Cognitive Behavioral Therapy vs Zopiclone for … · ORIGINAL CONTRIBUTION Cognitive Behavioral Therapy vs Zopiclone for Treatment of Chronic Primary Insomnia in Older Adults A Randomized

ORIGINAL CONTRIBUTION

Cognitive Behavioral Therapy vs Zopiclonefor Treatment of Chronic Primary Insomniain Older AdultsA Randomized Controlled TrialBørge Sivertsen, PsyDSiri Omvik, PsyDStale Pallesen, PhDBjørn Bjorvatn, MD, PhDOdd E. Havik, PhDGerd Kvale, PhDGeir Høstmark Nielsen, PsyDInger Hilde Nordhus, PhD

INSOMNIA IS USUALLY DEFINED AS SUB-jective complaints of poor sleep ac-companied by impairment in day-time function according to the

Diagnostic and Statistical Manual of Men-tal Disorders, Fourth Edition, compris-ing complaints of insufficient sleep, in-terrupted sleep, difficulty in initiating ormaintaining sleep, and poor-quality ornonrestorative sleep.1 Insomnia is com-mon in people older than 55 years (9%-25%2-5) and is associated with reducedquality of life,6,7 affective disorders,8 andincreased health service utilization.9 A re-cent analysis of the economic burden ofinsomnia in the United States estimatesthe direct medical costs to be $13.9 bil-lion annually.10 Despite these links to in-dividuals’ lives and societal costs, mostpeople with chronic insomnia—up to85%—remain untreated.11,12 Two thirdsof individuals with insomnia report poorknowledge of available treatment op-tions, and as many as one fifth resort toeither untested over-the-counter medi-cations or alcohol in attempts to im-prove their condition.13

Among primary care physicians, thetreatment of choice for insomnia hascommonly been pharmacological inter-

vention.14,15 The short-term efficacy ofsleep medications has been demon-strated in numerous studies.16,17 How-

See also Patient Page.

Author Affiliations: Department of Clinical Psychol-ogy (Drs Sivertsen, Omvik, Havik, Kvale, Nielsen, andNordhus); Department of Psychosocial Science (DrPallesen); Norwegian Competence Center for SleepDisorders (Drs Bjorvatn, Pallesen, and Nordhus); andDepartment of Public Health and Primary Health

Care (Dr Bjorvatn), University of Bergen, Bergen,Norway.Corresponding Author: Børge Sivertsen, PsyD, De-partment of Clinical Psychology, University of Ber-gen, Christiesgt 12, 5015 Bergen, Norway ([email protected]).

Context Insomnia is a common condition in older adults and is associated with a num-ber of adverse medical, social, and psychological consequences. Previous research hassuggested beneficial outcomes of both psychological and pharmacological treat-ments, but blinded placebo-controlled trials comparing the effects of these treat-ments are lacking.

Objective To examine short- and long-term clinical efficacy of cognitive behavioraltherapy (CBT) and pharmacological treatment in older adults experiencing chronic pri-mary insomnia.

Design, Setting, and Participants A randomized, double-blinded, placebo-controlled trial of 46 adults (mean age, 60.8 y; 22 women) with chronic primary in-somnia conducted between January 2004 and December 2005 in a single Norwegianuniversity-based outpatient clinic for adults and elderly patients.

Intervention CBT (sleep hygiene, sleep restriction, stimulus control, cognitive therapy,and relaxation; n=18), sleep medication (7.5-mg zopiclone each night; n=16), or pla-cebo medication (n=12). All treatment duration was 6 weeks, and the 2 active treat-ments were followed up at 6 months.

Main Outcome Measures Ambulant clinical polysomnographic data and sleep dia-ries were used to determine total wake time, total sleep time, sleep efficiency, andslow-wave sleep (only assessed using polysomnography) on all 3 assessment points.

Results CBT resulted in improved short- and long-term outcomes compared withzopiclone on 3 out of 4 outcome measures. For most outcomes, zopiclone did not dif-fer from placebo. Participants receiving CBT improved their sleep efficiency from 81.4%at pretreatment to 90.1% at 6-month follow-up compared with a decrease from 82.3%to 81.9% in the zopiclone group. Participants in the CBT group spent much more timein slow-wave sleep (stages 3 and 4) compared with those in other groups, and spentless time awake during the night. Total sleep time was similar in all 3 groups; at 6 months,patients receiving CBT had better sleep efficiency using polysomnography than thosetaking zopiclone.

Conclusion These results suggest that interventions based on CBT are superior to zopi-clone treatment both in short- and long-term management of insomnia in older adults.

Trial Registration clinicaltrials.gov Identifier: NCT00295386JAMA. 2006;295:2851-2858 www.jama.com

©2006 American Medical Association. All rights reserved. (Reprinted) JAMA, June 28, 2006—Vol 295, No. 24 2851

at N/A, on June 28, 2006 www.jama.comDownloaded from

http://www.jama.com

ever, a recent meta-analysis of sleepmedications in older patients withinsomnia concluded that effect sizes andclinical benefits both were small.18 Fur-thermore, a consensus statement by theNational Institutes of Health19 con-cluded that, while pharmacologicaltreatments may be useful for acute andsituational insomnia, long-term useinvolves risks of dependency and tol-erance. There is also some current con-troversy in the media about the increas-ing use of zolpidem in the United States,related to next-day sleepiness and traf-fic collisions.20,21

Cognitive behavioral therapy (CBT)is the most widely used psychologicalintervention for insomnia. Three meta-analyses22-24 haveconcluded that70%to80% of middle-aged adults with insom-nia benefit from interventions based onCBT, and a study by Jacobs and col-leagues suggested that young- andmiddle-aged individuals with insomniaimprovemore fromCBTcomparedwithpharmacotherapy.25 However, a Coch-rane review stated that there is limitedevidence available suggesting a cleareffect of CBT on insomnia in olderadults.26 Comparedwithplaceboandnotreatment,CBTwasassociatedwithsleepimprovements,27-29 but effects were notas great as for CBT used with youngeradults.29Also,arecentmeta-analysiscon-cludedthatbehavioralinterventionsweremore effective in the younger cohort inimprovingbothtotalsleeptimeandsleepefficiency.22 However, all studies in thismeta-analysis were based on self-reportor actigraphy. Surprisingly, only 1 ran-domized controlled clinical trial, byMorinandcolleagues,30hasdirectlycom-pared the clinical efficacy of both sleepmedication (temazepam) and psycho-logical interventions in older patientswithprimary insomnia.This studydem-onstrated that CBT and pharmacologi-cal interventionsproducedsimilarshort-term effects but that CBT was superiorat follow-up as indicated by sleep dia-ries.However,nostudieshavecomparedthe newer nonbenzodiazepine sleepmedications with nonpharmacologicaltreatments and, to our knowledge, nostudies have examined whether CBT af-

fects slow-wave sleep (stages 3 and 4)in treating insomnia. This is of particu-lar interest because lack of slow-wavesleep has been believed responsible formuch of the daytime impairment expe-rienced by patients with insomnia.31

The present study was a randomizedcontrolled trial conducted to evaluate theshort- and long-term clinical efficacy ofboth CBT and the non-benzodiazepinesleep medication zopiclone. In contrastto previous research, we used both poly-somnography (PSG) and sleep diaries atall 3 assessment points including follow-up. Providing independent estimates ofsleep and wake time in addition to clas-sification of sleep stages, the inclusion ofPSG at follow-up was of particular im-portance because patients’ subjective per-ceptions of actual sleep time have devi-ated from PSG-based recordings.32 Wealso wanted to compare the treatmentconditions in their ability to improveslow-wave sleep.

METHODSParticipants

Participants were recruited throughnewspaper advertisements that stated theaim of the study as comparing the ef-fects of sleep medications with psycho-logical treatment. No additional infor-mation about the study hypothesis ortype of interventions was provided.

Inclusion criteria were that partici-pants (1) be 55 years or older; (2) ful-fill the Diagnostic and Statistical Manualof Mental Disorders, Fourth Edition cri-teria for insomnia, including difficul-ties initiating sleep, maintaining sleep,and/or early morning awakenings withno ability of return to sleep; (3) haveinsomnia duration of at least 3 monthsinsomnia; and (4) complain of im-paired daytime functioning.

The following exclusion criteria wereused: (1) use of hypnotic medication inthe last 4 weeks before project incep-tion; (2) use of antidepressive or anti-psychotic medications; (3) signs of de-mentia or other serious cognitiveimpairment defined by a score of lessthan 23 on the Mini-Mental State Ex-amination33; (4) presence of a major de-pressive disorder or other severe men-

tal disorder as identified by a clinicalassessment based on the StructuredClinical Interview for the Diagnostic andStatistical Manual of Mental Disorders,Fourth Edition34; (5) presence of sleep ap-nea defined as apnea-hypopnea indexgreater than 15 or periodic limb move-ments during sleep (PLM index witharousal �15); (6) working night shiftsand unable or unwilling to discontinuethis work pattern; (7) unwillingness orinability to stop taking sleep medica-tion before study participation; or (8)having a serious somatic condition pre-venting further participation.

Procedure

Participants who responded to the ad-vertisement (N=92) underwent severalscreeningprocessesbefore theywere in-cluded.A15-minutetelephoneinterviewby 2 clinical psychologists ensured thatparticipants fulfilled thebasiccriteria forinclusion.Acceptedparticipants(n=75)met at the Department of Clinical Psy-chology,UniversityofBergen, forastruc-turedclinical interview(SCID-I) screen-ing for severepsychopathologyandcog-nitive impairment. The final screeningphase included 2 consecutive nights ofambulant polysomnography. Random-ization was performed by the projectleader using blocks of 3 with no strati-fication. After 12 participants had beenassigned to each of the 3 conditions,blocks of 2 were used to randomly as-sign the remaining participants into 1of the active treatments (CBT or zopi-clone). Allocation concealment wasimplemented using sealed, sequentiallynumbered boxes that were identical inappearance for the 3 treatment groups.All study personnel in contact with theparticipants were unaware of the ran-domization sequence. Double-blindingwasachievedwithpillswith identicalap-pearance, smell, and flavor containingeitherzopicloneorplacebo.Inall,48par-ticipants were randomized into eitherCBT(n=18),hypnotics (7.5mgofzopi-clone each night; n=18), or pharmaco-logicalplacebo treatment (n=12).How-ever, 2 participants withdrew from thezopiclone condition immediately afterrandomization and were excluded from

COGNITIVE BEHAVIORAL THERAPY VS PHARMACOTHERAPY FOR CHRONIC PRIMARY INSOMNIA

2852 JAMA, June 28, 2006—Vol 295, No. 24 (Reprinted) ©2006 American Medical Association. All rights reserved.


http://www.jama.com

themodifiedintent-to-treatanalysis.Theflowchart in FIGURE 1 outlines the de-sign of the study.

Instruments

PSG. Sleep variables were assessed byambulant clinical PSG performed in theparticipants’ homes. The PSG montageincluded electroencephalographic,electromyographic, and electro-oculographic monitoring.

Sleep stages, respiratory distur-bances, and limbmovementswere scoredaccording to standard criteria35 by 2 tech-nicians blinded to the participants’ con-dition. Respiration (air flow, tidal vol-ume,andoxygensaturation)andanteriortibialis electromyographic readings wererecorded to detect sleep apnea or peri-odic limb movements. Participants un-derwent 2 consecutive nights of PSG atpretreatment to allow for adaptation tothe PSG montage. At both posttreat-ment and follow-up assessment, only 1night of PSG was recorded because re-cent studies have demonstrated that theso-called “first night effect” is only pres-ent in the first night in the first assess-ment period.36 All electrophysiologicalsignals were acquired using Embla A10(Flaga-Medcare Somnologica 3.2 soft-ware package, Buffalo, NY).

The sleep outcome measures in-cluded total wake time (summation ofsleep-onset latency, wake time after sleeponset, and early morning awakening),total sleep time, sleep efficiency (ratio oftotal time spent asleep to the actual timespent in bed, multiplied by 100), andslow-wave sleep (time spent in sleepstages 3 and 4 registered by PSG).

Sleep Diaries. Participants com-pleted sleep diaries37 every morning for2 weeks at all 3 assessment points. Thesleep diary provided self-reported in-formation about the same sleep param-eters collected from PSG registration.To increase the reliability, data analy-sis was based on the mean scores com-piled during the 2-week period.

Treatment Conditions. CBT. Par-ticipants receiving CBT attended 6weekly individual treatment sessions,with each lasting approximately 50 min-utes. The rationale of this treatment

condition is based on a manualizedmulticomponent approach that in-cludes several modules introduced atdifferent stages in the treatment pro-cess.38 These components include sleephygiene education,39 sleep restric-tion,40 stimulus control,41 cognitive

therapy,42 and progressive relaxationtechniques.43 TABLE 1 provides an over-view of the treatment principles in theCBT condition. The therapy sessionswere facilitated by 2 clinical psycholo-gists (B.S. and S.O.) and administeredat the outpatient university clinic be-

Figure 1. Participant Flow in the Study

27 Excluded16 Sleep Apnea or Periodic

Limb Movements3 Psychiatric Condition1 Not Able to Stop Taking

Sleep Medication4 Spontaneous Improvement3 Withdrew

92 Potential ParticipantsScreened for Eligibility

75 Met Basic Inclusion Criteriafor Pretreatment Assessment

48 Randomized

18 Included in 6 mo Follow-up Analysis

16 Included in 6 mo Follow-up Analysis∗

2 Lost to Follow-up at 6 mo 5 Lost to Follow-up at 6 mo3 Withdrew2 Discontinued Treatment

(Adverse Effects)

18 Included in 6 wkPosttreatment Analysis

16 Included in 6 wk Posttreatment Analysis∗

12 Included in 6 wkPosttreatment Analysis

18 Completed Treatment Protocol 12 Completed Treatment Protocol15 Completed Treatment Protocol1 Discontinued Intervention

(Adverse Effects)

18 Assigned to Receive CognitiveBehavioral Therapy18 Received Intervention

as Assigned

12 Assigned to Receive Placebo12 Received Placebo

as Assigned

18 Assigned to Receive Zopiclone16 Received Intervention

as Assigned2 Withdrew Immediately After

Randomization

*Only 15 participants were included in polysomnographic analysis and 1 was excluded due to invalid poly-somnography data.

Table 1. Overview of Principles in Treatment Modules Included in the Cognitive BehavioralTherapy Condition

Module Description

Sleep hygieneeducation

The patient learns about the impact of lifestyle habits such as exercise;diet and alcohol use; and the influence of environmental factors suchas light, noise, and temperature

Sleep restriction Involves a strict schedule of bedtimes and rising times, restricting patients’allowed time in bed to the actual sleeping time according to thepatients’ sleep diary; the aim is to increase homeostatic sleep drivethrough partial sleep deprivation

Stimulus control The aim is to break associations between the sleep environment andwakefulness by teaching the participant not to engage in bedroomactivities incompatible with sleep and to stay in the bedroom onlywhen asleep or sleepy

Cognitive therapy The objective is to identify, challenge, and replace beliefs and fearsregarding sleep or the loss of sleep with realistic expectationsregarding sleep and daytime function

Progressive relaxationtechnique

The patient is taught how to recognize and control muscular tensionthrough use of exercise instructions on prerecorded tape or compactdisc, and to practice the technique at home on a daily basis




http://www.jama.com

tween March 2004 and June 2005. Be-cause both active treatment condi-tions were administered equally during14 to 15 months, seasonal variations indaylight were not likely to have con-founded the findings in the presentstudy. Due to the nature of CBT, nei-ther therapists nor participants wereblinded to it.

Zopiclone. Zopiclone, first intro-duced in 1988, is a cyclopyrrolone de-rivative that is chemically unrelated tobenzodiazepines or barbiturates. Zopi-clone works by enhancing the actions ofthe neurotransmitter �-aminobutyricacid and is a racemic mixture of 2 ste-reoisomers, only 1 of which is active. Theactive stereoisomer, eszopiclone, was in-troduced on the US market in April 2005,and although the dosages are different(7.5 mg of zopiclone is equivalent toabout 3.75 mg of eszopiclone), the 2drugs are identical in effect. Zopiclonehas demonstrated efficacy equivalent toand in some cases greater than both long-and short-acting benzodiazepines.44,45

Zopiclone is documented to be well-tolerated in elderly patients and is gen-erally less likely to produce adverse ef-fects than benzodiazepines.45 Zopiclonewas chosen because it has been the mostcommonly prescribed hypnotic in Nor-way during the last decade, and overallthis hypnotic agent has a market share

of 45% of the total sales of hypnotics andtranquilizers in Norway.46

Participants in the active sleep medi-cation group were administered 7.5 mgof zopiclone by a physician. Partici-pants met at the sleep laboratory ev-ery week for a 10-minute meeting to re-port any adverse effects and to obtainthe following week’s dosage of 7 pills.No behavioral recommendations re-garding sleep were given during theseshort meetings, and the main focus wason encouraging the participants to ad-here to the treatment program. Aftertreatment completion, the patients weregiven the opportunity to continue theirmedication for 6 additional months.

Placebo. Participants receiving pla-cebo treatment were subject to the sametreatment protocol as those in the ac-tive medication group. As with zopi-clone, the placebo capsules were madeof gelatin and there were no differ-ences in appearance, smell, or flavor be-tween the active and inactive pills. Af-ter 6 weeks, participants in the placebogroup were immediately randomizedinto 1 of the 2 active treatment condi-tions. Thus, the present study pro-vides no follow-up data after 6 monthsfor the placebo condition. Data fromparticipants who received an activetreatment following the placebo con-dition were not considered in the

6-month statistical comparisons. Therationale for omitting these data was tocompare solely the effects of CBT vszopiclone without including partici-pants who had received both placebomedication and subsequently eitherzopiclone or CBT. Because the zopi-clone and placebo medications were ad-ministered in a standard double-blindfashion, neither the patients nor thetherapists should have known whetherthe patient received the active or inac-tive medication, but patients and thera-pists were not asked to guess to whichtreatment patients were randomized.

Ethics

The study was approved by the NationalData Inspectorate, the Norwegian Medi-cines Agency, and the Regional Com-mittee for Medical Research Ethics inwestern Norway. Written informed con-sent was obtained from all participantsincluded in this study. Participantsreceivednopayment toparticipate in thestudy, and they were informed that theycould withdraw from the study at anytime without stating the reason.

Statistics

We used SPSS statistical software (SPSSInc, Chicago, Ill) for Windows 13 for allstatistical analyses. Analysis of variancewith Bonferroni posthoc comparison orPearson �2 tests were used to examinedemographicandclinicalvariablesatpre-treatment. Modified intention-to-treatanalyses(excludingthe2individualsran-domized to zopiclone who withdrewbefore the study began) based on endpoint data were used throughout thestudy. Pretreatment data were broughtforwardandusedas6-weekand6-monthdata for participants who dropped outduringtreatment(n=1),whereas6-weekdata were used as follow-up for indi-viduals lost during 6-month follow-up(n=7). A 2�3 (time� intervention)analysis of covariance analysis with Bon-ferroni-corrected post hoc comparisonswas used to investigate differencesbetween the interventions in terms oftreatment effects. Analysis of covari-ance was also used to examine the treat-ment effects at 6-month follow-up, and

Table 2. Pretreatment Demographic and Clinical Variables of Participants Included in the Study

Characteristic

CognitiveBehavioralTherapy(n = 18)

Zopiclone(n = 16)

Placebo(n = 12)

Total(N = 46)

PValue*

Age, mean (SD) 59.8 (4.3) 61.3 (6.9) 61.8 (5.0) 60.8 (5.4) .58Sex, women/men 7/11 6/10 9/3 22/24 .09Education, mean years (SD) 14.2 (2.1) 13.7 (2.1) 14.6 (3.3) 14.1 (2.4) .67Insomnia duration, mean y (range) 15.8 (2-43) 13.7 (3-35) 12.0 (1-30) 14.1 (1-43) .66Current smoker, No. (%) 7 (38.9) 10 (62.5) 3 (25.0) 20 (43.5) .12More than 2 cups coffee

daily, No. (%)6 (33.3) 5 (31.3) 4 (33.3) 15 (32.6) .99

Body mass index, mean (SD)† 30.8 (6.6) 32.3 (8.7) 29.9 (5.1) 31.1 (7.0) .70Previously treated for

insomnia, No. (%)12 (66.7) 4 (25.0) 4 (33.3) 20 (43.5) .04

Self-reported chroniccondition, No. (%)‡

7 (38.9) 9 (56.3) 3 (25) 19 (41.3) .24

Currently taking othermedication, No. (%)§

6 (33.3) 4 (25.0) 9 (75.0) 19 (41.3) .02

*P values are based on analysis of variance or Pearson �2 test.†Calculated as weight in kilograms divided by height in meters squared.†Heart disease, hypertension, chronic pain, urinary tract problems, headache, migraine.§Other than sleep medication or antidepressive or antipsychotic medications.




http://www.jama.com

paired-sample t tests were used to com-pare theposttreatmentandfollow-uplev-els. The clinical significance of the treat-ment effects was estimated by theproportion of participants who reachedPSG-recorded sleep efficiency level of atleast 85%,30 and Pearson �2 tests wereusedtotest forgroupdifferences.Within-group effect sizes (pooled SD) were cal-culated using the Cohen d formula.47

Multicomponent CBT-based treat-ments for older adults with insomniahave been shown to yield effect sizesvarying up to 2.0 on some sleep vari-ables, such as wake time after sleep on-set,29 while pharmacotherapy typi-cally yields effect sizes of approximately0.9 on most sleep variables.48 Hence, ex-pecting a difference between multicom-ponent CBT and pharmacotherapy at

posttreatment and follow-up equiva-lent to an effect size of approximately1.0, with a power of 80% at P=.05, thenumber of participants needed in eachgroup was estimated to be 17.

RESULTSPretreatment

Forty-five of the 46 participants origi-nally enrolled in the study completed the6-week treatment protocol (22 women,24 men). Mean age was 60.8 (SD, 5.4)years (median, 59; interquartile range 6),and the duration of insomnia was on av-erage 14.1 years (SD, 11.3) (median, 10;interquartile range 15). Age, sex, edu-cational level, insomnia duration, smok-ing, caffeine intake, body mass index, co-morbid chronic condition, or sleepmeasures did not differ significantly be-

tween the treatment groups at pretreat-ment assessment (TABLE 2). More par-ticipants in the CBT condition hadpreviously received treatment for insom-nia compared with the other condi-tions (P=.04), and more participants re-ceiving placebo treatment were alsotaking other non–sleep-related medica-tions compared with the CBT and zopi-clone groups (P=.02) (Table 2). MeanPSG-registered sleep efficiency at pre-treatment was 81.0 (SD, 10.5) across thetreatment conditions, while the partici-pants’ sleep diaries yielded a mean sleepefficiency of 66.2 (SD, 11.8).

6-Week Follow-up

PSG. Total wake time showed both a sig-nificant time effect (P�.001), indicat-ing that the participants spent less time

Table 3. Objective and Subjective Sleep Data for Each Treatment Condition at All 3 Assessment Points

Sleep Measureand Time

Cognitive Behavioral Therapy(n = 18)

Zopiclone(n = 16)*

Placebo(n = 12)† Time Effect‡

Time � GroupEffect§

Mean (SD)%

ImprovementEffectSizes � Mean (SD)

%Improvement

EffectSizes Mean (SD)

%Improvement

EffectSizes

FValue

PValue Post hoc

PValue

Total wake time, minPolysomnography

Pretreatment 107.8 (41.0) 102.8 (54.9) 153.6 (145.5)Posttreatment 51.4 (35.0)¶ 52 1.5 98.9 (42.3) 4 0.1 129.8 (64.1) 16 0.2 8.70 �.001 CBT�ZOP, PL �.0016-Mo follow-up 47.1 (31.0)¶ 56 1.7 92.9 (38.7) 10 0.2 7.22 .003 CBT�ZOP .001

Sleep diaryPretreatment 143.2 (63.4) 157.9 (75.1) 159.9 (67.9)Posttreatment 94.9 (88.2)# 34 0.6 132.1 (74.3) 16 0.3 126.4 (66.8) 21 0.5 6.52 .001 CBT, ZOP, PL .456-Mo follow-up 69.9 (46.7)¶ 51 1.3 115.7 (68.0)# 27 0.6 10.55 �.001 CBT�ZOP .03

Total sleep time, minPolysomnography

Pretreatment 370.0 (63.6) 388.3 (58.3) 346.0 (59.3)Posttreatment 343.8 (56.5) −7.1 −0.4 322.7 (57.2)# −17 −1.1 322.9 (60.2) −7 −0.4 0.47 .70 CBT, ZOP, PL .506-Mo follow-up 365.0 (56.1) −1.3 −0.1 332.1 (68.4)# −15 −0.9 1.27 .30 CBT, ZOP .13

Sleep diaryPretreatment 319.1 (60.7) 304.9 (67.6) 313.1 (54.1)Posttreatment 336.0 (55.5) 5 0.3 339.5 (75.9)** 11 0.5 334.2 (44.2) 7 0.4 5.49 .003 CBT, ZOP, PL .826-Mo follow-up 361.5 (57.9)** 13 0.7 345.4 (71.3)** 13 0.6 10.85 �.001 CBT, ZOP .69

Sleep efficiency, %Polysomnography

Pretreatment 81.4 (7.4) 82.3 (8.2) 78.9 (16.2) -Posttreatment 88.9 (8.4)** 9 1.0 81.5 (9.3) −1 −0.1 76.2 (11.3) −3 −0.2 5.03 .005 CBT, ZOP�PL .0046-Mo follow-up 90.1 (7.2)¶ 11 1.2 81.9 (9.0) −1 −0.0 4.42 .002 CBT�ZOP .008

Sleep diaryPretreatment 69.0 (12.4) 63.2 (12.5) 65.8 (9.9)

Posttreatment 80.8 (13.9)¶ 17 0.9 71.3 (14.7)** 13 0.6 71.7 (10.0) 9 0.6 15.61 �.001 CBT, ZOP, PL .176-Mo follow-up 83.2 (11.1)¶ 21 1.2 73.9 (13.8)** 17 0.8 16.97 �.001 CBT, ZOP .11

Slow-wave sleep, minPolysomnography

Pretreatment 63.1 (28.9) 76.8 (37.0) 79.5 (20.8)Posttreatment 80.3 (26.5)** 27 0.6 61.7 (32.4)# −20 −0.4 69.1 (25.2) −13 −0.4 7.53 �.001 CBT�ZOP, PL .0026-Mo follow-up 84.4 (34.1)** 34 0.7 59.2 (28.9)# −23 −0.5 11.60 �.001 CBT�ZOP .001

Abbreviations: CBT, cognitive behavioral therapy; ZOP, zopiclone; PL, placebo.*For polysomnographic analysis, n = 18 due to invalid polysomnographic data.†Placebo group was assessed only at baseline and 6 weeks.‡F- and P-values refer to overall time effect found using analysis of covariance.§Post hoc tests and P values refer to time � group interactions using analysis of covariance.�Denotes within-group effect size (pooled SD).¶P�.001 P value is based on paired-samples t tests to examine time effects within each treatment condition compared with pretreatment.#P�.05 P value is based on paired-samples t tests to examine time effects within each treatment condition compared with pretreatment.**P�.01 P value is based on paired-samples t tests to examine time effects within each treatment condition compared with pretreatment.




http://www.jama.com

awake during the night following treat-ment than prior to treatment (TABLE 3),and a significant time�group interac-tion (P�.001), indicating that treat-ment groups differed significantly. Thetotal wake time for the CBT group im-proved significantly more than both theplacebo group at 6 weeks and the zopi-clone group at 6 weeks. The zopiclonegroup did not differ significantly from theplacebo group (P=.62). Total wake timeat 6 weeks was reduced 52% in the CBTgroup compared with 4% and 16% in thezopiclone and placebo groups,respectively.

Total sleep time showed no signifi-cant time effects (P=.70), indicating thattotal sleep timedidnotchangewith treat-ment interventions. However, sleep effi-ciency demonstrated both a significanttime effect (P=.005), and time�groupinteraction(P=.004),withtheCBTgrouphaving significantly higher sleep effi-ciencyat6weeks than theplacebogroup(P=.004). CBT was not significantly dif-ferent fromzopiclone(P=.09), andzopi-clonewasnot significantlydifferent fromplacebo (P=.62) (FIGURE 2).

The amount of PSG-recorded slow-wave sleep (stage 3 and 4) improved sig-nificantly over time in the CBT groupcompared with both the placebo (P=.03)and zopiclone groups (P=.002). Thezopiclone group had significantly lessslow-wave sleep after treatment com-pared with before treatment (P=.01).

Sleep Diary. Total wake time(P=.001), total sleep time (P=.003),and sleep efficiency (P�.001) all im-proved over time as recorded in par-ticipants’ sleep diaries, but no differ-ences were seen by group (Table 3).

6-Month Follow-UpPSG. Total sleep time increased signifi-cantly in the CBT group at 6 monthscompared with 6 weeks (P=.05). Thezopiclone group showed no significantchange at 6 months, maintaining im-provements seen at 6 weeks (Table 3).Comparing the 2 active treatment con-ditions, total wake time, sleep effi-ciency, and slow-wave sleep were all sig-nificantly better in the CBT group thanin the zopiclone group; total sleep timewas not significantly different (Table 3).

Sleep Diary. Similar to PSG, the sleepdiaries showed an increase in total sleeptime in the CBT group at 6 monthscompared with 6 weeks of follow-up(P=.004). Total wake time declined inthe CBT group compared with the zopi-clone group (P=.03) (Table 3).

Clinical SignificanceThe clinical significance of the treat-ment effects in both active conditionswas examined by calculating the pro-portion of participants who reachedPSG-recorded sleep efficiency level ofat least 85%.30 In the CBT group, 13 in-dividuals (72%) had a sleep efficiency

level of at least 85% at the 6-week post-treatment assessment, while 14 (78%)fulfilled this criterion 6 months aftertreatment completion, compared with6 (33%) at pretreatment. In contrast,only 7 (47%) of the participants inthe zopiclone group had a sleep effi-ciency of at least 85% at the 6-weekposttreatment assessment (vs 6 [40%]at pretreatment), a proportion whichdeclined to 6 (40%) at 6-month follow-up. The group differences were statis-tically significant both at posttreat-ment (�2

2=8.94; P=.01), and follow-up(�2

1=4.89; P=.03).

Treatment Attendanceand Adherence

Participants rated their adherence toCBT,zopiclone,andplaceboona5-pointscale (ranging from0-“never” to5-“ev-ery night”) showing to what extent theytook the pills or followed the advice andinstructions in the treatment condition.The attendance rate in the CBT condi-tion was 100%, and participants in thezopiclone condition who cancelled anappointment were sent the week’s dos-age by mail. Overall level of adherenceacross all treatment groups was high(mean4.6 [SD,0.6]).Therewerenosig-nificant differences between the CBT(mean 4.8 [SD, 0.1]) and zopiclone(mean 4.5 [SD, 0.9]) condition on self-reported adherence at posttreatment,while participants in the placebo con-ditionscored lower than theCBTrecipi-ents (mean 4.3 [SD, 0.6]; P=.05). At6-month follow-up, the adherence ratedid not differ between the 2 active treat-mentgroups,buthaddeclinedto4.1(SD,0.3) (P�.001) in theCBTconditionand3.6 (SD, 1.3) in the zopiclone condition(P=.02comparedwith6weekposttreat-ment). All participants in both the CBTandthezopicloneconditionreportedthatthey had continued their treatment tosomeextentduring the6-month follow-upperiod.However,3participants(13%)in thezopiclonecondition reported thatthey only took their sleep medication“some of the days” after posttreatmentassessment, of which 1 participantstopped taking the drug 1 month aftertreatment completion. The remaining

Figure 2. Sleep Efficiency at Pretreatment, Posttreatment, and Follow-up as Measured WithPolysomnography and Sleep Diary

100

30

60

50

40

70

80

90

20

10

0

Sle

ep E

ffici

ency

, %

Polysomnography

Pretreatment Posttreatment Follow-up

Cognitive BehavioralTherapy(n = 18)

Zopiclone(n = 15)

100

30

60

50

40

70

80

90

20

10

0

Sle

ep E

ffici

ency

, %

Sleep Diary

Cognitive BehavioralTherapy(n = 18)

Zopiclone(n = 16)

Bars represent 95% confidence intervals.




http://www.jama.com

participants reported either “half of thedays,” “most days,” or “all days,” as wasalso the case for all participants in theCBTcondition. Inaddition,1participantin the zopiclone group did not use sleepmedication at follow-up assessment.

Adverse Effects. The following ad-verse effects were reported by partici-pants in the zopiclone condition: bit-ter taste (n = 6), dry mouth (n = 4),daytime drowsiness (n=4), light nau-sea (n=2), headache (n=2), and chestpain (n=1). One participant in the zopi-clone condition withdrew before post-treatment assessment due to adverseeffects, as did 2 participants in the fol-low-up period. One participant in theplacebo condition reported light nau-sea and dry mouth. No adverse effectswere reported in the CBT condition.

COMMENTWe found that CBT was more effectiveimmediately and long-term comparedwith both zopiclone and placebo in olderadults with chronic primary insomnia.On average, participants receiving CBTimproved their PSG-registered sleep ef-ficiency by 9% at posttreatment, com-pared with a decline of 1% in the zopi-clone condition, a difference that wasboth statistically and clinically signifi-cant. These improvements in the CBTgroup were maintained at 6-month fol-low-up. Furthermore, participants in theCBT group spent significantly more timein slow-wave sleep (stages 3 and 4) com-pared with the other conditions.

A Cochrane review concluded thatCBT had only a mild effect on sleep prob-lems in older adults.26 By contrast, ourfindings indicate a much stronger effect,with within-group effect sizes for CBTranging from 0.6 to 1.7 at follow-up ontotal wake time, sleep efficiency andslow-wave sleep. Although we found nosignificant changes in PSG-registeredtotal sleep time, the participants’ sleepdiaries yielded significant treatment gainsacross the treatment conditions. Extend-ing the findings by Morin et al,30 thepresent study provides additional evi-dence that CBT produces both short- andlong-lasting treatment effects in olderadults with insomnia. The clinical sig-

nificance of the CBT was underscored by72% of the participants having a PSG-registered sleep efficiency level of at least85% at posttreatment assessment, com-pared with 33% at pretreatment. Ourfindings that CBT recipients showed last-ing improvements in slow-wave sleep(63.1 minutes at study inception to 84.4minutes at follow-up) were striking, es-pecially as zopiclone resulted in a de-crease (76.8 minutes at study inceptionto 59.2 minutes at follow-up). This is anintriguing finding that needs to be rep-licated, as lack of slow-wave sleep maybe responsible for impaired daytimefunctioning and sleepiness.31

In contrast to the results by Morin etal,30 we found no significant treatmenteffects inthepharmacologicalgroup,nei-ther at 6-week posttreatment or at6-months follow-up.Zopicloneshowedno better effect than placebo and pro-ducedsignificantly less slow-wavesleepatposttreatmentcomparedwithpretreat-ment.This issomewhatsurprisingasnu-merous clinical trials have shown thatshort-term use of zopiclone is at least aseffective as the older benzodiazepinesin patients with insomnia.45 However,almost no studies have investigated theeffectsofzopiclonebeyond4weeks,andwe cannot rule out that participants inthis condition may have developed tol-erance when they were assessed after 6weeks.Ontheotherhand,5participantswithdrew by 6 months and 2 were nolonger taking the study drug, so theseresults shouldbereplicated inadditional6-month or longer-term studies.

The observed discrepancies betweenchanges in PSG and sleep diary–recorded sleep time in both active-treatment conditions should be noted.However, the reliability and validity ofsleep diaries have previously been ques-tioned, as patients’ self-reported sleeptime has been shown to deviate fromfindings based on PSG, ranging from un-derestimations to overestimations.32

There are some limitations to thepresent study. Only participants withchronicprimaryinsomniawere included,and thus, our results may not generalizetopatientswhosesleepproblemsare sec-ondary to psychiatric or medical condi-

tions. It remains to be seen whether CBTfor insomnia may yield similar positiveresults in primary care settings, in whichsleep problems may be part of a morecomplex clinical picture. Also, one mayargue that the average sleep quality ofincluded participants at pretreatmentassessment was relatively high (PSG-registered sleep efficiency of 81% acrossall treatment conditions). However, theparticipants’ subjective reports based onsleep diaries yielded a sleep efficiency of66%, indicating that they did experi-ence their sleep as impaired. Also, noinformationwasavailable toexamine theprolonged treatment effects beyond thelast follow-up assessment at 6 monthsafter treatment completion. However, asthe treatment effects in the CBT condi-tion were actually stronger at follow-upthan at posttreatment, our findingssuggest that the durability of CBT isconvincing.Furthermore, thegroupsizesin thepresentstudywererelativelysmall.Patientswhocompletedtheplacebotreat-ment were all randomized into an activetreatment, but these were excluded fromthe final analyses. However, whenconducting the statistical analyses ofall treated patients (CBT=23, zopi-clone=22), we found similar or highereffect sizes in the CBT group, while thezopiclone group remained mostlyunchanged (data available on requestfromauthor). It shouldalsobenoted thatwe were unable to blind the CBT con-dition, and that no nonpharmacologi-cal placebo group was used in thepresent study. We also have no data spe-cifically addressing daytime sleepiness,which would have been interesting tocompare with the observed changes inslow-wave sleep. Finally, care should betakenwithregard togeneralizing thepre-sent findings of zopiclone to other sleepmedications.

Regardlessof these limitations, thepre-sent findings have important implica-tions for the clinical management ofchronicprimaryinsomnia inolderadults.Given the increasing amount of evi-denceof the lastingclinical effectsofCBTand lack of evidence of long-term effi-cacy of hypnotics, clinicians should con-siderprescribinghypnoticsonly foracute




http://www.jama.com

insomnia. At present, CBT-based inter-ventions for insomnia are not widelyavailable in clinical practice, and futureresearch should focus on implementinglow-threshold treatment options forinsomnia in primary care settings. Asrecently demonstrated by Bastien et al,49

telephone consultations and CBT-basedgrouptherapy foryoungerpatientswith insomnia produced equally signifi-cant improvementsas individual therapysessions. In another study, CBT deliv-ered via the Internet in a self-help for-mat showedsignificant improvements inindividuals with chronic insomnia.50 Inaddition, preliminary findings suggestthat self-help programs for insomniabased on CBT delivered in the contextof community-based interventions mayoffersignificantclinicalbenefits.51 Finally,future research should seek to identifywhich single factors in the CBT regi-menproduce thebest results and towhatextentboostersessionsat1to2yearsafterinitial treatment may be necessary tomaintain improvements.

In conclusion, this study demon-strated superior benefits of CBT overzopiclonefortreatmentofchronic insom-nia inolderadultsat6-weekand6-monthfollow-up. Future research shouldrequire effects in slow-wave sleep anddefine effects on daytime sleepiness.

Author Contributions: Dr Nordhus had full access toall of the data in the study and takes responsibility forthe integrity of the data and the accuracy of the dataanalysis.Study concept and design: Sivertsen, Omvik, Pallesen,Bjorvatn, Havik, Kvale, Nielsen, Nordhus.Acquisition of data: Sivertsen, Omvik.Analysis and interpretation of data: Sivertsen, Omvik,Pallesen, Bjorvatn, Havik, Kvale, Nielsen, Nordhus.Drafting of the manuscript: Sivertsen.Critical revision of the manuscript for important in-tellectual content: Sivertsen, Omvik, Pallesen, Bjorvatn,Havik, Kvale, Nielsen, Nordhus.Statistical analysis: Sivertsen, Omvik, Pallesen, Havik,Nordhus.Obtained funding: Pallesen, Nordhus.Administrative, technical, or material support: Sivertsen,Omvik, Pallesen, Nordhus.Study supervision: Pallesen, Bjorvatn, Havik, Nor-dhus.Financial Disclosures: None reported.Funding/support: This research was funded by grantsfrom the University of Bergen, the Meltzer Fund, andthe EXTRA funds from the Norwegian Foundation forHealth and Rehabilitation.Role of the Sponsors: The funding organizations hadno role in the design and conduct of the study; col-lection, management, analysis, and interpretation ofthe data; and preparation, review, or approval of themanuscript.

REFERENCES

1. Sateia MJ, Doghramji K, Hauri PJ, Morin CM. Evalu-ation of chronic insomnia. Sleep. 2000;23:243-308.2. Ohayon MM. Prevalence of DSM-IV diagnostic cri-teria of insomnia. J Psychiatr Res. 1997;31:333-346.3. Ohayon MM. Epidemiology of insomnia. Sleep MedRev. 2002;6:97-111.4. Pallesen S, Nordhus IH, Nielsen GH, et al. Preva-lence of insomnia in the adult Norwegian population.Sleep. 2001;24:771-779.5. Roth T, Roehrs T. Insomnia: epidemiology, charac-teristics, and consequences. Clin Cornerstone. 2003;5:5-15.6. Zammit GK, Weiner J, Damato N, Sillup GP, McMillanCA. Quality of life in people with insomnia. Sleep. 1999;22(suppl 2):S379-S385.7. Avidan AY. Insomnia in the geriatric patient. ClinCornerstone. 2003;5:51-60.8. Ford DE, Kamerow DB. Epidemiologic study of sleepdisturbances and psychiatric disorders. JAMA. 1989;262:1479-1484.9. Simon GE, VonKorff M. Prevalence, burden, andtreatment of insomnia in primary care. Am J Psychiatry.1997;154:1417-1423.10. Walsh JK, Engelhardt CL. The direct economic costsof insomnia in the United States for 1995. Sleep. 1999;22(suppl 2):S386-S393.11. Mellinger GD, Balter MB, Uhlenhuth EH. Insom-nia and its treatment: prevalence and correlates. ArchGen Psychiatry. 1985;42:225-232.12. Riedel BW, Lichstein KL. Strategies for evaluatingadherence to sleep restriction treatment for insomnia.Behav Res Ther. 2001;39:201-212.13. Roth T, Ancoli-Israel S. Daytime consequences andcorrelates of insomnia in the United States. Sleep. 1999;22(suppl 2):S354-S358.14. Morin CM, Wooten V. Psychological and pharma-cological approaches to treating insomnia. Clin Psy-chol Rev. 1996;16:521-542. doi:10.1016/0272-7358(96)00027-x.15. Hohagen F, Kappler C, Schramm E, et al. Preva-lence of insomnia in elderly general practice attendersand the current treatment modalities. Acta PsychiatrScand. 1994;90:102-108.16. Nowell PD, Mazumdar S, Buysse DJ, Dew MA, Rey-nolds CF III, Kupfer DJ. Benzodiazepines and zolpidemfor chronic insomnia. JAMA. 1997;278:2170-2177.17. Holbrook A, Crowther R, Lotter A, Endeshaw Y. Therole of benzodiazepines in the treatment of insomnia.J Am Geriatr Soc. 2001;49:824-826.18. Glass J, Lanctot KL, Herrmann N, Sproule BA, BustoUE. Sedative hypnotics in older people with insomnia:meta-analysis of risks and benefits. BMJ. 2005;331:1169.19. National Institutes of Health. National Institutes ofHealth State of the Science Conference statement onManifestations and Management of Chronic Insomniain Adults, June 13-15, 2005. Sleep. 2005;28:1049-1057.20. Concern Over Ambien and Driving. “CBS News.”CBS television. March 8, 2006.21. Stuever H. Snooze alarm: the pill that drove us tosleep. Washington Post. May 6, 2006: C01.22. Irwin MR, Cole JC, Nicassio PM. Comparative meta-analysis of behavioral interventions for insomnia and theirefficacy in middle-aged adults and in older adults 55�years of age. Health Psychol. 2006;25:3-14.23. Morin CM, Culbert JP, Schwartz SM. Nonpharma-cological interventions for insomnia. Am J Psychiatry.1994;151:1172-1180.24. Murtagh DR, Greenwood KM. Identifying effec-tive psychological treatments for insomnia: ameta-analysis. J Consult Clin Psychol. 1995;63:79-89.25. Jacobs GD, Pace-Schott EF, Stickgold R, Otto MW.Cognitive behavior therapy and pharmacotherapy forinsomnia. Arch Intern Med. 2004;164:1888-1896.26. Montgomery P, Dennis J. Cognitive behavioural in-terventions for sleep problems in adults aged 60�. Coch-rane Database Syst Rev. 2003;(1)CD003161.27. Edinger JD, Means MK. Cognitive-behavioraltherapy for primary insomnia. Clin Psychol Rev. 2005;25:539-558.

28. Montgomery P, Dennis J. A systematic review ofnon-pharmacological therapies for sleep problems in laterlife. Sleep Med Rev. 2004;8:47-62.29. Pallesen S, Nordhus IH, Kvale G. Nonpharmaco-logical interventions for insomnia in older adults: a meta-analysis of treatment efficacy. Psychotherapy. 1998;35:472-482.30. Morin CM, Colecchi C, Stone J, Sood R, Brink D.Behavioral and pharmacological therapies for late-lifeinsomnia: a randomized controlled trial. JAMA.1999;281:991-999.31. Moldofsky H. Sleep and pain. Sleep Med Rev. 2001;5:385-396.32. Edinger JD, Fins A. The distribution and clinical sig-nificanceof sleep timemisperceptionsamong insomniacs.Sleep. 1995;18:232-239.33. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”: a practical method for grading the cog-nitive state of patients for the clinician. J Psychiatr Res.1975;12:189-198.34. First MB, Spitzer RL, Gibbon M, Williams JBW. Struc-tural Clinical Interview for DSM-IV Axis I Disorders. NewYork, NY: Biometrics Research Department; 1995.35. Rechtschaffen A, Kales A. A Manual of Standard-ised Terminology and Scoring System for Sleep Stagesof Human Sleep. Los Angeles: University of Californiaat Los Angeles, Brain Information Service/Brain Re-search Institute; 1968.36. Lorenzo JL, Barbanoj MJ. Variability of sleepparameters across multiple laboratory sessions inhealthy young subjects: the “very first night effect.”Psychophysiology. 2002;39:409-413.37. Lichstein KL, Riedel BW. Behavioral assessment andtreatment of insomnia: a review with an emphasis onclinical application. Behav Ther. 1994;25:659-688.38. Morin CM. Insomnia: Psychological Assessment andManagement. New York, NY: Guilford Press; 1993.39. Morgan K, Thompson J, Dixon S, Tomeny M,Mathers N. Predicting longer-term outcomes follow-ing psychological treatment for hypnotic-dependentchronic insomnia. J Psychosom Res. 2003;54:21-29.40. Spielman AJ, Saskin P, Thorpy MJ. Treatment ofchronic insomnia by restriction of time in bed. Sleep.1987;10:45-56.41. Bootzin RR, Nicassio PM Behavioral treatments forinsomnia. In: Hersen M, Eissler R, Miller P, eds. Progressof Behavior Modification Vol 6. New York, NY: Aca-demic Press; 1978:1-45.42. Beck AT. Cognitive Therapy and the EmotionalDisorders. New York, NY: International University Press;1976.43. Jacobsen E. You Can Sleep Well. New York, NY:McGraw-Hill Book Co; 1938.44. Hajak G. A comparative assessment of the risks andbenefits of zopiclone: a review of 15 years’ clinicalexperience. Drug Saf. 1999;21:457-469.45. Noble S, Langtry HD, Lamb HM. Zopiclone: anupdate of its pharmacology, clinical efficacy and toler-ability in the treatment of insomnia. Drugs. 1998;55:277-302.46. Norwegian Institute of Public Health. Drug Con-sumption in Norway. Oslo: Norwegian Institute of Pub-lic Health; 2004.47. Cohen J. Statistical Power Analysis for the Behav-ioral Sciences. 2nd ed. Hillsdale, NJ: Laurence Erlbaum;1988.48. Smith MT, Perlis ML, Park A, et al. Comparativemeta-analysis of pharmacotherapy and behavior therapyfor persistent insomnia. Am J Psychiatry. 2002;159:5-11.49. Bastien CH, Morin CM, Ouellet MC, Blais FC,Bouchard S. Cognitive-behavioral therapy for insom-nia: comparison of individual therapy, group therapy,and telephone consultations. J Consult Clin Psychol.2004;72:653-659.50. Strom L, Pettersson R, Andersson G. Internet-based treatment for insomnia: a controlled evaluation.J Consult Clin Psychol. 2004;72:113-120.51. Morin CM, Beaulieu-Bonneau S, LeBlanc M. Self-help treatment for insomnia: a randomized controlledtrial. Sleep. 2005;28:1319-1327.




http://www.jama.com

Documents

Cognitive Behavioral Therapy vs Zopiclone for … · ORIGINAL CONTRIBUTION Cognitive Behavioral Therapy vs Zopiclone for Treatment of Chronic Primary Insomnia in Older Adults A Randomized