Jouirnial of Neurology, Neurosurgery, and Psychiatry 1991;54:153-158

Cognitive behaviour therapy in chronic fatiguesyndrome

S Butler, T Chalder, M Ron, S Wessely

AbstractFifty patients fulfilling operationalcriteria for the chronic fatigue syndrome(CFS), and who had been ill for a mean

of five years, were offered cognitivebehaviour therapy in an open trial.Those fulfilling operational criteria fordepressive illness were also offered tri-cyclic antidepressants. The rationale wasthat a distinction be drawn between fac-tors that precipitate the illness and thosethat perpetuate it. Among the latter are

cognitive factors such as the belief thatphysical symptoms always imply tissuedamage, and behavioural factors such as

persistent avoidance of activities assoc-iated with an increase in symptoms.Therapy led to substantial improve-ments in overall disability, fatigue,somatic and psychiatric symptoms. Theprincipal problems encountered were a

high refusal rate and difficulties intreating affective disorders. Outcomedepended more on the strength of theinitial attribution of symptoms toexclusively physical causes, and was notinfluenced by length of illness. Theseresults suggest that current views on

both treatment and prognosis in CFS areunnecessarily pessimistic. It is also sug-

gested that advice currently offered tochronic patients, to avoid physical andmental activity, is counterproductive.

Department ofPsychiatry, NationalHospital forNeurology andNeurosurgery, QueenSquare, LondonS ButlerT ChalderM RonS WesselyCorrespondence to:Dr Wessely, WellcomeTraining Fellow inEpidemiology, Institute ofPsychiatry, De CrespignyPark, Camberwell, LondonSE5 8AF, UKReceived 27 November 1989and in final revised form13 June 1990.Accepted 29 June 1990.

The chronic fatigue syndrome, also known as

postviral fatigue or myalgic ence-

phalomyelitis, continues to attract attentionand controversy. The scale of the problem isunknown, largely due to difficulties in case

definition,' but it seems probable that increas-ing numbers of people are being diagnosed, or

diagnosing themselves, as suffering from thecondition. At the beginning of 1989 over 150people a week were joining the principalpatients' organisation.'Even nomenclature remains a problem in

this field. Myalgic encephalomyelitis (ME) isfrequently encountered, but is unsatisfactoryfor many reasons.3 Postviral fatigue syndromeimplies an aetiology that is by no means

certain. For the rest of this paper we havefollowed the international consensus of usingthe term "chronic fatigue syndrome" (CFS),as it is short, accurate and has no aetiologicalimplications.4The cause of CFS remains unknown.

153

Various factors, both medical and psychiatric,have been implicated.3 At present, however,evidence has only been presented from small,selected case control studies.' Furthermore,such studies are only able to report associa-tions, not aetiologies, and may not apply tothe majority of patients with the clinicalsyndrome.35 There are also no establishedtreatments for CFS. Suggested methods ofmanagement that focus on presumed chronicinfection have so far been found wanting.67Instead, most patients receive very little in theway of practical help. The mainstay of treat-ment remains rest, and to wait either forremission or a medical cure, and the adviceconsistently offered by the current self helpgroups is to avoid physical and mentalactivity. Others are told to accept disability as"living within their limits". This state ofaffairs has been succinctly described as"therapeutic nihilism"8 and is unsatisfactory.What little we know about the natural historyof these conditions9 shows that untreated theycan be severe and persistent. In the largestcurrent series "Most of the cases seen do notimprove, give up their work and become per-manent invalids".6 No satisfactory explan-ations have been advanced for this "alarmingtendency to chronicity'.10Most of the articles concerning CFS assume

a simple model in which an external agent,usually a virus, causes a fatiguing illness bypersistent viral infection." immune dysfunc-tion or a mixture of the two." "' We haveargued that a more complex model is neededto explain the peculiar clinical features of thisillness.'5 16

Recently attention has been given to therole of cognitive distortions, and the con-sequent maladaptive behaviour, as mediatorsof disability in many illnesses, both physicaland psychological.'7 Looking specifically atCFS, it is plausible that an initial infectivetrigger may begin a cycle in which both attri-butional and cognitive factors fuel avoidantbehaviour. The initial symptoms, in particularfatigue and myalgia, engender a state of"learned helplessness", being potent, aversiveand uncontrollable,'8 and may also trigger orexacerbate the mood disorder that is found inmany patients.'9 Continuing attribution of allsymptoms to a persistent, untreatable "virus",continues to increase helplessness, althoughpreserves self esteem.'8 Avoidant behaviour(which is reinforced by the advice currentlyoffered to patients) sustains symptoms, bydecreasing activity tolerance and increasing

on Novem

ber 30, 2021 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.54.2.153 on 1 February 1991. D

ownloaded from

Butler, Chalder, Ron, Wessely

sensitivity to any stimulation, as doesassociated mood disorder. Re-exposure toactivity causes more symptoms,20 and morefear. The result is a vicious circle of symp-toms, avoidance, fatigue, demoralisation anddepression-the clinical picture of CFS. Asimilar role for avoidance in perpetuatingsymptoms has been implicated in chronicpain." 22

Such a model implies treatments that differfrom those suggested by a simple externalagent/disease model. In particular,improvement should result from attention tomood disorder, cognitive distortions andfinally behavioural avoidance. Instead ofantiviral agents, therapeutic success shouldresult from reducing avoidant behaviour,decreasing the perception of helplessness andimproving mood,'7 all of which are already anestablished part of the treatment of bothchronic pain and fibromyalgia2324 (the latterbeing closely related to CFS25). In practicethis can involve using the techniques of cog-nitive behaviour therapy.26We report the results of an open trial of

cognitive behaviour therapy and antidepres-sants in patients who fulfilled the operationalcriteria for the syndrome and were referredfor a neurological assessment of fatigue to theNational Hospital for Neurology andNeurosurgery.'9

MethodsMost of the sample reported in this paper hasbeen described in detail in a previous paper.'9 Itconsists of fifty patients referred to our hospitalfor a medical assessment of severe fatigue forwhom no neurological explanation could befound and conventional neurological tests werenormal. The only difference is that the first 13patients in the previous communication werenot offered treatment, but have been replacedby 16 new patients. There are no clinicaldifferences between the previous sample andthe current extended sample.As an entry criterion all patients had a

principle complaint of physical fatigue. Inaddition 49/50 reported physical fatigability(that is, fatigue made worse by physical exer-tion), and 45/50 reported mental fatigue andfatigability. In addition to fatigue, 42 (84%)complained of muscle pain, 43 (86%) com-plained of post exercise myalgia. Over 50% ofthe sample also complained of the followingsymptoms: nausea, dizziness, daytime drowsi-ness, back pain, headaches, tremor, perspira-tion, eyestrain, paraesthesiae and insomnia.Three quarters (76%) reported that their ill-ness had begun after a "viral" infection, and42/50 (84%) believed they had postviral fatiguesyndrome or "ME." Twenty seven (54%) weremembers of the ME Association. Most hadmade their own diagnosis or had been diag-nosed by a doctor specialising in "ME".

All patients fulfilling the criteria werereferred by the neurological staff as part of aprevious study into the nature of fatigue in thepostviral syndrome. After completing selfreport questionnaires and a standardised

interview'9 all were given a brief description oftreatment by the psychiatrists. Those whofulfilled operational criteria for depressionwere also started on antidepressant medication(see later). A further assessment took placebetween four to six weeks later, and was carriedout by the two therapists. After further discus-sion an offer of treatment was made, andtherapy started on the next visit.

Thirty two patients accepted the offer oftreatment, of whom 26 were treated as out-patients. The mean therapist time was 7-5hours, with a range from two to 20 hours oftreatment. Six had such severe disability (beingconfined to bed or a wheelchair for most or allofthe time), that it was necessary to admit themto hospital for a period ofbetween three to eightweeks.Treatment followed cognitive behavioural

principles,3' derived from work on the man-agement of chronic pain,27 but adapted for theparticular problems of CFS. Full details andtechniques are reported elsewhere,'7 but a briefaccount follows. After a detailed assessment,patient and therapist identified prominent dis-abilities, and looked at the associations andreasons advanced for such disabilities. Cog-nitive distortions, when present, were noted,and patients were encouraged to look for alter-native explanations. Self monitoring of bothbehaviour and cognitions was also encouraged.Behavioural targets were jointly agreed upon,based on avoided activities which the patientwished to resume. These were practised bet-ween sessions. The intention was to introducepredictability into the pattern of rest andactivity, and to encourage self efficacy-thiswas phrased as "helping the patient, and not'ME', to be in control". Later in treatment thepatient was encouraged to break the associationbetween symptoms and ceasing activity, so thatsensitisation to symptoms and activity wasreplaced by tolerance.Twenty patients were offered tricyclic

antidepressants at the initial psychiatricinterview. Dothiepin, a conventional tricyclicantidepressant, was prescribed in doses rang-ing from 75 to 150 mg, depending upon abilityto tolerate side effects. All fulfilled theResearch Diagnostic Criteria' for probable ordefinite major depression. Three of thoseoffered treatment declined. Similar treatmentwas also given to an additional three patientswho fulfilled criteria for other psychiatric dis-orders (two somatisation, one anxiety), andwere already receiving antidepressants. Thosegiven antidepressants scored higher on scalesmeasuring psychiatric symptoms, somaticsymptoms, and mental fatigue, and were morefunctionally impaired. However, there were nodifferences in the length of illness, nor physicalfatigue, which is consistent with previousobservations on this sample.'9

After four to six weeks all patients were seenby the behaviour therapists, at which timemood disorder was reassessed with the BeckDepression Inventory (BDI).29 There were nolonger differences in the severity of mooddisorder between those who had been pres-cribed antidepressants and those who had not

154

on Novem

ber 30, 2021 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.54.2.153 on 1 February 1991. D

ownloaded from

Cognzitive behaviour therapy in chronic fatigue syndrome

(mean Beck Depression Score for those receiv-ing antidepressants was 10 2 (6-7-13 8), com-pared with 8-8 (4-1-13-5), although they stillhad significantly higher numbers of somaticsymptoms.Two different sets of measures were used.

The first comprised well validated self admin-istered questionnaires recording psychiatricsymptomatology. These were the 12 ItemGeneral Health Questionnaire (GHQ),30 theHospital Anxiety and Depression Scale(HAD),3' Modified Somatic DiscomfortsQuestionnaire,32 the Fear Questionnaire,33 theBeck Depression Inventory,29 and a new scalefor recording physical and mental fatigue.22 Allmeasures were completed between theneurological and psychiatric assessments, theexceptions were the Beck Depression Inven-tory and the Fear Questionnaire, which were

given at the start of therapy, and thus are onlyavailable for those in treatment.The second set of instruments recorded

functional disability. The principal measure ofdisability was a self assessment of functionalimpairment.33 This consists of four visualanalogue scales covering ability to work, homemanagement, social and private leisureactivities. Respondents are asked how muchtheir problems have affected each of the desig-nated areas. Ratings range from 0 ("not at all")to 8 ("very severe, I cannot do it"). The scores

of the four visual analogue ratings of functionalimpairment were summed and treated as a

single variable, labelled "impairment".The second measure was of "problems and

targets".37 The patient with the therapist iden-tifies which of their problems is the mostincapacitating, together with the associatedcognition. Examples include "inability toresume my job because I always feel tired", or

"I cannot get out of bed because of the pain".The appropriate target is then agreed bypatient and therapist as something to worktowards: it may be "to resume full time work",or "to be able to feed and dress myselfunaided".At the end of treatment all self report

measures were repeated, usually being com-

pleted at home, although the behaviouralmeasures of problems and targets were com-

pleted with the therapist. A five point self ratedoutcome scale was also completed, in whichpatients were asked to rate themselves along a

scale from much better to much worse. As theaim of the therapy was to improve themanagement of symptoms, the end point oftreatment was therefore when the patient was

able to deal with symptoms without furthertherapist contact, or when the therapist felt he/she had no more to offer. It was not when thepatient had become asymptomatic, as manycontinued to improve at follow up.

Parametric statistics were used whereappropriate, either on the initial variable or

following log transformation (length). Nonparametric statistics were used for the results ofself report of functional impairment, problems,targets and attitudes. All means are quotedwith 95% confidence intervals, and allsignificance testing is two tailed.

ResultsChanges were noted in all aspects of life (table1): social, private, work and leisure (table 2), aswell as psychological symptoms (table 3). Ofthose who completed treatment, there was anoverall self rated percentage improvement indisability of 60% (95% confidence interval 41-78%). In most, improvement occurredglobally. The decline in functional disabilitycorrelated with a decrease in mood and anxietyrelated symptoms and also somatic symptoms(data not shown). Some experienced a greaterthan 50% improvement in symptoms and dis-ability, but remained symptomatic, especiallywith a variety of somatic discomforts. In onlythree patients was there greater than a 50%decrease in psychiatric symptomatology with-out an equivalent improvement in functionalimpairment.Twenty three (70% of those starting

therapy) patients described themselves as "bet-ter" or "much better". Stricter criteria,however, were used to determine completerecovery. These were as follows: belowthreshold scores on all of the GHQ, HAD andBDI measures; a score of 0 or 1 on both thephysical fatigue scale (range 0-16) and themental fatigue scale (Range 0-10), a globalscore of 1 ("much better"), and a functionaldisability score of8 or below (range 0-32). Ninepatients met all these requirements (28%).

Eighteen patients (36%) declined the offer oftreatment (11 when interviewed by the psy-chiatrist, seven by the behaviour therapist).This contrasts with only 5/50 refusals amongthe last 50 patients referred by neurologists forcognitive behaviour therapy for conditionsother than CFS. Length of illness, presence ofmuscle pain, severity of mental or physicalfatigue, history of "viral" illness, currentpsychiatric illness (RDC criteria) and func-tional impairment were not associated with thedecision to accept or reject treatment. Therewas a trend for those refusing to be female(15/34 females refused compared to 3/16 males;Chi squared = 2-03; p = 0-15). There wasalso a trend for refusers to have lower scores onthe GHQ-12 (mean GHQ score for refusers =5.37 (3 39-7-35), accepted = 7.59 (6-27-8 9),t = 1 83, df = 46, p = 0 078). Although thisdid not reach conventional statistical signi-ficance, this may have been due to lack ofpower, since a similar non significant trend wasobserved for HAD scores. Attribution ofsymptoms also played a role. rhere was a trendfor those refusing treatment to bemore likely tobelieve their illness was entirely due to physicalfactors (Chi squared = 4 61, df = 2,p = 0-09).

Five patients withdrew after starting theprogramme. Two had already been unable to

Table I Outcome of treatment

Global ratings Completed Drop out

Much better 16 0Better 6 1About the same 4 2Worse 0 0Much worse 1 2Total 27 5

155 on N

ovember 30, 2021 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.54.2.153 on 1 F

ebruary 1991. Dow

nloaded from

Butler, Chalder, Ron, Wessely

Table 2 Changes in functional disability in those completing treatment (NR = 27)

Individual scores (mean, 95% CI)Range 0-8.Before After treatment

Ability to work 6-31 (5 57-7 05) 2-72 (1-77-3.65)*Home management 5-69 (5 04-6-34) 1-54 (0-78-217)*Social leisure activities 5-72 (5-08-6-35) 2-08 (1-17-2-99)*Private leisure activities 5-19 (4-52-5-86) 1-65 (0-89-241)*Total functional impairment 22-61 (20-42-24-8) 7-96 (4-92-11 00)*Principal target 6-63 (5-947-32) 1 52 (0-87-2-17)*Principal problem 6-57 (5-997-15) 1-61 (1-11-2-13)*

*Wilcoxon Matched-pairs signed ranks test: P < 0-001

Table 3 Changes in symptoms

Individual scores (mean, 95% CI)Before After treatment

GHQ-12 score: 7-7 (5 7-9 04) 2-09 (0-31-3-87)tMental fatigue 5-82 (4 82-682) 1-79 (0-62-2-96)tPhysical fatigue 12-48 (1116-13 8) 2 00 (0-491)tHAD score 17-91 (1459-21 23) 9-21 (5-73-12-69)tSomatic symptoms 14-33 (12-61-16-65) 8-12 (5-53-10-71)§Beck depression inventory 9-79 (7-2-12-37) 4-77 (2-22-7-32)*Fear questionnaire 27-93 (2035-35 51) 16-04 (9-61-22-47)$

*Wilcoxon Matched-pairs signed ranks test: P < 0-01tWilcoxon Matched-pairs signed ranks test: P < 0-001Paired t test (df 22), p = 0-014

§Paired t test (df = 26), p < 0-001

tolerate antidepressant therapy, and did notwish to continue. One improved, but did notwish to travel to the hospital, whilst two otherschanged their minds about treatment. None ofthe latter had experienced any change in symp-toms.One patient completed treatment, but was

clearly worse than at the outset. This patienthad a major affective disorder, which con-

tinued to deteriorate, despite additional con-

ventional treatment. Three completed therapybut reported no change on overall health (allhad increased their functional abilities, butwithout any marked change in symptoms). Onefurther patient completed the entire treatment,but was unchanged both symptomatically andfunctionally.As treatment produced changes in fatigue,

psychiatric symptoms and functional dis-ability, no single measure ofoutcome is entirelysatisfactory. Instead, those who started treat-ment were divided into two groups by theirresponse to a single question asking aboutglobal outcome. Those who rated themselves as

"better" or "much better" were classed as goodoutcome, whilst those who were "unchanged","worse") or "much worse" were classed as pooroutcome. The two patients who dropped out oftreatment were arbitrarily rated as "muchworseAs with the decision to accept/reject treat-

ment clinical features, such as severity offatigue, number of somatic symptoms, lengthof illness, presence of muscle pain after exer-

cise, and either the presence or severity ofpsychiatric disorder did not influence outcome.Unlike the decision to accept treatment, therewas also no sex difference in outcome.The first factor influencing outcome was

treatment resistant affective disorder. Thosewho failed to improve did not differ in theseverity of affective disorder before startingcognitive behaviour therapy (Mean pre-

Table 4 Influence of symptom attribution on treatmentoutcome

Physical (7) or Both (6) ormainly physical mainly(16) psychological (2)

Good outcome 14 8Poor outcome 9 0

P = 0-04(1 missing value)

treatment Beck Depression Score in the goodoutcome group was 8-25 (5-6-10-9), comparedwith 11-66 (5-1-18 2) in the poor outcomegroup; Mean HAD scores were 17-61 (13 2-22 0) compared with 1611 (10-0-22-2)).However, 3/3 patients who deteriorated werecases of affective disorder, as were 4/6 of thoseunchanged. The second variable associatedwith a poor outcome was the strength ofattribution to a physical cause (table 4). Of theseven patients who attributed their illness toexclusively physical causes six had a pooroutcome. Finally, the role of the VP-1 antigen"remains unclear. There was a trend for patientspositive for this group specific enteroviralantigen to do worse, although this did not reachconventional statistical significance (Fisher'sexact = 0 08). VP-1 antigen status was notobtained on eight cases.Of the twenty three patients eligible to

receive antidepressants, 16/23 (69%) had agood outcome. Of the seven with a pooroutcome, only one had received 150 mg ofdothiepin for six weeks. Of the other six, fourhad not tolerated antidepressants (defined asbeing unable to take a tricyclic antidepressantfor more than seven days), and two haddeclined to take them. Only one of the threedrug refusers therefore had a good outcome.The nine remaining patients who agreed totreatment did not fulfil the criteria for majordepression, and received no antidepressants;7/9 (77%) had a good outcome.

All patients who successfully completedtherapy have now been followed up for threemonths, as have 3/9 with poor outcomes. Ofthenine patients who finished symptom free, eightremain so. One has re-experienced some symp-toms, but remains able to work. Most of theremainder are stable, and have maintained theirimprovement (four have continued to improvefurther of whom one is now symptom free).Three have experienced a gradual worsening ofsymptoms, but without any change in func-tional disability (two were associated withdiscontinuing antidepressants). In conclusion,at three months follow up the 23 who improvedduring the treatment all remain better thanbefore.

DiscussionWe report an uncontrolled pilot study of treat-ment of the chronic fatigue syndrome. Themanagement of these patients followed ourstandard clinical management. The study isthus subject to all the deficiencies of a nonblind, non randomised study. In the currentclimate ofopinion, however, these results are ofinterest for several reasons.

156 on N

ovember 30, 2021 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.54.2.153 on 1 F

ebruary 1991. Dow

nloaded from

Cognitive behaviour therapy in chronic fatigue syndrome

First, we have shown that something can bedone for patients suffering from CFS andrelated syndromes. Most of those entering thetrial experienced an improvement in symptomsand functional activity, and many were able toresume their occupations, or if not employed,experienced a return to previous levels offunctioning.

Second, this study suggests that the view thatlittle can be done for patients with CFS isinaccurate. In particular, we have shown thateven those experiencing longer than two yearsof illness, and severe and persistent disability,can still return to premorbid functioning.

Third, the advice customarily given to thosewith illnesses of long duration, to avoidphysical and mental activity, may not only beunnecessary, but sometimes inadvertentlycounterproductive. Much of the current selfhelp literature and media reporting state thatthe approach we favour should lead to persis-tent illness and increased morbidity, especiallyin those with chronic disability.

Overall, only three out of 32 patients wereworse after entering the trial. All had mooddisorder on entry, which persisted throughouttreatment. Two of the three were unable totolerate conventional therapy in the form ofantidepressants. A major reason for a lack ofimprovement was therefore persistent mooddisorder. It is possible that others who wouldhave suffered adverse effects declined the offerof treatment, and such fears were indeed theprincipal reason expressed for refusal. Theaccuracy of that perception is impossible todetermine, but there were few clinical or symp-tomatic factors associated with refusal. Instead,the trend for those refusing treatment are morelikely to attribute all their symptoms to a purelyphysical problem, and less likely to accept aninteraction between physical and psy-chological, may reflect the external social pres-sures that complicate so much of CFS.34Both persistent mood disorder and the stren-

gth ofattributions of illness contributed to pooroutcome. More tentatively, there was also atrend for an association between poor outcomeand the persistence of enteroviral antigen. Thismay have been due to confounding factors, asthe antigen is found in both depression'5 andneurological disorders."6As in any non blind, uncontrolled study the

results must be interpreted cautiously. Theobserved improvements were, however, un-likely to be due to chance alone. The averageduration of illness before referral was justunder four years, after which time most authorsstate that little or no spontaneous recovery canbe expected.2'7 It is not possible to determinethe precise cause of improvement, nor toseparate out the effects of antidepressantmedication and cognitive behaviour therapy. Itseems unlikely that improvements were solelydue to medication, since a similar proportion ofthose not prescribed antidepressants achievedas good an outcome as those receiving medi-cation. We also doubt that improvement wassolely due to the physiological benefits ofincreased exercise,'8 since behavioural targetswere chosen on the basis of avoidance, not

because of their physiological or ergonomicproperties, which in practice were often min-imal. It is not even possible to rule out theinfluence of non specific factors, such astherapist time and interest.Demonstrating the benefits of cognitive

behaviour therapy sheds little light on thenature of the initial pathology in CFS. Thebalance between physical, psychological andsocial factors changes over time in many illnes-ses. Our results are compatible with currenttheories on either a viral or immune pre-cipitant. Such factors are not, however, theonly cause of long term disability. Even if allcases of CFS are initially the result of viralinfection, which seems unlikely,' the secondaryconsequences, both social and psychological,may be a more potent cause of long termdisability, analogous to the long term outcomeof chronic pain2 22 and head injury.'9 Fortun-ately, such secondary factors are more amena-ble to treatment at present.

Details of the clinical characteristics of thepatients confirm that the current sample wastypical of hospital samples currently viewed ashaving severe postviral fatigue/"ME", andresembled those reported from other special-ised referral centres. All fulfilled recent con-sensus criteria for CFS.'4 Our findings cantherefore be generalised to other CFS patientsseen in hospital practice with severe fatigue oflong duration and a diagnois ofpostviral fatiguesyndrome or its equivalent. The results, how-ever, cannot be assumed to apply to those seenin general practice, or with shorter durations ofillness.

Future trials are essential to clarify theseresults. In particular, randomisation ofpatientsto either the above therapy, rest or antidepres-sants is necessary. Other improvements,including independent ratings, should be used,although any trial of this nature will ultimatelydepend upon self report. Finally, it is hopedthat the current unwelcome "polarisation ofattitudes"'" concerning the nature of CFS willchange, to allow empirical treatments, asoutlined in this paper, to become more widelyavailable than at present.

In conclusion, these results confirm theintuition of Jerome Frank,4' who wrote that"Patients suffering from unfamiliar diseasestend to develop emotional reactions whichimpede recovery, such as anxiety, resentmentand confusion. To keep disability at a min-imum, therapeutic efforts must be directed notonly to overcoming the pathogenic agent but tomaintaining the patient's confidence in thephysician, and encouraging his'expectation of areturn to useful activity".

We thank the medical and nursing staff of the National Hospitalfor their continuing support. Most of all we thank the patientswho participated in the trial, often under trying circumstances.SW is supported by a Wellcome Training Fellowship inEpidemiology, MR is partly supported by the SCARFE Trust.

I David A, Wessely S, Pelosi A. Post-viral Fatigue: Time for aNew Approach. Br Med J 1988;296:696-9.

2 Smith D. Understanding ME. London: Robinson, 1989.3 Wessely S, Thomas PK. Chronic fatigue syndrome (myalgic

encephalomyelitis or postviral fatigue). In: Kennard C,ed. Recent advances in neurology, Vol 6. Edinburgh:Churchill Livingstone, 1990: 85-131.

157 on N

ovember 30, 2021 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.54.2.153 on 1 F

ebruary 1991. Dow

nloaded from

Butler, Chalder, Ron, Wessely

4 Holmes G, Kaplan J, Gantz N. et al. Chronic FatigueSyndrome: A Working Case Definition. Annals Int Med1988;108:387-9.

5 Swartz M. The chronic fatigue syndrome-one entity ormany? New Eng J Med 1988;319:1726-8.

6 Behan P, Behan W. The Postviral Fatigue Syndrome. CRCCritical Reviews in Neurobiology 1988;42:157-8.

7 Straus S, Dale J, Tobi M, et al. Acyclovir treatment of thechronic fatigue syndrome: lack of efficacy in a placebo-controlled trial. New Eng J Med 1988;319:1692-8.

8 Bayliss R, op cit Dawson J. Brainstorming the postviralfatigue syndrome. Br Med J 1988;297:1151.

9 Wessely S. The natural history of fatigue and myalgiasyndromes. In: Sartorius N, Goldberg D, de Girolamo G,Costa e Silva J, Lecrubier Y, Wittchen H, eds. Psycho-logical disorders in general medical settings. Bern: HansHuber, 1990:82-97.

10 Smith D. Myalgic encephalomyelitis. (Members ReferenceBook). R Coll Gen Pract. London, Sabre Crown,1989:247-50.

11 Yousef G, Bell E, Mann G, et al. Chronic enterovirusinfection in patients with postviral fatigue syndrome.Lancet 1988;i: 146-50.

12 Lloyd A, Wakefield D, Boughton C, Dwyer J. Immuno-logical Abnormalities in the Chronic Fatigue Syndrome.Med J Aust 1989;151:122-4.

13 Behan P, BehanW, Bell E. The postviral fatigue syndrome-an analysis of the findings in 50 cases. J Infection1985;1O:21 1-22.

14 Komaroff A. Chronic Fatigue Syndromes: relationship tochronic viral infections. J Virological Methods 1988;21:3-10.

15 Wessely S, David A, Butler S, Chalder T. The managementofthe chronic "post-viral" fatigue syndrome. J R Coll GenPract 1989;39:26-9.

16 Wessely S, Butler S, Chalder S, David A. The cognitivebehavioural management of the postviral fatigue syn-drome. In: Jenkins R, Mowbray J, eds. The postviralfatigue syndrome (ME). Chichester: John Wiley, (inpress).

17 Sensky T. Patients' reactions to illness. Brit Med J1990;300:622-3.

18 Powell R, Dolan R, Wessely S. Attribution and selfesteem indepression and chronic fatigue syndromes. J Psy-chosomatic Research, 1990;34:665-73.

19 Wessely S, Powell R. Fatigue Syndromes: A comparison ofchronic "postviral" fatigue with neuromuscular and affec-tive disorders. J Neurol Neurosurg Psychiatry 1989;52:940-8.

20 Editorial. Aching Muscles after Exercise. Lancet 1987;ii:1 123-5.

21 Keefe F, Gil K. Behavioural concepts in the analysis ofchronic pain syndromes. J Consult Clinical Psychology1986;54:776-83.

22 Philips H. Avoidance behaviour and its role in sustainingchronic pain. Behav Res Therapy 1987;25:273-9.

23 McCain G, Bell D, Mai F, Holliday P. A controlled study ofthe effects of a supervised cardiovascular fitness trainingprogram on the manifestations of primary fibromylagia.Arthritis Rheum 1988;31:1135-41.

24 Yunus M. Diagnosis, etiology and management offibromyalgia syndrome: an update ComprehensiveTherapy 1988;14:8-20.

25 Goldenberg D. Fibromyalgia and other chronic fatiguesyndromes: is there evidence for chronic viral disease?Seminars in Arthritis and Rheumatism 1988;18: 111-20.

26 Hawton K, Salkovskis P, Kirk J, Clark D. Cognitivebehaviour therapy for psychiatric problems: a practicalguide. Oxford: Oxford University Press, 1989.

27 Pither C. Treatment of persistent pain. Br Med J1989;299: 1239.

28 Spitzer R, Endicott J, Robins E. Research diagnostic criteria(RDC) for a selected group offunction disorders, 3rd ed.New York: New York State Psychiatric Institute, 1977.

29 Beck A, Ward C, Mendelson M, Mock J, Erbaugh J. Aninventory for measuring depression. Arch Gen Psychiatry1961;4:561-71.

30 Goldberg D. The detection ofa psychiatric illness by question-naire. London: Oxford University Press, 1972.

31 Zigmond A, Snaith R. The Hospital Anxiety and DepressionScale. Acta Psycha Scanda 1983;67:361-70.

32 Wittenborn J, Buhler R. Somatic Discomforts AmongDepressed Women. Arch Gen Psychiatry 1979;36:465-71.

33 Marks I. Behavioural psychotherapy: Maudsleypocket book ofclinical management. Bristol: Wright, 1986.

34 Wessely S. "Old wine in new bottles": neurasthenia and"ME". Psychological Medicine 1990;20:35-53.

35 Lynch S, Seth R. Postviral fatigue syndrome and the VP-1antigen. Lancet 1989;ii:1 160-1.

36 Halpin D, Wessely S. VP-I Antigen in chronic postviralfatigue syndrome. Lancet 1989;i:1028-9.

37 Shepherd C. Living with ME: a self-help guide. London:Heinemann, 1989.

38 Edwards RH. Muscle fatigue and pain. Acta Med Scand1986;Suppl. 711:179-88.

39 Lishman A. Physiogenesis and psychogenesis in the post-concussional syndrome. Br J Psychiatry 1988;153:460-9.

40 Sharpe M, Archand L, Banatvala J, et al. Chronic fatiguesyndrome: guidelines for research. J Roy Soc Med 1991;(in press).

41 Frank J. Emotional reactions of American soldiers to anunfamiliar disease. Am J Psychiatry 1946;144:631-40.

158 on N

ovember 30, 2021 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.54.2.153 on 1 F

ebruary 1991. Dow

nloaded from