Combined physiotherapy and education is efficacious for chronic low back pain

Manual therapy, exercise and education target distinct aspects of chronic low back pain and probably have distinct effects.This study aimed to determine the efficacy of a combined physiotherapy treatment that comprised all of these strategies. Byconcealed randomisation, 57 chronic low back pain patients were allocated to either the four-week physiotherapy program ormanagement  as  directed  by  their  general  practitioners.  The  dependent  variables  of  interest  were  pain  and  disability.Assessors were blind to treatment group. Outcome data from 49 subjects (86%) showed a significant treatment effect. Thephysiotherapy program reduced pain and disability by a mean of 1.5/10 points on a numerical rating scale (95% CI 0.7 to 2.3)and 3.9 points on the 18-point Roland Morris Disability Questionnaire (95% CI 2 to 5.8) , respectively. The number needed totreat  in  order  to  gain  a  clinically  meaningful  change  was  3  (95%  CI  3  to  8)  for  pain,  and  2  (95%  CI  2  to  5)  for  disability. Atreatment effect was maintained at one-year follow-up. The findings support the efficacy of combined physiotherapy treatmentin producing symptomatic and functional change in moderately disabled chronic low back pain patients.[Moseley L (2002):Combined  physiotherapy  and  education  is  efficacious  for  chronic  low  back  pain.Australian  Journal  ofPhysiotherapy 48: 297-302]

Introduction

Chronic low back pain is a multifactorial phenomenon andit is not surprising that many therapeutic approaches exist.Some  approaches  are  ineffective  but  manual  therapy,specific muscle training, cognitive behavioural therapy andmultidisciplinary pain management programs have all beensupported.  Distinct  approaches  tend  to  target  distincteffects. For example, manual therapy (mobilisation and/ormanipulation of the lumbar spine) is effective in reducingpain (Andersson et al 1999, Cherkin et al 1998, Triano et al1995), and specific exercise programs which aim to restorenormal  function  of  the  trunk  muscles  are  effective  inreducing  disability  and  increasing  performance  at  least  inmildly disabled patients (Mannion et al 1999, O’Sullivan etal  1997).  Although  they  may  be  effective  in  producingbenefit  across  outcome  domains,  these  approaches  do  notdirectly  address  psychosocial  aspects  of  pain.  In  somepatients, such aspects are thought to be enduring barriers toimprovement. These patients may obtain more benefit fromprograms  that  directly  address  cognitive  and  behaviouralaspects of pain.Multidisciplinary  pain  management  programs,  whichusually  focus  on  cognitive  and  behavioural  aspects,  areprimarily  effective  in  reducing  disability,  promoting  self-efficacy  and  normalising  pain  cognitions  (Guzman  et  al2001,  McQuay  et  al  1997).  Not  all  programs  are alike;more intensive programs, although more expensive, appearto  be  more  effective  (Guzman  et  al  2001,  Williams  et  al1996). The fact that they require substantial personnel andeconomic  resources  is  a  limitation  of  multidisciplinary programs.  An alternative  way  to target  cognitive andbehavioural aspects of chronic low back pain is through thetargeted provision of information. Education in this mannerattempts  to effect  change  through  reconceptualisation  ofthe problem. Employing  an  education  approach  may  appear  surprisingbecause  for  some  time,  there  has been a consensus  thatthere  is  no clinically  important  effect  of  educationprograms  for  chronic  low  back  pain  (Cohen  et  al  1994).Reviews   on   education-based   back   schools   appearconvincing  (Koes  et  al  1994),  however,  recent  evidencesuggests that the lack of effect is probably due to the typeof  information  that  has  been  presented.  Studies  that  haveemployed  an  approach  to  education that  emphasisescognitive-behavioural  (Burton  et  al  1999, Linton  andAndersson  2000, Symonds et  al 1995)     orneurophysiological  (Moseley  et  al  2001)  aspects  havereported reduced disability, reduced health care utilisation,normalisation  of  pain  cognitions,  and  increased  self-efficacy.

Manual  therapy,  specific  exercise  training  and  targetededucation all seem to promote therapeutic success throughtargeting  distinct  aspects  of  chronic  low  back  pain.Although  each  of  these  strategies  is  broadly  encompassedwithin  the  domain  of  physiotherapy,  the  effect  of  acombined physiotherapy treatment that consists of all threestrategies  is  not  known.  The  aim  of  this  study  was  todetermine  the  effect  of  such  a  combined  physiotherapytreatment  on  functional  and  symptomatic  parameters  ofmoderately disabled patients with chronic low back pain.

Methods

Experimental  design    This  study  was  a  randomisedcontrolled  trial  with  repeated  measures  comparison  ofmeans.  The  study  was  approved  by  the  InstitutionalMedical  Research  Ethics  Committee  and  all  proceduresconformed with the Declaration of Helsinki.Subjects    Sixty-two  subjects  volunteered  for  the  study  byresponding  to  a  note  that  advertised  the  project.  The  notewas included in the material given to each patient on initialattendance  at  participating  physiotherapy  clinics  or  thereferring general practitioner. Subjects were included if theprimary reason for presentation was a history of low backpain of greater than two months. Subjects were excluded ifthey  were  unable  to  understand,  read  and  speak  English,had  worsening  neural  signs,  had  any  neurological  ororthopaedic condition that would interfere with treatment,or were awaiting surgery. Five subjects were excluded. While each subject was undertaking the initial assessment,an  independent  person  allocated  them  to  experimentalgroup  using  a  coin  toss.  This  strategy  ensured  thatallocation was concealed from the subjects until after initialassessment,  and  from  the  assessors  throughout  the  study.Twenty-nine   and   28   subjects   were   allocated   to   thephysiotherapy  and  control  groups  respectively.  Figure  1presents the recruitment strategy and experimental plan.

Experimental protocol  The following items were used asoutcome  measures:  the  18-item  Roland  Morris  DisabilityQuestionnaire (RMDQ; Roland and Morris 1983) and the0-10 Numerical Rating Scale (NRS) for pain (“How wouldyou rate your low back pain, on average, over the last threedays?”). Initial and final assessment was performed by thesame two investigators, who were not otherwise involved inthe  study  and  were  blinded  to  experimental  group.  One-year  follow-up  data  were  collected  via  telephone  byseparate  assessors  who  were  also  blinded  to  experimentalgroup. The properties of the RMDQ and the NRS for painare  thought  to  be  maintained  when  administered  over  thephone  (Cherkin  et  al  1998).  A  further  question  estimatedthe  number  of  health  visits  for  low  back  pain  over  thecourse of the follow-up period: “Since your assessment on[date  of  final  assessment],  how  many  times  have  youconsulted  a  health  care  professional  for  your  low  backpain?”

Treatment    protocol        Each    subject    received    twophysiotherapy treatments per week for four weeks. Manualtherapy    treatment    involved    symptom    managementaccording to the discretion of the treating physiotherapist,who  chose  from  spinal  mobilisation/manipulation,  softtissue    massage,    and    muscle    and    neuromeningealmobilisation     techniques,     but     not     electrophysicalmodalities.

Each subject participated in specific trunk muscle trainingboth on an individualised level on two occasions per weekand  through a  standardised  home-exercise  program.  Thisprogram was  conducted  according to the protocol described by Richardson and colleagues (Richardson  andJull 1995). Subjects were instructed to maintain the homeprogram indefinitely. Compliance with the home programwas not assessed. Each subject participated in a one-hour education session,once per week for four weeks. The education session was ina  one-to-one seminar  format, was conducted  by  anindependent therapist, and focused on the neurophysiologyof pain with no particular reference to the lumbar spine. Inaddition,  the  subjects  completed  a  short  workbook  whichconsisted of one  page  of  revision  material and  threecomprehension exercises per day for 10 days.Subjects  in  the  control  group  received  ongoing  medicalmanagement as advised by their general practitioner. Thesesubjects  were also  advised  not  to seek physiotherapytreatment during the data collection period. Subjects in thecontrol group were questioned after the final assessment asto what intervention, if any, they had since initialassessment and how many visits to the general practitionerthey had made for their low back pain.Analysis  Two-factor repeated measures ANOVAs (groupxtime)  were used  to  identify  a treatment  effect on  thedependent  variables  at  final  assessment and at one-yearfollow-up.  Numerical  Rating  Scale and RMDQ  were  thedependent variables. Because two separate ANOVAs wereused in the analysis, the probability of a Type 1 error for thestudy was elevated.  To  adjust  for  this, a  Bonferronicorrection yieldedα= 0.025. Analysis was by intention totreat.

 For  those  dependent  variables  in  which  the  group  xtimeinteraction   was   significant,   treatment   affects   wereestimated  from  the  difference  in  group  means.  For  NRSand RMDQ, the number needed to treat (NTT) in order togain  a  clinically  significant  change  was  also  determined.The  threshold  for  a  clinically  significant  change  in  NRSand  RMDQ  was  set  a-priori at  2  points  and  4  points,respectively.  These  values  were  selected  according  toestimates in the literature, (eg Stratford et al 1994, Turk andMelzack 1992).

Results

Subject  details    Table  1  shows  the  subject  characteristics.There  were  no  pre-treatment  differences  between  thegroups  in  any  of  these  measures  or  in  the  dependentvariables (p> 0.31). On final assessment, the mean numberof visits to the general practitioner was 4 (SD 2). Eighteenof the subjects in the control group indicated that they hadbeen   prescribed   physical   exercises   by   their   generalpractitioner previous to the initial assessment, but only six subjects  had  these  exercises  reviewed  during  the  datacollection  period.  Six  subjects  indicated  that  they  hadreceived   weekly   manipulations   from   their   generalpractitioner and nine subjects indicated that their pain reliefmedications had been increased or altered during the datacollection  period.  Two  control  subjects  received  weeklyanalgesic injections during the data collection period.Pain and disability  Final assessment was performed 29 ±6  days  after  the  initial  assessment.  There  was  a  meanreduction  of  2.9/10  and  1.4/10  on  the  NRS  for  pain,  and8.2/18   and   4.3/18   points   on   the   RMDQ,   for   thephysiotherapy  treatment  and  control  groups  respectively(Figure  2).  Thus,  the  mean  improvement  effected  byphysiotherapy  treatment  was  1.5  points  on  the  NRS  forpain  (95%  CI  0.7  to  2.3)  and  3.9  points  on  the  RMDQ(95%  CI  2.0  to  5.8).  The  repeated  measures  ANOVAsindicated a significant treatment effect on NRS and RMDQ(p<0.01 for both). The number needed to treat (NNT) togain a clinically significant change was 3 (95% CI 2 to 8)for the NRS and 2 (95% CI 2 to 5) for the RMDQ. Twelve-month  follow-up    Nineteen  subjects  in  each  group(67% of total sample) were contactable at one year (mean ±SD = 352 ± 28 days) for follow-up. There were significanttreatment effects on NRS and RMDQ, and on the number ofhealth  care  visits  for  low  back  pain  during  the  follow-upperiod  (Figure  2).  The  treatment  effect  was  1.9  for  pain(95% CI 1 to 2.8) and 3.9 points on the RMDQ (2.3 to 5.8)corresponding to numbers needed to treat of 2 (95% CI 1 to4) and 2 (1 to 3) respectively. During the one year since finalassessment, subjects from the physiotherapy group made amean ± SD 3.6 ± 2 health care visits for their low back pain,which  was  fewer  than  the  control  group,  who  attended  amean ± SD 13.2 ± 5 health care visits (p< 0.001). Thus theeffect of treatment was to reduce the number of health carevisits by a mean of 9.6 (95% CI 6.9 to 11.9).Withdrawals,  dropouts  and  side  effects    Four  and  threesubjects,  from  the  physiotherapy  group  and  control  grouprespectively,  dropped  out  of  the  study  and  could  not  becontacted.  One  subject  from  the  physiotherapy  groupwithdrew due to urgent surgery unrelated to low back pain.Twenty-four  and  25  subjects  completed  the  physiotherapyand  control  programs  respectively.  The  pre-treatment  datafor  those  subjects  included  in  the  follow  up  showed  nodifferences      between the experimental groups (t-test, p> 0.21). 

Discussion

These  findings  show  that  a  combined  physiotherapytreatment  consisting  of  manual  therapy,  specific  exercisetraining,  and  neurophysiology  education  is  effective  inproducing  functional  and  symptomatic  improvement  inchronic  low  back  pain  patients.  This  is  evidenced  by  asignificant  treatment  effect  and  substantial  effect  size  forpain and disability, both of which appear to be maintainedfor at least one year.

The   effectiveness   of   the   physiotherapy   program   is substantiated  by  the  NNT  analysis.  One  advantage  of  theNNT  is  that  it  provides  a  clinically  relevant  indication  ofthe number of patients that need to be treated for one morepatient  to  achieve  a  particular  therapeutic  target.  In  short,an NNT of 1 suggests that the desired target is achieved inevery patient in the treatment group but in no patient in thecontrol group. Thus, the closer the NNT is to 1, the betterthe  treatment  is  at  achieving  the  targeted  outcome.  TheNNTs in the current work were 3 (pain) and 2 (disability),which   are   consistent   with   recommendations   in   theliterature that stipulate that, for chronic pain, NNTs of 2 or3 are indicative of an effective intervention (McQuay et al1997)

The    current    results    suggest    that    the    combinedphysiotherapy treatment is probably more effective than thecomponents  administered  in  isolation.  This  is  primarilyevidenced  by  the  fact  that  most  of  the  effects  of  soletreatments reported in the literature are small, particularlyin  those  studies  that  involved  subjects  with  high  initialdisability  levels.  For  example,  manipulation  has  beenreported  to  produce  effects  of  2/10  and  3  RMDQ  points(Cherkin  et  al  1998),  1.6/10  (Andersson  et  al  1999)  and2.5/10  (Triano  et  al  1995);  exercise  has  been  reported  toproduce  effects  of  1.2/10  and  2.9  RMDQ  points  (KlaberMoffett  et  al  1999);  and  education  has  been  reported  toproduce effects of (1/10 and 2.5 RMDQ points (Cherkin etal  1998),  0/10  and  1  RMDQ  point  (Moseley  et  al  2001).Even  so,  chronic  low  back  pain  is  heterogeneous  andsubjects  vary  across  studies  in  their  chronicity,  painintensity, functional level and pain impact. This means thatthe validity of a comparison between the current work andother studies is limited. For  this  study,  it  is  ultimately  impossible  to  isolate  thecontribution  that  each  component  treatment  made  to  theoutcome  of  the  combined  treatment.  In  future  studies,teasing   out   the   relative   contribution   of   componentstrategies  to  the  therapeutic  effect  may  allow  conclusionsabout the mechanisms involved. This may, in turn, enhancethe  efficacy  of  combined  physiotherapy  treatments  andpermit targeting of sub-groups of patients with chronic lowback pain.Considering  the  high  economic  cost  of  chronic  low  backpain, targeting of sub-groups may be beneficial. This studysuggests that a combined physiotherapy treatment is a costeffective  strategy  when  targeted  at  moderately  disabledpatients with chronic low back pain; based on A$60.00 persession, the estimated cost of the combined treatment wasA$720, which compares favourably with multidisciplinarypain  management  programs  that  can  cost  in  the  order  ofA$4000  (Moseley  1997,  unpublished  data).  Importantly,however,  the  combined  physiotherapy  treatment  may  beneither  cost-effective  nor  efficacious  in  more  disabledpatients with chronic low back pain or in other sub-groupsof  patients  with  chronic  pain.  Psychosocial  factors  arethought to be more important in some sub-groups and thereis  considerable  evidence  in  support  of  more  intensive,albeit more expensive, strategies for such patients (Guzmanet al 2001, Morley et al 1999). One aspect of the current study that is open to criticism isthe  lack  of  a  robust  control  group.  Although  “ongoingmedical  management”  is  used  widely  and  generallyaccepted   as   suitable   for   clinical   trials,   it   does   notadequately remove many sources of bias. By and large, thisshortcoming is ignored in the relevant literature (eg Bendixet  al  1997,  Deyo  1996,  Hides  et  al  1996,  Laclaire  et  al1996,  O’Sullivan  et  al  1998,  van  der  Heijden  et  al  1995).However,    non-treatment    factors    such    as    patientexpectations  (Carosella  et  al  1994,  Montgomery  andKirsch  1997),  health  provider  expectations  (Gracely  et  al1985, Shapiro et al 1954), patient-provider rapport (Egbertet  al  1964),  therapist  enthusiasm  and  perceived  level  of expertise  (Nordin  et  al  1998,  Shapiro  and  Shapiro  1984)are all considered to contribute to therapeutic effect. Bias  may  also  be  introduced  by  the  Hawthorne  effect,which is caused by knowledge that one is participating in aresearch  study  (Parson  1974).  However,  if  the  Hawthorneeffect  varies  according  to  how   much the  subject  isparticipating (this certainly seems reasonable), then in thepresent  work,  the  Hawthorne  effect  would  have  beengreater  in  the  physiotherapy  group.  Thus,  although  thecurrent findings appear potent, further trials incorporatinga  more  robust  control  group  probably  are  required  tosubstantiate the results of the current study. One  source  of  bias  that  may  limit  the  external  validity  ofthis  work  is  selection  bias  introduced  by  the  exclusioncriterion  that  subjects  have  an  ability  to  read,  speak  andunderstand English. Although 28% of subjects were from anon-English speaking background, broad application of thecurrent findings to linguistically diverse chronic low backpain patients would appear problematic. Notwithstanding  the  potential  limitations  of  the  currentwork,  the  results  strongly  suggest  that  the  combinedphysiotherapy  treatment,  consisting  of  manual  therapy,specific exercise training, and neurophysiology education,is  effective  in  producing  functional  and  symptomaticimprovement in chronic low back pain patients. The effectis  maintained  at  12  months  post-treatment  and  patientssubsequently  seek  substantially  fewer  health  care  visitsthan  those  under  ongoing  medical  care.  The  findingspresented  here  are  important  because  they  support  thelong-term  efficacy  of  this  approach  for  a  problematicpatient  group:  moderately  disabled  patients  with  chroniclow back pain. 

References

1. Andersson GBJ, Lucente T, Davix AM, Kappler RE, LiptonJAand Leurgans S (1999): A comparison of osteopathicspinal  manipulation  with  standard  care  for  patients  withlow  back  pain.New  England  Journal  of  Medicine341: 1426-1431.
2. Bendix  AF,  Bendix  T,  Lund  C,  Kirkbak  S  and  Ostenfeld  S(1997):  Comparison  of  three  intensive  programs  forchronic    low    back    pain    patients:   a    prospective,randomized, observer-blinded study with one-year follow-up.Scandinavian  Journal  of  Rehabilitation  Medicine29: 81-89.
3. Burton  AK,  Waddell  G,  Tillotson  KM  and  Summerton  N(1999): Information and advice to patients with back paincan have a positive effect. A randomized controlled trial ofa  novel  educational  booklet  in  primary  care.Spine24: 2484-2491
4. Australian Journal of Physiotherapy 2002  Vol. 48302Cherkin  DC,  Deyo  RA,  Battie  M,  Street  J  and  Barlow  W(1998):  A  comparison  of  physical  therapy,  chiropracticmanipulation, and provision of an educational booklet forthe treatment of patients with low back pain.New EnglandJournal of Medicine 339: 1021-1029