MINERVA MEDICA COPYRIGHT. One of the most promising techniques for the recovery

Transcription

1 EUR J PHYS REHABIL MED 2010;46: Constraint-induced movement therapy in stroke patients: systematic review and meta-analysis Aim. Upper extremity paresis is a leading cause of disability after stroke. A Cochrane review found an impact on disability of Constraint-Induced Movement Therapy (CIMT), its modified forms (mcimt) and Forced Use (FU), with a moderate significant effect and a large significant effect on arm motor function. This article aims to present an update of the Cochrane review and assess the effects of CIMT, mcimt and FU on disability and arm motor function. Methods. Electronic databases were searched for Randomised Controlled Trials (RCT) and quasi-rcts comparing CIMT, mcimt or FU with other rehabilitative techniques, or none, in adult stroke patients. The primary and secondary outcomes were disability and arm motor function. Two reviewers independently screened search results, documented the methodological quality and extracted data. Results. Four new studies were added to the previous review, for a total of 18 studies. The updated metaanalyses no longer indicate a benefit of CIMT mcimt and FU on disability (eight studies, 276 participants, Standardised Mean Difference (SMD) 0.21, 95% CI to 0.50), and a moderate benefit on arm motor function (14 studies, 479 participants, SMD 0.44, 95% CI 0.03 to 0.93). Conclusion. New evidence pushes the overall estimate of benefit toward the null effect. The majority of studies were underpowered and imprecise, exposing these analyses to small-study bias. This may explain why accumulation of evidence makes overall estimates in- D. CORBETTA 1, V. SIRTORI 1, L. MOJA 2, R. GATTI 1, 3 1Unit of Functional Recovery, Fondazione Centro San Raffaele del Monte Tabor, Milan, Italy 2Italian Cochrane Centre, Mario Negri Institute for Pharmacological Research, Milan, Italy 3School of Physiotherapy, Vita-Salute San Raffaele University, Milan, Italy consistent. Larger randomised trials to resolve these uncertainties are needed. Key words: Recovery of function - Physical therapy modalities - Stroke - Upper extremity - Review literature as topic. One of the most promising techniques for the recovery of upper extremity movement in stroke patients who has received much interest over the past 15 years is the constraint-induced movement therapy (CIMT). 1 CIMT is a techniques for patients with a good potential of recovery, not excessive spasticity and good cooperation. It is based on the theory of learning non-use 2-4 and has two fundamental principles: 1) forced use of the affected arm by restraining the unaffected arm, during dedicated exercise sections and usual activities of daily living; 2) massed practice (several hours of exercise) of the affected arm through a shaping method, where shaping involves a common operant conditioning method in which a behavioural objective (in this case movement ) is approached in small steps of progressively increasing difficulty. Over the years modified forms of CIMT have been developed, reducing the time of training during the period of restraint, 5, 6 or concentrating only on the use of restraint with no additional Conflicts of interest. None. Acknowledgments. The authors wish to thank Judith Baggott who helped with the revision of this manuscript. Corresponding author: V. Sirtori, Physiotherapist, Unit of Functional Recovery, Fondazione Centro San Raffaele del Monte Tabor, via Olgettina 60, Milan, Italy. sirtori.valeria@hsr.it Vol No. 4 EUROPEAN JOURNAL OF PHYSICAL AND REHABILITATION MEDICINE 537

2 CORBETTA Constraint-induced movement therapy in stroke patients treatment of the affected arm in a forced use condition. 7, 8 In 2009 we published a Cochrane review to evaluate the efficacy of CIMT, modified CIMT (mcimt) or forced used (FU) as rehabilitative techniques in the management of affected upper limb in hemiparetic patients after stroke. 9 Meta-analysis of the primary outcome disability (6 trials with 184 subjects) revealed a moderate, statistically significant impact of CIMT. A secondary meta-analysis focusing on arm motor function (11 studies with 373 participants), an outcome considered by the majority of studies included, showed a larger, statistically significant effect. Despite the apparently positive effect, however, most trials were small and some had methodological shortcomings, so the efficacy of this technique was considered with caution. This article presents the results of the update of the previously cited Cochrane review entitled Constraint-induced movement therapy for upper extremities in stroke patients 9 concerning the effect of the CIMT on disability and arm motor function. Materials and methods Inclusion/exclusion criteria We included randomised or quasi-randomised studies that compared CIMT or mcimt or FU with other rehabilitative techniques, occupational therapy or physiotherapy in adults with ischemic or haemorrhagic stroke. All interventions were considered irrespective of numbers of hours of training and number of hours of constraint per day, duration of treatment, type of exercise used in training sessions and were pooled and discussed under the heading of CIMT. Outcomes Primary outcome was disability, secondary outcome was arm motor function. CIMT is a behavioural approach to neurorehabilitation derived from basic neuroscience. 10, 11 It fosters the use of the paretic arm in daily life 12 with the aim of exploiting all residual motor potential. On the basis of this rationale, we preferred to use a measure of disability (i.e., functional independence measure) as primary outcome to better identify the arm use in real life. If a study included in the review used more than one scale for the assessing the same outcome category, the analysis was performed by considering the scale most frequently used across included studies. Search strategy An electronic search previously undertaken in any language for MEDLINE, EMBASE, CINAHL and the Physiotherapy Evidence Database (PEDro) have been updated (April 2010). To identify further published, ongoing and unpublished studies reference lists of relevant papers were searched and researchers in the field of CIMT have been contacted. Data selection and statistical analysis Two reviewers independently screened search results for eligibility, assessed methodological quality and extracted data. A contact with the authors of studies was activated for clarification or for the request of unpublished data. The quality of eligible studies was assessed using the Cochrane Handbook for Systematic Reviews of Interventions. 13 Results at the end of treatment were extracted. Standardised mean difference (SMD) was used to cumulate results across studies. The degree of heterogeneity among the trials in the analysis was assessed by the I 2 statistic for each outcome; 14 heterogeneity was judged as substantial with I 2 >50%. 13 Overall estimates were calculated using a random-effects model, irrespective of statistical heterogeneity, since the studies differ in intensity, duration of intervention and measure of outcomes. In the meta-analysis graphs studies were ordered by publication year. Cochrane Review Manager software (RevMan) was used for statistical analysis. 15 Results From January 2009 (when the search for eligible articles for the Cochrane review stopped) to April 2010, seven new potentially eligible RCTs have been published. Four 8, met the inclusion criteria. Eighteen studies with a total of 674 participants were considered in this update. Studies included patients with ischemic or haemorrhage stroke, with time since stroke between 14 days and 92 months. 538 EUROPEAN JOURNAL OF PHYSICAL AND REHABILITATION MEDICINE December 2010

3 Constraint-induced movement therapy in stroke patients CORBETTA Table I. Participants characteristics. Study Sample Size Time since stroke Motor inclusion criteria The motor inclusion criterion was the movement capacity of the arm, evaluated as the ability to extend the metacarpophalangeal and interphalangeal joints, or as arm movement assessed with the National Institutes of Health Stroke Scale (NIHSS), Brunnstrom scale or Chedoke McMaster Impairment Inventory (CMMII). Other inclusion criteria were the absence of cognitive impairment, no excessive spasticity and no Participants characteristics % Female Age range (years) Stroke details Intervention % Right side affected Atteya A Months Active extension 20 at wirst, 10 at metacarpophalangeal NR Brogardh Months Active extension 10 at wirst and two 33 Near NR C 16 fingers, abduction 10 of thumb Dhal AE > Months Active extension 20 at wirst, 10 at metacarpophalangeal 23 Near Ischemic or 78 D 58 D Dromerick Months Score 1 or 2 on the motor item of NIHSS Only ischemic AW 26 Dromerick Months Score 1-3 on the motor item of NIHSS, 60 Near ischemic or AW (b) 17 score 3 in the arm item of the MAS scale Hammer A Months Active extension 20 at wirst, 10 at 23 Near 31- NR fingers 83 Lin KC > Months Brunstrom stage >3 on arm section 34 Near Ischemic or Lin KC (b) > 6 Months Brunstrom stage >3 on arm section Ischemic or Myint M Months Active extension 20 at wirst, 10 at metacarpophalangeal 58 Near Ischemic or 48d 70d Page SJ Months Active extension 20 at wirst, 10 at metacarpophalangeal Only ischaemic Page SJ Months Active extension 20 at wirst, 10 at metacarpophalangeal Only ischaemic NR NR (b) 27 Page SJ (c 25 9> Months Active extension 20 at wirst, 10 at metacarpophalangeal Only ischaemic NR NR ) 28 Ploughman Months Stage 2 < CMMII < Stage 6 34 Near ischemic or M 7 Taub E > Months Active extension 20 at wirst, 10 at metacarpophalangeal 77 NR NR 100 d 100 d Wittenberg 16 9> Months Active extension 20 at wirst, 10 at metacarpophalangeal only ischaemic NR NR GF 29 Wolf SL * 3-9 Months Active extension 20 at wirst, 10 at metacarpophalangeal 36 Near ischaemic or 47.2d 51.7d haemorrhagic Wu CY > Months Active extension 20 at wirst, 10 at metacarpophalangeal 43 Near NR Wu CY 26 0,5-31 Months Brunstrom stage >3 on arm section ischaemic or 46d 54d (b) 24 NR: not reported; D: dominant side *= Based on statistical calculation Control balance problems, including walking. See Table I for participants characteristics. Intervention Interventions in studies are heterogeneous: the time of restraint ranged between 5 and 8 hours/day, the duration of exercise with the affected arm varied approximately from 5 to 45 hours/week with the Vol No. 4 EUROPEAN JOURNAL OF PHYSICAL AND REHABILITATION MEDICINE 539

4 CORBETTA Constraint-induced movement therapy in stroke patients Table II. Characteristics of studies. Study Type of intervention Time of restraint majority of studies from 10 to 20 hours/week. All studies used functional or ADL tasks (mainly done through shaping techniques). The intervention lasted between two and ten weeks. See Table II for more details of the included studies. Outcomes All studies considered pretreatment and post-treatment measures for outcomes. Disability was evaluated in nine studies using the Functional Independence Measure (FIM) or the Barthel Index (BI). Arm motor function was evaluated in 14 studies using the Action Research Arm Test (ARAT), Wolf Motor Function Test (WMFT), Emory Function Test (EMF) and the Motor Assessment scale (MAS). Anatomical region restraint Characteristic of treatment Time of exercise Type of exercise Dosage of practice Atteya A 25 mcimt 5 hours/day Arm plus hand Less than 5 hours/week PNF 30 hours Brogardh C 16 mcimt 90% of waking hours Hand 30 to 45 hours/week Shaping 30 hours Dhal AE 20 CIMT 90% of waking hours Hand 30 to 45 hours/week Shaping >30 hours Dromerick AW 26 CIMT 6 hours/day Hand 10 to 20 hours/week Task oriented >30 hours Dromerick AW mcimt 90% of waking hours (n=16) Hand 10 to 20 hours/week Shaping 30 hours (b) 17 and 6 hours/day (n=19) Hammer A 8 FU 6 hours/day Arm plus hand 10 to 20 hours/week Task oriented 30 hours Lin KC 21 mcimt 6 hours/day Hand 10 to 20 hours/week Shaping 30 hours Lin KC (b) 18 mcimt 6 hours/day Hand 10 to 20 hours/week Shaping 30 hours Myint M 22 CIMT 90% of waking hours Arm plus hand 10 to 20 hours/week Shaping >30 hours Page SJ 6 mcimt 5 hours/day Arm plus hand Less than 5 hours/week PNF 30 hours Page SJ (b) 27 mcimt 5 hours/day Hand Less than 5 hours/week PNF 30 hours Page SJ (c ) 28 mcimt 5 hours/day Arm plus hand Less than 5 hours/week Shaping 30 hours Ploughman M 7 FU Less than 3 hours/day Hand Less than 5 hours/week skilled-task 30 hours Taub E 12 CIMT 90% of waking hours Arm plus hand 30 to 45 hours/week Shaping >30 hours Wittenberg GF 29 CIMT 90% of waking hours Arm plus hand 30 to 45 hours/week Motor task perfor. >30 hours Wolf SL 19 CIMT 90% of waking hours Hand 30 to 45 hours/week Shaping >30 hours Wu CY 23 mcimt 6 hours/day Hand 10 to 20 hours/week Shaping 30 hours Wu CY (b) 24 mcimt 6 hours/day Hand 10 to 20 hours/week Shaping 30 hours Risk of bias Randomization was described and appropriate in 12 studies. The allocation concealment was described and appropriate in six RCTs. The outcome assessor was blinded in all studies except one. In all studies post-treatment withdrawals were less than 10%; only two studies reported information about of the reasons for withdrawing from treatment. All scales used in the studies for primary and secondary outcomes were supported by references about their psychometric properties. The majority of studies were small and likely to be underpowered: the median sample size was 15 patients (interquartile range 10 to 23). The sample size was based on prior statistical power calculations only in the trial by Wolf et al. 19 See Table III for details of methodological quality of the included studies. Primary outcome DISABILITY: CONSTRAINT versus CONTROL Eight studies recruiting 276 participants measured disability immediately after the experimental and control interventions. 7, 17, 18, On disability CIMT seemed to have no significant effect (SMD 0.21, 95% CI to 0.50; Figure 1). We found no relevant heterogeneity among studies (I 2 =29%). Secondary outcomes ARM MOTOR FUNCTION: CON- STRAINT versus CONTROL Fourteen studies recruiting 479 participants measured arm motor function. 6-8, 12, 16, 17, 19, 20, 22, CIMT had a moderate effect on upper limb function (SMD 0.44, 95% CI 0.03 to 0.84; Figure 2). We found substantial heterogeneity among studies (I 2 =63%). 540 EUROPEAN JOURNAL OF PHYSICAL AND REHABILITATION MEDICINE December 2010

5 Constraint-induced movement therapy in stroke patients CORBETTA Table III. Methodological quality of included studies. Study Type of intervention Sample size Description of randomisation Allocation concealment Blinded Assessor Scales Withdrowals <10% Atteya A 25 mcimt 4 No Unclear Yes Supported Yes Brogardh C 16 mcimt 24 Yes Unclear Yes Supported Yes Dhal AE 20 CIMT 30 Yes Unclear Yes Supported Yes Dromerick AW 26 CIMT 20 Yes Unclear Yes Supported Yes Dromerick AW (b) 17 mcimt 52 No Unclear Yes Supported Yes Hammer A 8 FU 30 Yes Unclear No Supported Yes Lin KC 21 mcimt 32 Yes Adequate Yes Supported Yes Lin KC (b) 18 mcimt 40 Yes Adequate Yes Supported Yes Myint M 22 CIMT 43 Yes Adequate Yes Supported Yes Page SJ 6 mcimt 4 No Unclear Yes Supported Yes Page SJ (b) 27 mcimt 10 Yes Unclear Yes Supported Yes Page SJ (c ) 28 mcimt 25 Yes Unclear Yes Supported Yes Ploughman M 7 FU 23 Yes Unclear Yes Supported Yes Taub E 12 CIMT 9 No Unclear Yes Supported Yes Wittenberg GF 29 CIMT 16 Yes Unclear Yes Supported Yes Wolf SL 19 CIMT 199 Yes Adequate Yes Supported Yes Wu CY 23 mcimt 30 No Unclear Yes Supported Yes Wu CY (b) 24 mcimt 26 Yes Unclear Yes Supported Yes CIMT: constraint-induced movementtherapy; mcimt: modified constraint-induced movementtherapy; FU: forced use; for the allocation concealment the therm Adequate is used when the method of allocation is clearly described while unclear when the authors do not report any allocation concealment approach. For the scales the therm supported is referred to the psychometric properties. Figure 1. Effect of CIMT compared with usual care on disability in adult stroke patients. Discussion Eighteen RCTs with a total of 674 participants are discussed in this updated review. The majority of studies happened to be in people who had compromised but residual ability of the upper arm and hand, with limited pain or spasticity and no cognitive impairment. While the meta-analysis on the secondary outcome (arm motor function) shows a moderate benefit, the primary outcome (disability) indicates no evidence of a benefit of the CIMT. The addition of four new studies with 146 patients partially contra- Vol No. 4 EUROPEAN JOURNAL OF PHYSICAL AND REHABILITATION MEDICINE 541

6 CORBETTA Constraint-induced movement therapy in stroke patients Figure 2. Effect of CIMT compared with usual care on arm motor function in adult stroke patients. dicts previous findings. In 2009 the Cochrane review concluded: Although CIMT appeared to have moderately positive effects on disability assessed at the end of treatment, many studies were underpowered and indeed at high risk of small-trial and publication bias. 9 The SMD values on disability changed from 0.36 (95% CI 0.06 to 0.65), a moderate effect, to 0.21 (95% CI to 0.50), not reaching statistical significance anymore. The SMD on motor function changed from 0.72 (95% CI 0.32 to 1.12) to 0.44, (95% CI 0.03 to 0.84), about halving the original benefit. This effect size can be considered a modest benefit. Against this background the evidence of a clinical benefit of CMIT should be considered cautiously. The meta-analysis on disability is inconclusive and means that the next published study may shift the overall estimate in one direction or the other. Meta-analyses of RCTs are generally considered the gold standard for intervention comparisons; however, they are not error free (i.e., they might risk reporting false positive results or important inaccuracies in treatment effects). This is generally because of random error ( the play of chance ), systematic errors ( bias ) or design errors ( wrong design to answer the question posed ) in the studies included. The risk of random error is higher when data come from small-sample trials and can be further exacerbated in meta-analysis also when statistical analysis are repeatedly performed as new trials provide additional data In order to reduce the risk of random error, before beginning a RCT, it is recommended to calculate the sample size to see what size is needed to observe a true significant effect. Moreover, studies with small samples tend to be prone to bias. 33 In small trials the randomisation is affected by a major threat: imbalances between intervention and control groups on prognostic factors, i.e., with few patients the distributions of the factors easily differ between groups and the factors are associated with outcome. Other imbalances are introduced because in some studies (including the larger study) authors compared CIMT with an unequal dose of exercise (i.e., duration of treatment always longer in the intervention group). Furthermore, the equal dose, in 542 EUROPEAN JOURNAL OF PHYSICAL AND REHABILITATION MEDICINE December 2010

7 Constraint-induced movement therapy in stroke patients CORBETTA terms of duration of treatment, may not be sufficient for balancing treatment and control groups because CIMT is a multifaceted intervention that includes a large amount and quality of exercise in addition to the restriction. Conclusions The effectiveness of CIMT regarding disability is still inconclusive, so it can be neither supported nor rejected. CIMT had a modest effect on upper limb function. To overcome the uncertainty related to this intervention the next RCTs need to have accurate characteristics in terms of methodological quality, larger samples, reliable and relevant measures (i.e., disability) and report of adverse events in order to avoid spurious results. It would be useful to consider the possibility of national or international multicenter studies. A single, large RCT could produce more reliable evidence than a meta-analysis of multiple small low-quality RCTs. 33 References 1. Langhorne P, Coupar F, Pollock A. Motor recovery after stroke: a systematic review. Lancet Neurol 2009;8: knapp HD, Taub E, Berman AJ. Movements in monkeys with deafferented forelimbs. Experimental Neurology 1963;7: Taub E, Heitmann RD, Barro G. Alertness, level of activity, and purposive movement following somatosensory deafferentation in monkeys. Ann N Y Acad Sci 1977;280: Taub E, Harger M, Grier HC, Hodos W. Some anatomical observations following chronic dorsal rhizotomy in monkeys. Neuroscience 1980;5: Page SJ, Sisto S, Johnston MV, Levine P. Modified constraintinduced therapy after subacute stroke: a preliminary study. Neurorehabil Neural Repair 2002;16: Page SJ. Sisto SA, Levine P, Johnston MV, Hughes M. Modified constraint induced therapy: a randomized feasibility and efficacy study. J Rehabil Res Dev 2001;38: Ploughman M, Corbett D. Can forced-use therapy be clinically applied after stroke? An exploratory randomized controlled trial. Arch Phys Med Rehabil 2004;85: Hammer A, Lindmark B. Is forced use of the paretic upper limb beneficial? A randomised pilot study during subacute post-stroke recovery. Clinical Rehabil 2009;23: Sirtori V, Corbetta D, Moja L, Gatti R. Constraint-induced movement therapy for upper extremities in stroke patients. Cochrane Database of Systematic Reviews 2009, Issue 4. Art. No.: CD DOI: / CD pub 10. Taub E, Somatosensory deafferentation research with monkeys: implications for rehabilitation medicine. In: Ince LP, editor. Behavioral psychology in rehabilitation medicine: clinical applications. New York, NY: Williams & Wilkins; p Taub E, Uswatte G, Elbert T. New treatments in neurorehabilitation founded on basic research. Nature Reviews Neuroscience 2002;3: Taub E, Miller NE, Novack TA, Cook EW 3rd, Fleming WC et al. Technique to improve chronic motor deficit after stroke. Arch Phys Med Rehabil 1993;74: Higgins JPT, Green S (editors). Cochrane Handbook for Systematic Reviews of Interventions Version [updated 2008 February; cited 2010 October 29]. The Cochrane Collaboration, Available at: Deeks JJ, Altman DG, Bradburn MJ. Statistical methods for examining heterogeneity and combining results from several studies in meta-analysis. In: Egger M, Davey Smith G, Altman DG (editors). Systematic Reviews in Health Care: Metaanalysis in Context. 2nd ed. London: BMJ Publication Group; Review Manager (RevMan) [Computer program]. Version 5.0. Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, Brogårdh C, Vestling M, Sjölund BH. Shortened constraint-induced movement therapy in subacute stroke no effect of using a restraint: a randomized controlled study with independent observers. J Rehabil Med 2009;41: Dromerick AW, Lang CE, Birkenmeier RL, Wagner JM, Miller JP, Videen TO et al. Very early constraint-induced movement during stroke rehabilitation (VECTORS): a single-center RCT. Neurology 2009;73: Lin KC, Chang Y, Chen Y. Effects of constraint-induced therapy versus bilateral arm training on motor performance, daily functions, and quality of life in stroke survivors. Neurorehabil Neural Repair 2009;23: Wolf SL, Winstein CJ, Miller JP, Taub E, Uswatte G, Morris D et al. Effect of constraint-induced movement therapy on upper extremity function 3 to 9 months after stroke: the EXCITE randomized clinical trial. JAMA 2006;286: Dahl AE, Askim T, Stock R, Langørgen E, Lydersen S, Indredavik B. Short- and long-term outcome of constraint-induced movement therapy after stroke: a randomized controlled feasibility trial. Clinical Rehabilitation 2008;22: Lin KC, Wu CY, Wei TH, Lee CY, Liu JS. Effects of modified constraint-induced movement therapy on reach-to-grasp movements and functional performance after chronic stroke: a randomized controlled study. Clin Rehabil 2007;21: Myint JMWW, Yuen GFC, Yu TKK, Kng CPL, Wong AMY, Chow KKC et al. A study of constraint-induced movement therapy in subacute stroke patients in Hong Kong. Clinical Rehabilitation 2008;22: Wu CY, Lin KC, Chen HC, Chen IH, Hong WH. Effects of modified constraint-induced movement therapy on movement kinematics and daily function in patients with stroke: a kinematic study of motor control mechanisms. Neurorehabil Neural Repair 2007;21: Wu CY, Chen CL, Tsai WC, Lin KC, Chou SH. A randomized controlled trial of modified constraint-induced movement therapy for elderly stroke survivors: changes in motor impairment, daily functioning and quality of life. Arch Phys Med Rehabil 2007;88: Atteya AAA. Effects of modified constraint induced movement therapy on upper limb function in subacute stroke patients. Neurosciences 2004;9: Dromerick AW, Edwards DF, Hahn M, Dromerick AW, Edwards DF, Hahn M. Does the application of constraint-induced movement therapy during acute rehabilitation reduce arm impairment after ischemic stroke? Stroke 2000;31: Page SJ, Levine P, Leonard AC. Modified constraint-induced therapy in acute stroke: a randomized controlled pilot study. Neurorehabil Neural Repair 2005;19: Vol No. 4 EUROPEAN JOURNAL OF PHYSICAL AND REHABILITATION MEDICINE 543

8 CORBETTA Constraint-induced movement therapy in stroke patients 28. Page SJ, Levine P, Leonard A, Szaflarski JP, Kissela BM. Modified constraint-induced therapy in chronic stroke: results of a singleblinded randomized controlled trial. Phys Ther 2008;88: Wittenberg GF, Chen R, Ishii K, Bushara KO, Eckloff S, Croarkin E et al. Constraint-induced therapy in stroke: magneticstimulation motor maps and cerebralvv activation. Neurorehabil Neural Repair 2003;17: Brok J, Thorlund K, Gluud C, Wetterslev J. Trial sequential analysis reveals insufficient information size and potentially false positive results in many meta-analyses. J Clin Epidemiol 2008;61: Wetterslev J, Thorlund K, Brok J, Gluud C. Trial sequential analysis may establish when firm evidence is reached in cumulative meta-analysis. J Clin Epidemiol 2008;61: Brok J, Thorlund K, Wetterslev J, Gluud C. Apparently conclusive meta-analyses maybe inconclusive - Trial sequential analysis adjustment of random error risk due to repetitive testing of accumulating data inapparently conclusive neonatal metaanalyses. Int J Epidemiol 2009;38: Rerkasem K, Rothwell PM. Meta-analysis of small randomized controlled trials in surgery may be unreliable. Br J Surg 2010;97: EUROPEAN JOURNAL OF PHYSICAL AND REHABILITATION MEDICINE December 2010