The new england journal of medicine

clinical practice

Generalized Anxiety Disorder

Gregory Fricchione, M.D.
This Journal feature begins with a case vignette highlighting a common clinical problem.
Evidence supporting various strategies is then presented, followed by a review of formal guidelines,
when they exist. The article ends with the author’s clinical recommendations.

During a routine visit, a 59-year-old woman, who describes herself as a lifetime “wor-
rier” and has a family history of depression, reports having restless sleep, muscle ten-
sion, and fatigue. Recently, her anxiety has intensified about her children, her job, and
her health, and it is having a negative effect on her family and work life. How should
she be treated?

the clinical problem

Anxiety disorders are the most prevalent psychiatric conditions in the United States aside From the Division of Psychiatry and Medi-
from disorders involving substance abuse.1 Generalized anxiety disorder has a lifetime cine, Massachusetts General Hospital, Bos-
ton. Address reprint requests to Dr. Fric-
prevalence of 5 percent. The criteria for diagnosis, as specified by the Diagnostic and Sta- chione at the Division of Psychiatry and
tistical Manual of Mental Disorders, fourth edition, are summarized in Table 1.2 The onset Medicine, Massachusetts General Hospital,
is usually before the age of 25 years, and the incidence in men is half that in women. Warren 6, 55 Fruit St., Boston, MA 02114,
or at [email protected].
Untreated, the typical course is chronic, with a low rate of remission and a moderate re-
currence rate.3 N Engl J Med 2004;351:675-82.
Copyright © 2004 Massachusetts Medical Society.
Risk factors for generalized anxiety disorder include a family history of the condi-
tion, an increase in stress, and a history of physical or emotional trauma.4,5 An associa-
tion has also been reported between smoking and anxiety, and the risk of generalized
anxiety disorder among adolescents who smoke heavily is five to six times the risk among
nonsmokers.6 Traits such as nervousness and social discomfort may predispose people
to both nicotine dependence and anxiety.7 Medical illnesses are often associated with
anxiety.8 For example, generalized anxiety disorder occurs in 14 percent of patients with
diabetes.9

coexisting psychiatric illnesses

Major depression is the most common coexisting psychiatric illness in patients with
generalized anxiety disorder, occurring in almost two thirds of such patients. Panic dis-
order occurs in a quarter of patients with generalized anxiety disorder, and alcohol abuse
in more than a third.1,10 Studies of twins suggest a shared genetic propensity to both
generalized anxiety disorder and major depression,11 and a recent report suggests that
a genetic variant of the serotonin-transporter gene may predispose people to both con-
ditions.12 In prospective studies, anxiety almost always appears to be the primary dis-
order and to increase the risk of depression.13 Patients who have generalized anxiety
disorder along with coexisting psychiatric illnesses have more impairment, seek more
medical attention, and have a poorer response to treatment than those without coexist-
ing illnesses.14

n engl j med 351;7 www.nejm.org august 12, 2004 675

 The new england journal of medicine

assessment of depression and suicide risk

strategies and evidence
When generalized anxiety disorder complicates
diagnosis major depression, there is an increased risk of sui-
Patients with generalized anxiety disorder often cide.18 Patients should be asked about depressive
have physical symptoms, and it may be difficult to symptoms, including suicidal ideation. If answers
distinguish the symptoms from those of medical ill- suggest that the patient is at risk for suicide, a psy-
nesses that are associated with anxiety.15 Factors chiatric evaluation should be performed as soon as
suggesting that anxiety is the symptom of a medical possible.
disorder include an onset of the symptoms after the
age of 35 years, no personal or family history of anx- therapy
iety, no increase in stress, little or no avoidance of
anxiety-provoking situations, and a poor response pharmacotherapy
to antianxiety medication.16 A physical cause should In part because depression is frequently associated
be suspected when anxiety follows recent changes in with generalized anxiety disorder, antidepressant
medication or accompanies signs and symptoms of agents are often used as first-line agents for treat-
a new disease. ing the latter condition.19 Commonly used agents
In evaluating patients for generalized anxiety and their doses and side effects are summarized in
disorder, practitioners should routinely consider Table 2.
medical conditions (e.g., cardiac, pulmonary, neu-
rologic, or endocrine illnesses, including hyperthy- Tricyclic Antidepressants
roidism), drug use (e.g., cocaine and other stimu- Tricyclic agents are effective for treating patients
lants, such as caffeine), drug withdrawal (e.g., who have generalized anxiety disorder alone or in
cessation of the use of alcohol, opiates, or benzodi- association with depression.20,21 In one random-
azepines), and both prescribed and over-the-coun- ized, controlled trial, the response to imipramine
ter medications (e.g., corticosteroids, sympathomi- was significantly greater than the response to pla-
metics, and herbal medicines, such as ginseng).10,15 cebo, with an improvement in symptoms that was
Before making a diagnosis of generalized anxi- more than 50 percent at eight weeks.21 When first
ety disorder, practitioners should take a history initiated, tricyclic agents can cause jitteriness and
and perform a physical examination in order to insomnia, and thus treatment should be initiated
rule out medical causes of anxiety. Laboratory test- at half the usual dose. Tricyclic therapy can be ham-
ing should be guided by the clinical presentation. pered by side effects, which may reduce a patient’s
The cost-effectiveness of specific laboratory test- adherence to treatment (Table 2).
ing is uncertain, but given the increased prevalence
of generalized anxiety disorder among patients with Selective Serotonin-Reuptake Inhibitors
hyperthyroidism, measurement of thyrotropin is The efficacy of selective serotonin-reuptake inhibi-
reasonable.17 tors (SSRIs) is similar to that of tricyclics, and SSRIs
Other psychiatric disorders that may be misdiag- have a more favorable side-effect profile. In an eight-
nosed as generalized anxiety disorder require con- week, randomized, placebo-controlled trial involv-
sideration. Whereas generalized anxiety disorder is ing 326 outpatients, those who received paroxetine
defined as worry that is present most of the time for (20 to 50 mg per day) had a significantly higher
at least six months, panic disorder is characterized response rate (defined as much or very much im-
by recurrent panic attacks, followed by at least one proved, with a reduction in anxiety and better func-
month of persistent anxiety about having more at- tioning) than those who received placebo (72 per-
tacks. Obsessive–compulsive disorder is manifested cent vs. 56 percent) and a significantly higher rate
as intrusive thoughts and ritualistic actions; post- of remission, which was defined as minimal or no
traumatic stress disorder as avoidance, numbing, anxiety and no functional impairment (42 percent
and hyperarousal; social phobia as severe anticipa- vs. 26 percent).22 These results have recently been
tory anxiety accompanying social situations; and so- replicated.23
matization disorder as multiple physical symptoms In a follow-up study, patients who had had a re-
in the absence of or out of proportion to underlying sponse to paroxetine in the 8-week trial were ran-
disease. domly assigned to continue to take paroxetine or to

676 n engl j med 351;7 www.nejm.org august 12 , 2004

 clinical practice

Table 1. Diagnostic Criteria for Generalized Anxiety Disorder.*

The patient reports having excessive anxiety and worry (apprehensive expectation), occurring more days than not for at
least 6 months, about a number of events or activities (such as work or school performance).
The patient has difficulty in controlling worry.
The anxiety and worry are associated with three or more of the following six symptoms (with at least some symptoms
present for more days than not for the previous 6 months): restlessness or feeling keyed up or on edge, being easily
fatigued, difficulty concentrating or mind going blank, irritability, muscle tension, sleep disturbance (difficulty falling
or staying asleep, or restless, unsatisfying sleep).†
The focus of the anxiety and worry is not confined to features of other types of psychiatric disorders (e.g., panic disorder,
social phobia, obsessive–compulsive disorder, separation anxiety disorder, anorexia nervosa, somatization disorder,
or hypochondriasis), and the anxiety and worry do not occur exclusively as part of post-traumatic stress disorder.
The anxiety, worry, or physical symptoms cause clinically significant distress or impairment in social, occupational, or
other important areas of functioning.
The disturbance is not due to the direct physiological effects of a medication, substance abuse, or a general medical con-
dition (e.g., hyperthyroidism) and does not occur exclusively during a mood disorder, a psychotic disorder, or a per-
vasive developmental disorder.

* Adapted from the American Psychiatric Association.2

† Only one item is required in children.

take placebo for an additional 24 weeks.24 The rate ized anxiety disorder and depression and is effective
of relapse during this period was 11 percent for pa- when a patient has both conditions.28 In a report of
tients receiving paroxetine, as compared with 41 two double-blind, placebo-controlled trials of ex-
percent for those receiving placebo. Among the pa- tended-release venlafaxine, the rates of response
tients who continued to receive paroxetine, the rate among patients with generalized anxiety disorder
of remission after six months (73 percent) was high- were 58 percent at eight weeks and 66 percent at six
er than the rate at the end of the eight-week study, a months (vs. 36 percent and 39 percent, respectively,
finding suggesting that a longer course of treatment for placebo).29 Almost two thirds of patients who
increases the likelihood of remission. had no response to venlafaxine at eight weeks did
Restlessness can occur with the initiation of SSRI have a response by six months. Remission rates for
therapy, so starting doses should be low. Paroxetine venlafaxine were 32 percent at eight weeks and 43
is the SSRI that has been studied most extensively percent at six months (vs. 15 percent and 19 percent,
for the treatment of generalized anxiety disorder, respectively, for placebo). Caution is warranted with
and it is approved by the Food and Drug Adminis- doses of extended-release venlafaxine that are great-
tration (FDA) for this indication, but other members er than 225 mg per day, given the infrequent com-
of this class of medications, such as citalopram and plication of sustained systolic hypertension, and in
escitalopram, have also been shown to have effica- patients with a history of conduction disturbance
cy.25,26 Concern about an increased risk of suicide or ventricular arrhythmias, given the risks of these
among adults taking SSRIs, in particular, is not complications with an overdose.
supported by a review of placebo-controlled studies
involving a total of 48,277 depressed patients and Duration of Therapy
nine antidepressants. The review showed no sig- Responses to the medications discussed above are
nificant difference in suicide rates among the study expected within eight weeks when patients are re-
groups.27 However, the initiation of treatment with ceiving a therapeutic dose, although longer cours-
any antidepressant in a depressed patient requires es may yield more favorable results, particularly in
monitoring for suicidal ideation and behavior. those patients who have had partial responses. Al-
though objective data on rates of relapse and remis-
Serotonin–Norepinephrine–Reuptake Inhibitors sion with longer use of the medications are sparse,
Extended-release venlafaxine (Effexor XR), a seroto- patients who have a response should generally be
nin–norepinephrine–reuptake inhibitor, is also ap- advised to continue taking the antidepressant for
proved by the FDA for treatment of both general- six months to a year. When a medication is ineffec-

n engl j med 351;7 www.nejm.org august 12, 2004 677

 Table 2. Psychopharmacologic Treatments for Generalized Anxiety Disorder.*

678
Risk during Target
Medication Class Pregnancy† Starting Dose‡ Titration Up Dose‡ Side Effects§
Antidepressant anxiolytics
Citalopram (Celexa) SSRI C 10 mg/day 10 mg after 7 days 10–40 mg Nausea, vomiting, dry mouth, headache, somnolence, in-
Maintain for 3–4 wk somnia, sweating, tremor, diarrhea, sexual dysfunction,
Escitalopram (Lexapro) SSRI C 10 mg/day 10 mg after 4 wk 10–20 mg syndrome of inappropriate antidiuretic hormone, cyto-
chrome P-450 2D6 substrate elevation due to enzyme in-
Paroxetine (Paxil) SSRI C 10 mg/day 10 mg after 1–2 wk 10–40 mg hibition (paroxetine especially; citalopram and escitalo-
Maintain for 3–4 wk pram are not significant inhibitors), discontinuation effects
Sertraline (Zoloft) SSRI C 25 mg/day 25 mg every 3–5 days 50–200 mg (fatigue, dysphoria, psychomotor changes)
Venlafaxine (Effexor) SNRI C 25–37.5 mg/day 25 mg every 4–7days 50–75 mg Nausea, somnolence, dizziness, dry mouth, nervousness,
3 times/day tremor, insomnia, constipation, sexual dysfunction, sweat-
Venlafaxine XR (Effexor XR) SNRI C 37.5 mg/day 37.5 mg every 4–7 75–225 mg ing, anorexia, blood pressure elevation, orthostasis, con-
days duction defects, ventricular arrhythmias, discontinuation
effects (fatigue, dysphoria, psychomotor changes); half
usual dose used in moderate hepatic or renal impairment
Imipramine (Tofranil) Tricyclic antidepres- D 10 mg/day 20–25 mg every 7 50–200 mg Orthostasis, conduction defects, ventricular arrhythmias, re-
The

sant days flex tachycardia, anticholinergic effects, weight gain, po-

Nortriptyline (Pamelor, Tricyclic antidepres- D 10 mg/day 10–25 mg every 7 20–150 mg tential lethality in overdose
Aventyl) sant days
Benzodiazepine anxiolytics
Clonazepam (Klonopin) Benzodiazepine D 0.25 mg 2 times/ 0.25–0.5 mg every 4 0.5–2 mg Sedation, ataxia, hypotonia, paradoxical agitation, memory

n engl j med 351;7

day days changes, withdrawal syndrome
Lorazepam (Ativan) Benzodiazepine D 0.5 mg 2 times/ 0.5 mg every 4 days 1–4 mg
day
Nonbenzodiazepine anxiolytic
Buspirone (BuSpar) Serotonin 1A agonist B 5 mg 2 times/day 5 mg every 2–3 days 10–60 mg Dizziness, headache, drowsiness, light-headedness, fatigue,

www.nejm.org
nausea, insomnia, restlessness
new england journal
of

Anticonvulsant anxiolytics
Gabapentin (Neurontin) Antiepileptic drug C 100 mg 2 times/ 100 mg every 3 days 100–1800 mg Somnolence, dizziness, ataxia, fatigue, nystagmus, nausea,
day dry mouth, constipation, peripheral edema, rhinitis,
pharyngitis, visual changes, myalgia
Tiagabine (Gabitril) Antiepileptic drug C 2 mg 2 times/day 2–4 mg every 7 days 2–16 mg Somnolence, nervousness, dizziness, tremor, abdominal
medicine

pain, diarrhea, vomiting, asthma, pharyngitis

august 12 , 2004
* Adapted from data from Pollack et al.,10 Goldberg and Posner,15 and Rosenbaum et al.16 More information is available at the Web site of the Anxiety Disorders Association of America at
www.adaa.org. SSRI denotes selective serotonin-reuptake inhibitor, and SNRI serotonin–norepinephrine reuptake inhibitor.
† The pregnancy-risk category is established by the FDA. B denotes that studies in animals do not indicate a risk to the fetus; however, there are no adequate, well-controlled studies in preg-
nant women, or studies in animals have shown an adverse effect on the fetus, but adequate, well-controlled studies in pregnant women have failed to demonstrate a risk to the fetus. De-
spite the findings in animals the possibility of fetal harm appears to be remote, if the drug is used during pregnancy. C denotes that studies in animals have shown that the drug has tera-
togenic or embryocidal effects, but there have been no adequate well-controlled studies in pregnant women, or no data are available in either animals or pregnant women. D denotes that
positive evidence of human fetal risk exists, but benefits in certain situations (e.g., life-threatening situations or serious diseases for which safer drugs cannot be used or are ineffective)
may make use of the drug acceptable despite its risks.
‡ In perinatal and geriatric patients, approximately half the usual doses are used.
§ Serotonin syndrome can occur when SSRIs, SNRIs, tricyclic antidepressants, serotonin antagonist–reuptake inhibitors, monoamine oxidase inhibitors, and serotonin 1A agonists are
used in various combinations.
 clinical practice

tive or intolerable, switching to another agent in the term use and is facilitated by treatment with anxio-
same class or another class is reasonable, although lytic antidepressants.20
this approach has not been extensively studied.30 Benzodiazepines can often be helpful as short-
term treatment when antidepressants are initiated,
Nonbenzodiazepine Anxiolytic Agents since benzodiazepines rapidly relieve symptoms
Buspirone has been shown in double-blind, ran- (whereas antidepressants typically take weeks to
domized, controlled trials to have efficacy in the work) and also help alleviate the restlessness or ner-
treatment of generalized anxiety disorder.31 It does vousness sometimes associated with the initiation
not cause sedation, physical dependency, or with- of antidepressant therapy. Benzodiazepines can be
drawal. However, it has no antidepressant effect, tapered over a period of several weeks after the an-
so it should not be used alone for anxiety with coex- xiolytic effects of an antidepressant have taken hold.
isting depression. A period of two to four weeks or Nevertheless, in selected patients who have a relapse
longer is generally needed for a response. or cannot tolerate tapering of the dose in order to
discontinue the drug, benzodiazepines may be used
Benzodiazepines long term. A meta-analysis of nine studies of com-
Double-blind trials have also shown the efficacy of bined benzodiazepine–antidepressant treatment,
benzodiazepines in treating generalized anxiety dis- including SSRIs and tricyclics, for coexisting anxi-
order, but the side effects are a concern (Table 2), ety and depression revealed that combined treat-
particularly in elderly patients.32,33 In one random- ment was more likely than antidepressant therapy
ized trial comparing paroxetine, imipramine, and a alone to reduce anxiety and depression and that it
benzodiazepine among nondepressed patients with was associated with a lower dropout rate.35
generalized anxiety disorder, the benzodiazepine
was the most effective of the three drugs during the psychotherapy
first two weeks, but at eight weeks it was less effec- The most extensively studied psychotherapy for
tive than either antidepressant.34 anxiety is cognitive behavioral therapy.36 This ther-
With long-term benzodiazepine use (which is apy, which teaches patients to substitute positive
generally defined as use of a therapeutic dose for thoughts for anxiety-provoking ones, usually in-
two months or more), patients may become physi- volves 6 to 12 individual sessions at weekly intervals.
cally dependent on the drug. However, addiction is Patients record their thoughts and feelings in dia-
infrequent and occurs mostly in patients at risk for ries, noting situations in which they feel anxious and
substance abuse. To avoid withdrawal symptoms behaviors that relieve the anxiety. They also role-play
(which include seizures, hypersympathetic tone, scenes and rehearse responses to anxiety.
and anxiety), the dose can be reduced gradually In one randomized, controlled study, 32 percent
(e.g., a 1.0-mg reduction in the dose of lorazepam of patients in the group that received cognitive be-
per week or a 0.5-mg reduction in the dose of al- havioral therapy had clinically significant improve-
prazolam per week). Slow tapering is especially im- ment at three months, and 42 percent had clinically
portant for patients taking high-potency and short- significant improvement at six months, whereas
acting benzodiazepines such as lorazepam and none of the patients in the control group had signif-
alprazolam. Alprazolam can be associated with a icant improvement at three months.37 In a recent
severe discontinuation syndrome, complicated by study that followed subjects from earlier random-
dysphoria and delirium. Switching to equipotent ized trials for 8 to 14 years, the condition of half of
doses of long-acting clonazepam, with alprazolam the patients who had an initial response to cognitive
available on an as-needed basis for one week, can behavioral therapy remained markedly improved,
make tapering easier.10 approximately one third still had some improve-
There are limited data to guide decisions about ment, and the rest had recurrent anxiety and dis-
the duration of benzodiazepine therapy. In one ability.38
study, among patients taking only diazepam for An alternative approach to cognitive behavioral
anxiety, discontinuation after six months resulted therapy is applied relaxation therapy, in which the
in a relapse rate of 63 percent within a year.33 How- patient imagines calming situations to induce mus-
ever, in most patients, tapering the dose as a means cular and mental relaxation. One randomized, con-
of gradual discontinuation is preferable to long- trolled trial comparing cognitive therapy with ap-

n engl j med 351;7 www.nejm.org august 12, 2004 679

 The new england journal of medicine

plied relaxation therapy showed no significant answer “yes” to the question, “During the past four
difference in improvement rates at one year.39 Sim- weeks, have you been bothered by feeling worried,
ilarly, in a review comparing the outcomes of two tense, or anxious most of the time?” Some practi-
studies of applied relaxation therapy with the out- tioners have supported using this question as a
comes of four studies of cognitive behavioral ther- screening technique.48
apy, the response rates at six months were similar
(52 percent for applied relaxation therapy and 41 perinatal management
percent for cognitive behavioral therapy).40 Psy- During pregnancy and the postpartum period, there
chotherapeutic additions to cognitive behavioral is an increased risk that generalized anxiety disor-
therapy are being tried to improve the response rate, der will develop or worsen. Psychotherapy may ob-
although the efficacy of these combined approach- viate the need for pharmacotherapy, but for severe
es is uncertain.36 anxiety (which may be associated with adverse
obstetrical outcomes, including premature birth),
areas of uncertainty medication may be necessary.49
There are no data from prospective controlled
resistant generalized anxiety disorder studies of antianxiety agents administered during
The optimal management of generalized anxiety pregnancy or lactation, and recommendations are
disorder that is resistant to either psychotherapy based on observational data alone. Buspirone may
alone or pharmacotherapy alone is uncertain.41 If be safer than some other medications during preg-
the patient’s initial response to medication is inad- nancy. Benzodiazepines should not be used in the
equate, the dose should be increased as tolerated first trimester because of the risk of oral clefts. Use
and then maintained for at least eight weeks. A pa- of benzodiazepines late in the third trimester may
tient’s partial response to a medication may war- cause the floppy infant syndrome or neonatal with-
rant a longer trial, given data that suggest further drawal. Sedation may occur in breast-fed infants of
improvement with longer use. Combining psycho- women taking benzodiazepines.49 If benzodiaz-
therapy and pharmacotherapy should also be con- epines are used perinatally, the rule is to use the
sidered, although data are lacking on whether the lowest effective dose for the shortest period.
combination results in a better outcome than either Although data are limited, SSRIs do not appear
approach alone.42 The possibility of an underlying to be teratogenic. However, there have been reports
medical condition or a coexisting psychiatric illness of neonatal toxicity, such as early delivery, that was
should be reconsidered in resistant cases.43 unrelated to the duration of fetal exposure and of
Little research has been done on the effects of low Apgar scores, including respiratory distress, in
combining medications. For patients who have a infants whose mothers were receiving the drugs in
partial response to one medication, combination the third trimester.49,50 When these medications are
treatment with antidepressants and benzodiaz- used prenatally, decisions about which drug to use
epines or buspirone, or augmentation with other may be based on apparent reproductive safety in
agents, including anticonvulsants (e.g., gabapen- practice (e.g., fluoxetine or citalopram) or on a low
tin or tiagabine), may be helpful.44,45 ratio of the cord drug concentration to the serum
drug concentration, suggesting minimal fetal expo-
prevention sure (e.g., sertraline or paroxetine).50,51 Neverthe-
In one study, college students who were thought to less, concern has been expressed about the use of
be at risk for depression were randomly assigned paroxetine near delivery because of reports of tran-
to an eight-week cognitive behavioral workshop. At sient neonatal withdrawal symptoms, including res-
three years, they had fewer episodes of generalized piratory distress, which in some cases has required
anxiety than students who had undergone only an intensive treatment.49 Breast-fed infants are also
initial assessment.46 More research on prevention exposed to antidepressants, and use of the lowest
is warranted. effective dose possible is advised during both preg-
nancy and lactation. A detailed discussion of the
screening use of anxiolytic medications in pregnancy and lac-
Screening for generalized anxiety disorder is not tation is beyond the scope of this review but is
routinely advocated in practice.47 However, approx- available elsewhere.52,53 Ultimately, the risk of a
imately 90 percent of patients with the condition medication should be weighed against the poten-

680 n engl j med 351;7 www.nejm.org august 12 , 2004

 clinical practice

tial benefits of relief of serious illness and improved a low dose and then increase it during the next three
caretaking abilities, and decisions should be indi- weeks or so until the target dose is reached (Table 2).
vidualized. A four-to-five-week course of a benzodiazepine
(e.g., clonazepam, given at a dose of 0.25 to 0.5 mg
guidelines twice daily) may also be useful in reducing restless-
ness related to antidepressant therapy and in rapidly
There are currently no formal guidelines from U.S. controlling anxiety. One would taper this dose dur-
or European professional societies for the manage- ing the next two to four weeks. Patients who prefer a
ment of generalized anxiety disorder. nonpharmacologic approach should be referred for
cognitive behavioral therapy and relaxation train-
ing. These therapies may also be useful in patients
conclusions
and recommendations who take medication, although data are limited.
During the initiation of drug therapy, patients
In patients who present with anxiety as a symptom, can be seen at intervals of two to four weeks, with the
medical causes, such as hyperthyroidism, require frequency decreased to every three to four months
consideration, as do common coexisting illnesses during maintenance therapy. If medication is effec-
such as major depression, panic disorder, and sub- tive, it should be continued for six months to a year
stance abuse. For generalized anxiety alone or anx- and then tapered off, with monitoring for the re-
iety that is associated with depression, a reasonable currence of anxiety or depression, a finding that
first-line approach is to administer an SSRI or ex- would require reinitiation of treatment. If a psychia-
tended-release venlafaxine on the basis of the dem- trist has not been consulted earlier, referral is ad-
onstrated efficacy of these drugs and their generally visable after two failed medication trials or when
favorable side-effect profiles. To minimize side ef- the patient has complex coexisting illnesses or sui-
fects, particularly restlessness, one would start with cidal ideation.

references
1. Kessler RC, McGonagle KA, Zhao S, et 8. Rogers MP, White K, Warshaw MG, et 16. Rosenbaum JF, Pollack MH, Otto MW,
al. Lifetime and 12-month prevalence of al. Prevalence of medical illness in patients Bernstein JG. Anxious patients. In: Cassem
DSM-III-R psychiatric disorders in the Unit- with anxiety disorders. Int J Psychiatry Med NH, Stern TA, Rosenbaum JF, Jellinek MS,
ed States: results from the National Comor- 1994;24:83-96. eds. Massachusetts General Hospital hand-
bidity Survey. Arch Gen Psychiatry 1994;51: 9. Grigsby AB, Anderson RJ, Freedland book of general hospital psychiatry. 4th ed.
8-19. KE, Clouse RE, Lustman PJ. Prevalence of St. Louis: Mosby–Year Book, 1997:173-210.
2. Diagnostic and statistical manual of anxiety in adults with diabetes: a systematic 17. Simon NM, Blacker D, Korbly NB, et al.
mental disorders, 4th ed.: DSM-IV. Wash- review. J Psychosom Res 2002;53:1053-60. Hypothyroidism and hyperthyroidism in
ington, D.C.: American Psychiatric Associa- 10. Pollack MH, Smoller JW, Lee DK. Ap- anxiety disorders revisited: new data and lit-
tion, 1994:435-6. proach to the anxious patient. In: Stern TA, erature review. J Affect Disord 2002;69:209-
3. Wittchen HU, Carter RM, Pfister H, Herman JB, Slavin PL, eds. The MGH guide 17.
Montgomery SA, Kessler RC. Disabilities to psychiatry in primary care. New York: 18. Olfson M, Weissman MM, Leon AC,
and quality of life in pure and comorbid gen- McGraw-Hill, 1998:23-37. Sheehan DV, Farber L. Suicidal ideation in
eralized anxiety disorder and major depres- 11. Noyes R Jr. Comorbidity in generalized primary care. J Gen Intern Med 1996;11:
sion in a national survey. Int Clin Psycho- anxiety disorder. Psychiatr Clin North Am 447-53.
pharmacol 2000;15:319-28. 2001;24:41-55. 19. Kapczinski S, Lima MS, Souza JS,
4. Brantley PJ, Mehan DJ Jr, Ames SC, 12. Hariri AR, Mattay VS, Tessitore A, et al. Schmitt R. Antidepressants for generalized
Jones GN. Minor stressors and generalized Serotonin transporter genetic variation and anxiety disorder. Cochrane Database Syst
anxiety disorders among low-income pa- the response of the human amygdala. Sci- Rev 2002;2:CD003592.
tients attending primary care clinics. J Nerv ence 2002;297:400-3. 20. Rickels K, DeMartinis N, Garcia-Espana
Ment Dis 1999;187:435-40. 13. Wittchen HU, Kessler RC, Pfister H, F, Greenblatt DJ, Mandos LA, Rynn M. Imip-
5. Brown ES, Fulton MK, Wilkeson A, Pet- Lieb M. Why do people with anxiety disor- ramine and buspirone in treatment of pa-
ty F. The psychiatric sequelae of civilian trau- ders become depressed? A prospective-lon- tients with generalized anxiety disorder for
ma. Compr Psychiatry 2000;41:19-23. gitudinal community study. Acta Psychiatr discontinuing long-term benzodiazepine
6. Johnson JG, Cohen P, Pine DS, Klein DF, Scand Suppl 2000;406:14-23. therapy. Am J Psychiatry 2000;157:1973-9.
Kasen S, Brook JS. The association between 14. Olfson M, Fireman B, Weissman MM, et 21. Rickels K, Downing R, Schweizer E,
cigarette smoking and anxiety disorders al. Mental disorders and disability among Hassman H. Antidepressants for the treat-
during adolescence and early adulthood. patients in a primary care group practice. ment of generalized anxiety disorder: a pla-
JAMA 2000;284:2348-51. Am J Psychiatry 1997;154:1734-40. cebo controlled comparison of imipramine,
7. Breslau N, Kilbey MM, Andreski P. Vul- 15. Goldberg RJ, Posner DA. Anxiety in the trazodone, and diazepam. Arch Gen Psychi-
nerability to psychopathology in nicotine- medically ill. In: Stoudemire A, Fogel BS, atry 1993;50:884-95.
dependent smokers: an epidemiologic study Greenberg DB, eds. The psychiatric care of 22. Pollack MH, Zaninelli R, Goddard A, et
of young adults. Am J Psychiatry 1993;150: the medical patient. New York: Oxford Uni- al. Paroxetine and treatment of generalized
941-6. versity Press, 2000:165-80. anxiety disorder: results of a placebo-con-

n engl j med 351;7 www.nejm.org august 12, 2004 681

 clinical practice

trolled flexible-dosage trial. J Clin Psychiatry 33. Rickels K, Case WG, Downing RW, Frid- 43. Yonkers KA, Dyck IR, Warshaw M,
2001;62:305-7. [Erratum, J Clin Psychiatry man R. One-year follow-up of anxious pa- Keller MB. Factors predicting the clinical
2001;62:658.] tients treated with diazepam. J Clin Psy- course of generalized anxiety disorder. Br J
23. Rickels K, Zaninelli R, McCafferty J, chopharmacol 1986;6:32-6. Psychiatry 2000;176:544-9.
Bellew K, Iyengar M, Sheehan D. Paroxetine 34. Rocca P, Fonzo V, Scotta M, Zanalda E, 44. Pollack MH, Matthews J, Scott EL. Ga-
treatment of generalized anxiety disorder: a Ravizza L. Paroxetine efficacy in the treat- bapentin as a potential treatment for anxiety
double-blind, placebo-controlled study. Am ment of generalized anxiety disorder. Acta disorders. Am J Psychiatry 1998;155:992-3.
J Psychiatry 2003;160:749-56. Psychiatr Scand 1997;95:444-50. 45. Schwartz TL. The use of tiagabine aug-
24. Stocchi F, Nordera G, Jokinen RH, et al. 35. Furukawa TA, Streiner DL, Young LT. mentation for treatment-resistant anxiety
Efficacy and tolerability of paroxetine for the Antidepressant and benzodiazepine for ma- disorders: a case series. Psychopharmacol
long-term treatment of generalized anxiety jor depression. Cochrane Database Syst Rev Bull 2002;36:53-7.
disorder. J Clin Psychiatry 2003;64:250-8. 2002;1:CD001026. 46. Seligman MEP, Schulman P, DeRubeis
25. Varia I, Rauscher F. Treatment of gener- 36. Borkovec TD, Newman MG, Caston- RJ, Hollon SD. The prevention of depression
alized anxiety disorder with citalopram. Int guay LG. Cognitive-behavioral therapy for and anxiety. Prev Treat 1999;2:1-22. (Also
Clin Psychopharmacol 2002;17:103-7. the treatment of generalized anxiety disor- available at http://www.journals.apa.org/
26. Waugh J, Goa KL. Escitalopram: a re- der with integrations from interpersonal prevention.)
view of its use in the management of major and experiential therapies. CNS Spectr 47. Lang AJ, Stein MB. Screening for anxiety
depressive and anxiety disorders. CNS 2003;8:382-9. in primary care: why bother? Gen Hosp Psy-
Drugs 2003;17:343-62. 37. Butler G, Fennel M, Robson P, Gelder chiatry 2002;24:365-6.
27. Khan A, Khan S, Kolts R, Brown WA. M. Comparison of behavior therapy and 48. Wittchen H-U, Boyer P. Screening for
Suicide rates in clinical trials of SSRIs, other cognitive behavior therapy in the treatment anxiety disorders: sensitivity and specificity
antidepressants, and placebo: analysis of of generalized anxiety disorder. J Consult of the Anxiety Screening Questionnaire
FDA reports. Am J Psychiatry 2003;160: Clin Psychol 1991;59:167-75. (ASQ-15). Br J Psychiatry Suppl 1998;34:
790-2. 38. Durham RC, Chambers JA, MacDonald 10-7.
28. Pollack MH. Optimizing pharmaco- RR, Power KG, Major K. Does cognitive- 49. Cohen LS, Nonacs R, Viguera AC. The
therapy of generalized anxiety disorder to behavioural therapy influence the long-term pregnant patient. In: Stern T, Fricchione GL,
achieve remission. J Clin Psychiatry 2001; outcome of generalized anxiety disorder? Cassem NH, Jellinek MS, Rosenbaum JF,
62:Suppl 19:20-5. An 8-14 year follow-up of two clinical trials. eds. The Massachusetts General Hospital
29. Montgomery SA, Sheehan DV, Meoni P, Psychol Med 2003;33:499-509. handbook of general hospital psychiatry.
Haudiquet V, Hackett D. Characterization of 39. Ost L-G, Breitholtz E. Applied relax- 5th ed. St. Louis: Mosby, 2004:593-611.
the longitudinal course of improvement in ation vs. cognitive therapy in the treatment 50. Simon GE, Cunningham ML, Davis RL.
generalized anxiety disorder during long- of generalized anxiety disorder. Behav Res Outcomes of prenatal antidepressant expo-
term treatment with venlafaxine XR. J Psy- Ther 2000;38:777-90. sure. Am J Psychiatry 2002;159:2055-61.
chiatr Res 2002;36:209-17. 40. Fisher PL, Durham RC. Recovery rates 51. Hendrick V, Stowe ZN, Altshuler LL,
30. Thompson PM. Generalized anxiety in generalized anxiety disorder following Hwang S, Lee E, Haynes D. Placental pas-
disorder treatment algorithm. Psychiatr psychological therapy: an analysis of clini- sage of antidepressant medications. Am J
Ann 1996;26:227-32. cally significant change in the STAI-T across Psychiatry 2003;160:993-6.
31. Sramek JJ, Tansman M, Suri A, et al. Ef- outcome studies since 1990. Psychol Med 52. Wisner KL, Parry BL, Piontek CM. Post-
ficacy of buspirone in generalized anxiety 1999;29:1425-34. partum depression. N Engl J Med 2002;347:
disorder with coexisting mild depressive 41. Coplan JD, Tiffon L, Gorman JM. Thera- 194-9.
symptoms. J Clin Psychiatry 1996;57:287- peutic strategies for the patient with treat- 53. Levine RE, Oandasan AP, Primeau LA,
91. ment-resistant anxiety. J Clin Psychiatry Berenson AB. Anxiety disorders during
32. Chouinard G, Annable L, Fontaine R, 1993;54:Suppl:69-74. pregnancy and postpartum. Am J Perinatol
Solyom L. Alprazolam in the treatment of 42. Foa EB, Franklin ME, Moser J. Context in 2003;20:239-48.
generalized anxiety and panic disorders: the clinic: how well do cognitive-behavioral Copyright © 2004 Massachusetts Medical Society.
a double-blind placebo-controlled study. Psy- therapies and medications work in combina-
chopharmacology (Berl) 1982;77:229-33. tion? Biol Psychiatry 2002;52:987-97.

682 n engl j med 351;7 www.nejm.org august 12 , 2004