This systematic review will identify and summarize evidence on physical therapies used after Botulinum toxin-A injection to improve motor function in adults with neurological impairments. In summarizing evidence, the review will evaluate the effectiveness of physical therapies used post-BoNT-A injection for improving motor function in this patient population. Inclusion criteria for studies in the review now follow.
Types of studies
Eligible studies will be randomizsed controlled trials and quasi-randomized controlled trials. Only the first arm of cross-over trials will be included.
Types of participants
Eligible participants will be aged 16 years and over with a neurological impairment (central nervous system damage) and have received BoNT-A injections in their limb muscles up to three months prior to study rehabilitation commencement for spasticity management.
Types of interventions
Physical rehabilitation interventions provided to patients who have received BoNT-A to the upper or lower limb, including casting, splinting, stretching, movement training, exercises, strengthening or electrical stimulation, are compared with sham or no physical therapy.
Types of outcome measures
Measures of range of movement, such as goniometry or torque-controlled range of movement.
Measures of improved functional limb use or active movement of the affected limb as measured on the Action Research Arm Test, the Motor Activity Log (assessment or self report), Wolf Motor Function Test, Fugl-Meyer assessment of motor function, Box and Blocks test, velocity of gait (meters/second) and step length.
Measures of reduced spasticity as measured by the Modified Ashworth Scale.
Health-related quality of life
Caregiver/attendant care time and/or number
Discharge destination/living situation
Rehabilitation length of stay
Studies will be excluded if:
The study design investigated diagnostic or prognostic factors/relationships
Less than 50% of the interventions were applied to the upper or lower limb
Outcomes from physical therapies were not able to be differentiated from the outcomes of other therapies provided simultaneously to participants.
The following databases will be electronically searched for all available years: PEDro, CENTRAL, Pub Med, EMBASE, CINAHL, National Research Register, Meta Registry of Controlled Trials and Occupational Therapy Systematic Evaluation of Effectiveness database (OTseeker). The search will not be limited by date or publication status. We will check the reference lists of any eligible studies identified for further relevant studies. Unpublished, non-peer reviewed sources, such as conference abstracts, will not be included (refer to appendix 1.)
Two authors will independently review all potential studies for inclusion against the eligibility criteria. They will examine the title and abstract and, where necessary, the full text of studies to assess if they are eligible for inclusion. If they cannot reach agreement by discussion, a third author will make the final decision about eligibility.
Two authors will independently use a standard form to extract study characteristics and outcome data from the studies. Discrepancies will be checked against the original data. A third author will make the final decision if there is a disagreement. One author (BK) will enter data in Revman meta-analysis software (Review Manager (RevMan) [Computer program]. Version 5.1. Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2011)
. Data will be reported during therapy as well as at end of therapy. Any outcomes measured after therapy has finished will be grouped as less than one month, one to six months, and over six months.
Methodological quality will be assessed using the PEDro scale by one reviewer (BK). The PEDro scale has established reliability and provides a score out of 10
. Studies which attain a PEDro score of 7 or greater are considered ‘high quality’, those with a PEDro score of 5 or 6 are considered ‘moderate quality’ and those with a PEDro score of 4 or less are considered ‘poor quality’ in terms of study methods and susceptibility to bias
. Adequacy of concealment will additionally be rated using the procedure outlined by Schulz
. The methodological quality of included observational studies will be assessed independently by two raters using the MOOSE (Meta-analysis Of Observational Studies in Epidemiology) guidelines
. Scores will be based on all information available from both the published version and the authors themselves. No trial will be excluded on the basis of poor quality.
Analysis of covariance-adjusted between-group means and standard deviations will be extracted in preference to between-group differences in change scores, and between-group differences in change scores will be extracted in preference to between group differences in final scores. If studies reported data as medians and interquartile ranges, medians will be used as a surrogate for means and standard deviations were estimated as 80% of the interquartile range (studies which do not report an interquartile range (IQR) will be excluded from the meta-analysis). For continuous outcome measures, a pooled estimate of treatment effect will be determined by calculating the mean difference and the corresponding 95% CIs. For dichotomous outcome measures, a pooled estimate of treatment effect will be calculated for each outcome across studies using risk ratio where appropriate and the corresponding 95% confidence intervals (CIs). Time-to-event data will be analyzed using the hazard ratio and 95% CIs as required. When conducting a meta-analysis combining results from crossover studies, the first-arm data only will be used. In the event of missing, incomplete or unclear data, the original investigators will be contacted. If it is not possible to obtain the necessary data for analysis, the study results will be described in the text and will not be included in the meta-analysis.
When there are at least two clinically homogenous studies (studies which investigated the effect of similar interventions on similar populations and reported similar outcomes) meta-analysis will be considered. In such circumstances, the I2 statistic will be used to quantify the heterogeneity of outcomes and informed decisions about whether to pool data. Where I2 is greater than 30% in the presence of significant chi-squared test result (P-value < 0.10), this will be interpreted as indicating heterogeneity
. If there is significant heterogeneity (over 50%), data will not be pooled.
We will enter data extracted from included studies into RevMan software
. Meta-analyses of clinically homogenous therapies will be conducted using a fixed-effects model. Where heterogeneity is substantial (I2 > 50%), the possible causes of heterogeneity will be explored in sensitivity analyses in which individual studies will be omitted one at a time. Should heterogeneity not be able to be explained, a random-effects model will be used.
The robustness of our results will be tested through sensitivity analyses excluding unpublished studies, small studies and studies with a PEDro score less than 5.