There have been many research studies that have tried to answer this question.A variety of treatments have been studied, including orthopedic surgery (tendon transfer, muscle lengthening, etc.), spasticity control (rhizotomy, BTX, or phenol blocks, etc.), orthotic prescription (some type of ankle–foot orthosis, etc), and others (e.g. serial casting, muscle strengthening exercise, biofeedback, etc).

When clinicians are considering potential treatments for an individual, they are often faced with a multitude of previous research studies. There might be conflicting results among the many studies, often making it difficult to draw some conclusion as to whether a particular treatment is effective. One of the techniques that is sometimes used in an attempt to evaluate a large number of research studies is called “meta-analysis.” Meta-analysis is a statistical technique for combining the results of many studies. By combining the results of many studies, a meta-analysis can often produce a more accurate estimate of the efficacy of a treatment than is possible with a single study. Such an estimate may be more broadly generalized to the larger population than just one estimate from a single study.

Performing a meta-analysis can also sometimes give a rough estimate of the overall quality of research that has been performed.

Researchers almost always publish the results of their studies in professional scientific, medical, and engineering journals. The largest freely available database of the articles in these journals is maintained by the National Library of Medicine (it is referred to as Medline or PubMed). Each article in this database is assigned some appropriate indexing terms, called Medical Subject Headings. In Medline, there are over 11,000 articles under the Medical Subject Heading “Cerebral Palsy,” and over 12,000 articles under the Medical Subject Heading “Gait” and its various subheadings (Gait is a term used interchangeably with walking). At the time we did our meta-analysis, there were approximately 650 citations that have both “Cerebral Palsy” and “Gait” as Medical Subject Headings. We reviewed the abstracts of all of these 650 articles, and found that the speed of walking was the most frequently measured quantity to describe how well individuals walked. It is generally assumed among most researchers and clinicians that any treatment that increases walking speed means that the treatment improved walking.

One of the problems in conducting a meta-analysis is that researchers often measure things in different ways. This is true of studies of human walking. Walking has been described using measures of speed, stride length, number of steps per unit of time or distance, the angles of the segments of the leg and foot, the forces between the feet and the ground, and the activities of muscles in the leg, among others. However, how fast someone walks seems to be the most common measure among the studies we reviewed.

On the basis of our literature review, we decided to conduct our meta-analysis by asking the question “Do treatments intended to improve walking in CP result in an increase in walking speed?”

There were 41 of these 650 articles that reported the effect of a treatment on gait velocity. Of these 41 articles, many reported more than one intervention, so the total number of studies in our meta-analysis was 63.

We chose to do the type of meta-analysis that based on a quantity called the “effect size.” In our meta-analysis, effect size was a measurement of how much the speed of walking changed as a result of the treatment. An effect size of zero means that the treatment had no effect. The larger the effects size, the better the treatment worked. Effect sizes of 0.2 are considered small (but still significant) effects, while effect sizes of 0.8 are considered large effects. The effect size is sometimes negative, which means the treatment actually made things worse.

The majority of the 63 studies were small sample prospective studies using samples of convenience. The number of studies in the four categories of treatments were orthopedic surgery (17), spasticity control (15), orthotic prescription (26), and other (5). The number of people participating in a particular study ranged from four to 115. The elapsed time between the pre-treatment and post-treatment testing varied between hours and years. Statistical tests (if used) included parametric tests, most commonly t-tests and analysis of variance (ANOVA). All studies appeared to include participants of both sexes, typically combining results for both sexes; a few studies did not explicitly report gender. Results for participants of different ages were also combined, although some studies differentiated between two age groups. Patients with hemiplegia and diplegia were also analyzed together in most studies, although some studies used only patients with diplegia. No study appeared to differentiate participants by ethnic background. Most studies did not explicitly specify whether patients were receiving other interventions concurrently.

This meta-analysis showed that taken as a whole, treatments intended to improve walking in individuals with CP do have a statistically significant effect on walking speed. The overall effect size for all 63 studies was 0.17, which would be considered small. There were differences in effect size between the different types of interventions. However, we do not believe our meta-analysis results are sufficient to warrant clinical recommendations regarding efficacy of various treatment modalities. This is because we believe additional research is needed.

This meta-analysis led us to suggest some important considerations for future research in this field that are addressed primarily to researchers. There were six recommendations.

(1) Studies trying to determine if interventions actually result in improved walking need to be sufficiently powered to detect improvements. Larger sample sizes are needed, and this suggests researchers need to cooperate in multicenter studies to increase sample size (“power” in a statistical sense means that if an improvement has been made, the study will be able to detect the improvement). Of the 63 studies in our meta-analysis, few were adequately powered.

(2) When quantitative data about walking are reported by researchers, both raw and scaled/normalized data need to be reported to allow for easier comparison and use in future meta-analyses that will be undertaken. We had to exclude 14 studies from our meta-analysis because only normalized data were reported.

(3) In addition to describing the treatment that is being studied to improve walking, researchers need to describe any other interventions which their research subjects might also have received or be receiving which might influence their walking ability.

(4) Future studies are needed to determine if some treatments might be more effective in different subgroups (such as hemiplegia versus diplegia) or different age groups.

(5) Future studies are also needed to determine if there is an optimum time or age at which to provide the treatment, and the time it takes for the treatment to become effective.

(6) Finally, studies need to be performed to determine what level of improvement in the various measures of walking constitute improvements that are not only reported by the patient, but also recognized as clinically meaningful rather than just statistically significant results. While the major focus in this meta-analysis was speed of walking, we realize that there are other measures that may also need to be considered.

In conclusion, our meta-analysis shows that treatments to improve walking ability in CP do work – overall, they do increase the speed of walking. Future research in this field needs to attend primarily to concerns regarding sample size, statistical power, and subject inclusion criteria.

Note: This fact sheet is based on an article “Evaluating interventions to improve gait in cerebral palsy: a meta-analysis of spatiotemporal measures” by Paul, et. al, which appeared in Developmental Medicine & Child Neurology 2007, 49: 542–549.