4 <strong>Australasian</strong> <strong>Anaesthesia</strong> <strong>2011</strong> 5Table 1. Previous (anachronistic) names for complex regional pain syndromeReflex sympathetic dystrophySudeck’s atrophySudeck’s osteodystrophyCausalgiaAlgodystrophyShoulder-h<strong>and</strong> syndromeTable 2. Diagnosis of CRPS9Clinical use: 1 or more symptoms from 3 or more categories <strong>and</strong> one or more signs from 2 or more categories –sensitivity 0.85, specificity 0.6Research use: 1 or more symptoms from all categories <strong>and</strong> one or more signs from 2 or more categories –sensitivity 0.70, specificity 0.96Sensory abnormalitiesVascular abnormalitiesOedema or sweating abnormalitiesMotor or trophic changesSpontaneous painMechanical hyperalgesiaThermal hyperalgesiaDeep somatic hyperalgesiaVasodilatationAsymmetric skin temperaturesSkin colour changesSwellingHyperhidrosisWeaknessTremorDystoniaImpaired co-ordinationNail or hair changesSkin atrophyJoint stiffnessSoft tissue changesREFERENCES1. http://pdver.atcomputing.nl/pdf/CRPS_I_Guidelines.pdf Guidelines complex regional pain syndrome type 1Netherl<strong>and</strong>s Society of Rehabilitation Specialists <strong>and</strong> Netherl<strong>and</strong>s Society of Anaesthesiologists.2. Zollinger P, Tuinebreijer W, Breederveld R, Kries R Can vitamin C prevent complex regional pain syndrome inpatients with wrist fractures? J Bone Joint Surg am 2007;1424-1431.3. Kemmler MA, De Wet HC et al The effect of spinal cord stimulation in patients with reflex sympathetic dystrophy:two years’ follow-up of the r<strong>and</strong>omized control trial Am Neurol 2004;55:13-18.4. Marinus J, Moseley L, Birklein F, et al Clinical features <strong>and</strong> pathophysiology of complex regional pain syndromewww.thelancet.com/neurology<strong>2011</strong>;10:637-648.5. Bruehl S An Update on the Pathophysiology of Complex Regional Pain Syndrome Anesthesiology2010; 113: 713-725.6. Maihofner C, Seifert F, Markovic K Complex regional pain syndromes: new pathophysiological concepts <strong>and</strong>therapies European J Neurology 2010; 17: 649-660.7. Moseley GL.Graded motor imagery for pathologic pain: a r<strong>and</strong>omized controlled trial. Neurology. 2006;67:2129-34.8. Daly AE, Bialocerkowski AE. Does evidence support physiotherapy management of adult Complex RegionalPain Syndrome Type One? A systematic review. Eur J Pain. 2009; 13: 339-53.9. Baron R, Janig W Complex regional pain syndromes-how do we escape the diagnostic trap? Lancet2004;364:1739-1741.
The flimsy framework of methodology in the acute pain literature – shaky structure in need of repair? 7The flimsy framework of methodology in the acute pain literature –shaky structure in need of repair?MARK REEVES, MBBS, FANZCA, PGDIPBIOSTAT.Senior Clinical Research Fellow, University of Tasmania <strong>and</strong> Visiting Medical Officer, North West Regional Hospital,Burnie TasmaniaDr Reeves has been an anaesthetist in rural public <strong>and</strong> private practice in north-west Tasmania for 10 years. He isa Primary examiner <strong>and</strong> Councillor for ANZCA. His research interests include acute pain <strong>and</strong> statistics.INTRODUCTIONThe revelation that a respected author had been fraudulently altering data in his clinical trials sent shockwavesthrough the anesthesia research community. 1,2 Scott Reuben’s particular field of research had been acute <strong>and</strong>chronic pain medicine. Whilst his motivations are not public knowledge the undeniable possibility is that he falsifieddata to get positive results <strong>and</strong> that he wanted positive results because these would be more easily published.PUBLICATION BIAS IN THE ACUTE PAIN LITERATUREPublication bias is alive <strong>and</strong> well, as evidenced by the simple fact that the vast majority of published trials in acutepain do have a positive result. 3 A negative trial tends to attract the obvious criticism of “Type 2 error” – the treatmenteffect was not statistically significant because there were not enough subjects in the groups to reliably detecta difference between them. Thus, the trial does not add to our knowledge if the treatment was novel <strong>and</strong> does notst<strong>and</strong> as a refutation if the trial was a repeat of work previously published. Ironically, no such criticism appears tobe leveled at positive trials. One rarely reads a clinical trial with a discussion of “Type 1 error” – the possibility thatthe finding of a treatment effect was merely a chance occurrence <strong>and</strong> the treatment has, in reality, no discernableeffect.At first blush, this may not appear to be a major problem, other than to give the impression that the weight ofpublished evidence favours a specific intervention. However, when these trials are collated in a meta-analysis, anyType 1 error can be immediately conflated as it is difficult to source the unpublished negative trials. Now the apparenttreatment effect has narrower confidence limits <strong>and</strong> increased “significance”. Perhaps the more dangerousinterpretation is that an intervention may now have the imprimatur of a meta-analysis <strong>and</strong> be considered “Level 1”evidence. To my thinking this is bizarre, as the included trials are not necessarily a cross-section of all trials in thearea but just those that have been published. Also, those performing the analysis rarely have equipoise as theywould have to be interested in an area to wish to perform such an analysis. Equally they can hardly be blinded tothe provenance of a paper when determining methodological quality as the outcome is available <strong>and</strong> the literatureis often familiar to any researcher with an interest in that area. I have had the honour of having a single clinical trial(4) examined in three separate meta-analyses, all examining the impact of ketamine on postoperative analgesia.In one it received a quality score of 5/5 5 , in a second 3/5 (using the same system) 6 <strong>and</strong> in the third it was excludeddue to “methodological flaws” 7 . I cannot believe my experience is isolated. Meta-analyses have an important placein the literature, but mostly in the area of hypothesis generation (if positive) or to refute treatments (if negative).CHOICE OF ENDPOINTS IN THE ACUTE PAIN LITERATUREThe most frequent type of clinical trial on acute postoperative pain in the major anaesthesia journals is a comparisonof one analgesic strategy to another. The commonest overall design is the comparison of an analgesic regime orstrategy against placebo, measuring the endpoints of opioid requirements <strong>and</strong> pain scores between the groups,usually after major surgery. The theory is undeniably attractive – an intervention leads to reduced opioid requirements(<strong>and</strong> possibly therefore reduced opioid-related side-effects) or reduced pain scores (or even both). If only the formerthen this is because every participant is seeking a specific pain score <strong>and</strong> those in the treatment group require lessopioid to achieve this. If only the latter, then somehow the intervention enhances the quality of the analgesia, evenif not the requirement for opioid.The particular attraction of these endpoints – <strong>and</strong> these are the ones on which virtually all such trials are predicatedin the a priori power analysis – is that the variables are continuous, ratio data <strong>and</strong> can thus be compared withparametric analyses such as Students t-test or analysis of variance (ANOVA). Or are they?