Occupational and Environmental Medicine 1994;51:145-151

Cognitive-behavioural treatment for workers withchronic spinal pain: a matched and controlledcohort study in Sweden

I B Jensen, A Nygren, A Lundin

AbstractAn ambulatory multimodal cognitive-behavioural treatment programme(MMCBT) for Swedish workers withchronic spinal pain was formally evalu-ated. The design was a matched cohortstudy with three repeated measures donein two groups of 35 referred patients (oneexposed to MMCBT and controlsexposed to usual care). The MMCBTpackage includes standardised modulesof physical treatment, cognitive-behav-ioural treatment, education of workerpatients, and education of subjects'supervisors. The primary outcome vari-ables assessed were absenteeism, disabil-ity, pain, and depression. Because of achange in Swedish sick leave compensa-tion laws affecting records of absenceduring the study period, absenteeismcould not be reliably measured amongcontrols. The trend, however, suggesteda reduction of absenteeism among thesubjects in the MMCBT cohort. Thefindings for disability, pain, and depres-sion all showed clinically important andsignificant beneficial changes (ANOVAfor repeated measures: disability p =0 05; pain p = 0-001; depression p =0.01). The direction of the improvementsand the size of effect were coherent andclinically plausible. The benefits wereonly among the women in the study.These comprised 74% of each group.Further research on larger sample sizesand in cohorts more representative of thewhole country are needed to confirmthese encouraging findings and toexplore how the benefits might beextended to men.

(Occup Environ Med 1994;51:145-15 1)

Department ofpersonal injuryprevention,Karolinska Institute,StockholmI B JensenA NygrenNarRehab,HalsoInvest i Orebro,SwedenA LundinRequests for reprints to:Dr I B Jensen, Nar Rehab,Halso Invest i Orebro,Ribbingsg 1-3, 70363Orebro, Sweden.

Accepted 17 May 1993

Chronic long term neck, shoulder, and backpain is becoming more prevalent in manywestern countries and, increasingly, is costlyboth socially and economically. Several stud-ies have underscored the difficulties in diag-nosing these problems and in institutingtreatment that is demonstrably effective. 1 2The Quebec Task Force on Spinal Disordersin the Workplace in particular showed thedevastating effects that chronic spinal painhas on the chances of rehabilitation and on

the long term quality of life of individual sub-jects and families.' Virtually nothing is knownabout primary prevention or about the pre-

vention of pain becoming chronic. A newdevelopment in the past 10 years has been theintroduction of treatments for chronic spinalpain syndromes that are based on a cognitivebehavioural approach. Such intervention pro-grammes have been shown to be effective,particularly in the management of pain.3 4

Multimodal cognitive-behavioural treat-ment, including physical components, hasalso been developed in Sweden in recentyears. Typically, in the past, the therapeuticprogrammes have been in inpatient settingsand the clinics are located geographicallydistant from the patients' everyday environ-ment.56 (And Jensen et al, unpublished data.)Other models of intervention, such as the wellknown Swedish Back School, have beenalmost exclusively educational depending ondidactic classroom teaching without specifictechniques of behavioural change. The pas-sive approach in the Back School strategiesmight explain the disappointing results whenthey have been evaluated.7 Even though theoutcomes in some evaluative studies haveshown positive results these tend to decline.after six and 18 months of follow Up.4 (AndJensen et al, unpublished data.) Importantfactors in cognitive-behavioural interventionare the maintenance programmes that aim toprevent relapses and protect initial gains. Wespeculate that the declining gains of treat-ments developed in Sweden are partlyexplained by the lack of opportunity forpatients to practice their newly acquired skillsin their daily environment. Lack of opportu-nity and lack of maintenance programmesmay be the keys affecting long term results.Other factors associated with the mainte-nance of initial improvement in perceivedpain and in residual impairment are compli-ance and social support available in the sur-roundings where the behavioural changes areintended to take place (Jensen et al, unpub-lished data).We report a study wherein we evaluated an

outpatient treatment programme designed forlong term neck, shoulder, and back pain ofemployed persons, which was based on cog-nitive behavioural theory. A distinctive fea-ture of the strategy was that the workplacesupervisors of the employees were taught toreinforce the behavioural changes ofemployed patients. The interventions evalu-ated in this study included a one day trainingcourse for the patients' supervisors within themain training programme for the patients,which lasts four weeks. Thus one objective ofthe project was to augment the supervisors'

145 on A

pril 29, 2022 by guest. Protected by copyright.

http://oem.bm

j.com/

Occup E

nviron Med: first published as 10.1136/oem

.51.3.145 on 1 March 1994. D

ownloaded from

6Jensen, Nygren, Lundin

understanding of pain and to teach them howto reintegrate employees who had partici-pated in a cognitive-behavioural programmeof rehabilitation. The purpose of the studywas to assess the effectiveness of multimodalcognitive-behavioural treatment by measuringphysical and psychological endpoints amongpatients.

MethodsGENERAL PLAN OF THE STUDYThis study compares two small cohorts of 35subjects each. The cohorts were formed bymatching non-referred controls to referredpatients by criteria listed later. Anothercohort of 88 patients, including 53 who couldnot be successfully matched, were used as areference to monitor bias. The two cohortswere contrasted, having been measured threetimes each.As we did not assign the subjects to treat-

ment and control groups this prospectivestudy is non-experimental or observational.Nevertheless, the groups assembled for com-parison proved to be similar.

DESCRIPTION OF INTERVENTIONSThe treatment consists of five modules: phys-ical treatment, cognitive behavioural modifi-cation, education, promotion of patients'interaction with their usual occupational andsocial situation, and training of work supervi-sors to enhance reintegration to the job.

PHYSICAL TREATMENTPhysical treatment emphasises selfadminis-tered care and is designed to enhanceendurance more than strength. Specifically, itincludes progressive training with machinesfor physical training and specially designedgymnastic exercises aimed to improve thefunction of the neck, shoulders, and back.Leg muscles and abdominal muscles are alsotargets. More natural and spontaneous physi-cal activities such as walking in the forest,bowling, swimming, cycling, and badmintonare also strongly promoted and practise underobservation is part of the programme.Optionally, the physical therapists can usetheir clinical judgement and give traditionalmanoeuvres such as passive exercises, ultra-sound, heat, massage, and others.

THE COGNITIVE-BEHAVIOURAL MODULEThe cognitive-behavioural module consists ofeight sessions where problem solving, goalsetting, acquisition of skills (stepwise goal ful-filment), and relaxation applied to eachpatients' life context (applied relaxation) aretaught and practised under supervision.During the treatment period a contract wasdeveloped and consisted of goals of behav-ioural changes to enhance health promotingbehaviour and pertained to all the modules,not just the cognitive-behavioural module.The latent goal of the cognitive-behaviouralmodule is to enable patients to perform allnew skills learnt without supervision in theirhome and work settings and to induce a sus-

tained change of health-illness behaviourthrough a procedure that rewards and rein-forces positive change (also known as operanttechnique). In the skills acquisition compo-nent, the patients have homework that is indi-vidually tailor made to each person's problemsolving and goal setting. The cognitive-behav-ioural module sessions follow a predeter-mined standardized plan. (Available uponrequest from I B Jensen.)

THE EDUCATIONAL MODULEThe educational module involves all cate-gories of professionals and clinicians of theinterdisciplinary team: psychologist, physi-cian, nurse, and physical therapist. It consistsof 12 lectures that consider anatomy, thenature of stress, the use and effect of medica-tions, pain behaviour and its consequences,cognition, the dynamics of psychosomaticresponses, and the consequences of overstim-ulation or understimulation of the body andthe psyche. All lectures are followed by smallgroup sessions where the participants discussthe issues raised in class and answer questionsabout them in writing. Spouses or others areinvited to all lectures particularly to the onesabout pain driven behaviour.

INTERACTION OF THE TREATMENT TEAM WITHTHE PATIENTS' MEDICAL, OCCUPATIONAL, ANDSOCIAL SUPPORT NETWORKSpecifically this important interactioninvolves one or more of the referring physi-cian, a human resources officer (personneldepartment) of the employer, an occupationalcounsellor, a representative of a governmentalsocial insurance authority. The purpose ofthis module is to exchange information aboutthe patient and to acquaint relevant people inthe patient's environment of the individu-alised goals of the treatment programmedeveloped and adopted by the patient. Thusit becomes possible for all persons involved inthe reintegration of the patients to be conflu-ent with them in pursuing exactly the samegoals.

TRAINING OF PATIENTS' SUPERVISORSTraining of supervisors is, in our view, a criti-cally important module orientated to smoothreintegration of the subjects to their employ-ment. It is a one day session offered in twoparts: in the morning the supervisors hear anintense and concentrated synopsis of the edu-cational syllabus that the patients themselveswill have learned; in the afternoon, they aretaught rehabilitation planning, which incor-porates the techniques of problem solvingand goal setting taught to the patients. Theythen have to practise the techniques throughsimulated problems of the work environmentprepared specifically for them. Supervisorshave an opportunity to meet the patients inthe clinic that day and discuss common con-cerns.

All the modules described become an inte-grated and interdependent package(MMCBT). Apart from the individualizationof each patient's goals and the principle of

146

on April 29, 2022 by guest. P

rotected by copyright.http://oem

.bmj.com

/O

ccup Environ M

ed: first published as 10.1136/oem.51.3.145 on 1 M

arch 1994. Dow

nloaded from

Cognitive-behavioural treatmentfor workers with chronic spinal pain

discretion in clinical judgement of the physio-therapists there is no deviation from patientto patient. Accordingly, the MMCBT as awhole is the intervention under assessment inthis study (independent variable) although inlater analyses we will examine selected mod-ules in isolation, notably the supervisors' role.

STUDY SUBJECTSThere were 101 patients originally referred toan outpatient facility in the south east ofSweden called NarRehab/Halsoinvest (NRH)located in the centre of the city of Orebro.Patients were referred primarily by the gov-ernmental social insurance authority(F6rsdkringskassan) and by physicians prac-tising occupational health. The patients wereall from the Orebro area, drawn from a catch-ment not more than 50 km from the city lim-its. Eight patients refused to participate. Fivepatients were referred with insufficient infor-mation to establish their eligibility. Eightyeight subjects were eligible. Table 1 showsavailable data on non-participants.

CRITERIA OF ELIGIBILITY FOR PATIENTS ANDMATCHING PROCEDURES FOR CONTROLSAll subjects in the study were first screenedbefore being assigned to the patient group inthe MMCBT programme or to the group ofmatched controls. The screening processcomprised a standardized and documentedhistory and physical examination by anorthopaedic surgeon, a functional assessmentby a physical therapist, and a psychologicalevaluation by a registered psychologist.Eligibility criteria were then applied. Thesame three specialists examined all patientsand controls. Informed written consent wasobtained from all participants after an expla-nation of the treatment and of the study toeach patient. The orthopaedic surgeon andthe physical therapist were also part of theteam that administered the MMCBT pro-gramme. A different psychologist was in thetreatment team than the one who screenedfor measured outcome variables.

PATIENTS' INCLUSION CRITERIAInclusion criteria included, (1) Pain locatedin the neck and shoulder, or vertebral column(either dorsal or lumbar or both). (2)Diagnosis of "non-specific pain syndrome"without objective neurological signs (con-firmed in all patients at the NRB rehabilita-

Table 1 Characteristics of refusers, ineligible patients, and losses to follow upRefusers Ineligible Losses to follow up(n = 8) (n =5) (n = 4)

Age (mean; y) 45 36 38Women (n) 7 2 3Occupation(n):

Unskilled blue collar 6 4 2Skilled blue collar 1 1 1Unskilled white collar 1 - -Skilled white collar - - 1

Topography of pain(n):Neck and shoulder 5 2 2Dorsal and lumbar 1 2 -

Mixed pain 2 1 2

tion clinic by the same orthopaedic surgeonas part of the screening procedures). (3) Age20-55. (4) Fluent in the Swedish language.(5) Currently employed. (6) On full or partialleave of absence (sicklisted) for six months orless.

It is important to stress that shoulder painwas never in isolation.

CRITERIA USED FOR MATCHINGA list of all concurrently sicklisted workers inthe Orebro area was available from the socialinsurance authority. We sought to pair all 88eligible patients to an eligible control. Thematching criteria were: (1) age, plus or minusfive years; (2) currently employed; (3) broadcategory of work (unskilled blue collar includ-ing manufacturing, craftsman, service andcare, white collar, academics); (4) location ofpain and diagnosis that excludes neurologicalsigns; (5) about the same amount of time onsick leave.We located 48 pairs but at least one mem-

ber of five of the pairs became ineligible (forexample, through new diagnoses or lostemployment). Eight controls refused to par-ticipate (table 1). Refusers tended to beunskilled blue collar workers who had neckand shoulder pain.

BLINDINGFor insurmountable logistic reasons, it wasnot possible to blind the measuring psycholo-gist to whether study subjects were patients orcontrols. As there were at least twice as manypatients in the MMCBT programme than inthe study, the treating clinicians were, how-ever, never sure which of the patients in theirclasses or groups were being measured forpurposes of controlled evaluation.

MEASUREMENT OF DEPENDENT VARIABLESThere were two categories of endpoints thatwere dependent variables-namely priorityendpoints and explanatory variables. The pri-ority endpoints were selected before theanalysis was begun and so were absenteeism,physical functional capacity, pain intensity,and depression. The explanatory variableswere anxiety, perceived helplessness, andcoping strategies. In this paper we report pri-marily the findings of the priority endpoints.Explanatory variables are only discussedwhen illuminating but will be reported indetail in a subsequent report. The detailedanalyses of the explanatory variables are avail-able on request.

Target variables used as endpoints wereassessed at baseline, before treatment, at theend of treatment (about four weeks afterbaseline), and at six months of follow up.

ABSENTEEISMDetails on leave of absence were obtainedfrom the national health insurance (NHI)authority for one year before the treatmentand for six months afterwards. The maincomparisons made before and after treatmentwere possible for the same six month periodbefore treatment and the same six month

147

on April 29, 2022 by guest. P

rotected by copyright.http://oem

.bmj.com

/O

ccup Environ M

ed: first published as 10.1136/oem.51.3.145 on 1 M

arch 1994. Dow

nloaded from

8Jensen, Nygren, Lundin

follow up period. Moreover, total absen-teeism was the hardest variable used. It isnoteworthy that in Sweden all employees are

covered by the national insurance scheme.Regulations require that both the employerand the insurance authority must be notifiedby the end of the same business day of theabsence so that benefits can be received. Thatmeans that virtually all absences are on

record. Employees also can be absent for a

full or a part of a day due to illness.Accordingly absenteeism was recorded inhours rather than days.

PAIN INTENSITY AND ANXIETYThese data were recorded in a diary by thepatients for seven days leading up to the timeof the other tests. Pain intensity and anxietywere recorded three times daily (morning,noon, and evening) on a 100 mm visual ana-logue scale (VAS). High score indicateshigher pain/anxiety level.

DEPRESSIONWe used the Beck depression inventory.8Scores range from 0-60 with the higherscores more severe.

DISABILITYThis 12 item self administered questionnairewith semantic differential questions measures

patients' satisfaction with their ability to per-form daily activities (for example, walkingand dressing). It is the Swedish version of theHealth Assessment Questionnaire, and we

found it to be reliable and valid.9 Scoresrange from 0-60; higher means less satisfac-tion with functional disability.

PERCEIVED HELPLESSNESSThe 15 item measure of perceived helpless-ness is the Swedish version of the arthritishelplessness index,'0 slightly modified tofocus on neck, shoulder, and back pain. Itrequires subjects to rate the extent to whichthey believe they can control the conse-quences of pain. It is self administered. Ahigh score indicates a high level of perceivedhelplessness (range 0-48).

COPING STRATEGIES

This was a Swedish version of the copingstrategies questionnaire with 48 items"; scoresrange from 0-36 for each of eight strategies

Table 2 Selected sociodemografic variables

Controls Matched patients Reference group(n = 35) (n = 35) (n = 88)

Age (mean, (SD) 44 (9 2) 43 (8 8) 40 (9-5)Female(%) 74 74 74Married(%) 60 89 77Number of children(Mean, (SD)) 2 (0 72) 2 (0 77) 2 (0 80)Occupation(%):

Unskilled blue collar 34 37 36Skilled blue collar 37 43 41Unskilled white collar 23 23 22Skilled white collar 6 0 1

Workers compensation:claim(% yes) 66 74 67

Unfinishedworkers compensation(%): 26 31 34

reviewed. Higher scores indicate greater useof the particular strategy concerned.

CONTRASTS AND ANALYSESFor the priority endpoints, the most impor-tant contrast is between the 35 patients whowere matched successfully and their respec-tive 35 controls. To detect bias we also com-pared the two cohorts of the 35 patients andthe 35 controls with the reference cohort ofall 88 patients by analysing them as indepen-dent samples. We did no tests of significancein the second set of comparisons; it was doneto find bias and for descriptive purposes only.Parametric paired statistics were used in themain contrast. To assess the trends in thedata over time we used analysis of variancewith repeated measures.'2

ResultsSIMILARITY OF COHORTS COMPAREDTable 2 shows that the two main matchedcohorts compared and the reference cohort ofall patients are sufficiently similar in theirsociodemographic characteristics at baselineto enable comparisons of the effect of theintervention. Data in table 3 show that theyare also similar in respect to the clinical andexplanatory variables measured initially.

NON-PARTICIPANTSTable 1 suggests that there was no majorselection bias in excluding subjects from thecohorts with the possible exception of asomewhat greater high refusal rate amongunskilled blue collar workers.

PRIORITY ENDPOINTSTable 4 presents the principal findings of thisstudy. The means of the four outcomesselected before the start are shown for thebaseline measurements, and for the two fol-low ups (at four weeks, and six months). Themeans for the within pair differences frombaseline to each follow up are also shown. Weshow all patients as the reference cohort to

Table 3 Explanatory and clinical variables at baseline

Control Matched Referencepatients group

(n = 35) (n = 35) (n = 88)

Participated in rehabilitationprogramme (% of controls) 34 - -

Mean duration of pain(months) 44 37 45

Topography of pain:Neck and shoulder(%) 20 11 18Dorsal and Lumbal (%) 0 6 11Mixed pain (%) 80 83 71

Spinal mobility (median) 98 90 90Neck mobility (median) 329 280 290Global self rating (mean) 3-4 3 6 3-8Perceived helplessness (mean) 26-6 27 1 27 2Anxiety (mean) 20-1 21 6 23 4Strategies for coping (mean):

Diverting attention 12-7 12-0 12-1Reinterpreting pain 5-1 3-1 3-4Coping self statement 16-3 15-9 16-2Ignoringsensations 13 2 12-1 11-7Praying and hoping 11-5 10-8 11-5Imagining catastrophy 11-5 11-8 12-0Increasing behavioural 15 7 14-2 15-0

activityPain behaviours 17 3 16 2 16 4

148

on April 29, 2022 by guest. P

rotected by copyright.http://oem

.bmj.com

/O

ccup Environ M

ed: first published as 10.1136/oem.51.3.145 on 1 M

arch 1994. Dow

nloaded from

Cognitive-behavioural treatmentfor workers with chronic spinal pain

Table 4 Priority endpoints at baseline (BA) after treatment (4 weeks) and atfollow up (6 months) (means and means of differences within pairs)

Means of differences Means of differences

(BA- (BA-BA 4 Weeks 6 Months 4 weeks) p Value 6 months) p Value

Absenteeism total (six monthsbefore andsix months after treatment):*

Matched patients 82-6 77-9Reference group 85-6 81 7Disability:

Controls 29-1 29-1 26-3Matched patients 33-6 29-9 29-7Reference group 32-9 28-9 28-5Within pair differences 4-3 0 049 -3-3 0 091

Pain intensity:Controls 44-8 40 9 40-6Matched patients 48-0 36-9 37-3Reference group 51-7 38-7 39-8Within pair differences -6-8 0-0112 -7 5 0-060

Depression:Controls 9-8 9.1 9-6Matched patients 10-7 7-2 8-6Reference group 11-2 7 2 8-3Within pair differences -2-9 0 011 -1-8 0-275

*Mean days of absence.

insure that the patients we succeeded inmatching were not unrepresentative of all thepatients referred and entered in the MMCBTprogramme. For all four endpoints the threecohorts are similar at baseline.

During the time from when the field workbegan until the end of record linkage at theNHI authority one year later (August 1992)the sick leave data were altered by changes inSwedish legislation. Since January 1992, thegovernment is no longer responsible for shortterm sick leave. Instead, employers recordand pay for the first 14 days of absence due tosickness and all such short term sick leavehad only to be reported to the NHI authority.It was soon discovered, however, thatemployers neither recorded nor reported andthus the NHI authorities data base was miss-ing up to 30% of its information.Accordingly, the before and after measuresfor the matched controls are meaningless forsickness related absenteeism. We do, how-ever, have the before and after measures forthe matched patients and the reference groupof patients (table 4) after six months. Thetrends are towards less absenteeism and theextent to which they are sustained will bereassessed after another 12 months. For theother three outcomes that we have assessed atthe first follow up an improvement is evidentand is significant after four weeks for disabil-ity. Notably the improvements seem to besustained through to the sixth month. Therepeated measures analysis of variance showsthat the reduction of disability over time issignificant (F = 3-87; p = 0-05). For pain, thedifference from baseline to four weeks and tosix months showed more than a 15% reduc-tion. The trend over time again showed a sus-tained improvement and with repeatedmeasures analysis of variance it was highlysignificant (F = 10-39; p = 0001). Depressionshowed an important improvement in thatthe patients changed from a measured mildlydepressed state (Beck depression inventoryscore >11) to a score within normal limits(Beck depression inventory <10). The trendover time was clearly significant (F = 6-97; p= 0.01).

A very important finding (despite the factthat the hypothesis was not declared inadvance) showed that all the improvementsseem limited to the women in the study. Thisphenomenon, to be reported extensively in alater paper, was characterised by both meansand trends over time as significant for the74% of subjects who were women and not atall for the men.

DiscussionInnovative treatments, such as the MMCBTprogramme that we adapted and standardizedin Sweden, tend to be implemented withmuch enthusiasm but with little evaluativerigour about their effectiveness.34 That isespecially true when the new approach makesclinical and common sense and deals with anintractable type of personal and social prob-lem. The enthusiasts commonly retreat to theargument that "you can't lose!". The innova-tion then takes on a life of its own, attractszealous supporters and society indeed canlose. Many ineffective treatments are costlyand divert scarce resources from other moreeffective strategies of care, prevention, orrehabilitation. Unfortunately once a newintervention moves into the mainstream ofusual care it is judged unethical to evaluate itwith controlled studies. In the case of theMMCBT we were determined to do as rigor-ous an evaluation of the programme new toSweden as we could soon after the MMCBTintervention was packaged and pretested infeasibility studies.Why not a randomised controlled double

blind clinical trial? Regulations and agree-ments between the NHI authority and ourclinic do not permit us to withhold treatmentfrom any referred patient. When weattempted to do a prereferral randomisedcontrolled trial on eligible patients the planwas not acceptable to the authority on ethicalgrounds. Double blinding is not possible dueto the nature of the intervention; completesingle blinding became impossible becauseof the small budget available to engage dupli-cate research staff for the observations. Still

149

on April 29, 2022 by guest. P

rotected by copyright.http://oem

.bmj.com

/O

ccup Environ M

ed: first published as 10.1136/oem.51.3.145 on 1 M

arch 1994. Dow

nloaded from

0Jensen, Nygren, Lundin

wishing to do as good an evaluation as possi-ble we opted for an observational compara-tive cohort study with the advantagesafforded by it being prospective and less vul-nerable to bias than other non-experimentaldesigns such as an uncontrolled before andafter study or a case control method.We did sample size calculations but they

proved to be irrelevant; we were dependenton the referral rates, refusal rates, ineligibility,and matching impossibilities within the timeimplicitly allotted to the study by the fundingagency. Our two small matched cohorts of 35subjects were the largest we could accrue inone year (although we assembled a total of 88eligible referred patients who were treatedand measured). To our surprise, several ofthe priority endpoints did show significantdifferences in trends over time pointing toeffectiveness of the MMCBT programmedespite the small sample size. Moreover, allthe variables pointed in a consistent directionand with a similar size of effect. We recognisethat these four outcomes are likely to be cor-related. Nevertheless, our findings have clini-cal plausibility.

Biases can never be completely eliminatedfrom an observational study. Yet, by design,this prospective cohort study assured thatexposure to treatment was established beforethe outcomes could be known. Beyond that,we believe that we have minimised the biasesin several ways. Firstly, the patients in thestudy were less than a third of the patientload of the clinic during the year of accrualand the treatment team was unaware ofwhich patients were in the study or thematched cohort. Secondly, the patients andcontrols were only aware of the general pur-pose of the study and not the specifichypotheses. Thus they could not alter theirresponses to please their therapists or expressanimosity to anyone in any way. Thirdly,most of the data gathering was self adminis-tered so the opportunity of any informationbias mediated by the evaluating personnel ofthe study was reduced. Fourthly, for the mostimportant variable, absenteeism, the informa-tion was obtained by record linkage with theNHI authority's data banks. Total absen-teeism thus obtained is hard data. For someof the softer data we used validated scales orquestionnaires whenever possible.

External validity and the appropriatenessof extrapolating this experience to all Swedishworkers affected by chronic spinal pain is yetto be established. This is a promising begin-ning and we would not wish that clinicians orpolicy makers overinterpret the findings asdefinitive evidence that the MMCBT shouldbe widely implemented without further evalu-ation or at least surveillance.We were most encouraged that gains

observed at four weeks seemed to be pre-served at six months. To achieve such sus-tained improvement had been a centralobjective of our group when we designed andadapted the cognitive behavioural strategiesfor workers. The improvements, however,were seen only in women among our study

subjects with intractable pain of the spine.We found the extent of the sex differencessurprising, but sex specific effects have beendescribed before. Lindstrom et al 13 forinstance, found improvements in men only ina randomised controlled trial of graded activ-ity intervention. In other studies of health andillness women have been shown to have amore complex pattern of factors associatedwith selected illnesses.14 15 (And Jensen et al,unpublished data.)Our cohort study has some important limi-

tations-namely, uncertain generalisablevalue, lack of power in instances where thedifferences of means were not significant, andthe impossibility of setting aside completelyconcerns about potentially important biasesin a non-experimental or observational study.Comparative cohort studies, however, aresecond only to randomised controlled trials inthe weight that the derived evidence carries.In this evaluation we believe we have uncov-ered good evidence (not opinion) that anMMCBT programme for workers withchronic spinal pain can work to reintegratethem to their jobs and improve their wellbe-ing. To rebut this evidence would require amuch larger cohort study or, preferably, anexperimental controlled trial. We think thatsuch a major study should become a researchpriority not only in Sweden but in othercountries where occupational health researchis well developed. The burden of illness, suf-fering, disability, and indirect, and directcosts of spinal ailments occurring in theworkplace is too high for financial, regulatory,and administrative constraints to prevent animportant definitive study with adequatesample size. The required investment inremodelling and refining the therapeuticintervention so that benefits might even begreater and so that observed effectiveness ofthe MMCBT might extend to men is also ofcompelling importance. This study has shownthat such research can be done in the realworld of the workers' environment and of therehabilitation centre.

We acknowledge the advice given in design and analysis byProfessor WO Spitzer of the Department of Epidemiology andBiostatistics of McGill University, Montreal, Canada. We alsoacknowledge the assistance of Irene Larsson in the collectionof data. Further we thank Anders Hfgg for his excellent workwith the computerised data base. We graciously acknowledgethat without financial support from AMF, Trygghets-forsakringar this research would not have been possible.

1 Spitzer WO, LeBlanc FE, Dupuis M. Scientific approachto the assessment and management of activity relatedspinal disorders. Spine 1987;12:7S.

2 Nachemson A. Ont i ryggen, Statens beredning for utvarder-ing av medicinsk metodik. Stockholm: SBU report, 1991.

3 Flor H, Fydrich T, Turk D. Efficacy of multidisciplinarypam treatment centres: a meta-analytic review. Pain1992;49:221-30.

4 Corry A, Linssen G, Spinhoven P. Multimodal treatmentprogrammes for chronic pain: a quantitative analysis ofexisting research data. JPsychosm Res 1992;36:2275-86.

5 Linton SJ, Bradley LA, Jensen I, et al. The secondary pre-vention of low back pain: a controlled study with follow-up. Pain 1989;36:197-207.

6 Nygren A. Areprojektet. Rehabilitering av patienter medsmirttillstAnd i nacke/skuldra. Stockholm: Folksam,1992.

7 Kamwendo K. Neck and shoulder disorders in secretaries.

150

on April 29, 2022 by guest. P

rotected by copyright.http://oem

.bmj.com

/O

ccup Environ M

ed: first published as 10.1136/oem.51.3.145 on 1 M

arch 1994. Dow

nloaded from

Cognitive-behavioural treatmentfor workers with chronic spinal pain

Prevalence, risk factors and neck school intervention.Lund: University of Lund, 1991. (Doctoral thesis.)

8 Beck AT. An inventory for measuring depression, ArchGen Psychiatty, 1961;4;53-63.

9 Pincus T, Summey JA, Soraci SA, et al. Assessment ofpatient satisfaction in activities of daily living using amodified Stanford Health Assessment Questionnaire.Arthritis Rheum, 1983;26; 1346-53.

10 Nicassio PM, Wallstone KA, Callahan LF, et al. Themeasuring of helplessness in rheumatoid arthritis. Thedevelopment of the arthritis helplessness index.J Rheumatol 1985;12:462-7.

11 Rosenstiel AK, Keefe FJ. The use of coping strategies inchronic low back pain patients: relationship to patientcharacteristics and current adjustment. Pain 1983;17:33-44.

12 Roscoe JT. Fundamental research statistics for the behav-ioral sciences. USA: Holt, Rinehart, and Winstone, Inc,1975.

13 Lindstrom I, Ohlund C, Eek C, et al. The effect of gradedactivity on patients with sub acute low back pain: a ran-domized prospective clinical study with an operant-con-ditioning behavioural approach. Phys Ther 1992;72:279-93.

14 Frankenhauser M, Lundberg U., Fredrikson M, et al.Stress on and off the job as related to sex and occupa-tional status in white-collar workers. Journal ofOrganizational Behaviour 1989;10:321-46.

15 Schumaker S, Hill RD. Gender differences in socialsupport and physical health. Health Psychol 1991;10:102-11.

Vancouver styleAll manuscripts submitted to Occup EnvironMed should conform to the uniformrequirements for manuscripts submitted tobiomedical journals (known as theVancouver style.)

Occup Environ Med, together with manyother international biomedical journals, hasagreed to accept articles prepared in accor-dance with the Vancouver style. The style(described in full in the BM7, 24 February1979, p 532) is intended to standardizerequirements for authors.

References should be numbered consec-utively in the order in which they are firstmentioned in the text by Arabic numeralsabove the line on each occasion the refer-ence is cited (Manson' confirmed otherreports2-5 . . .). In future references topapers submitted to Occup Environ Med

should include: the names of all authors ifthere are seven or less or, if there are more,the first six followed by et al; the title ofjournal articles or book chapters; the titlesof journals abbreviated according to thestyle of Index Medicus; and the first andfinal page numbers of the article or chap-ter. Titles not in Index Medicus should begiven in full.

Examples of common forms of refer-ences are:

1 International Steering Committee of Medical Editors,Uniform requirements for manuscripts submitted tobiomedical journals. BrMedJ7 1979;1:532-5.

2 Soter NA, Wasserman SI, Austen KF. Cold urticaria:release into the circulation of histamine and eosino-phil chemotactic factor of anaphylaxis during coldchallenge. N EnglJ Med 1976;294:687-90.

3 Weinstein L, Swartz MN. Pathogenic properties ofinvading micro-organisms. In: Sodeman WA Jr,Sodeman WA, eds. Pathologic physiology, mechanismsof disease. Philadelphia: W B Saunders, 1974:457-72.

151

on April 29, 2022 by guest. P

rotected by copyright.http://oem

.bmj.com

/O

ccup Environ M

ed: first published as 10.1136/oem.51.3.145 on 1 M

arch 1994. Dow

nloaded from