INTRODUCTION AND PURPOSE
Balance is the ability to maintain one's projected center of mass with respect to one's base of support to orient and align the body in space.1 Balance is a requisite component for successful completion of functional activities including locomotor and manipulative skills. For the purpose of this article, functional balance in children is defined as the ability to maintain the center of mass with respect to the base of support during typical childhood activities of daily living, school, and play.
Development of balance begins in infancy with the establishment of head and trunk control, and by 12 months most infants are mastering standing and walking.2 Toddlers and preschoolers continue to develop walking, running, jumping, and climbing skills.3 Balance abilities continue to be refined into young adulthood as advanced athletic and performing arts skills are mastered on the basis of the individual's talents and interests.3
Although it is accepted that balance abilities improve as a result of maturation,2,3 the influences of gender and anthropometric measures such as height, weight, and body mass index (BMI) on the acquisition and refinement of balance skills in children are less well understood. The specific relationships between gender and development of balance skills appear to be dependent on the testing tools employed, subjects' ages, and the specific balance skills examined.4–12 Several investigators have used force platforms and dynamic posturography to examine gender differences in the development of standing postural sway,4,5 and in responses to self-induced perturbations with limits of stability (LOS) testing.5 Boys younger than 10 years tend to sway more than girls in static standing, potentially indicating greater instability.4,5 Girls younger than 10 years also outperform boys on most parameters associated with self-induced perturbations in limits of stability analyses.5
Single-item clinical balance measures such as the Functional Reach Test (FRT)6,7 and single limb stance (SLS)8 have been used to examine gender effects on children's balance skills. Gender does not appear to affect FRT scores6,7; however, Lee8 found that girls had significantly smaller mean radius of center-of-pressure measurements while maintaining SLS than did boys, indicating greater stability.
Gender differences in balance have also been investigated using standardized developmental tests.9–12 Jonaid and Fellowes9 examined gender effects on Movement Assessment Battery for Children performance in children aged 7 to 8 years. They found no significant differences between boys and girls in balance skills including timed SLS and tandem walking on a beam. Davies and Rose10 found similar results when examining relationships between age, gender, and acquisition of motor skills as measured by selected items from the Bruininks Oseretsky Test of Motor Proficiency (BOT). Although gender differences were found in many of the gross and fine motor skills tested, no significant gender differences in balance skills (SLS time and beam walking) were observed in children aged 7 to 18 years. These findings are in contrast with those of other investigators, who examined children's static and dynamic balance skills as part of the Physical and Neurological Assessment of Subtle Signs11 and the Zurich Neuromotor Assessment12 tests. In these studies, girls aged 7 to 14 years outperformed boys in SLS tasks11,12 and in dynamic balance tasks, including tandem walking forward and backward,11 heel walking,12 and toe walking.12
Balance tends to degrade as BMI increases in children. Increased spontaneous sway in static standing has been observed in boys who are obese and aged 8 to 10 years13 and in teens who are obese when compared with peers of normal weight.14 Goulding et al15 examined the effect of BMI on BOT balance subtest scores and LOS in boys aged 10 to 21 years. Postural sway and LOS did not change with increasing BMI; however, scores on the balance subtest of the BOT were significantly lower in the boys who were overweight (BMI > 85th percentile) than in the boys of normal weight (BMI < 85th percentile). These authors noted that SLS on a beam was most challenging for boys who were obese.15 Using International Obesity Task Force criteria, D'Hont et al16 compared Movement Assessment Battery for Children performance in children who were obese and children who were of normal weight, aged 5 to 10 years. Children with obesity scored significantly lower on balance and ball manipulation tasks than their peers who were of normal weight. Although evidence indicates that obesity may adversely affect balance, the relationship between balance and BMI in children who are not obese is less clear.
The influences of height and weight on balance must be cautiously interpreted because of the interrelationships with age and BMI. Donahoe et al7 examined the effects of age, gender, height, and weight on anticipatory balance control, using an FRT in children aged 5 to 15 years. Functional reach (FR) distances increased with height and weight, although age was the only significant factor related to a child's FR ability. Norris et al6 examined the influence of height, weight, and age on FR in 3- to 5-year-old children. Functional reach increased with age, but weight was the only significant predictor of FR. Habib and Westcott17 examined the effects of height, weight, and age in 180 Afghan children aged 5 to 13 years on the FRT, the Timed Up and Go (TUG), and the BOT Running Speed and Agility and Balance subtests. Height had a significant effect on FRT, TUG, and running speed and agility scores in younger children but had a significant influence only on TUG scores in older children. The results of these studies indicate that anthropometric factors have varying influences on balance, depending on the specific balance tasks and age. A greater understanding of the effect of age, BMI, gender, height, and weight may prove helpful to guide decision making when examining children's balance abilities.
Pediatric physical therapists use a variety of functional balance screening tools developed for adults including the FRT,6,7 the TUG,18 the Timed Up Down Stairs,19 and the Berg Balance Scale (BBS).20 A review of the content and construct of these tools leads one to believe that they may be useful in assessing balance dysfunction in children, but each one has limitations. Although there is considerable support for use of the FRT in children,6,7 the main disadvantage is that it examines only one aspect of functional balance, anticipatory control in forward reach. The TUG and Timed Up Down Stairs are multistep single-item assessments.18,19 Both emphasize dynamic balance in walking and movement transitions but neither examines static standing balance. Williams18 examined TUG performance in children with typical development and in children with motor impairments resulting from cerebral palsy and spina bifida. The TUG discriminated between children developing typically and those with moderate to severe impairments, but it did not identify children with mild motor impairments.
Kembhavi et al20 examined the use of the BBS in children, aged 8 to 12 years, with cerebral palsy and in children developing typically. The BBS discriminated between children classified at various levels on the Gross Motor Function Classification System, but like the TUG, BBS scores did not differentiate children with mild impairments, (Gross Motor Function Classification System I), from children without motor impairments. Franjoine et al21 investigated test-retest reliability of the BBS in children aged 4 to 13 years, who were developing typically. Reliability could not be assessed because children were uncooperative with many of the static balance items due to the lengthy time requirements for maintenance of stationary positions. Children also appeared challenged by the hierarchical organization of the scale. Although the BBS has excellent reliability in adult populations,22 a reliable multiconstruct balance assessment is lacking for children.
As a result of difficulties using the BBS in children, the Pediatric Balance Scale (PBS) was adapted from the BBS. The systematic process of modification of the BBS for use with children has been previously reported.21 The PBS is a 14-item, criterion-referenced measure, which examines functional balance in the context of everyday tasks (Appendix available online at http://links.lww.com/PPT/A15). It can be administered and scored in less than 20 minutes using equipment commonly found in schools and clinics. The PBS also has excellent test-retest (intraclass correlation coefficient, ICC [2,1] = 0.923), interrater reliability (ICC [2,1] = 0.972), and intrarater reliability (ICC [2,1] = 0.895-0.998) (M.R.F., unpublished data, 2010). The PBS also has the potential to discriminate between children developing typically and children with mild motor impairments,23 as well as those with moderate to severe motor impairments.21
For clinicians to use the PBS as a balance screening tool or a clinical outcomes measure, typical performance standards must be established on the basis of key characteristics of children including age, gender, and anthropometrics. Therefore, the purposes of this study were to (1) investigate PBS performance in pre–school- and school-aged children with typical development and (2) explore the relationships among age, gender, height, weight, and BMI, and PBS performance.
Six hundred forty-three healthy children, who were developing typically per parent report, participated in this study. This sample of convenience included 321 boys and 322 girls, aged 2 years 4 months to 13 years 7 months. Subjects were recruited from public and private preschools, day care centers, schools, and the community. The majority of subjects were from the Midsouth, Central New York, and the eastern Great Lakes regions; however, a small number were from the Northeast, Louisiana, and the Washington, District of Columbia, areas. Five of the participating test sites were urban inner-city sites, and the others were in urban, suburban, and small-town locations.
Children were eligible to participate if they were able to stand unsupported for 4 seconds and follow simple 1-step directions. Children were excluded from participation if they had any of the following conditions: (1) history of any communicable illness within the previous 2 weeks, (2) inability to follow 1-step directions, (3) any medical conditions that resulted in a physician-recommended activity restriction, (4) history of any orthopedic injuries within 3 months before data collection, (5) identified developmental delays as indicated by parent on health history form, and (6) uncontrolled seizures (participants had to be free of seizures for the previous 6 months).
Thirty-three raters (4 physical therapy faculty and 29 entry-level physical therapist students) participated in data collection over a period of 12 years. All raters were formally trained in administration and scoring of the PBS by the test author (M.R.F.) before participating in data collection. Each student rater participated in a minimum of 20 hours of training that included the following activities: expert demonstration, simulated administration and scoring activities, administration and scoring from video recordings, and test administration and scoring with children who were not part of the data set. At the conclusion of training, each rater achieved a mastery-skill level for administration and scoring of the PBS established by the test author; interrater reliability was established within rater groups and with the test author (ICC [3,2] = 0.90-0.92). Student raters collected data under the supervision of a faculty investigator.
This study was approved by the Belmont University institutional review board, the Daemen College Human Subjects Research Review Committee, and the State University of New York, Upstate Medical University institutional review board. Before data collection, each subject's parent/guardian completed an informed consent and a health history form, which were used to identify children who were not eligible to participate in the study due to medical or developmental conditions. Children, 7 years and older, signed an age-appropriate pediatric assent form. Each child was then asked to stand for 4 seconds, and parents and/or teachers were asked whether the child could follow 1-step directions to ensure that subjects met the minimum inclusion criteria. Height (inches) and weight (pounds) were measured and recorded for each child.
The PBS was administered to each child individually using the protocol for PBS test administration and scoring described by Franjoine et al.21 Each subject participated in one testing session that lasted 10 to 20 minutes. Testing locations included classrooms, gymnasiums, hallways, homes, cafeterias, and playgrounds.
Data from a total of 641 subjects were analyzed (data from 2 subjects were eliminated because of refusal to attempt greater than 50% of the test items). Children were divided into 11 groups on the basis of 6-month age increments for data analysis. Each age group was further divided into 2 gender subgroups, creating a total of 22 groups. Children, 7 years and older, were collapsed into one age group for analysis because there were no statistically significant differences in PBS scores for these children (P > .05). The numbers of subjects per group in children younger than 7 years varied by age and gender but were adequate for statistical analysis. Descriptive statistics including PBS total score means, standard deviations (SDs), ranges of performance, and 95% confidence intervals (CIs) of the means were calculated for each of the 22 groups. To further describe each group, means and SDs were calculated for height, weight, and BMI.
A 2-way analysis of variance was performed to determine the effects of gender and age on PBS total scores. Pediatric Balance Scale total scores were converted to a proportion of total possible points for each child. To meet the assumptions of equal variances as tested by Levene's test of equality, scores were converted to a proportion with an arcsine-square root transformation applied to proportions. Post hoc Bonferroni multiple comparison tests were used to determine significant age-group differences.
To explore age and gender differences among static and anticipatory items, scores were totaled across the 5 static items (numbers 4, 6, 7, 8, and 9) and 3 anticipatory items (numbers 11, 12, and 14) (see the Appendix online at http://links.lww.com/PPT/A15.) The scores were converted to a proportion of the total possible score and transformed by arcsine-square root before analysis. A 2-way analysis of variance was used to determine age and gender differences. When significant interactions were detected, a separate 1-way analysis of variance was conducted for both boys and girls. Post hoc Bonferroni tests were used to identify significant pairwise differences.
Spearman rank correlation analyses were performed to determine the associations between PBS scores and age, height, weight, and BMI; Pearson rank correlation coefficients were determined to represent the association of variables relating to body size.
Data were further analyzed, using the 95% CI to identify ranges of PBS typical performance within each age and gender group. This method was selected, as the PBS scores did not follow a normal distribution especially in the older-age groups where most children achieved close to the maximal score of 56. The lower bound of the 95% CI was used to identify outliers and determine a “critical cut score” for each group. The numbers and percentages of children scoring above and below this critical cut score were calculated to further describe the sample.
Mean PBS total test scores, standard deviations, ranges of performance, and 95% CIs of the means, stratified by age and gender for 641 children, are presented in Tables 1 and 2. Pediatric Balance Scale scores increased with advancing age from ages 2 through 7 years. Increases were most pronounced in younger age groups (Figure 1). Performance on the PBS was significantly affected by both age (F10,619 = 71.32, P < .0001) and gender (F1,619 = 6.76, P < .01) with girls achieving higher mean scores across the age groups.
Effect of Age on PBS Performance
Post hoc Bonferroni multiple comparisons indicated statistically significant differences in PBS scores between the ages of 2 years 6 months and 3 years, ages 3 and 4 years, and both ages 4 years and 4 years 6 months as compared with 5 years (P < .05), with older children demonstrating higher scores. Scores continued to differ across age groups after 5 years of age with older age groups scoring higher than younger groups, but the differences were more modest and not statistically significant with the exception of a statistically significant difference between the ages of 6 and 7 years (Figure 1).
Effect of Gender on PBS Performance
Girls had significantly higher scores across age groups (F1,619 = 6.76, P = .01). This effect was most pronounced in children younger than 5 years. The greatest gender differences in mean total PBS scores occurred in the 2 years, 2 years 6 months, 3 years, and 4 years age groups (Figure 1; Table 1). Items were further examined by balance construct groupings (ie, static, anticipatory). Both age (F1,614 = 15.68, P < .001) and gender (F10,614 = 56.48, P < .001) significantly affected performance on static items; however, the interaction between age and gender (F10,614 = 2.72, P = .003) was also significant, so separate analyses on girls and boys were performed. An analysis of girls' performance alone showed significant age effects (F9,308 = 25.3, P < .001) with girls 4 years 6 months and older showing similar performance. Girls in both the 2 years and the 2 years 6 months age groups performed at a significantly lower level on these static items than girls 3 years and older (Figure 2). Boys' performance differed by age (F10,306 = 38.4, P < .001), with lowest scores in the 2 years and the 2 years 6 months age groups. Scores of static standing items for boys, aged 3 through 4 years, were similar (Figure 2).
Analysis of anticipatory items indicated no gender effect (F1,614 = .02, P = .88), although performance differed significantly by age (F10,614 = 34.4, P < .001). Children in the 3 years 6 months age group and those younger performed similarly on anticipatory items (Figure 3). No differences were observed between the 3 years 6 months and the 4 years age groups or between the 4 years and the 4 years 6 months age groups. Children 5 years and older had similar scores with 53% of 5-year-olds attaining a perfect score on all of these items.
Relationships of Height, Weight, and BMI With PBS Performance
Subjects' height, weight, BMI, and PBS total test scores, stratified by age and gender, are presented in Table 1. As would be expected, there were strong correlations between age and height (r = 0.897, P < .01) and age and weight (r = 0.833, P < .01). Body mass index was not strongly correlated with age (r = 0.210, P = .01) or height (r = 0.112, P < .01) but was moderately correlated with weight (r = 0.558, P < .01). There were moderately strong correlations between age and PBS score (rs = 0.689), height and PBS score (rs = 0.650), and weight and PBS score (rs = 0.642). The correlation between BMI and PBS scores was weak (rs = 0.182).
Variability of Performance
Variability of performance was lower in older age groups (Figure 4). The largest variability was seen in the 2 years 6 months age group (mean ± SD = 34.3 ± 7.72). Variability was lower compared to the youngest groups but remained high in the 3 years, 3 years 6 months, 4 years, and 4 years 6 months age groups (SD3 = 6.55, SD3.5 = 5.02, SD4 = 5.76, SD4.5 = 5.07). Variability then gradually decreased with age (Table 1).
Establishment of “Cut Points”
Data were further analyzed to establish PBS total score typical performance and outlier ranges within each age and gender group using 95% CIs of the means. Table 2 provides the lower boundary of the 95% CI of the means of PBS total scores by age and gender. Numbers and percentages of children scoring below the 95% CI of the means for each age and gender group are also presented in Table 2.
Progressively higher PBS scores were most marked between the ages of 2 and 5 years and were more modest between the ages of 5 years 6 months and 7 years (Figure 1). Scores ranged from 15 to 53 in children younger than 4 years (Table 1). By the age of 5 years, 30.3% of the children obtained a maximal score of 56, and 79.8% scored 53 or higher. In the oldest age group (7 years and older), 69.1% of the children obtained the maximal score of 56, and 95% of these children scored 53 or greater (Figure 5).
A nonlinear increase in PBS scores occurred in children aged 2 through 7 years (Figure 1). The largest increases in mean PBS scores occurred between the 2 years 6 months and the 3 years age groups (increase of 12 points) and between the 2 years and the 2 years 6 months age groups (increase of 8 points). Most children aged 7 years and older demonstrated mastery of PBS items, which is consistent with previous studies.23 The PBS appears to measure acquisition of functional balance skills in young children; however, further investigation of validity is recommended.
In the present study, girls outperformed boys in PBS total test scores. Gender differences were most pronounced in children 4 years and younger. In examining the content of the PBS, it is noted that 5 PBS items (36%) examine static standing tasks of varying difficulty (standing unsupported, standing with eyes closed, standing with feet together, standing with 1 foot in front, and standing on 1 foot). Three items (21.4%) examine anticipatory standing balance (turning to look behind, retrieving object from floor, and reaching forward with outstretched arm). Previous studies support gender differences in balance with girls outperforming boys on tasks requiring static standing,4,5 assuming and maintaining standing with a narrowed base of support,8,10–11 and tasks requiring a self-induced challenge to limits of standing stability.5 The present study supports previous works indicating that girls perform better on static standing balance; however, gender differences were not detected in balance tasks requiring anticipatory control.
As expected, height and weight were correlated with age, and age, height, and weight were all correlated with PBS total score. Body mass index correlations with balance as measured by PBS scores were very weak. This is in contrast to previous studies suggesting that BMI adversely affects postural sway (static balance) tested on dynamic posturography13–15 and on functional balance tasks such SLS and tandem standing.15,16 Although children's PBS performance in the present study was not correlated with BMI, it should be noted that the mean age in the present study (5 years 9 months) was considerably lower than that in previous studies investigating BMI and balance (age range = 5-21 years). Using Centers for Disease Control and Prevention guidelines,24 the majority of children (59.7%) in the present study had BMIs within the healthy range for age (BMI in the 5th to less than 85th percentiles), while 14.7% and 17.6% were overweight (BMI in the 85th to less than 95th percentiles) and obese, respectively (BMI in the 95th and greater percentiles). Interestingly, 8% of the children in this sample were underweight (BMI < 5th percentile). The present study is a descriptive study, and it was not the intention to compare the performances of children of normal weight and those who were obese on the PBS.
Height and weight were moderately correlated with PBS scores, and both were correlated with age. Previous authors have found that the relative effects of height and weight on balance vary with the age of test group and the specific balance tasks.6–8,17 The PBS contains 14 distinct functional balance tasks, each of which may be affected differently by age, height, and weight. Although it is not possible to absolutely differentiate effects of stature versus age in many of the PBS tasks, the strategies employed in the following PBS items appeared to be most influenced by height: sit to stand, transfers, turning to look behind, retrieving an object from the floor, and reaching forward. Item number 3 (transferring from a bench to an adult-sized chair) provides an example of strategy selection on the basis of both height and age. A variety of strategies for completing this task were observed. Most 2-year-old children (mean height = 33 in) climbed into the chair and rotated into a sitting position, and many 3-year-olds (mean height = 38 in) added vertical propulsion to this strategy by leaping into the chair. By the age of 4 years (mean height = 42 in) to 5 years (mean height = 44.5 in), most children transferred to the chair buttocks first, but many needed to wiggle back into the chair because of short stature. It must be acknowledged that many factors other than age, gender, height, weight, and BMI may have influenced a child's performance on the PBS. Cognition, language comprehension, behavior, and willingness to participate are variables that may influence a child's PBS performance. The results of this study do not allow one to provide clear clinical guidelines regarding the degree of the effect of height and weight, as both were correlated with age and PBS scores.
A wide range in PBS performance (range = 30 points) was observed in the 2 years, 2 years 6 months, and 3 years age groups (Table 1). Caution must be used in examining the 2-year-old group, as the sample size was smaller than other groups (5 subjects). Variability, as described by the standard deviation, was greatest in the youngest age groups. The 2 years, 2 years 6 months, and 3 years age groups had the greatest variability within each group (Table 1). Variability was lower throughout the remainder of the preschool years. The variability seen in the youngest children (2 years through 2 years 11 months) may have been in part due to the inability of this group to follow directions and comply with testing. Anecdotally, raters reported that many 2-year-old children had difficulty remaining within the confines of the testing space and understanding and following directions. Two of 7 children in the 2-year-old age group were excluded from analysis because they refused to attempt more than 50% of the test items. Many of the PBS items (ie, standing with 1 foot in front, standing on 1 foot) are developmentally challenging for a typical 2-year-old, which may contribute to his or her unwillingness to participate in testing. Although variability is still high in the 3-year-old group, language acquisition and preschool social skills are greater so that the majority of 3-year-old children are willing to attempt all or most PBS items. On the basis of these findings, the authors recommend caution when interpreting PBS scores in children younger than 3 years.
Many developmental tests use cut points to identify children who fall below expected age-based performance and may need further assessment or intervention. These cut points have been established using a variety of methods. Critical reach values were described for the FRT by Donahoe et al,7 using 95% CIs of the means. The Alberta Infant Motor Scale employs the use of cutoff points set at the 5th and 10th percentiles.25 The BOT-2 manual26 contains tables that describe below average performance as 1.0 SD below the mean and well-below performance as 2.0 SD below the mean. Table 1 provides the lower bounds of 95% CI of the means to indicate the mathematical boundaries of what may be considered typical performance on the PBS. Because of the lack of normality in the data in the present study and the high percentage of children reaching the maximum PBS score of 56 by the age of 7 years, the lower bounds of the 95% CI of the means were used as an indicator of performance that may warrant further investigation. Although the data support the use of CIs to identify children at risk of balance dysfunction, scores falling above the lower bound of 95% CI of the means may not imply absence of balance dysfunction or presence of age-appropriate balance skills. The authors suggest that while these cut points may provide guidelines for interpreting PBS scores, clinicians are cautioned to interpret the results of the PBS in conjunction with a clinical examination of balance skills. Further investigation of the sensitivity and specificity of the PBS is recommended.
Although typical performance ranges and cut points based on 95% CI of the means have been suggested, differentiation of performance is limited in children developing typically beyond the age of 6 years. Twenty-nine percent of children in the 6 years 6 months age group achieved the maximum score of 56, and 90% achieved a score of 53 or greater (Figure 5), indicating a ceiling effect; however, the percentage of children achieving PBS scores of 55 and 56 continues to increase beyond 7 years of age. The authors recommend that clinicians consider using tests developed specifically for older children, such as the BOT-2, to investigate balance dysfunction in children older than 6 years.
The performance expectations of children presented in this study should not be interpreted as norm-referenced values. This large sample size does not follow a normal distribution, as scores are clustered at the upper limit in children aged 5 years and older. Ethnicity can affect performance on other motor tests and may also influence a child's performance on the PBS.27 Children in this study were from diverse backgrounds; however, ethnicity was not formally documented during the first 4 years of data collection. To obtain a truly normative sample, the authors recommend expanding the sample to include children representative of the US population in terms of ethnicity, geography, and socioeconomic status.
The PBS is a 14-item, criterion-referenced screening tool of functional balance, which is most appropriate for children aged 3 through 6 years. This study provides the clinician with guidelines for interpretation of PBS performance on the basis of age and gender, although the effects of factors such as ethnicity and socioeconomic status on PBS performance are not known. The PBS must always be interpreted in conjunction with a clinical examination of balance.
We thank all the children and their families who participated in this study, and program coordinators and staff at the community sites who allowed us to collect data. In addition, we thank numerous entry-level physical therapy students from Daemen College and Belmont University who assisted with data collection.