Invited commentary.

Time after time, parents of children with cerebral palsy (CP) ask clinicians, "Will my child walk?" To support this goal, a wide variety of therapeutic interventions and surgeries are offered. Research studies on the effectiveness of these approaches, however, often fail to provide outcome measures that are meaningful to children with CP and their families. To be meaningful, it is essential that the outcomes are ecologically valid and measure what the children do routinely in the context of their daily life and the extent to which the children are able to participate in age, gender, cultural, and community life roles and responsibilities.

Bjornson and colleagues' study, therefore, is timely and a valuable extension of previous work on daily ambulatory measures for children. The primary purpose of their study was to compare day-to-day ambulatory performance of youth with CP in comparison with youth developing typically, utilizing a StepWatch monitor. (1) A second purpose was to examine the validity and feasibility of ambulatory activity monitoring of children with CP. As defined in the International Classification of Functioning, Disability and Health terminology, (2) Bjornson and colleagues' measurement of performance (ie, "what the child really does" in the child's natural situation) unequivocally meets the criteria for a meaningful outcome. The focus of my commentary, therefore, will be on the magnitude of homogeneity in ambulatory performance within Gross Motor Function Classification System (GMFCS) categories I, II, and III (3) and the feasibility of monitoring ambulatory activity of children with CP.

One of the barriers to understanding the effectiveness of physical therapy for children with CP has been our inability to summarize disparate evidence across studies due to wide variations in the characteristics of the participants and the lack of ecologically valid outcome measures. Bjornson and colleagues' examination of ambulatory performance based on GMFCS level further elucidates the utility of the GMFCS in forming homogeneous groups. Having predicted that "activity performance would be ordered by functional walking activity level as measured by the GMFCS," the authors reported statistically significant differences in average daily total steps between youth developing typically and youth in GMFCS levels I, II, and III; between youth in GMFCS levels I and III; and between youth in GMFCS levels II and III.

In Bjornson and colleagues' article, Figure 2 shows large interquartile ranges for average daily total steps within groups. Furthermore, there is much overlap among the groups. Therefore, it is surprising to me that statistically significant differences were found. Large interquartile ranges within groups and overlap among groups also are evident in Figures 3, 4, and 5, which present data on percentage of time active and ratio of activity level. In addition, the authors use median scores, which are usually indicative of non-normal distribution and heterogeneous variances requiring nonparametric statistical tests.

My interpretation of the findings, in contrast to that of the authors, is that we should anticipate relatively wide variety in ambulatory performance of children with CP within GMFCS levels. Given the study's small sample size and division of participants with CP into 3 groups, it is essential for all ambulatory performance data that the researchers present confidence intervals, estimates of the boundaries within which the true population values are covered. The authors address this, in part, by providing confidence intervals for average daily total steps for youth developing typically and for children with CP. They did not, however, provide separate confidence intervals for each of the 3 GMFCS categories. As a result, estimates of the true population values for these data are obscured.

Effect size and amount of overlap also are important statistics that present the strength of association between variables--in this case, the strength of association between ambulatory performance and GMFCS category. Therefore, to provide the reader with a measure of whether individuals in one GMFCS category differ moderately or substantially from those in the other groups, nonparametric statistics such as the Cliff Delta or the Cohen d, when the data are normally distributed, also are warranted. (4)

It is critical that differences in ambulatory performance reflect real differences and not variations in monitor usage. Bjornson and colleagues defined inadequate monitoring as "wearing the monitor upside down, not wearing the monitor, or wearing the monitor incorrectly on the ankle," but the researchers did not present the extent of non-adherence or whether it was different across GMFCS levels or other factors, such as, age, sex, or socioeconomic status. Given the busy lives of families during the school year, especially those with children with CP, it is plausible that parents may forget and children may refuse to wear a monitor.

To strengthen validity of monitor use in future studies, tracking of adherence is suggested through logs, 24-hour recall surveys, or other such strategies. In addition, I recommend that logs of after-school and weekend activities that involve walking be collected to provide the researchers with triangulation data on ambulatory performance. These data may further explain ambulatory performance variations within groups based on the children's interests and motivation to be physically active as well as the family lifestyle.

Bjornson and colleagues provide several viable applications for use of the StepWatch monitor. For many children with CP, antigravity muscles are not sufficiently strong to surmount the Increasing demands that accompany physical growth. An inverted "U" curve reflects the common pattern of increases in ambulatory performance during early to middle childhood, which is often followed by a period of plateau and then ambulatory performance decline. Therefore, as the authors concluded, in addition to monitoring ambulatory performance to document the effectiveness of physical therapy interventions, longitudinal monitoring can elucidate the natural history of ambulatory skills within each of the GMFCS levels as children transition to adolescence and adulthood. National data on physical activity for children who are developing typically show decreases in activity from the elementary to the high school years and then even less activity throughout adulthood. (5) Changes in ambulatory status of children with CP, therefore, should be evaluated within this context. Along this avenue of research, another application is to monitor ambulatory performance as an outcome measure for programs promoting increased participation in physical activity for children with CP or other physical disabilities.

At present, I do not envision routine use of the StepWatch monitor by clinicians. Instead, use of the StepWatch monitor may be restricted to research due to cost of the monitor, software, and loading unit (more than $2,000) as well as time and training in calibrating the unit to establish reliability, downloading the data, and analyzing results. One note of caution not brought to the readers' attention in the article is that decreases in daily steps should not automatically be interpreted as decline. For example, in response to an intervention, increased stride length may occur as a result of decreased number of steps required to cover the same distance, or increased speed may decrease ambulatory duration.

In conclusion, Bjornson and colleagues' findings provide a valuable step (pun intended) to providing an ecologically valid outcome measure for examination of the effectiveness of interventions for children with CP. I look forward to reading future studies by these researchers.

DOI: 10.2522/ptj.20060157.ic

References

(1) Cyma Inc. Make every step count: Step Watch3.Available at: http://www.cymatech. com. Accessed December 8, 2006.

(2) International Classification of Functioning, Disability and Health (ICF). Geneva, Switzerland: World Health Organization; 2002.

(3) Palisano RJ, Rosenbaum PL, Walter S, et al. Gross Motor Function Classification System. Hamilton, Ontario, Canada: Neurodevelopmental Clinical Research Unit, McMaster University; 1995.

(4) Hess MR, Kromrey JD. Robust confidence intervals for effect sizes: a comparative study of Cohen's d and Cliffs Delta under non-normality and heterogeneous variances. Paper presented at: Annum Meeting of the American Educational Research Association; April 12-16, 2004; San Diego, Calif. Available at: www.coedu.usf.edu/ main/departments/me/documents/cohen. pdf. Accessed December 8, 2006.

(5) National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and prevention. Promoting better health for young people through physical activity and sports: a report to the President. Available at: www.cdc.gov/ healthyyouth/physicalactivity/pdf/facts. pdf. Accessed December 8, 2006.

S Goodgold, PT, ScD, is Professor and Interim Chair of Physical Therapy, Simmons College, Boston, MA 02115 (USA). Address all correspondence to Dr Goodgold at: [e-mail].