International Journal of Neuropsychopharmacology (2005), 8, 293–302.

Copyright f 2004 CINP

S E R I ES
S P E C IA L
doi :10.1017/S1461145704004870

Evidence-based pharmacotherapy of
generalized anxiety disorder

David S. Baldwin1 and Claire Polkinghorn2

1
Clinical Neurosciences Division, Faculty of Medicine, Health and Life Sciences, University of Southampton, UK
2
Department of Psychiatry, Royal South Hants Hospital, West Hampshire NHS Trust, Southampton, UK

Abstract

Downloaded from https://academic.oup.com/ijnp/article-abstract/8/2/293/755456 by FMRP user on 26 April 2019

Generalized anxiety disorder (GAD) is a common and often disabling disorder. This paper reviews the
pharmacological treatment of GAD, based on the findings of published meta-analyses and randomized
placebo-controlled studies. In doing so, it aims to address three fundamental questions : What is the first-
line treatment for GAD? How long should treatment continue? What is the best intervention in patients
who do not respond to first-line and second-line treatments? Due to their efficacy in GAD and comorbid
anxiety and depressive disorders, their tolerability and safety, certain selective serotonin re-uptake inhibi-
tors (escitalopram, paroxetine, sertraline) should be considered the first-line treatment for most patients,
although the serotonin-noradrenaline re-uptake inhibitor venlafaxine is a reasonable alternative. Little
is known about the optimal length of therapy after response to acute treatment but relapse-prevention
studies with paroxetine suggest that continuation treatment should last for at least 6 months. The man-
agement of patients who do not respond to first-line treatment is uncertain, but some patients may benefit
from certain tricyclic antidepressants, buspirone, or pregabalin.
Received 24 February 2004 ; Reviewed 13 July 2004 ; Accepted 13 July 2004
Key words : Drug treatment, Generalized Anxiety Disorder.

Introduction The lifetime prevalence of GAD in the general

population is around 5–6 %, the 12-month prevalence
Generalized anxiety disorder (GAD) is a common, ty-
varying according to diagnostic criteria – from 1.5 %
pically chronic and disabling mental disorder
with DSM-IV to 3.1 % with DSM-III-R criteria
associated with substantial medical and psychiatric
(Wittchen, 2002). The age of onset of GAD differs from
comorbidity and occupational impairment. It is
that with other anxiety disorders, the majority of
characterized by inappropriate or excessive anxiety
cases presenting aged between 35–45 yr (Carter et al.,
and worrying that persists over time and is not re-
2001 ; Wittchen et al., 2000 ; Yonkers et al., 2000). It is
stricted to a particular set of circumstances. Common
probably the most common anxiety disorder among
features include apprehension, with worries about
the older population (55–85 years) (Beekman et al.,
future misfortune ; inner tension and difficulty in con-
1998). Typically, symptoms fluctuate in intensity over
centrating ; motor tension, with restlessness, tremor
time, but GAD is usually a chronic condition (Weiller
and headache ; and autonomic anxiety symptoms, with
et al., 1998). The functional impairment associated
excessive perspiration, dry mouth and epigastric dis-
with GAD is similar in severity to that seen with
comfort. The last decade has seen major increases
major depression (Kessler et al., 1999 ; Wittchen et al.,
in understanding of the epidemiology and neuro-
2000).
biology of the condition, together with development
GAD is amongst the most common mental dis-
and widespread availability of a range of treatment
orders seen in primary care. The point prevalence
approaches.
(defined according to ICD-10 criteria) in European
primary-care settings was 4.8 % for GAD without
Address for correspondence : Dr D. S. Baldwin, University
comorbid depressive or anxiety disorders, and 3.7 %
Department of Psychiatry, Royal South Hants Hospital, Graham
Road, Southampton SO14 0YG, UK.
for GAD with depression : a further 4.1 % had ‘ sub-
Tel. : 0044 2380 825533 Fax : 0044 2380 234243 threshold ’ GAD (Weiller et al., 1998). Comorbid GAD
E-mail : dsb1@soton.ac.uk is associated with more severe symptoms, greater
294 D. S. Baldwin and C. Polkinghorn

functional impairment, a more prolonged course and serotonin-noradrenaline re-uptake inhibitor (SNRI)
decreased productivity (Kessler et al., 1999 ; Weiller venlafaxine, certain benzodiazepine anxiolytics, and
et al., 1998) and higher use of health services the novel anxiolytic pregabalin. The results of these
(Greenberg et al., 1999 ; Maier and Falkai, 1999). Co- studies are reviewed below, the focus being on ad-
morbid depressive symptoms are associated with dressing three fundamental questions :
an improved chance of a patient being recognized as
$ What is the first-line treatment for GAD?
having a psychological problem, though not necess-
$ How long should treatment continue?
arily as having GAD (Weiller et al., 1998).
$ What is the best intervention after non-response
The pathophysiology of GAD is uncertain, but
to first-line and second-line treatments?
disturbances in neurotransmission of serotonin (5-
hydroxytryptamine, 5-HT), noradrenaline, gamma-
aminobutyric acid (GABA), cholecystokinin, and Search strategy

Downloaded from https://academic.oup.com/ijnp/article-abstract/8/2/293/755456 by FMRP user on 26 April 2019

corticoptropin-releasing factor may all be important.
We conducted a computerized literature search of
Serotonin is integrally involved in the mediation
electronic databases (MEDLINE, Embase and PsychInfo)
of anxiety, through serotonergic innervation of the
for the years 1980–2003 using a strategy which com-
limbic system, hypothalamus and thalamus. Levels
bined the terms (generalised/generalized anxiety
of the serotonin metabolite 5-hydroxyindoleacetic
disorder) with (randomised/randomized controlled
acid in cerebrospinal fluid in patients with GAD are
trial). In addition, we consulted with colleagues about
low ; and anxiety symptoms may be worsened by
other potential treatment studies, not identified by
administration of the 5-HT1/5-HT2 receptor agonist,
the search ; examined published systematic reviews
m-chlorophenylpiperazine. In addition, patients with
in the Cochrane Collaboration database ; and at-
GAD show a reduction in binding of the selective
tempted to identify recently completed treatment
serotonin re-uptake inhibitor (SSRI) paroxetine to
studies, currently only available as scientific conference
platelets (Connor and Davidson, 1998).
abstracts.
Disturbances of the noradrenergic system may also
be important. The a2-adrenoceptor antagonist yohim-
bine increases noradrenergic cell firing and induces
Which is the first-line treatment for GAD?
anxiety, whereas the a2-adrenoceptor agonist cloni-
dine reduces firing rates and inhibits anxiety. In GAD Summary of published clinical trials of SSRIs
patients, challenge with yohimbine is associated with and venlafaxine
a blunting of the increase in plasma 3-methoxy-
Citalopram
4-hydroxyphenylglycol, compared to that seen in
healthy controls (Charney et al., 1989). In addition, No randomized controlled trials with citalopram in
[3H]yohimbine binding to platelets is reduced, com- GAD have been published. A randomized controlled
pared to controls or patients with major depression trial comparing two dose ranges of citalopram and
(Sevy et al., 1989). imipramine in 472 primary-care depressed patients
The GABA/benzodiazepine receptor system also found that the treatments were associated with a
appears implicated in the pathophysiology of GAD. similar reduction in mean score on the Hamilton
Untreated patients have a reduced number of Rating Scale for Depression (HAMD ; Hamilton, 1967)
benzodiazepine-binding sites on platelet membranes, anxiety factor (Rosenberg et al., 1994). Another study
the number increasing after diazepam treatment comparing citalopram and paroxetine in 104 patients
(Weizman et al., 1987) ; an investigation of lympho- with either DSM-IV major depression or mixed
cyte membrane benzodiazepine-binding sites has anxiety-depressive disorder (Jefferson and Greist,
produced similar findings (Rocca et al., 1991). The 2000) found that both treatments were associated with
reduction in binding sites may explain the reduction significant reductions in total score on the Hamilton
in saccadic eye movement velocity seen in GAD, Rating Scale for Anxiety (HAMA ; Hamilton, 1959). By
as this velocity has been used as a marker of func- contrast, the results of a 24-wk double-blind, placebo-
tional integrity of the GABA/benzodiazepine system controlled treatment study comparing citalopram and
(Connor and Davidson, 1998). sertraline in 323 patients with DSM-IV major depress-
The best evidence for a role of disturbances in these ive disorder indicate that citalopram, but not sertra-
neurotransmitter systems in the pathophysiology line, was significantly more efficacious than placebo
of GAD comes from randomized placebo-controlled in reducing the mean score on the HAMD anxiety
treatment studies involving certain SSRIs, the cluster (p<0.01) (Stahl, 2000).
 Evidence-based pharmacotherapy of GAD 295

Escitalopram compared to imipramine and the benzodiazepine 2k-

chlorodesmethyl-diazepam. Paroxetine was superior
Citalopram is a racemic mixture of two enantiomers,
to 2k-chlorodesmethyl-diazepam and had similar
of which only the S-isomer (escitalopram) has signifi-
efficacy to imipramine : in addition, paroxetine treat-
cant serotonin reuptake inhibitory properties (Hyttel
ment differed significantly (p<0.05) from 2k-chloro-
et al., 1992). Escitalopram is more selective and more
desmethyl-diazepam from week 4 onwards, while
potent than citalopram, and has been found to be
imipramine only did so at the end of the study (Rocca
significantly more efficacious than citalopram on
et al., 1997).
some measures in a pooled analysis of randomized
The second investigation was an 8-wk, fixed-dose
controlled trials in patients with major depressive
study involving 566 patients, performed in the USA
disorder (Gorman et al., 2002). The results of three
(Rickels et al., 2003). Paroxetine treatment (20 or
8-wk randomized, double-blind placebo-controlled,
40 mg/d) was significantly superior to placebo (p<
parallel-group trials in patients with DSM-IV GAD all

Downloaded from https://academic.oup.com/ijnp/article-abstract/8/2/293/755456 by FMRP user on 26 April 2019

0.001) in reducing both the mean HAMA total score,
indicate that escitalopram is significantly more effi-
and the mean scores on HAMA items 1 (anxious
cacious than placebo, in reducing anxiety symptoms
mood) and 2 (tension), considered by some to reflect
as measured by the HAMA (Davidson et al., 2002).
the most important symptoms of GAD. There was
Further investigations of the efficacy of fixed doses of
no dose–response relationship in the mean change in
escitalopram and paroxetine, and of escitalopram in
HAMA scores, but overall response rates were 68 %
the prevention of relapse are currently underway.
and 81 % with paroxetine 20 and 40 mg/d respect-
ively, compared with 52 % of patients in the placebo
Fluoxetine
group (p<0.001). By the end of the study, the mean
There are no published controlled treatment studies change from baseline on a health-related quality-of-
in adults with DSM-IV GAD. An open pilot treatment life questionnaire (EuroQoL-5D) and visual analogue
study in 16 children and adolescents (aged between scale was significantly greater for both paroxetine-
9–18 yr) with mixed anxiety disorders showed that treatment groups, indicating a significant improve-
fluoxetine was of only limited benefit (Fairbanks et al., ment in quality of life.
1997). Double-blind treatment studies in depressed The third randomized controlled trial was an 8-wk,
patients have indicated that fluoxetine is as efficacious flexible-dose study conducted in 326 US patients
as imipramine, clomipramine or amitriptyline in re- with GAD. Paroxetine (20–50 mg/d) was significantly
lieving anxiety symptoms in depression, but the superior to placebo (p<0.05) in reducing both the
efficacy of fluoxetine in patients with comorbid de- mean HAMA total score, and the mean scores on
pression and GAD is not proven (Hurst and Lamb, HAMA items 1 and 2, and was generally well toler-
2000). ated (Pollack et al., 2001). A fourth study of similar
design conducted in 372 patients in Europe has re-
Fluvoxamine vealed similar reductions in HAMA total score and
HAMA items 1 and 2.
The efficacy of fluvoxamine as a treatment for GAD is
The effects of paroxetine treatment appear to extend
not established. A small (n=30) open study in patients
beyond simple symptom reduction or improved qual-
with comorbid major depression and GAD showed
ity of life. A small (n=29) uncontrolled study showed
that fluvoxamine treatment was associated with sig-
that paroxetine treatment was associated with a re-
nificant improvement in both anxiety and depressive
duction in maladaptive personality traits, with sig-
symptoms (Sonawalla et al., 1999), but this result
nificant decreases in harm avoidance (p=0.0001) and
needs to be replicated in patients with GAD before
novelty seeking (p=0.006), and a significant increase
the efficacy of fluvoxamine in that disorder can be
in self-directedness (p=0.0004) (Allgulander et al.,
assumed.
1998). The placebo-controlled paroxetine-treatment
studies also show that as symptoms of GAD resolve
Paroxetine
there is an associated improvement in symptom-
The efficacy of paroxetine in the short-term treatment related disability, assessed using the patient-rated
of patients with GAD has been evaluated in four Sheehan Disability Scale (SDS ; Sheehan et al., 1996)
randomized, double-blind, placebo-controlled studies which covers symptom-related impairment in social,
(Baldwin, 2001). The first evaluation was an 8-wk work and family life. At end-point in all three studies,
comparator-controlled trial involving 81 patients there was a statistically significant difference be-
with a DSM-IV diagnosis of GAD, paroxetine being tween paroxetine and placebo in the SDS total score
296 D. S. Baldwin and C. Polkinghorn

(flexible-dose study 1 difference=x1.8, p=0.037 ; ment response, venlafaxine was superior to placebo
study 2 difference=x2.4, p=0.001 ; fixed-dose study in both younger (p<0.001) and older (p<0.01) sub-
for 20 mg/d regimen, difference=x3.1, p<0.001 ; groups of patients (Katz et al., 2002).
for 40 mg/d regimen, difference=x3.6, p<0.001) In the first comparator-controlled study, which
(Baldwin et al., 2002). included 564 patients, there was no significant
advantage for either active treatment (75 mg/d venla-
Sertraline faxine XL, or 15 mg/d diazepam) over placebo in
the intention-to-treat last observation carried forward
Double-blind treatment studies indicate that sertraline
analysis. However, in a second analysis, which
is efficacious in relieving anxiety symptoms within
omitted those study centres that had been unable to
depression and the symptoms and impairment
distinguish diazepam from placebo (designated the
associated with panic disorder, social phobia, post-
‘ verum-sensitive population ’), there were significant

Downloaded from https://academic.oup.com/ijnp/article-abstract/8/2/293/755456 by FMRP user on 26 April 2019

traumatic disorder and obsessive–compulsive dis-
advantages for both venlafaxine and diazepam over
order. The efficacy of sertraline in adult patients with
placebo, in the reduction of HAMA total score and
GAD has been examined only recently, although a
other efficacy measures (Hackett et al., 2000).
small (n=22) randomized, placebo-controlled 12-wk
In the second comparator-controlled study, with
trial in children and adolescents aged 5–16 yr found
405 participating patients, there were numerical
significant advantages for sertraline over placebo on
advantages for venlafaxine XL (75 or 150 mg/d) over
both the HAMA and CGI [Clinical Global Impression
both placebo and buspirone (30 mg/d) across a range
(Guy, 1976)] scales (Rynn et al., 2001).
of primary efficacy variables, but none of these
A preliminary report of a randomized, placebo-
reached statistical significance. There were significant
controlled parallel-group, flexible-dose (50–150 mg/d,
advantages (p<0.05) for venlafaxine over placebo on
mean dose 95.1 mg) 12-wk study in 370 patients
both the HAMA psychic anxiety item scores at study
with DSM-IV GAD found that sertraline was signifi-
end-point, and on the HAMA anxious mood item
cantly more efficacious than placebo across a variety
at weeks 2, 4, 6 and 8 (p<0.05). Venlafaxine was associ-
of measures, including the HAMA and CGI (Morris
ated with significantly greater overall improvement,
et al., 2003). The reduction in mean HAMA score be-
compared to buspirone, on the CGI scores at week 3
tween baseline and end-point was 11.7 with sertraline,
[CGI-S (Severity), week 4 (CGI-I (Improvement) and
compared to 8.0 with placebo ; furthermore, 37 % of
CGI-S)] and week 8 (CGI-S) (Davidson et al., 1999).
sertraline-treated patients entered symptomatic re-
The dose–response relationship with venlafaxine
mission (defined as a HAMA score of 7 or less),
was examined in two of the randomized, placebo-
compared to 23.0 % with placebo (Sjodin et al., 2003).
controlled treatment studies. In the first, lasting 8 wk,
daily doses of 75, 150 and 225 mg all showed superior
Venlafaxine
efficacy to placebo (p<0.05), the most positive efficacy
A preliminary study (Feighner et al., 1998) in de- results being seen with the highest dosage (Rickels
pressed outpatients indicated that once-daily treat- et al., 2000). In the second, lasting 24 wk, daily doses
ment with the SNRI venlafaxine was efficacious in of 75 and 150 mg showed significantly greater im-
relieving anxiety symptoms, and suggested it might, provement than patients allocated to placebo on all
therefore, have a role in the management of patients outcome measures (p<0.01) whereas a daily dose of
with GAD. The evidence for the efficacy of venlafaxine 37.5 mg was only superior on one measure at some
in the short-term and long-term treatment of GAD time-points. Furthermore, there were significant ad-
is now based upon the results of five randomized vantages for the two higher doses over the lowest dose
placebo-controlled trials, two of which involved an on some outcome measures (Allgulander et al., 2001).
active comparator (diazepam or buspirone).
Pooled analysis of these five placebo-controlled Summary of published clinical trials of other
trials, which includes 1839 patients with a DSM-IV psychotropic compounds
diagnosis of GAD, provides good evidence for venla-
Benzodiazepines
faxine in short-term treatment. In these studies, the
effect size for venlafaxine (i.e. difference from placebo A systematic review of the findings of randomized
in mean HAMA score at study end-point) ranges controlled trials has established that benzodiazepines
between 1.6 and 4.2, with a mean effect size from the are an effective and rapid treatment for many patients
pooled data of 2.78 (Gorman, 2002). In a further with GAD, having similar efficacy to cognitive therapy
analysis, which examined the effects of age on treat- (Gould et al., 1997). However, the benzodiazepines
 Evidence-based pharmacotherapy of GAD 297

are far from ideal in the treatment of GAD, having such as imipramine have a rather poor tolerability
limited efficacy against comorbid depressive symp- profile due to blockade of histamine H1 receptors,
toms. The unwanted effects of benzodiazepines in- a1-adrenoceptors and muscarinic receptors, which
clude sedation, memory disruption and psychomotor limits their long-term use in treating patients with
impairment, with an associated increased risk of traffic GAD.
accidents. Other problems include the development
of tolerance, abuse or dependence, and distressing 5-HT1A agonists
withdrawal symptoms on stopping the drug. Many
Buspirone is an azapirone anxiolytic drug, with partial
authorities counsel that benzodiazepines should be
agonist properties at 5-HT1A receptors, which has
reserved for short-term use (up to 4 wk), and pre-
proven efficacy in the treatment of patients with GAD
scribed only at low dosage (Lader, 1999). Others have
(Goa and Ward, 1986). An early study (Goldberg and
argued that benzodiazepines are clearly efficacious,
Finnerty, 1979) established that buspirone had com-

Downloaded from https://academic.oup.com/ijnp/article-abstract/8/2/293/755456 by FMRP user on 26 April 2019

and that withholding treatment from patients on the
parable efficacy to diazepam in patients with general-
basis of a potential risk of dependence is unjustified
ized anxiety. Not all studies with buspirone have been
and probably detrimental to overall well-being
positive (Ansseau et al., 1990) but a meta-analysis of
(Argyropoulos and Nutt, 1999).
eight controlled treatment studies has indicated that
buspirone has comparable efficacy to benzodiazepines
Pregabalin
in the management of GAD (Gammans et al., 1992).
Pregabalin is a novel psychotropic drug with anti- Buspirone appears efficacious in reducing associated
convulsant, anxiolytic and analgesic properties. The depressive symptoms in patients with GAD, but it
mechanism of action is largely unknown, although is not an accepted treatment for patients with major
it binds to an auxiliary subunit (a2d. of voltage-gated depression, and therefore, is not a suitable first-line
calcium channels, thereby increasing whole brain treatment in patients with comorbid GAD and de-
GABA. The results of a recently published placebo- pression (Sramek et al., 1996).
controlled study indicate that pregabalin was superior Flesinoxan is a related drug, which acts as a full
to placebo in reducing mean HAMA score from the agonist at somatodendritic 5-HT1A receptors : it too
first week of double-blind treatment (Feltner et al., has been found efficacious in the treatment of GAD.
2003). In three randomized, double-blind, placebo- A five-arm study comparing three doses of flesinoxan,
controlled treatment studies (two involving an active alprazolam and placebo found that both the highest
comparator) pregabalin was significantly more effi- dose of flesinoxan and alprazolam were significantly
cacious in relieving symptoms than placebo and had more efficacious than placebo in reducing anxiety
similar overall efficacy to either alprazolam (Pande symptoms, rated by the HAMA (Bradford and Stevens,
et al., 2000) or venlafaxine (Rickels et al., 2002). 1994, but there is little additional information on the
Further studies with pregabalin are complete, and efficacy of the drug, and its further development has
demonstrate it has anxiolytic effects within 1 wk been halted.
(Montgomery, 2003).
Hydroxyzine
Imipramine
The efficacy of the antihistamine hydroxyzine in
The tricyclic antidepressant (TCA) imipramine has acute treatment of patients with GAD has been exam-
proven efficacy in the treatment of patients with GAD, ined in three randomized placebo-controlled trials.
defined according to DSM-III criteria. In a seminal A preliminary French general practice study (using
8-wk double-blind, placebo-controlled study involv- observed case analysis only) found some evidence for
ing 230 patients with symptoms lasting least 4 months, efficacy of hydroxyzine (Darcis et al., 1995), this being
a baseline HAMA score of 18 or more, and no indi- supported by the findings of a UK primary-care study
cation of current depression or panic disorder, subjects involving 244 patients with DSM-IV GAD, with or
were randomized to treatment with imipramine, without comorbid depressive symptoms, which also
diazepam, trazodone or placebo. During the first 2 wk included buspirone, although only hydroxyzine was
of treatment, diazepam was associated with most superior to placebo (p<0.02) on the primary outcome
improvement in anxiety symptoms, but from the measure (HAMA) (Lader and Scotto, 1998). In the third
third week to the study end-point imipramine had study, hydroxyzine had similar efficacy to the benzo-
significantly greater anxiolytic efficacy, compared diazepine bromazepam, and superior efficacy to
to diazepam (Rickels et al., 1993). However, TCAs placebo, with numerically fewer patients experiencing
298 D. S. Baldwin and C. Polkinghorn

discontinuation symptoms with hydroxyzine (40.2 %) How long should treatment continue?
than with placebo or bromazepam (each 51 %) (Llorca
et al., 2002). There have been few randomized controlled trials
of the treatment of patients with GAD beyond early
Propranolol response to acute treatment, although the findings of
studies with paroxetine, venlafaxine and escitalopram
A double-blind, randomized placebo- and comparator-
all suggest that treatment should probably continue
controlled, 3-wk dose-ranging, parallel-group study
for at least a further 6 months. Two different ap-
has examined the efficacy of the beta-blocker propra-
proaches to assessing maintenance of effect have been
nolol (80, 160 or 320 mg/d) and the benzodiazepine
utilized : either a relapse-prevention design, in which
chlordiazepoxide (30 or 45 mg/d) in GAD. Both pro-
responders to open-label acute treatment or randomly
pranolol and the active comparator showed signifi-
allocated to either continue with active treatment, or
cantly greater efficacy than placebo after 1 wk of

Downloaded from https://academic.oup.com/ijnp/article-abstract/8/2/293/755456 by FMRP user on 26 April 2019

to switch to placebo ; or a continuation treatment
double-blind treatment, but there were no advantages
design, in which responders to double-blind acute
at study end-point (Meibach et al., 1987).
treatment continue with double-treatment for a fur-
ther 6 months.
Trifluoperazine
In a double-blind relapse-prevention study, parox-
The results of a randomized placebo-controlled, etine was found efficacious in long-term treatment
flexible-dose acute treatment study indicate that the of patients with GAD, there being significantly (p<
antipsychotic drug trifluoperazine had superior effi- 0.001) fewer relapses with paroxetine (10.9 %) than
cacy from the first week of double-blind treatment, with placebo (33.9 %) (Stocchi et al., 2003). By contrast,
although there were markedly more treatment- the placebo-controlled relapse-prevention study with
emergent adverse events with trifluoperazine (62 %, venlafaxine, over a period of 4 months, did not find
compared to 46 % with placebo) (Mendels et al., 1986). greater efficacy than placebo in preventing relapse.
Taken together, the findings of the randomized However, the long-term efficacy of venlafaxine XR
placebo-controlled trials with escitalopram, parox- capsules has been shown through the results of a
etine, sertraline and venlafaxine indicate that SSRI pooled analysis of the results of two 6-month ran-
or SNRI treatment can be efficacious in the acute domized, double-blind, placebo-controlled parallel-
management of patients with GAD. There is also some group studies. With venlafaxine, 61 % of the patients
evidence for the efficacy of certain benzodiazepines, who had responded but not remitted by week 8
buspirone, imipramine, hydroxyzine and trifluoper- showed remission by 6 months, whereas only 39 %
azine. Systematic reviews support these observations : of non-remitting placebo-responders at 8 wk were
a recent but already outdated Cochrane Collabora- in symptomatic remission at study end-point
tion review concluded that imipramine, paroxetine (Montgomery et al., 2002a). The advantage for con-
and venlafaxine all had superior efficacy to placebo tinuing with venlafaxine was also seen in a survival
(Kapczinski et al., 2003). Two comprehensive but analysis of patients stopping treatment due to lack
similarly outdated reviews both inferred that the of efficacy, in which patients allocated to placebo
SSRI paroxetine offered advantages over comparator stopped double-blind treatment at a fairly constant
benzodiazepines, such as its efficacy in comorbid de- rate from the first month of treatment, whereas few
pression, and now-disputed low risk of causing venlafaxine-treated patients discontinued after the
discontinuation symptoms (Davidson et al., 2001 ; second month. After 6 months of double-blind treat-
Wagstaff et al., 2002). The most recent consensus ment, discontinuation rates were 10 % for venlafaxine
statement and guidelines on treating anxiety dis- and 21 % for placebo (Montgomery et al., 2002b).
orders states that SSRIs are the preferred first-line A preliminary report of a relatively small (n=123)
drug treatment in patients with GAD (Bandelow et al., randomized, 24-wk controlled flexible-dose compari-
2002). There are of course many further research son of escitalopram (10–20 mg/d) and paroxetine
needs, such as establishing the comparative efficacy (20–50 mg/d) in GAD shows the two SSRIs had simi-
and acceptability of differing drugs in both short- and lar overall efficacy, at 8 and 24 wk of double-blind
long-term treatment ; examining the effects of com- treatment. The proportion of patients who responded
bining drug treatment with psychological approaches to treatment increased in both groups over time
such as cognitive–behavioural therapy (CBT) ; and (escitalopram, from 52 % to 70 % ; paroxetine, from
evaluating the effectiveness of SSRI or SNRI treatment 46 % to 61 %), as did the proportion of patients entering
in patients with comorbid or resistant GAD. symptomatic remission (escitalopram, from 30 % to
 Evidence-based pharmacotherapy of GAD 299

55 % ; paroxetine, from 28 % to 46 %) (data on file, treatment with drugs of unproven efficacy. Whilst
Lundbeck Ltd). some TCAs and benzodiazepines have been found
efficacious in patients with GAD, tolerability problems
and other risks limit their use in clinical practice. By
What is the best treatment in resistant patients?
contrast, buspirone, the SSRIs escitalopram, parox-
Our literature search did not identify any placebo- etine and sertraline, the SNRI venlafaxine, and the
controlled studies in patients who have not responded novel anxiolytic pregabalin all have established effi-
to first-line or second-line treatments. As such, the cacy in placebo-controlled trials. At present, SSRIs
management of patients with resistant GAD is based are probably the first-line treatment, with venlafaxine
largely on experience and anecdotal case reports. being used in those patients who do not respond.
Approaches which could be employed include all
those compounds described above for which there is
Acknowledgements

Downloaded from https://academic.oup.com/ijnp/article-abstract/8/2/293/755456 by FMRP user on 26 April 2019

evidence of efficacy from placebo-controlled studies,
and various forms of psychological treatment, both None.
alone and in combination with psychotropic drugs.
A systematic review of 35 randomized controlled
Statement of Interest
trials has found that CBT (using a combination of ap-
proaches including exposure, cognitive restructuring, The University of Southampton has received research
anxiety management and relaxation) is more effi- support from a number of pharmaceutical companies
cacious than anxiety management alone, non-directive that manufacture compounds that have proven or
therapy or staying on a waiting list (Gould et al., 1997). potential efficacy in the treatment of Generalized
In one long-term follow-up study, cognitive therapy Anxiety Disorder. Dr Baldwin has acted as an advisor
appeared associated with better long-term outcomes to some of these pharmaceutical companies and has
than either placebo or drug treatment ; in another, received personal honoraria for attending advisory
there was no difference in outcome between cognitive boards and speaking at company-sponsored satellite
therapy, analytic psychotherapy and anxiety manage- symposia at scientific meetings. Dr Baldwin is co-
ment (Durham et al., 1999). chairman of the British Association for Psycho-
It is unclear whether it is more helpful to combine pharmacology Working Group that is charged with
pharmacological and psychological approaches in producing a consensus statement on the treatment
the management of patients with GAD, compared to of anxiety disorders.
using single approaches alone (Lader and Bond, 1998).
Concomitant treatment with benzodiazepines was
References
associated with markedly reduced response rates,
compared to cognitive or behaviour therapy alone, in Allgulander C, Cloninger CR, Przybeck TR, Brandt L (1998).
one small study (Durham and Turvey, 1987). However, Changes on the temperament and character inventory
in another study performed in primary care, the com- after paroxetine treatment in volunteers with generalized
anxiety disorder. Psychopharmacology Bulletin 34, 165–166.
bination of diazepam and CBT was more efficacious
Allgulander C, Hackett D, Salinas E (2001). Venlafaxine
than either given alone although the study was unable
extended release (ER) in the treatment of generalised
to detect a difference between diazepam and placebo anxiety disorder : twenty-four-week placebo-controlled
(Power et al., 1990). dose-ranging study. British Journal of Psychiatry 179, 15–22.
Ansseau M, Papart P, Gerard MAA, von Frenckell R,
Franck G (1990). Controlled comparison of buspirone
Conclusions
and oxazepam in generalised anxiety. Neuropsychobiology
GAD is a common and disabling anxiety disorder : 24, 74–78.
epidemiological indicates a lifetime prevalence of Argyropoulos SV, Nutt DJ (1999). The use of
approx. 5 %, and substantial associated social and benzodiazepines in anxiety and other disorders. European
Neuropsychopharmacology 9 (Suppl. 9), S407–S412.
occupational impairment, comparable to that with
Baldwin DS (2001). SSRIs in the treatment of generalised
major depression. GAD has considerable comorbidity,
anxiety disorder. In : Montgomery SA, den Boer JA (Eds.),
with depression, other anxiety disorders and physical SSRIs in Depression and Anxiety, (pp. 193–209). Chichester :
illness. Many patients with GAD are not recognized Wiley.
as suffering from a potentially treatable anxiety dis- Baldwin DS, McCafferty J, Bellew K, Bryson H, Adams A,
order : others are recognized as having some form of Hewett K, Iyengar M (2002). Improving the disability and
mental disorder, but are either not treated or receive impairment associated with generalised anxiety disorder
300 D. S. Baldwin and C. Polkinghorn

with paroxetine treatment. International Journal of Open fluoxetine treatment of mixed anxiety disorders in
Neuropsychopharmacology 5 (Suppl. 1), S213. children and adolescents. Journal of Child and Adolescent
Bandelow B, Zohar J, Hollander E, Kasper S, Moller HJ, Psychopharmacology 7, 17–29.
WFSBP Task Force on Treatment Guidelines for Anxiety, Feighner JP, Entsuah AR, McPherson MK (1998). Efficacy
Obsessive-Compulsive and Posttraumatic Stress of once-daily venlafaxine extended release (XR) for
Disorders (2002). World Federation of Societies of symptoms of anxiety in depressed outpatients. Journal
Biological Psychiatry (WFSBP) guidelines for the of Affective Disorders 47, 55–62.
pharmacological treatment of anxiety, obsessive Feltner DE, Crockatt JG, Dubovsky SJ, Cohn CK,
compulsive and posttraumatic stress disorders. World Shrivastava RK, Targum SD, Liu-Dumaw M, Carter CM,
Journal of Biological Psychiatry 3, 171–199. Pande AC (2003). A randomized, double-blind, placebo-
Beekman AT, Bremmer MA, Deeg DJ, van Balkom AJ, controlled, fixed-dose, multicenter study of pregabalin in
Smit JH, de Beurs E, van Dyck R, van Tilburg W (1998). patients with generalized anxiety disorder. Journal of
Anxiety disorders in later life : a report from the Clinical Psychopharmacology 23, 240–249.

Downloaded from https://academic.oup.com/ijnp/article-abstract/8/2/293/755456 by FMRP user on 26 April 2019

Longitudinal Aging Study Amsterdam. International Gammans RE, Stringfellow JC, Hvidzos AJ, Seidehamel RJ,
Journal of Geriatric Psychiatry 13, 717–726. Cohn JB, Wilcox CS, Fabre LF, Pecknold JC, Smith WT,
Bradford LD, Stevens G (1994). Double-lind, placebo- Rickels K (1992). Use of buspirone in patients with
controlled fixed dose study of flesinoxan in generalized generalised anxiety disorder and coexisting depressive
anxiety disorder. American College of Neuro- symptoms : a meta-analysis of eight randomised controlled
psychopharmacology 167. trials. Neuropsychobiology 25, 193–201.
Carter RM, Wittchen H-U, Pfisster H, Kessler RC (2001). Goa KL, Ward A (1986). Buspirone : a preliminary review
One-year prevalence of subthreshold and threshold of its pharmacological properties and therapeutic efficacy
DSM-IV generalised anxiety disorder in a nationally as an anxiolytic. Drugs 32, 114–129.
representative sample. Depression and Anxiety 13, 78–88. Goldberg HL, Finnerty RJ (1979). The comparative efficacy
Charney DS, Woods SW, Heninger GR (1989). of buspirone and diazepam in the treatment of anxiety.
Noradrenergic function in generalized anxiety disorder : American Journal of Psychiatry 136, 1184–1187.
effects of yohimbine in healthy subjects and patients with Gorman J (2002). Treatment of generalised anxiety disorder.
generalized anxiety disorder. Psychiatry Research 27, Journal of Clinical Psychiatry 63 (Suppl.), 17–23.
173–182. Gorman JM, Korotzer A, Su G (2002). Efficacy comparison
Connor KM, Davidson JRT (1998). Generalized anxiety of escitalopram and citalopram in the treatment of major
disorder : neurobiological and pharmacotherapeutic depressive disorder : pooled analysis of placebo-controlled
perspectives. Biological Psychiatry 44, 1286–1294. trials. CNS Spectrums 7 (Suppl. 1), S40–S44.
Darcis T, Ferreri M, Natens J, Burtin B, Deram P and Gould RA, Otto MW, Pollack MH, Yap L (1997). Cognitive
the French GP Study Group for Hydroxyzine (1995). behavioural and pharmacological treatment of generalised
A multicentre double-blind placebo-controlled study anxiety disorder : a preliminary meta-analysis. Behavior
investigating the anxiolytic efficacy of hydroxyzine in Therapy 28, 285–305.
patients with generalized anxiety disorder. Human Greenberg PE, Sisitsky T, Kessler RC, Finkelstein SN,
Psychopharmacology 10, 181–187. Berndt ER, Davidson JR, Ballenger JC, Fyer AJ (1999).
Davidson J, Bose A, Su G (2002). Escitalopram in the The economic burden of anxiety disorders in the 1990s.
treatment of generalized anxiety disorder. International Journal of Clinical Psychiatry 60, 427–435.
Journal of Neuropsychopharmacology 5 (Suppl. 1), S214. Guy W (1976). CDEU Assessment Manual for Psychopharma-
Davidson JRT, Ballenger JC, Lecrubier Y, Nutt DJ (2001). cology. US Department of Health, Education and Welfare.
Pharmacotherapy of generalized anxiety disorder. DHEW Publication no. ADM 218–222. Kensington, MD :
Journal of Clinical Psychiatry 62 (Suppl. 11), 46–52. The George Washington University.
Davidson JR, DuPont RL, Hedges D, Haskins JT (1999). Hackett D, Haudiquet V, Salinas E (2000). Controlling
Efficacy, safety, and tolerability of venlafaxine extended for the placebo response rate : a methodological approach
release and buspirone in outpatients with generalised to data from double-blind studies with a high placebo
anxiety disorder. Journal of Clinical Psychiatry 60, 528–535. response. Presented at British Association for
Durham RC, Fisher PL, Trevling LR (1999). One year Psychopharmacology, Harrogate.
follow-up of cognitive therapy, analytic psychotherapy Hamilton M (1959). The assessment of anxiety states by
and anxiety management training for generalised anxiety rating. British Journal of Medical Psychology 32, 50–55.
disorder : symptom change, medication usage and Hamilton M (1967). Development of a rating scale for
attitudes to treatment. Behavioural and Consulting primary depressive illness. British Journal of Social and
Psychotherapy 27, 19–35. Clinical Psychology 6, 278–296.
Durham RC, Turvey AA (1987). Cognitive therapy versus Hurst M, Lamb HM (2000). Fluoxetine : a review of its use
behaviour therapy in the treatment of chronic generalised in anxiety disorders and mixed anxiety and depression.
anxiety. Behavior Research and Therapy 25, 229–234. CNS Drugs 14, 51–80.
Fairbanks JM, Pine DS, Tancer NK, Dummit 3rd ES, Hyttel J, Bsego KP, Perregard JA, Sanchez C (1992). The
Kentgen LM, Martin J, Asche BK, Klein RG (1997). pharmacological effect of citalopram resides in the
 Evidence-based pharmacotherapy of GAD 301

(S)-(+)-enantiomer. Journal of Neural Transmission General Montgomery SA, Sheehan DV, Meoni P, Haudiquet V,
Section 88, 157–160. Hackett D (2002b). Characterization of the longitudinal
Jefferson J, Greist J (2000). A double-blind comparison of course of improvement in generalized anxiety disorder
citalopram and paroxetine in the treatment of patients during long-term treatment with venlafaxine XR.
with depression and anxiety. Presented at the Journal of Psychiatric Research 36, 209–217.
American College of Neuropsychopharmacology, Morris PLP, Dahl AA, Kutcher SP, Sogaard JA,
December. Allgfulander C, Burt T (2003). Efficacy of sertraline for
Kapczinski F, Schmitt R, Lima MS (2003). The use of the acute treatment of generalized anxiety disorder
antidepressants for Generalized Anxiety Disorder. (GAD). European Neuropsychopharmacology 13 (Suppl. 4),
Cochrane Database of Systematic Reviews. S375.
Katz IR, Reynolds C, Alexopoulos GS, Hackett D (2002). Pande AC, Crockatt JG, Janney C, et al. (2000). Pregabalin,
Venlafaxine ER as treatment for generalized anxiety a novel agent, in the treatment of generalized anxiety
disorder in older adults : pooled analysis of five disorder. Presented at American Psychiatric Association,

Downloaded from https://academic.oup.com/ijnp/article-abstract/8/2/293/755456 by FMRP user on 26 April 2019

randomized placebo-controlled clinical trials. Journal San Diego, May.
of the American Geriatric Society 50, 18–25. Pollack MH, Zaninelli R, Goddard A, McCafferty JP,
Kessler RC, DuPont RL, Berglund P, Wittchen H-U (1999). Bellew KM, Burnham DB, Iyengar MK (2001). Paroxetine
Impairment in pure and comorbid generalized anxiety in the treatment of generalized anxiety disorder : results
disorder and major depression at 12 months in two of a placebo-controlled, flexible-dosage trial. Journal of
national surveys. American Journal of Psychiatry 156, Clinical Psychiatry 62, 350–357.
1915–1923. Power KG, Simpson RJ, Swanson V, Wallace LA (1990).
Lader M, Scotto JC (1998). A multicentre double-blind Controlled comparison of pharmacological and
comparison of hydroxyzine, buspirone and placebo in psychological treatment of generalized anxiety disorder
patients with generalized anxiety disorder. Psycho- in primary care. British Journal of General Practice 40,
pharmacology 139, 402–406. 289–294.
Lader MH (1999). Limitations on the use of benzodiazepines Rickels K, Downing R, Schweizer E, Hassman H (1993).
in anxiety and insomnia : are they justified? European Antidepressants for the treatment of generalised anxiety
Neuropsychopharmacology 9 (Suppl. 9), S399–S405. disorder. A placebo-controlled comparison of imipramine,
Lader MH, Bond AJ (1998). Interaction of pharmacological trazodone and diazepam. Archives of General Psychiatry
and psychological treatments of anxiety. British Journal of 50, 884–898.
Psychiatry 173 (Suppl. 34), 42–48. Rickels K, Pollack MH, Lydiard RB, Bielski DE, Feltner AC,
Llorca PM, Sapdone C, Sol O, Danniau A, Bougerol T, Pande AC, Kavoussi RJ (2002). Comparison of the efficacy
Corruble E, Faruch M, Macher JP, Sermet E, Servant D and safety of pregabalin and alprazolam in generalised
(2002). Efficacy and safety of hydroxyzine in the anxiety disorder. International Journal of Neuro-
treatment of generalized anxiety disorder : a three-month psychopharmacology 5 (Suppl. 1), S213.
double-blind study. Journal of Clinical Psychiatry 63, Rickels K, Pollack MH, Sheehan DV, Haskins JT (2000).
1020–1027. Efficacy of extended-release venlafaxine in non-depressed
Maier W, Falkai P (1999). The epidemiology of comorbidity outpatients with generalized anxiety disorder. American
between depression, anxiety disorders and somatic Journal of Psychiatry 157, 968–974.
diseases. International Clinical Psychopharmacology 14 Rickels K, Zaninelli R, McCafferty JP, Bellew K, Iyengar M,
(Suppl. 2), 51–56. Sheehan D (2003). Paroxetine treatment of generalized
Meibach RC, Dunner D, Wilson LG, Ishiki D, Dager SR anxiety disorder : a double-blind, placebo-controlled
(1987). Comparative efficacy of propranolol, study. American Journal of Psychiatry 160, 749–756.
chlordiazepoxide, and placebo in the treatment of Rocca P, Ferrero P, Gualerzi A, Zanalda E, Maina G,
anxiety : a double-blind trial. Journal of Clinical Psychiatry Bergamasco B, Ravizza L (1991). Peripheral-type
84, 355–358. benzodiazepine receptors in anxiety disorders. Acta
Mendels J, Krajewski TF, Huffer V, Taylor RJ, Secunda S, Psychiatrica Scandinavica 84, 537–544.
Schless A, Sebastian JA, Semchyshyn G, Durr MJ, Rocca P, Fonzo V, Scotta M, Zanalda E, Ravizza L (1997).
Melmed AS (1986). Effective short-term treatment of Paroxetine efficacy in the treatment of generalized anxiety
generalized anxiety with trifluoperazine. Journal of Clinical disorder. Acta Psychiatrica Scandinavica 95, 444–450.
Psychiatry 47, 170–174. Rosenberg C, Damsbo N, Fuglum E, Jacobsen LV, Horsgard
Montgomery SA (2003). Pregabalin in generalized anxiety S (1994). Citalopram and imipramine in the treatment of
disorder : speed of onset. European Neuro- depressive patients in general practice. A Nordic
psychopharmacology 13 (Suppl. 4), S376. multicentre clinical study. International Clinical
Montgomery SA, Mahe V, Haudiquet V, Hackett D (2002a). Psychopharmacology 9 (Suppl. 1), 41–48.
Effectiveness of venlafaxine, extended release formulation, Rynn MA, Siqueland L, Rickels K (2001). Placebo-controlled
in short-term and long-term treatment of generalized trial of sertraline in the treatment of children with
anxiety disorder : results of a survival analysis. Journal generalised anxiety disorder. American Journal of Psychiatry
of Clinical Psychopharmacology 22, 561–567. 158, 2008–2014.
302 D. S. Baldwin and C. Polkinghorn

Sevy S, Papadimitriou GN, Surmont DW, Goldman S, Disorder Study Team (2003). Efficacy and tolerability of
Mendlewicz J (1989). Noradrenergic function in paroxetine for the long-term treatment of generalised
generalized anxiety disorder, major depressive anxiety disorder. Journal of Clinical Psychiatry 64, 250–258.
disorder, and healthy subjects. Biological Psychiatry 25, Wagstaff AJ, Cheer SM, Matheson AJ, Ormrod D, Goa KL
141–152. (2002). Paroxetine – an update of its use in psychiatric
Sheehan DV, Harnett-Sheehan K, Raj BA (1996). disorders in adults. Drugs 62, 655–703.
The measurement of disability. International Clinical Weiller E, Bisserbe JC, Maier W, Lecrubier Y (1998).
Psychopharmacology 11 (Suppl. 3), 89–95. Prevalence and recognition of anxiety syndromes in
Sjodin I, Kutcher SP, Ravindran A, Burt T (2003). Efficacy five European primary care settings. A report from the
of sertraline in improving quality of life and functioning WHO study on Psychological Problems in General Health
in generalized anxiety disorder. European Care. British Journal of Psychiatry 173 (Suppl. 34), 18–23.
Neuropsychopharmacology 13 (Suppl. 4), S365. Weizman R, Tanne Z, Granek M, Karp L, Golomb M,
Sonawalla SB, Spillmann MK, Kolsky AR, Alpert JE, Tyano S, Gavish M (1987). Peripheral benzodiazepine

Downloaded from https://academic.oup.com/ijnp/article-abstract/8/2/293/755456 by FMRP user on 26 April 2019

Nierenberg AA, Rosenbaum JF, Fava M (1999). Efficacy of binding sites on platelet membranes are increased during
fluvoxamine in the treatment of major depression with diazepam treatment of anxious patients. European Journal
comorbid anxiety disorders. Journal of Clinical Psychiatry of Pharmacology 138, 289–292.
60, 580–583. Wittchen HU (2002). Generalized anxiety disorder :
Sramek JJ, Tansman M, Suri A, Hornig-Rohan M, prevalence, burden and cost to society. Depression and
Amsterdam JD, Stahl SM, Weisler RH, Cutler NR (1996). Anxiety 16, 162–171.
Efficacy of buspirone in generalised anxiety disorder Wittchen HU, Carter RM, Pfisster H, Montgomery SA,
with coexisting mild depressive symptoms. Journal of Kessler RC (2000). Disabilities and quality of life in
Clinical Psychiatry 57, 287–291. pure and comorbid generalized anxiety disorder and
Stahl S (2000). Placebo-controlled comparison of the selective major depression in a national survey. International
serotonin reuptake inhibitors citalopram and sertraline. Clinical Psychopharmacology 15, 319–328.
Biological Psychiatry 48, 894–901. Yonkers KA, Dyck IR, Warshaw M, Keller MB (2000).
Stocchi FG, Nordera G, Jokinen RH, Lepola UM, Hewett K, Factors predicting the clinical course of generalised
Bryson H, Iyengar MK, Paroxetine Generalized Anxiety anxiety disorder. British Journal of Psychiatry 176, 544–549.