Efficacy and tolerability of duloxetine in elderly patients with generalized anxiety disorder: a pooled analysis of four randomized, double-blind, placebo-controlled studies

Efficacy and tolerability of duloxetine in elderly patientswith generalized anxiety disorder: a pooled analysis of fourrandomized, double-blind, placebo-controlled studies

Jonathan Davidson1*, Christer Allgulander2, Mark H. Pollack3, James Hartford4,Janelle S. Erickson5, James M. Russell5, David Perahia6,7, Madalaine M. Wohlreich5,Janice Carlson5 and Joel Raskin8

1Department of Psychiatry and Behavioral Sciences, Duke University, Durham, NC, USA2Karolinska Institutet, Huddinge University Hospital, Huddinge, Sweden3Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA4Community Research, Cincinnati, OH, USA5Lilly Research Laboratories, Indianapolis, IN, USA6Lilly Research Centre, Windlesham, UK7The Gordon Hospital, London, UK8Lilly Research Laboratories, Eli Lilly Canada, Toronto, ON, Canada

Objective To assess the efficacy and tolerability of duloxetine in elderly patients with generalized anxiety disorder (GAD).Methods Acute-phase data from a subset of patients (�65 years) with GAD were pooled from four randomized,double-blind, placebo-controlled trials of duloxetine (3 flexible, 1 fixed dosing). Patients were treated with duloxetine60–120mg once daily or placebo for 9–10 weeks. The primary outcome measure was the mean baseline-to-endpoint changein Hamilton anxiety scale (HAMA) total score. Secondary measures included the HAMA psychic and somatic anxietysubscales and the Hospital Anxiety Depression Scale (HADS).Results Of 1491 patients randomly assigned to treatment, 4.9% (duloxetine, n¼ 45; placebo, n¼ 28) were� 65 years old.Compared with placebo-treated patients, duloxetine-treated patients experienced significantly greater improvements on theHAMA-total (p¼ 0.029), the HAMA-psychic anxiety factor (p¼ 0.034), HADS-anxiety (p¼ 0.049) and -depression scales(p¼ 0.026), but not the HAMA somatic anxiety factor (p¼ 0.074). Nausea was reported significantly more often induloxetine-treated patients (30.0% vs. 7.1%, p¼ 0.023); duloxetine-treated patients experienced greater weight loss(p¼ 0.018). More duloxetine-treated patients discontinued treatment due to an adverse event (22.2% vs. 0%; p¼ 0.006).Conclusion Duloxetine was effective in an elderly patient subset with GAD, although there was a high rate ofdiscontinuations due to adverse events. Copyright # 2008 John Wiley & Sons, Ltd.

key words—duloxetine; elderly; generalized anxiety disorder; efficacy; tolerability

INTRODUCTION

Generalized anxiety disorder (GAD) is common in theelderly (Beekman et al., 2000; Copeland et al., 1987;Flint and Rifat, 1997). Symptoms of GAD oftendevelop earlier in life (Le Roux et al., 2005), but in a

substantial number of elderly patients, GAD appearsfor the first time after age 60 (Blazer et al., 1991). Thesymptoms in elderly patients may include insomnia,hypervigilance, and difficulties with cognition, andmay be confused for depression or dementia(Schneider, 1996).Treating GAD or high-trait anxiety symptoms in

the elderly is important because GAD can affectphysical health (Himmelfarb and Murrell, 1984) andpsychological health (Wetherell et al., 2004). These

human psychopharmacologyHum. Psychopharmacol Clin Exp 2008; 23: 519–526.

Published online 14 May 2008 in Wiley InterScience

(www.interscience.wiley.com) DOI: 10.1002/hup.949

*Correspondence to: J. Davidson, 3068 Baywood Drive, SeabrookIsland, SC 29455, USA. Fax: 919-684-8866.E-mail: [email protected]

Copyright # 2008 John Wiley & Sons, Ltd.

Received 13 March 2008

Accepted 19 March 2008

patients may be at a greater risk for mortality(Allgulander and Lavori, 1991; Allgulander and Lavori,1993) and suicide (Allgulander and Lavori, 1993;Alwahhabi, 2003; Lenze et al., 2000). Treatingelderly patients can be difficult because they mayhave higher rates of polypharmacy (Linjakumpu et al.,2002), including benzodiazepine use (Bogunovic andGreenfield, 2004; Schneider, 1996). Furthermore, theefficacy and tolerability of medications in the elderlymay differ from those in younger patients because ofage-related changes in drug metabolism (Turnheim,2003).Duloxetine is a selective serotonin and norepi-

nephrine reuptake inhibitor (SNRI) that is efficaciousand well tolerated in treating patients with GAD andelderly patients with major depressive disorder (MDD;Raskin et al., 2007; Russell et al., 2007; Wise et al.,2007). As duloxetine’s efficacy and tolerability inelderly patients with GAD have not been investigated,we conducted post hoc analyses on data from elderlypatients by pooling such data from four randomized,double-blind, placebo-controlled studies of duloxetinein GAD. The studies represent all of the Lilly-conducteddouble-blind, placebo-controlled trials. To our knowl-edge, this manuscript is the first publication to focus ondata from duloxetine-treated patients� 65 years oldwith GAD.

MATERIALS AND METHODS

Study selection

Four separate studies (Study 1, Koponen et al., 2007;Study 2, Rynn et al., 2008; Study 3, Hartford et al.,

2007; and Study 4; Nicolini et al., 20081 conducted byEli Lilly and Company (from July 2004 to January2007) were combined for these post hoc analyses. Allwere multicenter, randomized, double-blind, place-bo-controlled, parallel-group studies of outpatientswith a diagnosis of GAD (Table 1).

Patients

All patients in these analyses participated in one of thefour studies of duloxetine treatment in GAD. Studyparticipants included male and female out-patients� 18 years old with a diagnosis of GAD.Diagnoses were based on the Mini InternationalNeuropsychiatric Interview (MINI; Sheehan et al.,1998) from the Diagnostic and Statistical Manual ofMental Disorders, 4th edition, text revision(DSM-IV-TR; APA, 2000) and were confirmed bythe study investigators. Raters were trained in the useof the Structured Clinical Interview Guide (Shearet al., 2001) for the Hamilton anxiety scale total score(HAMA; Hamilton, 1959). Inter-rater scoringreliability was assessed using a modified version ofthe rater applied performance scale (Lipsitz et al.,2004). Study protocols were approved by the ethicscommittee of each individual site and in accordancewith the principles of the Declaration of Helsinki.Patients provided written informed consent beforeparticipation in any study-related procedures.

Patients were required to have at least a moderatedisease severity, as defined by a hospital anxiety

Table 1. Summary of the four double-blind, placebo-controlled duloxetine studies in patients with GADa

Study N

Includedin thisanalysis

Duration ofthe active

phase (weeks)Control/comparator

Dosing regimen(mg/day, once/day)

Age range(years)

Mean age(SD, years)

1 513b 31 9 Placebo Fixed, 60 or 120 18–83 43.8 (13.0)2 327c 19 10 Placebo Flexible, 60–120 19–77 41.6 (14.1)3 487d 15 10 Placebo

venlafaxineFlexibleduloxetine: 60–120venlafaxine: 75–225

19–83 40.8 (13.7)

4 581e 9 10 Placeboduloxetine (20mg)venlafaxine

Flexibleduloxetine: 60–120venlafaxine: 75–225

18–78 42.3 (12.8)

Abbreviations: GAD, generalized anxiety disorder.aAccording to DSM-IV criteria.bDuloxetine, n¼ 338; placebo, n¼ 175 (Koponen et al., 2007).cDuloxetine, n¼ 168; placebo, n¼ 159 (Rynn et al., 2008).dDuloxetine, n¼ 162; venlafaxine, n¼ 164; placebo, n¼ 161 (Hartford et al., 2007).eDuloxetine, n¼ 158; venlafaxine, n¼ 169; placebo, n¼ 170; duloxetine (20mg), n¼ 84 (Nicolini et al., 2008). The velafaxine and 20mgduloxetine arms were not included in this analysis.

1Studies 1–4 are F1J-HMBR, -HMDT, -HMDU, and -HMDW,respectively.

Copyright # 2008 John Wiley & Sons, Ltd. Hum. Psychopharmacol Clin Exp 2008; 23: 519–526.DOI: 10.1002/hup

520 j. davidson ET AL.

depression scale (HADS; Zigmond and Snaith, 1983)anxiety subscale score of� 10, a Covi anxiety scale(CAS; Lipman et al., 1981) score of� 9, and noitem> 3 on the Raskin depression scale (RDS; Raskinet al., 1969). The CAS score must also have beengreater than the RDS score at baseline. Patients werealso required to have a clinical global impressions–severity of illness (CGI-S; Guy, 1976) score� 4 atboth baseline and randomization. Reasons for exclu-sion are provided in detail elsewhere (Hartford et al.,2007; Koponen et al., 2007; Rynn et al., 2008;Nicolini et al., 2008).

Treatments

Study 1 consisted of a 9-week acute therapy phase,followed by a 2-week discontinuation phase. After a1-week single-blind placebo lead-in, patients wererandomly assigned (1:1:1) to receive either duloxetine(60 or 120mg once daily) or placebo. Patients in bothduloxetine treatment groups were started at 60mg/day, which could initially and temporarily be loweredto 30mg/day if tolerability concerns arose. Allpatients were required to be receiving a dose of atleast 60mg/day by the second-week visit.

Study 2 consisted of a 10-week acute therapy phase,followed by a 2-week discontinuation phase. After a1-week single-blind, placebo lead-in, patients wererandomly assigned (1:1) to receive either duloxetine(60–120mg once daily) or placebo. Patients inthe duloxetine treatment group were started at 60mg/day, which could initially and temporarily be lowered to30mg/day if tolerability concerns arose. All patientswere required to be receiving a dose of at least 60mg/day by the second-week visit. After titration to 60mg/day, flexible dosing was allowed in weekly incrementsof 30mg/day up to a maximum dose of 120mg/day.

Study 3 consisted of a 10-week acute therapy phase,followed by a 2-week discontinuation phase. Patientswere randomly assigned in a 1:1:1 ratio to duloxetine(60–120mg once daily), venlafaxine (75–225mg oncedaily), or placebo. Duloxetine treatment was initiatedat 30mg/day for one week, followed by an increase to60mg/day. After titration to 60mg/day, flexibledosing was allowed in weekly increments of 30mg/day up to a maximum dose of 120mg/day. Patients inthe venlafaxine arm were not included in this pooledanalysis.

Study 4 consisted of a 10-week acute therapy phase,followed by a 2-week tapering phase. Patients wererandomly assigned in a 2:1:2:2 ratio to duloxetine(60–120mg once daily), duloxetine (20mg once daily),venlafaxine (75–225mg once daily), or placebo.

Duloxetine treatment was initiated at 30mg/day forone week, followed by an increase to 60mg/day.After titration to 60mg/day, flexible dosing wasallowed in weekly increments of 30mg/day up to amaximum dose of 120mg/day. Patients in the 20mgduloxetine and venlafaxine arm were not includedin this pooled analysis, as these arms had too fewpatients� 65 years.For Studies 2–4, dose increases to maximize

efficacy were allowed based on investigator judgment;however, the protocols required that the dose beincreased if a patient’s CGI–improvement of illness(CGI-I; Guy 1976) score was� 3 (minimal improve-ment, no change, or worse) during the first four weeksof treatment, unless the patient was unable to toleratean increased dose. For the same studies, a total of twodownward dose adjustments for tolerability concernswere allowed, with a minimum allowable dose of60mg/day duloxetine.

Outcome measures

Efficacy assessments. In all four studies, the primaryefficacy measure was the HAMA total score. Theprimary objective was to determine if duloxetine wasstatistically significantly superior to placebo, asmeasured by change from baseline to endpoint inthe HAMA total score during the acute therapy phases.Secondary efficacy measures included the HADS; theCGI-I and Patient Global Impressions of Improvement(PGI-I; Guy, 1976) rating scales; and the functionaloutcome measure, the Sheehan disability scale (SDS;Sheehan, 1983) impairment scores. Response wasdefined as a� 50% reduction from baseline in HAMAtotal score, or a CGI-I score of� 2 at endpoint;sustained improvement was defined as a � 30%reduction from baseline in HAMA total score at anyvisit before endpoint, sustained until the last visit; andremission was defined as a HAMA total score of� 7 atendpoint.

Tolerability assessments. Data on treatment-emergentadverse events (TEAEs) were collected at baselineand at each visit. An adverse event was consideredtreatment-emergent if its onset occurred after random-ization, or if the event was present before but increasedin severity after randomization.

Statistical analyses

All randomly assigned patients� 65 years old wereincluded in the safety analyses, and all patients with abaseline and at least one postbaseline measurement


duloxetine treatment for elderly patients with gad 521

were included in the efficacy analyses. In all the fourstudies, patients assigned to receive duloxetine weretreated with duloxetine at either 60mg/day, 120mg/day, or dosages within that range; in this analysis, datafrom all duloxetine-treated patients aged � 65 yearswere pooled to form one treatment group.Treatment group comparisons on baseline clinical

and demographic variables were examined using x2

statistics for categorical variables and analysis ofvariance (ANOVA, with treatment and study as termsin the model) for continuous variables. For continuousefficacy variables (except CGI-I and PGI-I scores),treatment group differences were examined using ananalysis-of-covariance (ANCOVA) model, with treat-ment and study as main effects, and the baseline scoreas the covariate (last observation carried forward). TheCGI-I and PGI-I endpoint scores were analyzed usingan ANOVA model with treatment and study as fixedeffects. The categorical efficacy and tolerabilityvariables in the duloxetine and placebo treatmentgroups were compared using a Cochran–Mantel–Haenszel test, controlling for study.Treatment-emergent abnormally low and high

laboratory values were defined using Covance referenceranges (Princeton, NJ). Sustained elevation in bloodpressure was defined as a systolic blood pressure� 140mmHg and an increase of� 10mmHg from

baseline, and/or a diastolic blood pressure� 90mmHgand an increase of � 10mmHg from baseline forthree consecutive visits. Serious adverse events werethose events that resulted in death, initiated or prolongedinpatient hospitalization, resulted in a life-threateningexperience or a persistent or significant disability, orwere judged to be significant for any other reason.

For the efficacy measures, mean refers to least-squares mean. No adjustments for multiple compari-sons were made, as this is a post hoc analysis of asubset of a larger dataset that was not prospectivelydesigned to address this issue. Statistical comparisonswere based on 2-sided, 0.05 significance levels. SAS1

software version 8.2 (SAS Institute Inc., Cary, NorthCarolina) was used to perform all statistical analysesat Eli Lilly and Company.

RESULTS

Patient characteristics

Of the 1491 patients who were randomly assigned totreatments in the four studies, 73 (4.9%; duloxetine,n¼ 45; placebo, n¼ 28) were included in this pooledanalysis. The treatment groups did not significantlydiffer in any demographic variable or severity of illnessat baseline (Table 2). Thirty duloxetine- (66.7%) and 20

Table 2. Baseline characteristics of duloxetine-treated patients aged� 65 and<65 years with GAD

Patient characteristic

�65 years <65 years

Duloxetinea

(N¼ 45)Placebo(N¼ 28) p-valuea

Duloxetinea

(N¼ 781)Placebo(N¼ 637) p-valueb

Sex, female, n (%) 27 (60.0) 13 (46.4) 0.335 501 (64.2) 410 (64.4) 0.956Age, mean (SD), years 70.1 (4.3) 70.9 (5.1) 0.489 40.9 (11.7) 41.2 (12.3) 0.403Ethnic origin, n (%) 1.00 0.042

African descent 2 (4.4) 1 (3.6) 56 (7.2) 48 (7.5)Caucasian 43 (95.6) 27 (96.4) 637 (81.7) 498 (78.2)Eastern Asian 0 (0.0) 0 (0.0) 17 (2.2) 21 (3.3)Hispanic 0 (0.0) 0 (0.0) 58 (7.4) 67 (10.5)Native American 0 (0.0) 0 (0.0) 1 (0.1) 1 (0.2)Western Asian 0 (0.0) 0 (0.0) 11 (1.4) 2 (0.3)

HAMA, mean (SD)Total 20.9 (7.7) 23.3 (7.0) 0.173 24.6 (6.8) 25.6 (7.3) 0.723Psychic anxiety 12.9 (3.5) 13.9 (3.6) 0.187 14.3 (3.6) 14.6 (13.7) 0.792Somatic anxiety 8.0 (5.1) 9.3 (4.5) 0.276 10.4 (4.5) 11.0 (4.7) 0.732

HADS, mean (SD)Anxiety 11.1 (2.8) 12.1 (3.2) 0.436 13.4 (3.5) 13.7 (3.6) 0.490Depression 6.9 (3.2) 7.4 (2.3) 0.560 8.3 (4.0) 8.3 (4.0) 0.663

SDS Global functioning, mean (SD) 10.5 (7.1) 12.7 (8.0) 0.317 16.1 (7.0) 16.6 (7.1) 0.842

Some percentages do not add up to 100 due to rounding.Abbreviations: GAD, generalized anxiety disorder; HAMA, Hamilton anxiety scale; HADS, hospital anxiety depression scale; CGI-S,clinical global impressions–severity of illness; SDS, Sheehan disability scale.aDuloxetine doses were 60 or 120mg/day, or fell within that range.bFrequencies were analyzed using x2 test; means, using Type III sum of squares ANOVA, model¼ study treatment.



(71.4%) placebo-treated patients completed treatment(p¼ 0.610). Significantly more duloxetine-treatedpatients discontinued treatment due to an adverse event(10 (22.2%) vs. 0 (0.0%), p¼ 0.011), but the treatmentgroups did not differ significantly with respect to otherreasons for discontinuation. No single adverse eventled to discontinuation significantly more often induloxetine- than in placebo-treated patients (Table 3).

Efficacy

Patients treated with duloxetine had statistically greaterbaseline-to-endpoint improvements on the primary

efficacy measure, the HAMA total score, comparedwith patients treated with placebo (�10.1 vs. �5.9,p¼ 0.029). Duloxetine-treated patients also had stat-istically greater improvements, compared with placebo-treated patients, on the HAMA-psychic anxiety factor(p¼ 0.034), the HADS anxiety (p¼ 0.049), anddepression scales (p¼ 0.026), but not the HAMAsomatic anxiety factor (p¼ 0.074; Table 4). Thetreatment groups did not significantly differ in CGI-I(p¼ 0.143) or PGI-I scores at endpoint (p¼ 0.064).Lastly, the treatment groups did not significantly differin the incidence of HAMA response (20/42 (48%) vs.8/28 (29%), p¼ 0.149); and approached but did not

Table 3. Reasons for discontinuation of patients aged� 65 years with GAD treated with duloxetine or placebo pooled from four studies

Reason for discontinuationDuloxetine

(N¼ 45) n (%)aPlacebo

(N¼ 28) n (%) p-valueb

For any reason 15 (33.3) 8 (28.6) 0.610Adverse event 10 (22.2) 0 (0.0) 0.006Nausea 3 (6.7) 0 (0.0) 0.103Anxiety 1 (2.2) 0 (0.0) 0.394Dizziness 3 (6.7) 0 (0.0) 0.103Orthostatic hypotension 1 (2.2) 0 (0.0) 0.491Somnolence 2 (4.4) 0 (0.0) 0.215Urinary hesitation 1 (2.2) 0 (0.0) 0.394Vomiting 1 (2.2) 0 (0.0) 0.394

Lack of efficacy 1 (2.2) 2 (7.1) 0.324Lost to follow up 1 (2.2) 0 (0.0) 0.386Patient decision 2 (4.4) 4 (14.3) 0.114Protocol violation 1 (2.2) 2 (7.1) 0.298

Abbreviation: GAD, generalized anxiety disorder.aIncludes patients randomly assigned to duloxetine treatment who did not begin treatment.bCochran–Mantel–Haenzsel test controlling for study.

Table 4. Least-squares mean changes on efficacy measures and least-squares mean endpoint improvement scores in patients aged� 65years with GAD treated with duloxetine or placebo pooled from four studies

ScaleDuloxetine (n¼ 42) Placebo (n¼ 28)

t-score, p-valueMean S.E. Mean S.E.

Changes from baseline to endpointHAMATotala �10.1 1.2 �5.9 1.5 t1¼�2.2, p¼ 0.029Psychic anxietya �6.1 0.8 �3.6 0.9 t1¼�2.2, p¼ 0.034Somatic anxietya �4.0 0.6 �2.4 0.7 t1¼�1.8, p¼ 0.074

HADSAnxietya �4.1 0.6 �2.2 0.7 t1¼�2.0, p¼ 0.049Depressiona �2.6 0.5 �0.8 0.6 t1¼�2.3, p¼ 0.026

SDS global functioninga �4.1 1.2 �2.2 1.5 t1¼�1.0, p¼ 0.324Scores at endpointCGI-Ib 2.6 0.2 3.1 0.3 t1¼�1.5, p¼ 0.143PGI-Ib 2.7 0.2 3.4 0.3 t1¼�1.9, p¼ 0.064

Abbreviations: GAD, generalized anxiety disorder; HAMA, Hamilton anxiety scale; HADS, hospital anxiety depression scale; SDS,Sheehan disability scale; CGI-I, clinical global impressions–improvement of illness; PGI-I, patient global impressions of improvement.ap-values were derived using Type III sum of squares ANOVA, model¼ treatment study baseline.bp-values were derived using Type III sum of squares ANOVA, model¼ treatment study.



achieve significance for remission (10/42 (24%) vs. 2/28 (7%), p¼ 0.053), or sustained improvement (26/42(62%) vs. 10/28 (36%), p¼ 0.05).

Tolerability

One duloxetine-treated patient experienced the seriousadverse event of anxiety (none were reported in theplacebo group). Significantly more duloxetine-treatedpatients experienced nausea (15/50 (30.0%) vs. 2/28(7.1%), p¼ 0.023). The incidence of other TEAEs didnot significantly differ between the treatment groups(Table 5), with the exception of nasopharyngitis inplacebo-treated patients (p¼ 0.014). The incidence ofabnormal laboratory parameters at endpoint did notsignificantly differ between the treatment groups.Duloxetine-treated patients lost significantly more

baseline-to-endpoint weight compared with placebo-treated patients (�1.1 vs.þ0.0 kg, p¼ 0.018). Thetreatment groups did not significantly differ in meanbaseline-to-endpoint changes in pulse (p¼ 0.976),supine systolic blood pressure (p¼ 0.831), or supinediastolic blood pressure (p¼ 0.317).

DISCUSSION

Compared with elderly patients treated with placebo,elderly patients treated with duloxetine experiencedsignificantly greater baseline-to-endpoint improve-ments on the primary efficacy measure (HAMA total),the HAMA psychic anxiety factor, and the HADSanxiety and depression subscales compared withpatients treated with placebo. The treatment groupsdid not significantly differ in improvement on theHAMA somatic anxiety factor. The results are some-

what similar to the overall findings of the duloxetinestudies from which these patients’ data were pooled(Hartford et al., 2007; Koponen et al., 2007; Rynnet al., 2008; Nicolini et al., 2008), which demonstratedsignificant changes on most studied anxiety measuresin duloxetine-treated patients. These results are alsosomewhat similar to those observed in studies ofelderly patients with MDD (Raskin et al., 2007;Russell et al., 2007; Wise et al., 2007). The treatmentresponse seen on the primary efficacy measure, theHAMA total score, was comparable with thosereported in the four independent studies from whichthe data were pooled (Hartford et al., 2007; Koponenet al., 2007; Rynn et al., 2008; Nicolini et al., 2008); itis possible that the failure to achieve statisticaldifferences between treatment groups on the dichot-omous response and remission rate outcomes may bein part attributable to the relatively small sample size.

The types and incidence of adverse events reportedin our pooled analysis of the elderly patients weresimilar to those in the source studies: in all the fourstudies, nausea was the most common adverse event(Hartford et al., 2007; Koponen et al., 2007; Rynnet al., 2008; Nicolini et al., 2008). The greaterlikelihood of polypharmacy and altered drug metab-olism (Linjakumpu et al., 2002) in the elderly suggestscaution in extrapolating conclusions regarding safetyand tolerability of duloxetine beyond this studypopulation, in which concomitant medication usewas restricted.

This pooled analysis has several limitations inaddition to the small sample size. Although the foursource studies were similar in design, there were somedifferences, including study duration (9 or 10 weeks)and dosing schedules (fixed vs. flexible). These

Table 5. Frequency of treatment-emergent adverse events occurring in� 5% of duloxetine-treated patients aged� 65 years with GADpooled from four studies

Adverse eventDuloxetine,

N¼ 50a n (%)Placebo,

N¼ 28 n (%) p-valueb

�1 TEAE 35 (70.0) 15 (53.6) 0.218Nausea 15 (30.0) 2 (7.1) 0.023Dizziness 7 (14.0) 5 (17.9) 0.747Hyperhidrosis 6 (12.0) 3 (10.7) 1.00Dry mouth 6 (12.0) 1 (3.6) 0.411Constipation 5 (10.0) 4 (14.3) 0.715Diarrhea 4 (8.0) 0 (0.0) 0.291Fatigue 4 (8.0) 1 (3.6) 0.649Somnolence 3 (6.0) 0 (0.0) 0.549Vomiting 3 (6.0) 1 (3.6) 1.00

Abbreviations: GAD, generalized anxiety disorder; TEAE, treatment-emergent adverse event.aIncludes patients randomly assigned to duloxetine treatment who did not begin the treatment.bCochran–Mantel–Haenszel test, controlling for study.



short-term studies also required a forced titration tothe highest tolerated dose, which could have lead tomore discontinuations than would be seen in clinicalpractice. This analysis also included only short-termacute data, whereas new efficacy and safety signals maybecome evident only after months of treatment. Anadditional limitation is the strict inclusion/exclusioncriteria of the source studies, such as comorbid majordepressive, post-traumatic stress, and substance abusedisorders limiting the generalizability of these find-ings. Finally, these analyses are post hoc in nature, andthe initial studies were not designed with this analysisin mind; the results should be interpreted with thecontext of these limitations.

Elderly patients treated with duloxetine experiencedimprovement in the overall symptoms of anxiety,and some secondary symptoms of psychic anxiety anddepression. Given the high rate of discontinuations dueto adverse events (�22%), the issue of the tolerability ofduloxetine in this sensitive population should beexplored further. Duloxetine may be a useful treatmentoption for GAD in the elderly patient population.

ACKNOWLEDGEMENTS

This work was sponsored by Eli Lilly and Company andBoehringer Ingelheim. The authors thank the patientswho agreed to participate in the clinical studies, as wellas the clinical staff, and principal investigators. We alsothank the Duloxetine Antidepressant Team for theirefforts in the execution of the four studies; MelissaSpann, Wenqi You, Huifang Chen, and Steve Gelwicksfor their assistance with the statistical analyses in thismanuscript; Susan Ball and Daniel Walker for theirassistance with the development of this project; andSvetlanaDominguez for her editorial assistancewith themanuscript.

REFERENCES

Allgulander C, Lavori PW. 1991. Excess mortality among 3302patients with ’pure’ anxiety neurosis. Arch Gen Psychiatry 48:599–602.

Allgulander C, Lavori PW. 1993. Causes of death among 936 elderlypatients with ’pure’ anxiety neurosis in StockholmCounty, Sweden,and in patients with depressive neurosis or both diagnoses.ComprPsychiatry 34: 299–302.

Alwahhabi F. 2003. Anxiety symptoms and generalized anxietydisorder in the elderly: a review. Harv Rev Psychiatry 11: 180–193.

American Psychiatric Association (APA). 2000. Diagnostic andStatistical Manual of Mental Disorders, 4th edition, text revision(DSM-IV-TR).

Beekman AT, de Beurs E, van Balkom AJ, Deeg DJ, van Dyck R,van Tilburg W. 2000. Anxiety and depression in later life:

co-occurrence and communality of risk factors. Am J Psychiatry157: 89–95.

Blazer D, Burchett B, Service C, George LK. 1991. The associationof age and depression among the elderly: an epidemiologicexploration. J Gerontol 46: M210–M215.

Bogunovic OJ, Greenfield SF. 2004. Practical geriatrics: use ofbenzodiazepines among elderly patients. Psychiatr Serv 55:233–235.

Copeland JR, Dewey ME, Wood N, Searle R, Davidson IA,McWilliam C. 1987. Range of mental illness among the elderlyin the community. Prevalence in Liverpool using theGMS-AGECAT package. Br J Psychiatry 150: 815–823.

Flint AJ, Rifat SL. 1997. Anxious depression in elderly patients.Response to antidepressant treatment. Am J Geriatr Psychiatry 5:107–115.

Guy W. 1976. ECDEU Assessment Manual for Psychopharma-cology, Revised (DHEW Publ No ADM 76–338). U.S. Depart-ment of Health, Education, and Welfare, Public Health Service,Alcohol, Drug Abuse, and Mental Health Administration, NIMHPsychopharmacology Research Branch, Division of ExtramuralResearch Programs, Rockville, MD.

Hamilton M. 1959. The assessment of anxiety states by rating. Br JMed Psychol 32: 50–55.

Hartford J, Kornstein S, Liebowitz MR, et al. 2007. Duloxetine as anSNRI treatment for generalized anxiety disorder: results from aplacebo and active-controlled trial. Int Clin Psychopharmacology22: 167–174.

Himmelfarb S, Murrell SA. 1984. The prevalence and correlates ofanxiety symptoms in older adults. J Psychol 116: 159–167.

Koponen H, Allgulander C, Erickson J, et al. 2007. Efficacy ofduloxetine for the treatment of generalized anxiety disorder:implications for primary care physicians. Prim Care CompanionJ Clin Psychiatry 9: 100–107.

Le Roux H, Gatz M,Wetherell JL. 2005. Age at onset of generalizedanxiety disorder in older adults. Am J Geriatr Psychiatry 13:23–30.

Lenze EJ, Mulsant BH, Shear MK, et al. 2000. Comorbid anxietydisorders in depressed elderly patients. Am J Psychiatry 157:722–728.

Linjakumpu T, Hartikainen S, Klaukka T, Veijola J, Kivela SL,Isoaho R. 2002. Use of medications and polypharmacy areincreasing among the elderly. J Clin Epidemiol 55: 809–817.

Lipman RS, Covi L, Downing RW, et al. 1981. Pharmacotherapy ofanxiety and depression: rationale and study design. Psychophar-macol Bull 17: 91–103.

Lipsitz J, Kobak K, Feiger A, Sikich D, Moroz G, Engelhard A.2004. The rater applied performance scale: development andreliability. Psychiatry Res 127: 147–155.

Nicolini H, Bakish D, Duenas H, et al. 2008. Improvement ofpsychic and somatic symptoms in adult patients with generalizedanxiety disorder: examination from a duloxetine, venlafaxineextended-release, and placebo-controlled trial. Psychol Med.(in press).

Raskin A, Schulterbrandt J, Reating N, McKeon JJ. 1969. Replica-tion of factors of psychopathology in interview, ward behaviorand self-report ratings of hospitalized depressives. J Nerv MentDis 148: 87–98.

Raskin J, Wiltse CG, Siegal A, et al. 2007. Efficacy of duloxetine oncognition, depression, and pain in elderly patients with majordepressive disorder: an 8-week, double-blind, placebo-controlledtrial. Am J Psychiatry 164: 900–909.

Russell JM,Weisberg R, FavaM, Hartford JT, Erickson JS, D’SouzaDN. 2007. Efficacy of duloxetine in the treatment of generalized



anxiety disorder in patients with clinically significant pain symp-toms. Depress Anxiety Jun 22, [Epub ahead of print].

Rynn M, Russell J, Erickson J, et al. 2008. Efficacy and safety ofduloxetine in the treatment of generalized anxiety disorder: aflexible-dose, progressive-titration, placebo-controlled trial.Depress Anxiety Feb 20, [Epub ahead of print].

Schneider LS. 1996. Overview of generalized anxiety disorder in theelderly. J Clin Psychiatry 57: 34–45.

Shear MK, Vander BJ, Rucci P, et al. 2001. Reliability and validityof a structured interview guide for the Hamilton anxiety ratingscale (SIGH-A). Depress Anxiety 13: 166–178.

Sheehan DV. 1983. Sheehan disability scale. In Handbook ofPsychiatric Measures. American Psychiatric Association:Washington, DC; 113–115.

Sheehan DV, Lecrubier Y, Harnett-Sheehan KH, et al. 1998. Themini-international neuropsychiatric interview (M.I.N.I.): the

development and validation of a structured diagnostic psychiatricinterview for DSM-IV and ICD-10. J Clin Psychiatry 59: 22–33.

Turnheim K. 2003. When drug therapy gets old: pharmacokineticsand pharmacodynamics in the elderly. Exp Gerontol 38: 843–853.

Wetherell JL, Thorp SR, Patterson TL, Golshan S, Jeste DV,Gatz M. 2004. Quality of life in geriatric generalized anxietydisorder: a preliminary investigation. J Psychiatr Res 38: 305–312.

Wise TN, Wiltse CG, Iosifescu DV, Sheridan M, Xu JY, Raskin J.2007. The safety and tolerability of duloxetine in depressedelderly patients with and without medical comorbidity. Int J ClinPract 61: 1283–1293.

Zigmond AS, Snaith RP. 1983. The hospital anxiety and depressionscale. Acta Psychiatr Scand 67: 361–370.



Documents

Efficacy and tolerability of duloxetine in elderly patients with generalized anxiety disorder: a pooled analysis of four randomized, double-blind, placebo-controlled studies