE questionnaire score. A similar
0.66, p<0.001) for both study groups. result was obtained for the PCS-12, although
PCL-S caseness was more prevalent in the the relationship was less marked. Perceived ex-
older age groups, and in the lower ranks. posure to an increasing number of psychological
stressors, as indicated by an increasing score
Effects of stressful military experiences on the MSE questionnaire, was significantly
For Gulf War veterans only, Table 3 sum- (p<0.001) associated with increasing preva-
marizes the relationship between scores on the lence of both GHQ-12 and PCL-S caseness. The
1426 D. P. McKenzie et al.

expected increase in the odds of caseness, per implications for veterans of the recent war in
unit increase in MSE questionnaire score, was Iraq. The impact of traumatic events on long-
13 % for the GHQ-12 and 22 % for the PCL-S. term psychological health tends to be under-
estimated in both defence (McFarlane, 2003),
The effect of other deployments and general (McFarlane, 2000 a) populations.
Approximately one third of the comparison In addition to causing suffering and distress,
group (n=514) reported that they had been on psychological ill-health has adverse effects on
at least one active deployment. Of these, 507 cognitive processing, including target detection
completed the SF-12, 513 completed the GHQ- and reaction times (Clark et al. 2003; Farrin
12 and 488 completed the PCL-S. Mean MCS- et al. 2003 ; Hammar et al. 2003), with major
12 scores were significantly lower in Gulf implications for the capability of those in-
War veterans (mean=47.4, S.D.=11.2) than dividuals who may be unwell, yet remain in the
comparison group members who had active services.
deployments (mean=50.4, S.D.=9.6) (adjusted We also found Australian Gulf War veterans
difference between means=x2.5, 95 % CI x3.6 to have lower physical health status, as mea-
tox1.4, p<0.001). The difference between mean sured by the SF-12 physical health component
PCS-12 scores (mean=49.2, S.D.=9.2 ; versus summary scale (PCS-12), than the comparison
mean=49.5, S.D.=9.3 respectively) for the above group members, although the difference in
two groups was not statistically significant (p= physical health was not as great as the difference
0.501). in SF-12 mental health.
PCL-S caseness remained significantly elev- Younger Gulf War veterans, and older
ated (OR 1.9, 95 % CI 1.1–3.1, p=0.015) in comparison group members, were more likely
Gulf War veterans (7.9 %), compared to com- to exhibit psychological distress as measured
parison group members with active deployments by the GHQ-12. McCabe et al. (1996) observed
(4.5%). The difference in GHQ-12 caseness that GHQ-12 scores increased with age in the
(39.6% v. 36.1 %) between the above two groups general population. Goyne (2001) found GHQ-
was not statistically significant (p=0.370). 12 (Likert-scaled) scores to be lowest for the
The adjusted odds ratios given above are only youngest Australian Army officers, although
slightly smaller than those obtained when Gulf there was no age relationship for the non-officer
War veterans were compared with all compari- group (corresponding to our other ranks –
son group members. As expected, however, the non-supervisory and other ranks – supervisory
confidence intervals are slightly wider than categories combined).
those previously presented, as a result of re- We found that PCL-S caseness increased with
duced sample sizes. age, for both study groups. The group differ-
ences (Gulf War veterans having lower values)
in MCS-12 scores decreased as age increased,
DISCUSSION
although the reasons for this are unclear.
Our study found significantly lowered psycho- The group difference in MCS-12 scores (Gulf
logical health, as measured by the SF-12 mental War veterans having lower values) was much
health component summary scale (MCS-12), larger for the lowest ranks. Our results further
GHQ-12 and PCL-S, in Australian veterans suggested that the group difference in GHQ-12
of the 1991 Gulf War, compared with the mili- and PCL-S caseness was strongest for the lowest
tary comparison group. This finding supports ranks, with the findings for the GHQ-12, but
and extends that obtained for Australian Gulf not the PCL-S, narrowly missing statistical
War veterans by Ikin et al. (2004) using more significance (p=0.088). Subjects in the lowest
comprehensive 12-month CIDI diagnoses. The ranks typically recorded the highest levels of
veteran group exhibited high levels of cur- GHQ-12 and PCL-S caseness in both study
rent (past month) psychological ill-health, as groups. Goyne (2001) did not find any dif-
measured using the GHQ-12 and PCL-S, and ferences in GHQ-12 Likert scores between
low levels of mental health status, as measured Australian soldiers and officers, but found dif-
by the MCS-12, more than a decade after the ferences within the officers. The highest-ranking
end of the Gulf War. This has important officers had the highest scores (higher levels
 Psychological health of Australian Gulf War veterans 1427

of psychological distress), reflecting possible of recall bias cannot be ruled out in our study.
differences in age, experience, or level of Veterans reporting poor psychological health
responsibility. potentially exhibited heightened recall of
Ismail et al. (2000) found an inverse relation- psychological stressors, as reflected in their
ship between rank and GHQ-12 caseness for MSE scores.
UK Gulf War veterans (a comparison group We found no statistically significant differ-
was not used), with lower ranks exhibiting ence in GHQ-12 caseness between the Gulf War
higher caseness. In addition, Ismail et al. (2000) veterans and those members of the comparison
found such a relationship between rank and group who had been on at least one active
PTSD, as measured using several items from the deployment, in contrast to Unwin et al. (1999)
Mississippi Scale for Combat-Related Post- who found that UK Gulf War veterans had
traumatic Stress Disorder (Keane et al. 1988), significantly higher levels of GHQ-12 caseness
and suggested that rank could be a proxy for than Bosnia veterans. We did, however, find
socio-economic status. The latter is associated significant differences, in both PCL-S caseness
with psychological morbidity in civilian popu- and MCS-12 scores, between Gulf War veterans
lations (McCabe et al. 1996). We found, how- and members of the comparison group who had
ever, that the interaction between study group been on at least one active deployment.
and rank for MCS-12 scores, as well as the The percentages of subjects currently (past
overall inverse relationship between rank and month) with PTSD as measured by the self-
GHQ-12 and PCL-S caseness within both study administered PCL-S (7.9% for Gulf War vet-
groups, persisted after adjustment for demo- erans and 4.6% for the comparison group) were
graphic and socio-economic variables including higher than the percentages diagnosed as having
age, marital status and education. recent (12 month) PTSD (5.1 % and 1.7 %
We found strong relationships between in- respectively) using the clinician-administered
creasing numbers of psychological stressors ex- CIDI, reported by Ikin et al. (2004). Further
perienced during the Gulf War, as measured by analysis of these data indicates that the percent-
the MSE questionnaire, and decreased psycho- ages obtained for past month PTSD diagnosis
logical health. The link between perceived trau- using the CIDI (4.6% and 1.3 %) are similar to
matic or stressful exposure and subsequent those obtained for past 12-month CIDI PTSD
psychopathology needs to be interpreted diagnoses, and remain lower than the percent-
cautiously (McFarlane, 2000 b), particularly ages obtained using the PCL-S. The difference
because the health problems and possible ex- between the level of PCL-S caseness and the
posures of Gulf War veterans have been the level of CIDI PTSD diagnosis is larger than ex-
subject of high media coverage (Wessely et al. pected, both instruments being based upon the
2003). It is reasonable to assume that such DSM-IV diagnostic criteria for PTSD. Forbes
persistent reports of increased health risks would et al. (2001) found, however, that Australian
have a negative psychological impact. Further, Vietnam veterans self-rated their PTSD symp-
as PTSD symptoms increase, memories of trau- toms as slightly more severe than the clinician
matic or stressful events may be affected ratings.
(McNally, 2003). Amplification of such mem- The differences in prevalence between the self-
ories has been demonstrated in Gulf War vet- reported PCL-S and the more comprehensive
erans by Southwick et al. (1997) and King et al. CIDI PTSD diagnosis may reflect a similar
(2000). Wessely et al. (2003) demonstrated, over-reporting of symptom severity, differences
however, that changes in the recall of perceived between the two instruments in their application
exposures by Gulf War veterans were more of the DSM-IV criteria, or simply that the cut-
associated with changes in the perception of off score we employed may not be suitable for
health than with changes in the number of this population. These issues will form the basis
PTSD symptoms or the level of GHQ-12 case- of future research.
ness. Such findings question the validity of any The use of brief instruments such as the ones
retrospectively determined relationship between applied here, may be problematic. Self-report
level of perceived exposure to trauma and de- can be misleading, especially in defence ap-
gree of PTSD or other symptoms. The absence plications (Schei, 1994 ; Goyne, 2001). For
1428 D. P. McKenzie et al.

example, respondents may use questionnaires as cautiously, it is worth noting that our overall
an opportunity to express their dissatisfaction results bear out those found by Ikin et al. (2004)
with military life. In a similar vein, ques- using the CIDI. Brief instruments such as the
tionnaires may exacerbate the effects of re- ones used in our study may aid in the assessment
sponse style or frame (Anastasi & Urbina, of the psychological health of current and future
1997). As participants were aware of the study’s veterans.
purpose, they may have tended towards a re-
sponse set of how Gulf veterans are ‘ expected ’
to appear. Regrettably, there is little that can be ACKNOWLEDGEMENTS
done to eliminate such biases in large survey The Australian Gulf War Veterans’ Health
designs of this kind. Finally, a search for opti- Study was funded by the Australian Govern-
mal cut-off scores in screening tests, rather than ment – Department of Veterans’ Affairs. The
optimal combinations of symptoms, may en- study was overseen by a Scientific Advisory
courage a focus on the quantity, and not type, of Committee, headed by Professor Terry Dwyer,
symptoms present (Kraemer, 1992 ; Clarke & and by a veterans’ Consultative Forum. We
McKenzie, 1994). are grateful to members of both groups for
Possible problems not withstanding, more their contribution and support. We gratefully
research is required into optimal thresholds for acknowledge the contribution of Dr Keith
the PCL-S in different populations. Although it Horsley, Dr Warren Harrex, Mr Bob Connolly
has been suggested that the GHQ-12 may lack and his contact and recruitment team at the
face validity as a general measure of psycho- Department of Veterans’ Affairs, and the staff at
logical stress for the Australian Defence Force Health Services Australia who conducted the
(Chapman, 2001 ; Goyne, 2001), this should medical and psychological assessments. We
not preclude its use as a general measure of thank Dr Leigh Blizzard for his valuable stat-
psychological health. The use of such a measure istical advice, and thank Associate Professor
is important, as a focus on PTSD alone may David Clarke for his incisive reading of this
miss other psychological disorders. Finally, manuscript. Lastly, but most importantly, we
the use of the SF-12 allows mental, as well sincerely and wholeheartedly thank the Gulf
as physical, health status to be assessed and War veterans and comparison group members
compared. for their participation.

CONCLUSIONS DECLARATION OF INTEREST

Australian Gulf War veterans have slightly This research was funded by the Australian
poorer physical health status, but markedly Government – Department of Veterans’ Affairs.
poorer mental health status, as measured by the Professor A. C. McFarlane is Chair of the
SF-12, than members of the comparison group. Mental Health Consultative Group to the Di-
Similarly, Gulf War veterans have higher levels rector General of the Health Service Branch of
of psychological ill-health, as measured by the the Australian Defence Force. The Australian
GHQ-12 and PCL-S, than the comparison Centre for Posttraumatic Mental Health is par-
group. The results for the MCS-12 and PCL-S, tially funded by the Department of Veterans’
but not the GHQ-12, persisted when the Gulf Affairs.
War veterans were compared with only those
comparison group members who had been REFERENCES
actively deployed. This suggests that there is Adler, A. B., Vaitkus, M. A. & Martin, J. A. (1996). Combat
something unique to the Gulf War experience. exposure and posttraumatic stress symptomatology among US
Screening for psychiatric disorders, whether soldiers deployed to the Gulf War. Military Psychology 8, 1–14.
Anastasi, A. & Urbina, S. (1997). Psychological Testing (7th edn).
performed before or after stressful events, can Prentice-Hall: Upper Saddle River, NJ.
be highly problematic (Wessely, 2003) and care APA (1994). Diagnostic and Statistical Manual of Mental Disorders
must be taken not to extrapolate a need for (4th edn). American Psychiatric Association : Washington, DC.
Australian Bureau of Statistics (1998). Mental Health and Wellbeing :
widespread treatment from the current results. Profile of Adults, Australia 1997. Australian Bureau of Statistics :
While self-report data should always be treated Canberra.
 Psychological health of Australian Gulf War veterans 1429

Barrett, D. H., Doebbeling, C. C., Schwartz, D. A., Voelker, M. D., Herrman, H., Patrick, D. L., Diehr, P., Martin, M. L., Fleck, M.,
Falter, K. H., Woolson, R. F. & Doebbeling, B. N. (2002). Post- Simon, G. E. & Buesching, D. P. (2002). Longitudinal investigation
traumatic stress disorder and self-reported physical health status of depression outcomes in primary care in six countries : the LIDO
among U.S. military personnel serving during the Gulf War study. Functional status, health service use and treatment of
period. Psychosomatics 43, 195–205. people with depressive symptoms. Psychological Medicine 32,
Blanchard, E. B., Jones-Alexander, J., Buckley, T. C. & Forneris, 889–902.
C. A. (1996). Psychometric properties of the PTSD checklist Hyams, K. C. & Scott, K. (2002). Adding to our understanding of
(PCL). Behaviour Research and Therapy 34, 669–673. Gulf War health issues. Psychological Medicine 32, 1335–1337.
Chapman, S. (2001). The use of the General Health Questionnaire in Ikin, J. F., Sim, M. R., Creamer, M. C., Forbes, A. B., McKenzie,
the Australian Defence Force: a flawed but irreplaceable measure? D. P., Kelsall, H. L., Glass, D. C., McFarlane, A. C., Abramson,
In The Management of Stress in the Australian Defence Force : M. J., Ittak, P., Dwyer, T., Blizzard, L., Delaney, K. R., Horsley,
Human Factors, Families, and the Welfare of Military Personnel K. W. A., Harrex, W. K. & Schwarz, H. (2004). War-related
away from the Combat Zone (ed. G. Kearney, M. Creamer, R. psychological stressors and risk of psychological disorders in
Marshall and A. Goyne), pp. 35–45. Department of Defence : Australian veterans of the 1991 Gulf War. British Journal of Psy-
Canberra. chiatry 185, 116–126.
Clark, C. R., McFarlane, A. C., Morris, P., Weber, D. L., Sonkkilla, Iowa Persian Gulf Study Group (1997). Self-reported illness
C., Shaw, M., Marcina, J., Tochon-Danguy, H. J. & Egan, G. F. and health status among Gulf War veterans: a population-
(2003). Cerebral function in posttraumatic stress disorder during based study. Journal of the American Medical Association 277,
verbal working memory updating : a positron emission tomo- 238–245.
graphy study. Biological Psychiatry 53, 474–481. Ishoy, T., Suadicani, P., Guldager, B., Appleyard, M., Hein, H. O. &
Clarke, D. M. & McKenzie, D. P. (1994). A caution on the use of Gyntelberg, F. (1999). State of health after deployment in the
cut-points applied to screening instruments or diagnostic criteria. Persian Gulf – The Danish Gulf War Study. Danish Medical
Journal of Psychiatric Research 28, 185–188. Bulletin 46, 416–419.
Clarke, D. M., Smith, G. C. & Herrman, H. E. (1993). A comparative Ismail, K., Blatchley, N., Hotopf, M., Hull, L., Palmer, I., Unwin, C.,
study of screening instruments for mental disorders in general David, A. & Wessely, S. (2000). Occupational risk factors for
hospital patients. International Journal of Psychiatry in Medicine ill health in Gulf veterans of the United Kingdom. Journal of
23, 323–337. Epidemiology and Community Health 54, 834–838.
Cytel Software Corporation (2000). LogXact 4 for Windows [com- Joseph, S. C. & Comprehensive Clinical Evaluation Program Evalu-
puter software]. Cytel Software Corporation, Cambridge, MA. ation Team (1997). A comprehensive clinical evaluation of 20,000
Donath, S. (2001). The validity of the 12-item General Health Ques- Persian Gulf War veterans. Military Medicine 162, 149–55.
tionnaire in Australia : a comparison between three scoring Kang, H., Natelson, B., Mahan, C. M., Lee, K. Y. & Murphy, F. M.
methods. Australian and New Zealand Journal of Psychiatry 35, (2003). Post-traumatic stress disorder and Chronic Fatigue
231–235. Syndrome-like illness among Gulf War veterans : a population
Farrin, L., Hull, L., Unwin, C., Wykes, T. & David, A. (2003). Effects based survey of 30,000 veterans. American Journal of Epidemiology
of depressed mood on objective and subjective measures of atten- 157, 141–148.
tion. Journal of Neuropsychiatry and Clinical Neurosciences 15, Keane, T. M., Caddell, J. M. & Taylor, K. L. (1988). Mississippi
98–104. Scale for Combat-Related posttraumatic stress disorder : three
Forbes, D., Creamer, M. & Biddle, D. (2001). The validity of studies in reliability and validity. Journal of Consulting and Clinical
the PTSD checklist as a measure of symptomatic change in Psychology 56, 85–90.
combat-related PTSD. Behaviour Research and Therapy 39, Keane, T. M., Fairbank, J. A., Caddell, J. M., Zimering, R. T.,
977–986. Taylor, K. L. & Mora, C. A. (1989). Clinical evaluation of a
Gandek, B., Ware, J. E., Aaronson, N. K., Apolone, G., Bjorner, J. B., measure to assess combat exposure. Psychological Assessment 1,
Brazier, J., Bullinger, M., Kaasa, S., Leplege, A., Prieto, L. & 53–55.
Sullivan, M. (1998). Cross-validation of item selection and scoring King, D. W., King, L. A., Erickson, D. J., Huang, M. T., Sharkansky,
for the SF-12 Health Survey in nine countries : results from E. J. & Wolfe, J. (2000). Posttraumatic stress disorder and retro-
the IQOLA project. Journal of Clinical Epidemiology 51, spectively reported stressor exposure : a longitudinal prediction
1171–1178. model. Journal of Abnormal Psychology 109, 624–633.
Goldberg, D. P., Gater, R., Sartorius, N., Ustun, T. B., Piccinelli, M., Kraemer, H. C. (1992). Evaluating Medical Tests : Objective and
Gureje, O. & Rutter, C. (1997). The validity of two versions of the Quantitative Guidelines. Sage : Newbury Park, CA.
GHQ in the WHO study of mental illness in general health care. Mackinnon, A. (2000). A spreadsheet for the calculation of com-
Psychological Medicine 27, 191–197. prehensive statistics for the assessment of diagnostic tests and
Goldberg, D. P., Oldehinkel, T. & Ormel, J. (1998). Why GHQ inter-rater agreement. Computers in Biology and Medicine 30,
threshold varies from one place to another. Psychological Medicine 127–134.
28, 915–921. McCabe, C. J., Thomas, K. J., Brazier, J. E. & Coleman, P. (1996).
Goldberg, D. P. & Williams, P. (1988). A User’s Guide to the General Measuring the mental health status of a population : a comparison
Health Questionnaire. NFER-Nelson: Windsor. of the GHQ-12 and the SF-36 (MHI-5). British Journal of
Goss Gilroy Inc. (1998). Health Study of Canadian Forces Personnel Psychiatry 169, 516–521.
Involved in the 1991 Conflict in the Persian Gulf. Goss Gilroy Inc. : McFarlane, A. C. (2000a). Traumatic stress in the 21st century.
Ottawa. Australian and New Zealand Journal of Psychiatry 34, 896–902.
Goyne, A. (2001). The measurement of organisational stress in the McFarlane, A. C. (2000 b). Can debriefing work ? In Critical Ap-
Australian Regular Army. In The Management of Stress in the praisal of Theories of Interventions and Outcomes, with Directions
Australian Defence Force: Human Factors, Families, and the Wel- for Future Research in Psychological Debriefing, Theory, Practice
fare of Military Personnel away from the Combat Zone (ed. G. and Evidence (ed. B. Raphael and J. P. Wilson), pp. 327–336.
Kearney, M. Creamer, R. Marshall and A. Goyne), pp. 21–33. Cambridge University Press : London.
Department of Defence : Canberra. McFarlane, A. C. (2003). Military mental health in the 21st century.
Hammar, A., Lund, A. & Hugdahl, K. (2003). Long-lasting cognitive ADF Health 4, 1–2.
impairment in unipolar major depression: a 6 month follow-up McKenzie, D. P., Vida, S., Mackinnon, A. J., Onghena, P. & Clarke,
study. Psychiatry Research 118, 189–196. D. M. (1997). Accurate confidence intervals for measures of test
Hanley, J. A. & McNeil, B. J. (1982). The meaning and use of performance. Psychiatry Research 69, 207–209.
the area under a receiver operating characteristic (ROC) curve. McNally, R. J. (2003). Remembering Trauma. Harvard University
Radiology 143, 29–36. Press : Cambridge, MA.
1430 D. P. McKenzie et al.

Mehta, C. R. & Patel, N. R. (1995). Exact logistic regression : theory and psychological impact of subsequent hurricane. Journal of
and examples. Statistics in Medicine 14, 2143–2160. Psychopathology and Behavioural Assessment 24, 25–37.
Pevalin, D. J. & Goldberg, D. P. (2003). Social precursors to onset Unwin, C., Blatchley, N., Coker, W., Ferry, S., Hotopf, M., Hull, L.,
and recovery from episodes of common mental illness. Psycho- Ismail, K., Palmer, I., David, A. & Wessely, S. (1999). Health of UK
logical Medicine 33, 299–306. servicemen who served in Persian Gulf War. Lancet 353, 169–178.
Robins, L. N., Wing, J., Wittchen, H. U., Helzer, J. E., Babor, Voelker, M. D., Saag, K. G., Schwartz, D. A., Chrischilles, E.,
T. F., Burke, J., Farmer, A., Janlenski, A., Pickens, R., Regier, Clarke, W. R., Woolson, R. F. & Doebbeling, B. N. (2002). Health-
D. A., Sartorius, N. & Towle, M. S. (1988). The Composite related quality of life in Gulf War era military personnel. American
International Diagnostic Interview : an epidemiological instrument Journal of Epidemiology 155, 899–907.
suitable for use in conjunction with different diagnostic Ware, J. E., Kosinski, M. A. & Keller, S. D. (1998). SF-12 : How to
systems and in different cultures. Archives of General Psychiatry Score the SF-12 Physical and Mental Health Summary Scales (3rd
45, 1069–1077. edn). Quality Metric Incorporated : Lincoln, RI.
Sampson, M. J., Kinderman, P., Watts, S. & Sembi, S. (2003). Ware, J. E., Snow, K. K., Kosinski, M. & Gandek, B. (1993). SF-36
Psychopathology and autobiographical memory in stroke and Health Survey : Manual and Interpretation Guide. B1-B5. The
non-stroke hospitalized patients. International Journal of Geriatric Health Institute, New England: Boston, MA.
Psychiatry 18, 23–32. Weathers, F. W., Litz, B. T., Herman, D. S., Huska, J. A. & Keane,
Sanderson, K. & Andrews, G. (2002). The SF-12 in the Australian T. M. (1993). The PTSD Checklist (PCL) : reliability, validity, and
population : cross validation of item selection. Australian and New diagnostic utility. Paper presented at the 9th Annual Conference
Zealand Journal of Public Health 26, 343–345. of the International Society for Traumatic Stress Studies : San
Schei, E. (1994). A strengthening experience ? Mental distress during Antonio, TX.
military service. A study of Norwegian army conscripts. Social Wessely, S. (2003). The role of screening in the prevention of
Psychiatry and Psychiatric Epidemiology 29, 40–45. psychological disorders arising after major trauma : pros and cons.
Southwick, S. M., Morgan, C. A., Nicolaou, A. L. & Charney, D. S. In Terrorism and Disaster : Individual and Community Mental
(1997). Consistency of memory for combat-related traumatic Health Interventions (ed. R. J. Ursano, C. S. Fullerton and A. E.
events in veterans of Operation Desert Storm. American Journal of Norwood), pp. 121–145. Cambridge University Press : Cambridge.
Psychiatry 154, 173–177. Wessely, S., Unwin, C., Hotopf, M., Hull, L., Ismail, K., Nicolaou, V.
StataCorp (2001). Stata Statistical Software, Release 7.0 [computer & David, A. (2003). Stability of recall of military hazards over
software]. StataCorp : College Station, TX. time : evidence from the Persian Gulf War of 1991. British Journal
Stimpson, N. J., Thomas, H. V., Weightman, A. L., Dunstan, F. & of Psychiatry 183, 314–322.
Lewis, G. (2003). Psychiatric disorder in veterans of the Persian WHO Collaborating Centre for Mental Health and Substance Abuse
Gulf War of 1991. Systematic review. British Journal of Psychiatry (1997). Composite International Diagnostic Interview : CIDI-Auto
182, 391–403. Version 2.1 Administrator’s Guide and Reference. World Health
Sutker, P., Corrigan, S. A., Sundgaard-Riise, K., Uddo, M. & Allain, Organization Collaborating Centre for Mental Health and
A. N. (2002). Exposure to war trauma, war-related PTSD, Substance Abuse : Sydney.