Selective dorsal rhizotomy: meta-analysis of three randomized controlled trials

Cerebral palsy (CP) has an incidence and prevalence rate ofabout two per 1000 in children (Grether et al.1992, Yeargin- Allsopp et al. 1992). Spasticity is a major clinical feature of over75% of cases of CP and is conventionally considered to be amajor cause of discomfort, gait abnormalities, and function- al limitations for persons with CP. Many resources have beendirected at the treatment of spasticity and the current treat- ment choices include oral medications, orthotics, orthopaedicsurgical procedures, electrical stimulation, intramuscular injections, physiotherapy, ‘selective’ dorsal rhizotomy, con-tinuous intrathecal baclofen infusion, and a host of alterna-tive treatment methods.
Many of the treatments introduced for spasticity and other aspects of CP were inadequately evaluated when first intro- duced (McKenzie 1988) and have become enshrined in clini- cal practice without benefit of acceptable evidence of efficacy Nancy Temkin PhD, Children’s Hospital and Regional and effectiveness. Over the last 15 years, the rigor of clinical research methodology has advanced considerably (Spilker Paul Steinbok MBBS FRCSC, Children’s and Women’s 1991). Valid and reliable outcome measures have been devel- oped for many relevant aspects of CP ranging from electro- Virginia Wright BSc (PT) MSc, Bloorview MacMillan Centre, mechanical torque measures of spasticity (Price et al. 1991) and instrumented gait analysis (Sutherland 1988, Gage Ann Reiner BSR MA, Children’s and Women’s Health 1991, Perry 1992) to measures of functional ability (Russell et al. 1989, Haley et al. 1991, Msall et al. 1997), and quality of Theodore Roberts MD, Children’s Hospital and Regional ‘Selective’ dorsal rhizotomy (SDR) emerged as a popular James Drake MBBCh, FRCSC, Hospital for Sick Children, surgical intervention for spasticity in North America and else- where since the early 1980s (Peacock and Staudt 1991, Park Maureen O’Donnell MD, Children’s and Women’s Health and Owen 1992, Abbott et al. 1993). History of the develop- ment of this procedure, including the introduction of elec- Peter Rosenbaum MD, Bloorview MacMillan Centre, trophysiological criteria for ‘selecting’ dorsal root tissue for transection, has been reviewed repeatedly in the literature Jason Barber BS, University of Washington; (Albright 1992, Park and Owen 1992, Hays et al. 1997). Many Anne Ferrel BS, Children’s Hospital and Regional Medical published results of open consecutive surgical series strongly suggested that SDR reduced spasticity substantially, improvedfunction, and involved no unacceptable short-term risks when *Correspondence to first author at Children’s Hospital and carried out by experienced interdisciplinary clinical teams Regional Medical Center, Mail Stop 47, 4800 Sand Point Way (Peacock and Staudt 1991, Steinbok et al. 1992, McLaughlin et al. 1994, Nishida et al. 1995). Enthusiasm was tempered by concerns about long-term effects, questions about the validityof the root tissue selection process (Landau and Hunt 1990,Cohen and Webster 1991, Logigian et al. 1994, Steinbok et al.
1994, Hays et al. 1998), and recognition that uncontrolledclinical series are open to bias in favor of a positive outcome This study is a comparative analysis and meta-analysis of three (Neville 1988, McLaughlin et al. 1994).
randomized clinical trials. Children with spastic diplegia received Three randomized clinical trials have been published: either ‘selective’ dorsal rhizotomy (SDR) plus physiotherapy Steinbok et al. 1997 (Vancouver), McLaughlin et al. 1998, (SDR+PT) or PT without SDR (PT-only). Common outcome (Seattle), and Wright et al. 1998 (Toronto). Main results of measures were used for spasticity (Ashworth scale) and these three studies are presented in Table I. All three studies function (Gross Motor Function Measure [GMFM]). Baseline confirmed that SDR consistently reduces or eliminates spastici- and 9- to 12-month outcome data were pooled (n=90). At
ty. There is a difference in the functional outcome with two baseline, 82 children were under 8 years old and 65 had Gross studies showing a statistically significant advantage for SDR Motor Function Classification System level II or III disability.
(Steinbok et al. 1997, Wright et al. 1998) and the third showing Pooled Ashworth data analysis confirmed a reduction of no advantage (McLaughlin et al. 1998). The major similarities spasticity with SDR+PT (mean change score difference –1.2; and differences regarding the design of these studies are out- Wilcoxon p<0.001). Pooled GMFM data revealed greater
functional improvement with SDR+PT (difference in change This paper presents a comparative analysis and meta-analy- score +4.0, p=0.008). Multivariate analysis in the SDR+PT
sis of these three studies. The primary aim of these analyses group revealed a direct relationship between percentage of dorsal was to identify the factors responsible for the different out- root tissue transected and functional improvement. SDR+PT is comes of the studies and to clarify whether SDR makes a statis- efficacious in reducing spasticity in children with spastic diplegia tically significant contribution to the functional improvement and has a small positive effect on gross motor function. of children with spastic CP up to 1 year after the operation. Developmental Medicine & Child Neurology 2002, 44: 17–25 Function Classification System (GMFCS) provides a means of A search of Medline, the Cochrane Collaboration database, rank-ordering the functional severity of cerebral palsy (CP) recent scientific meeting abstracts, and contacts with other based on age-adjusted clinical descriptors (Palisano et al.
researchers in the field did not reveal the existence of any 1997). A developmental pediatrician familiar with the GMFCS other randomized clinical trials of SDR as of December 2000.
and masked with respect to participant identification andstudy group assignment, but not study site, retrospectively reviewed selected clinical descriptions of all participants. Data The senior study biostatistician reviewed the research proto- from each child were assigned a GMFCS baseline score, which cols from each site to compare details of the study designs was entered into the multivariate analyses.
(Tables III and IV). Common definitions were agreed uponfor key variables. The three studies all used the Ashworth scale for spasticity (Bohannon and Smith 1987) and the All three centers provided children in both the SDR with Gross Motor Function Measure (GMFM; Russell et al. 1989) physiotherapy (SDR+PT) groups and the PT without SDR as major outcome variables. The 12-month data from Seattle (PT-only) groups with physiotherapy programs of stretching, and Toronto and the 9-month outcome data from Vancouver strengthening, and training in functional movements intend- ed to enhance mobility (Tables IV and Va and b). The greatest The data were entered into SPSS software (version 7.5.3).
number of planned treatment hours for both groups was pro- Complete data were obtained for all variables reported vided in Seattle, and the fewest in Vancouver. Seattle and except baseline ambulation status. No new data were collect- Vancouver physiotherapy protocols stressed equivalency of ed for the present report. The original three studies were treatments for both groups. The Toronto protocol called for a approved by the ethics committees of the respective institu- stronger emphasis on strengthening and on postural control for the SDR+PT group as compared with the PT-only group The developers of the GMFM have published an improved based on the rationale that the anticipated weakening effect scaling method (GMFM-66) that addresses the linearity of of the surgery would provide a negative bias. Treatment plan- the individual items scores across the entire range of GMFM ning for the PT-only group in Toronto was based on written scores (Russell et al. 2000). In these analyses we calculated goals set by community therapists before the randomized group assignment. Postoperative protocols for the surgical In the initial descriptions and single-center analyses, the groups in all three centers varied in emphasis and timing markers of severity were the initial diagnosis (spastic diple- gia), qualitative descriptors (mild, moderate, or severe), ambu-latory status, and the baseline GMFM score. The Gross Motor DATA ANALYSISCharacteristics of the three samples were compared usingKruskal–Wallis distribution-free analysis of variance (ANOVA)for continuous or ordered variables and χ2 or Fisher’s exact Table I: Selective dorsal rhizotomy randomized clinical trials tests for unordered categorical variables. For each outcome, the change in the SDR+PT group was compared with that inthe PT-only group using two methods: blocked Wilcoxon’s test (blocking on site; Marascuilo and McSweeney 1997) and ANOVA including factors for treatment group, site, and atreatment by site interaction. To evaluate the effect of charac- teristics or the effect of SDR on outcome as measured by Ashworth, GMFM, and GMFM-66, separate multiple regres- sion analyses were performed. These analyses used backward selection and the following additional predictors: age, sex, birthweight, ambulatory status, baseline GMFM-66, and base- line lower-extremity Ashworth score. Regardless of signifi- Vancouver, Steinbok et al. 1997; Toronto, Wright et al. 1998; Seattle, cance, site was included in the models; all other variables required a significance level of p<0.05 to be retained. Oncesignificant main effects were identified, two-way interactionsamong the included variables were evaluated. Effects of per-cent of root tissue transected were studied in the same fash- Table II: Selective dorsal rhizotomy randomized clinical trials ion, except site was not automatically retained as a factor. A descriptive summary of the samples from the three studies and the pooled data is presented in Table VI. The Seattle study was the only one to include children (n=8) who were 8 years of age or older at baseline. The Vancouver study had an unusual sex ratio (17/28 were female) for a sample of chil- dren with CP. The three samples were similar with respect to prematurity. The children in the Seattle sample were slightly more heterogeneous with respect to intellectual function Developmental Medicine & Child Neurology 2002, 44: 17–25 and cause of CP. At baseline, the children in the Toronto sam- scores and had less dorsal root tissue transected (25%) than ple had what appeared to be somewhat more severe CP as was the case in the other two studies (41 and 45%). In each of judged by the GMFCS, GMFM, and clinical rating of ambula- the three original studies, there were no differences in base- tory status, although the amount of spasticity as judged by line characteristics between treatment groups so the corre- the baseline Ashworth scale was not greater. The children in sponding analysis is not presented here for the pooled data.
the Seattle sample had noticeably lower baseline Ashworth The main results comparing the magnitude of change in Table III: Selective dorsal rhizotomy study design comparison 2-sample t-test on change –4 measures, 2-sample t-test, Wilcoxon on change, χ2, Fisher’s exact (both intent-to-treat and aNote: first number is number of children in SDR+PT group, second number is number of children in PT-only group; bButler et al. 1984;cStokes et al. 1990. SDR, selective dorsal rhizotomy; PT, physiotherapy; OT, occupational therapy; ROM, range of motion. Selective Dorsal Rhizotomy Meta-analysis John McLaughlin et al. weaker direct correlation that remained statistically signifi- and all differences among studies as we have endeavoured to cant (p=0.02). The potential effects of outliers and of section- do in this report. The strengths of this meta-analysis should ing S2 root tissue were examined by eliminating relevant data not be overlooked. Many meta-analyses are carried out by from the analyses. No appreciable effects were identified. investigators not involved in any of the original studies and There was a consistent and statistically significant (p=0.0002) who may not have clinical experience with either the disor- inverse correlation between the baseline GMFM-66 score and der or the intervention being studied. As a consequence, the the percent of dorsal root tissue transected (Fig. 7), i.e. the authors of a meta-analysis may have access only to published lower the score, the more dorsal root tissue was cut, regard- data, which are inevitably a summary of the raw data. This study less of study site. Inclusion of baseline GMFM-66 score in the is the product of close collaboration among the teams who regression model did not change the relations of outcome published the original results of all randomized trials of SDR variables to per cent of dorsal root tissue transected. conducted and published to date. Unpublished raw data werepooled and all data was available for reanalysis. It was possibleto ensure that all variables had common definitions across studies and to recalculate variables when necessary. In short, Meta-analysis is a well established, if still evolving, methodolo- the circumstances of this meta-analysis assured that compara- gy for synthesizing outcomes from existing studies of a single ble data were pooled from the three studies. Many meta-analy- research question (Moher et al. 1999). The procedure remains ses suffer from diminished specificity of outcomes arising from counter-intuitive for clinicians and suffers from both uncritical the need to reduce disparate outcome variables across multi- acceptance and categorical disregard. The applicability of a ple studies to a common proxy variable. In this study, identical given meta-analysis is best revealed by setting forth limitations outcome variables were used in the three original studies for Table VI: Selective dorsal rhizotomy study sample characteristics Baseline lower extremity Ashworth scale (SD) aInformation missing on one patient; binformation missing on three patients; c information missing on eightpatients (some of these were adoptions or other circumstances where we were unable to determine cause. Table VII: Selective dorsal rhizotomy multivariate analysis: main results Developmental Medicine & Child Neurology 2002, 44: 17–25 the SDR+PT versus PT-only groups for the pooled three- weak inverse correlation (p=0.03) between percent of dorsal study sample are presented in Figures 1, 2, and 3. Ashworth root tissue transected and change in Ashworth scores (Fig. 4).
scale change scores were clinically robust and similar across There was a statistically significant direct relation (p<0.001) all three studies with a mean difference in change between between GMFM change score and percent of dorsal root tis- groups of –1.2 points which was statistically significant using sue transected (Fig. 5). The regression lines for the individual the Wilcoxon’s blocked on site (p<0.001). Mean difference sites appear to be very similar. There was an apparent direct in change score in the GMFM (4.0) was statistically significant effect of site (p=0.005) that disappeared after adjustment for (p=0.008) by Wilcoxon’s blocked on site. When the GMFM- percent of dorsal root tissue transected (p=0.6). There was 66 scores were used, the mean difference in change score for no interaction effect with study site. The same analysis was the pooled data was smaller (2.6) and remained statistically performed for GMFM-66 change score (Fig. 6) and revealed a significant (p=0.002) by Wilcoxon’s blocked on site.
Multivariate analysis was carried out to detect effects attributable to site-related differences in baseline variablespresented in Table VI and to look for interactions suggesting Table IV: Physiotherapy by treatment group subgroup effects (e.g. younger versus older participants).
Baseline characteristics were chosen as clinically plausiblecandidate variables. The analyses were done with and with- out the data for the eight children from the Seattle study, who were 8 years of age or more at baseline. No differences in results were seen, and age was not a significant covariate.
Subsequent analysis excluded these older children. As shown in Table VII, the effect of SDR was significant on all outcomes.
There was no evidence that this effect differed by baseline characteristics or by study site (i.e. there was no significant treatment by site interaction). Based on the lack of interac-tional effects in the multivariate model, no subgroup defined by baseline characteristics was identified for which SDR is particularly effective. This was confirmed by looking at mean effects within and across sites in subgroups defined posthoc(analysis not presented). Retrospective GMFCS classification of baseline severity was not related to outcome.
Regression analysis of the outcomes for the SDR+PT groups at the three study sites and the pooled data reveal a Table Va: Comparison of postoperative physiotherapy protocols: timing of intervention Start transition movements (to sit, side-sit, all fours, Table Vb: Comparison of physiotherapy protocols: muscle groups treated Hip adductors, flexors, hamstrings, and heel-cords Hip adductors, flexors, hamstrings, and heel-cords Hip adductors, flexors, hamstrings, and heel-cords Hip adductors, flexors, hamstrings, and heel-cords Hip adductors, flexors, hamstrings, and heel-cords Lower extremity muscle groups, abdominals, and trunk extensors Strengthening through functional activities Selective Dorsal Rhizotomy Metaanalysis John McLaughlin et al. both spasticity and functional outcomes. The only loss of statistically significant advantage to SDR+PT compared with rigor is that the Ashworth scale measurements were not col- PT-only that is independent of any site-specific differences lected by masked observers in two of the studies. The critical- among the studies. It is difficult to know the clinical impor- ly important functional outcomes were measured by masked tance of a mean difference in change score of 4 percentage assessment of GMFM scores in all participants. points on the GMFM. It was not possible to identify any sub- The primary conclusion supported by the meta-analysis is groups of children with specific characteristics who benefit- that the pooled results of the three studies show a small but ed more than others included in the studies.
It was expected that the multivariate analysis would iden- tify a specific reason for the difference in outcome among thethree studies. Several of the differences in the studies seemedlikely to explain the divergent outcomes. The most obvious difference is the surgical technique. The Seattle study used the electrophysiological monitoring technique and criteria for transecting dorsal root tissue popularized by Peacock (Staudt et al. 1995) and other clinicians. The Vancouver study utilized similar criteria for root tissue assessment, but the primary basis for root tissue transection was preopera- tive clinical assessment. The Toronto study used electrophys- iological monitoring only to differentiate dorsal from ventral roots. The amount of dorsal root tissue transected was 25% in the Seattle study and was 41% and 45% respectively in the other two studies. If these apparently major differences were responsible for the difference in outcome, one would expect the study site by treatment interaction variable to be statisti-cally significant in the multivariate analysis model. No such Ashworth change score. PT Only and SDR+PT The regression analysis of outcome change scores on per- group data for individual studies and pooled meta- cent dorsal root tissue transected (Figs 4 to 6) of the SDR+PT analysis. Interval is from baseline to 12 months after group data provides further insight. Patients who had a larg- beginning treatment (9 months for Vancouver). Boxes er amount of dorsal root tissue transected had more improve- represent the 25th, 50th, and 75th centiles. Whiskers ment on GMFM (p<0.001), GMFM-66 (p=0.02), and Ashworth represent minimum and maximum values excluding score (p=0.03). As seen in the figures, the effect is modest outliers beyond 1.5 times the interquartile range. for GMFM, smaller for GMFM-66, and perhaps inconsistent *p values are based on Wilcoxon’s tests, with blocking on across sites for the Ashworth scale. Adjustment for per cent Figure 2: Gross Motor Function Measure (GMFM) change Figure 3: GMFM-66 change score. PT Only and SDR+PT score. PT Only and SDR+PT group data for individual studies group data for individual studies and pooled meta-analysis. and pooled meta-analysis. Interval is from baseline to 12 Interval is from baseline to 12 months after beginning months after beginning treatment (9 months for Vancouver). treatment (9 months for Vancouver). Boxes represent 25th, Boxes represent 25th, 50th, and 75th centiles. Whiskers 50th, and 75th centiles. Whiskers represent the minimum represent minimum and maximum values excluding outliers and maximum values excluding outliers beyond 1.5 times beyond 1.5 times the interquartile range. *p values are based the interquartile range. *p values are based on Wilcoxon’s on Wilcoxon’s tests, with blocking on site for combined test. tests, with blocking on site for combined test. Developmental Medicine & Child Neurology 2002, 44: 17–25 of dorsal root tissue eliminated evidence of site effect (see the biggest mean percent of dorsal root tissue transected Fig. 5) on GMFM change in the SDR+PT pooled data (p=0.6 (45%) but was not dramatically different from the Toronto after adjustment). The per cent of dorsal root tissue transect- study (41%). The Seattle study (25%) was clearly different ed is related to baseline GMFM-66 (Fig. 7, p<0.001) which from the other two in this regard. There are at least two pos- suggests that the initial functional level may have influenced sible interpretations. First, the Ashworth scale is not known the extent of transection. Including baseline GMFM-66 score to have good interrater or test–retest reliability, so perhaps it in the regression did not change the models substantially.
is simply a problem of measurement. Using a quantitative Although the relation of percentage of dorsal root tissue electromechanical torque measure of spasticity in a larger transected to outcome has biological plausibility, caution is group of children undergoing SDR, no relation could be essential in ascribing a causal effect to a factor that was not demonstrated between spasticity and the percent of dorsal assigned at random. It is not possible to rule out a chance dif- root tissue transected (Hays et al. 1998). Second, perhaps it ference in functional outcome or some other undetected takes a smaller amount of dorsal root transection to abolish deep tendon reflexes, clonus, and clinically detectable spastic- The inverse relation of Ashworth score to percent of dor- ity than it does to influence function. Perhaps the Vancouver sal root tissue transected is statistically significant (p=0.03, investigators performed sufficiently extensive rhizotomies to see Fig. 4). However, the interaction with site showed a trend get beyond a range of reduction in spasticity that would be (p=0.08) and inspection of the individual regression lines in expected to have a dose-response curve while the other two Figure 4 suggests that the Vancouver data are different from studies still show such a response. The amount of functional the other two in this respect. The Vancouver study reported change may still show a strong dose–response correlation in Figure 4: Regression analyses of Ashworth change score by Figure 6: Regression analyses of GMFM-66 change score by % dorsal root tissue transected. Linear regression analysis % dorsal root tissue transected. Linear regression analysis (p=0.03). "––, Seattle; #– –, Toronto; $---, Vancouver. (p=0.02). "––, Seattle; #– –, Toronto; $---, Vancouver. Figure 5: Regression analyses of GMFM change score by % Figure 7: Regression analyses of baseline GMFM-66 by % dorsal root tissue transected. Linear regression analysis dorsal root tissue transected. Linear regression analysis (p<0.001). "––, Seattle; #– –, Toronto; $---, Vancouver. (p<0.001). "––, Seattle; #– –, Toronto; $---, Vancouver. Selective Dorsal Rhizotomy Meta-analysis John McLaughlin et al. the range of tissue transection described in the Toronto and ed primarily ambulatory children with spastic diplegia and excluded those with dystonia, athetosis, and ataxia. Children The severity of CP was retrospectively reclassified accord- with severe visual impairment were excluded. All the chil- ing to the GMFCS system from baseline clinical data. The dren undergoing SDR in these studies had access to peri- amount and quality of clinical data available differed across operative and postoperative care at a regional children’s studies, which limits the ability to make inferences based on hospital in affluent societies, and access to expert pediatric GMFCS scores in the pooled sample. Changes in GMFCS physiotherapists for postoperative rehabilitation. Regular classification determined prospectively might be another follow up by an experienced clinical team may have mitigat- method for capturing global changes in response to specific ed adverse events and motivated families in ways that were not measured. The effects of SDR on spasticity can probablybe safely extrapolated to most children with spasticity, but this meta-analysis provides no comparison to other available This meta-analysis of the 9- to 12-month outcomes of three effective treatments. The effects of SDR on function are mod- randomized clinical trials confirms a clinically important change in spasticity. With regard to function, a small but sta- Based on clinical experience in addition to these data, tistically significant advantage to SDR+PT (additional mean we speculate that SDR might be most effective for a child GMFM change of 4 percentage points) was shown when between 3 and 8 years of age whose functional level falls into compared with PT-only. There appears to be a direct relation GMFCS levels III and IV. Children in this age range are easy to between the percent of dorsal root tissue transected and the assess, tolerate the physiotherapy regimens, and have acade- magnitude of gain in function. The sectioning rate was not mic and social demands that can accommodate an intensive randomized and the sectioning technique varied across stud- intervention. Children with more severe CP may have more ies. This statistical correlation may not reflect a true cause potential gain from an invasive procedure.
and effect relation. There are no detectable subgroups thatbenefit more than others. The three original studies did notreport any worrisome problems with adverse events. These Accepted for publication 17th August 2001. are short-term outcomes applicable to a child who meets thecriteria used for enrollment in these studies. These results suggest that the decision whether or not to Funding was generously provided by the United Cerebral PalsyResearch & Educational Foundation. Funding for the original perform SDR on a similar child partly rests on whether or not studies was provided by the National Institute for Neurological an anticipated mean GMFM change score increment of 4 per- Disease and Stroke, RO1-NS27867 (Seattle), the British Columbia centage points above the amount of change with non-invasive Health Care Research Foundation (Vancouver), and the Easter Seal care justifies the time, effort, and risk involved. While an addi- Research Institute of Canada and the United Cerebral Palsy tional change of 4 percentage points over a year might be Research & Educational Foundation (Toronto). No author has afinancial interest in any aspect of this study. Elaine Smith provided twice the change expected with non-invasive care, it is still a invaluable assistance with the manuscript. The authors gratefully relatively small amount of change compared with some of the acknowledge the assistance of colleagues at each of the three claims in uncontrolled studies (Peacock and Staudt 1991, centers and the cooperation of the families.
Park and Owen 1992, Abbott et al. 1993, Nishida et al. 1995).
These modest improvements in gross motor function aresomewhat disappointing, but it should be noted that these results are short term and do not take into account other pos- Abbott R, Johann-Murphy M, Shiminski-Maher T, Quartermain D, sible positive and negative long-term outcomes. In addition to Forem SL, Gold JT, Epstein FJ. (1993) Selective dorsal rhizotomy: improving ambulation, an effective surgical treatment for outcome and complications in treating spastic cerebral palsy.
spastic diplegia should have beneficial long-term effects on gait parameters, pain reduction, and rate of acquired muscu- Albright AL. (1992) Neurosurgical treatment of spasticity: selective posterior rhizotomy and intrathecal baclofen. Stereotactic and loskeletal deformity, function, participation, and quality of Functional Neurosurgery 58: 3–13.
life. As was the case in these studies, there should be a low rate Bohannon RW, Smith MB. (1987) Interrater reliability of a modified of serious adverse events attributable to the intervention.
Ashworth scale of muscle spasticity. Physical Therapy 67: 206–7.
Answers to these important questions will not become avail- Butler P, Engelbrecht M, Major RE, Tait JH, Stallard J, Patrick JH.
(1984) Physiologic cost index of walking for normal children and able until a sizable well-studied cohort of individuals with its use as an indicator of physical handicap. Developmental spastic diplegia is followed to adulthood. There is serious Medicine & Child Neurology 26: 607–12.
doubt regarding the validity of the electrophysiological moni- Cohen AR, Webster HC. (1991) How selective is selective posterior toring criteria that formed the basis for the ‘selectivity’ of dor- rhizotomy? Surgical Neurology 35: 267–72.
sal rhizotomy (Cohen and Webster 1991, Steinbok et al. 1994, Gage JR. (1991) Gait Analysis in Cerebral Palsy. London: Mac Keith Logigian et al. 1996, Hays et al. 1998), yet this meta-analysis Grether JK, Cummins SK, Nelson KB. (1992) The California suggests that the percent of dorsal root tissue transected may Cerebral Palsy Project. Paediatric Perinatal Epidemiology be important. The correct method to determine where and how much dorsal root tissue to transect must be elucidated by Haley SM, Coster WJ, Faas RM. (1991) A content validity study of the Pediatric Evaluation of Disability Inventory. Pediatric Physical Clinicians are faced with the practical problem of deciding Hays RM, McLaughlin JF, Bjornson KB, Stephens K, Roberts TS, whether or not to recommend treatments such as SDR for Price R. (1998) Electrophysiologic monitoring during selective individual children. The results of this meta-analysis should dorsal rhizotomy, and spasticity and GMFM performance.
be used with care in the clinical setting. These studies includ- Developmental Medicine & Child Neurology 40: 233–8.
Developmental Medicine & Child Neurology 2002, 44: 17–25 Hays RM, McLaughlin JF, Geiduschek JM, Bjornson KF, Graubert CS.
— Keyes R, Langill L, Cochrane DD. (1994) The validity of (1997) Evaluation of the effects of selective dorsal rhizotomy.
electrophysiological criteria used in selective functional Mental Retardation and Developmental Disabilities Research posterior rhizotomy for treatment of spastic cerebral palsy.
Journal of Neurosurgery 81: 354–61.
Landau WM, Hunt CC. (1990) Dorsal rhizotomy, a treatment of — Reiner AM, Beauchamp RD, Armstrong RW, Cochrane DD, Kestle unproven efficacy. Journal of Child Neurology 5: 174–8.
J. (1997) A randomized clinical trial to compare selective Landgraf J, Abetz L. (1996) Measuring health outcomes in pediatric posterior rhizotomy plus physiotherapy with physiotherapy populations: issues in psychometrics and application. In: Spilker alone in children with spastic diplegic cerebral palsy.
B, editor. Quality of Life and Pharmacoeconomics in Clinical Developmental Medicine & Child Neurology 39: 178–84.
Trials. Philadelphia: Lippincott-Raven. p 793–802.
Stokes NA, Dietz JR, Crowe TW. (1990) The Peabody Developmental Logigian EL, Wolinsky JS, Soriano SG, Madsen JR, Scott RM. (1994) Fine Motor Scale: an interrater reliability study. American H reflex studies in cerebral palsy patients undergoing partial Journal of Occupational Therapy 44: 334–40.
dorsal rhizotomy. Muscle and Nerve 17: 539–49.
Sutherland D, Olshen R, Biden E, Wyatt M. (1988) The Development — Shefner JM, Goumnerova L, Scott RM, Soriano SG, Madsen J.
of Mature Walking. Clinics in Developmental Medicine Nos (1996) The critical importance of stimulus intensity in 104/5. London: Mac Keith Press.
intraoperative monitoring for partial dorsal rhizotomy. Muscle Wright FV, Sheil EM, Drake JM, Wedge JH, Naumann S. (1998) Evaluation of selective dorsal rhizotomy for the reduction of Marascuilo LA, McSweeney M. (1997) Nonparametric and spasticity in cerebral palsy: a randomized controlled trial.
Distribution Free Methods for the Social Sciences. Monterey, CA: Developmental Medicine & Child Neurology 40: 239–47.
Yeargin-Allsopp M, Murphy CC, Oakley GP. (1992) A multiple-source McKenzie S. (1988) Clinical trials, statistics, and dilemmas. Archives method for studying the prevalence of developmental disabilities of Disease in Childhood 63: 231–4.
in children: the Metropolitan Atlanta Developmental Disabilities McLaughlin JF, Bjornson KF, Astley SJ, Hays RM, Hoffinger SA, Study. Pediatrics 89: 624–30.
Armantrout EA, Roberts TS. (1994) The role of selective dorsalrhizotomy in cerebral palsy: critical evaluation of a prospectiveclinical series. Developmental Medicine & Child Neurology36: 755–69.
— — — Graubert C, Hays RM, Roberts TS, Price R, Temkin N. (1998) Selective dorsal rhizotomy: efficacy and safety in an investigator-masked randomized clinical trial. Developmental Medicine &Child Neurology 40: 220–32.
Moher D, Cook DJ, Eastwood S, Olkin I, Rennie D, Stroup DF.
(1999) Improving the quality of reports of meta-analyses ofrandomised controlled trials: the QUOROM statement. Lancet354: 1896–900.
Msall ME, Rogers BT, Ripstein H, Lyon N, Wilczenski F. (1997) Measurements of functional outcomes in children with cerebralpalsy. Mental Retardation and Developmental DisabilitiesResearch Reviews 3: 194–203.
Neville BG. (1988) Selective dorsal rhizotomy for spastic cerebral palsy. Developmental Medicine & Child Neurology 30: 395–8.
Nishida T, Thatcher SW, Marty GR. (1995) Selective posterior rhizotomy for children with cerebral palsy: a 7-year experience.
Childs Nervous System 11: 374–80.
Palisano R, Rosenbaum P, Walter S, Russell D, Wood E, Galuppi B.
(1997) Development and reliability of a system to classify grossmotor function in children with cerebral palsy. DevelopmentalMedicine & Child Neurology 39: 214–23.
Park TS, Owen JH. (1992) Surgical management of spastic diplegia in cerebral palsy. New England Journal of Medicine 326: 745–9.
Peacock WJ, Staudt LA. (1991) Functional outcomes following selective posterior rhizotomy in children with cerebral palsy.
Journal of Neurosurgery 74: 380–5.
Perry J. (1992) Gait Analysis: Normal and Pathological Function.
Price R, Bjornson KF, Lehmann JF, McLaughlin JF, Hays RM. (1991) Quantitative measurement of spasticity in children withcerebral palsy. Developmental Medicine & Child Neurology 33: 585–95.
Russell D, Rosenbaum P, Cadman D, Gowland C, Hardy S, Jarvis S.
(1989) The Gross Motor Function Measure: a means to evaluatethe effects of physical therapy. Developmental Medicine & ChildNeurology 31: 341–52.
Russell DJ, Avery LM, Rosenbaum PL, Raina PS, Walter SD, Palisano RJ. (2000) Improved scaling of the Gross Motor FunctionMeasure for children with cerebral palsy: evidence of reliabilityand validity. Physical Therapy 80: 873–85.
Spilker B. (1991) Guide to Clinical Trials. New York: Raven Press.
Staudt LA, Nuwer MR, Peacock WJ. (1995) Intraoperative monitoring during selective posterior rhizotomy: technique andpatient outcome. Electroencephalography and ClinicalNeurophysiology 97: 296–309.
Steinbok P, Reiner AM, Beauchamp RD, Cochrane DD, Keyes R.
(1992) Selective functional posterior rhizotomy for treatment ofspastic cerebral palsy in children. Review of 50 consecutive cases.
Pediatric Neurosurgery 18: 34–42.
Selective Dorsal Rhizotomy Meta-analysis John McLaughlin et al.

Source: http://www.mackeith.co.uk/annivApr08OA.pdf