CB&M Properties

The CB&M was developed in response to the needs of clinical practice to identify postural instability and evaluate change following interventions in patients who are ambulatory and functioning at a high level of functional mobility, yet who have persistent balance problems. It was initially developed for the adult population with traumatic brain injury (TBI) and has since been validated with the paediatric brain injury population. The following is a summary of measurement properties across the studies.

Clinical Feasibility
Adults (Howe et al 2006)
The CB&M can be administered within 20-30 minutes depending on the patient's ability and the rating therapist's familiarity with the administration of the measure. Much of the testing is designed to occur within a clinic setting upon a measured track (requiring approximately 2 X 10 metre area). The therapist must also have access to a full flight of stairs. The following materials are required for testing: stop watch, laundry basket, 2 lb and 7.5 lb weights, bean bag and a paper circle 20 cm in diameter used as a visual target. Details of the track set-up can be found in the CB&M Administration and Scoring Guidelines. For some items and/or for patients with more compromised balance, the rating therapist may find value in having an assistant to monitor patient safety. This will allow the therapist to appropriately observe the patient's performance. For example, during the 'Walking & Looking' task, the therapist must be positioned away from the patient to observe whether they are looking at a visual target while walking. A second person may therefore be needed to walk alongside the patient to ensure safety.

Children and Youth (Wright et al 2010)
The time to complete the CB&M in the pediatric reliability sample varied from 19 to 46 minutes (mean time = 29 minutes [SD= 6.8]). Children as young as 7 years of age can be assessed on the CB&M. The CB&M test process does not allow practice trials, as it is meant to assess a person's immediate response to a particular task/demand in a community setting. However, the test administrator may use his/her discretion and repeat the instructions and/or demonstration of a test item if the child clearly does not understand what to do in order to complete that item. When younger children are assessed, several demonstrations of an item may be required to ensure that they clearly understand the multi-step instructions before they attempt the task. As well, prior to testing, the treating therapist should be comfortable that poor performance will not unduly upset the child who is likely still coming to terms with the impact of brain injury on his/her physical skills. Children with balance issues should be given some stand-by guarding (but no hands-on assistance) for the more difficult items to ensure safety, e.g., descending the stairs, running, and step-ups. As noted above with the adult testing, it may be prudent to have a second physiotherapist or physiotherapy assistant present to perform this role when testing the less mobile children.

This assessment is meant to assess performance in a functional setting and therefore should be tested in the footwear a person would use in the community. Given the wide diversity of footwear that children and youth tend to wear, they do not necessarily come to the test wearing the appropriate shoes. Parents need to be informed prior to the test that the child should wear a running style shoe. Ideally, they should be reassessed in the same footwear in order to create the most accurate picture of change in ability.

Content Validity
Adults (Howe et al 2006)
Content validity was examined in the first phase of research and was supported by the involvement of patients with brain injury and clinicians in the item generation phase. Initial item generation for the CB&M involved open-ended interviews with 17 physical therapists and occupational therapists with neurorehabilitation experience and 7 ambulatory patients with brain injury who were living in the community. The therapist interviews probed their perceptions of the balance problems experienced by the ambulatory patient and identified the standardized and non-standardized methods used to evaluate this population. The patient interviews probed the balance difficulties experienced during their normal daily activities and the specific circumstances surrounding these situations. From the information obtained, some common themes and difficulties regarding balance within community mobility emerged. Nineteen items were initially generated.

To further investigate content validity, 13 physical therapists from the continuum of care settings (acute, rehab and community), with 5.9 ±4.4 years of practice experience (range 1.5±16 years) working with the acquired brain injury patient population, each assessed 4 patients within their care setting to gain familiarity with the scale. Therapists were asked their perceptions of the patient's balance performance within mobility and activities of daily living using a 5-point global balance rating scale for each patient. There was a moderate to good relationship (r=0.62; p<0.001) between the physical therapists' global rating scales and CB&M total scores.

In a follow-up focus group, through questionnaires and discussion, the therapists were asked to identify the items that were most relevant to measuring balance, items that should be considered for exclusion and any additional areas of balance and mobility that were not represented and should be included. The final measure was comprised of 13 items with six items performed on both the right and left side resulting in a total of 19 tasks.

Reliability
Adults (Howe et al 2006)
Intra-rater, inter-rater and test-retest reliability of the CB&M was established among 4 teams of 2 physical therapists who assessed a total of 32 patients (age 34 ± 12 years ) with traumatic brain injury (TBI) from both inpatient and outpatient rehabilitation settings. All physical therapist assessors completed a training session prior to the studies. To examine inter-rater reliability, each team assessed 8 patients and the scores of the 2 team members were compared. Assessments were again conducted in 5 days to evaluate test-retest reliability. Three weeks later, a videotape of the initial assessment was reviewed and scored independently by the same physical therapist to evaluate intra-rater reliability. In an earlier phase of research, test-rest reliability was also examined with a different methodology: immediate test-rest reliability was established with 13 physical therapists who assessed a total of 36 patients with TBI where each patient immediately performed two consecutive trials of each scale item. Internal consistency was also evaluated within both phases of research.

The CB&M demonstrated intraclass correlation coefficients (ICCs) of 0.977 (95% CI =0.957-0.986), 0.977 (95% CI =0.972-0.988), 0.898 (95% CI =0.624-0.953) and 0.975 (95%CI =0.810-0.991) for intra-rater, inter-rater and test-retest (5 day apart) and test-retest (immediate) reliability, respectively. Internal consistency was evaluated with a Cronbach's alpha of 0.96 and 0.95 in the earlier and latter phases of research, respectively suggesting that the items correlate highly and reflect the same construct. This level of reliability suggests that the CB&M can be used with confidence in clinical and research settings.

Children and Youth (Wright et al 2010)
Intra-rater, inter-rater and test-retest reliability of the CB&M was established by testing with 7 pediatric physical therapists. They worked in rater pairs to assess a total of 32 children and youth (range from 8 years, 2 months – 18 years, 1 month, mean age 14 years, 1 month) with an acquired brain injury (ABI). The sample was composed of 14 children with a TBI and 18 with a non-TBI. Three children's treatment centres (inpatient and outpatient rehabilitation) were involved in the study. All physical therapist assessors completed an extensive CB&M training session and passed a criterion test prior to being allowed to assess participants for the study. To examine inter-rater reliability, one PT assessor (PT1) administered and scored the CB&M, and a second PT (PT2) observed and scored independently. Participants were reassessed 3 to 10 days later by PT1 to evaluate test-retest reliability. All assessments were videotaped to permit a comparative evaluation of reliability of rating from videotape. At least one month later, the videotape of the initial assessment was reviewed and scored by PT1 (intra-rater reliability). Inter-rater reliability of the video versions of the child's CB&M test was also evaluated with the two PT raters who had done the live rating.

The CB&M demonstrated intraclass correlation coefficients (ICCs) of 0.93 (95% CI =0.87-0.97), and 0.90 (95% CI =0.81-0.95) for inter-rater and test-retest reliability (3 to 14 days apart, mean interval = 8.4 days) respectively in the live-rating situation. The ICCs were 0.96 (95% CI=0.91-0.98) for intra-rater reliability in the live- versus video-rating situation, and 0.95 (95% CI=0.89-0.97) and 0.90 (95%CI =0.79-0.95) for inter-rater and test-retest reliability in the video-rating situation. Bland-Altman plots revealed a slight indication of greater test-retest differences for mid-range CB&M scores. Internal consistency was evaluated for the baseline reliability study data. Cronbach's alpha was 0.89, indicating that the items correlate highly and reflect the same construct. Overall, the reliability results suggest that the CB&M can be used with confidence by trained PT raters in clinical and research settings in children and youth with ABI both for evaluation of status at a single point in time and for follow-up over time.

Construct Validity
Adults (Howe et al 2006, Inness et al 2011)
Construct validity was established over two studies. In a study of a convenience sample of 36 patients with traumatic brain injury (Howe et al 2006), mean CB&M scores were statistically different (p<0.03) across care settings (acute, inpatient, community) thus supporting the ability of the CB&M to differentiate between the patients along the continuum of care. In a study with a convenience sample of 27 adult patients with traumatic brain injury, associations were determined between the CB&M and other measures thought to have related theoretical constructs, including laboratory measures of postural control, gait and dynamic stability (Inness et al 2011). There were no significant correlations between the CB&M total scores and measures of centre-of-pressure amplitude or velocity. The lack of association between the CB&M and static measures of balance may suggest that underlying mechanisms of postural control may be different in these tasks and that measure of static balance may not effectively evaluate dynamic stability. In contrast, the relationship of CB&M total scores and most spatiotemporal measures of gait, measured on a computerized pressure-sensitive mat, were of a moderate to good or excellent magnitude (Colton 1974). For example, significant associations with the CB&M were demonstrated with walking velocity (r=0.77; p< 0.001), with step length (r=0.80; p< 0.001), with step width (r=-0.55; p=0.018) and step time (r=-0.55; p=0.003). Significant negative correlations of a moderate to good magnitude were also demonstrated between CB&M scores and measures of dynamic instability; for example, significant associations with the CB&M were demonstrated with step length variability (r=-0.66; p<0.001) and step time variability (r=-0.70; p<0.001).

Using a sample of community-dwelling patients with TBI, associations were also investigated between CB&M scores and measures that would reflect the individual's self-efficacy and participation in the community environment (Inness et al 2011). Significant correlations of a moderate to good magnitude between total CB&M total scores and the Activities-Specific Balance Confidence (ABC) scale were demonstrated in 18 patients (r=0.60; p =0.011) when one outlier (who had a low CB&M but high ABC score) was removed. Significant correlations of a moderate to good magnitude between the CB&M and CIQ were also revealed with a larger pooled data set of 47 community-dwelling patients with TBI (r=0.54; p<0.001)

Children and Youth
Construct validity work has not been done with the CB&M in paediatrics. Future evaluations should focus on comparison of CB&M scores with other measures of paediatric balance such as the Pediatric Berg Balance Scale (Franjoine et al., 2003), the motor subtest of the Bruininks Oseretsky Test of Motor Proficiency (Bruininks al., 2005) or the Hi-MAT (Williams et al., 2006) . Evaluations could also be done to determine association with the 6 minute walk test or with kinematic gait variables (step length, width, velocity) as has been done with the CB&M in adults. At the level of participation, the association between the CB&M and measures such as the Activity Scale for Kids (Young et al., 2000) or the Community Mobility Assessment (Moody et al., 2007) could be evaluated.

Sensitivity
Adults (Inness et al 2011)
A convenience sample of 26 adult patients with traumatic brain injury were evaluated with the CB&M and the Berg Balance Scale (BBS) by the treating therapist at admission to inpatient or outpatient rehabilitation services and again 10 weeks later, whichever was earlier (mean time between assessments was 7.6 SD 4.8 weeks). On admission, scores for this patient group were at or near the ceiling for the BBS (mean score= 53.6/56 SD 4.3) in contrast to the wide distribution of CB&M admission scores (mean score= 57.8/96 SD 23.3). Change scores from admission to discharge were 1.2 ± 3.1 for the BBS versus 13.0 ± 9.9 for the CB&M. Standardized response means for the BBS and CB&M were 0.40 and 1.3 respectively. These findings support the rationale for the use of the CB&M, with its more challenging items, to reveal and quantify deficits that persist even in the highly ambulatory individual.

In un-published work, the minimum detectable change (MDC) of the CB&M at the 90% confidence level (MDC90) has been calculated at 7.5 points. For more detail on the MDC, please see the Frequently Asked Questions (FAQ) section. Further study is required to establish the minimal clinically important difference.

Children and Youth (Wright et al 2010)
Within the reliability study by Wright et al. (2010), the CB&M's minimum detectable change at the 90%CI level (MDC90) was 12.7 points (note: all CB&M scores have been converted in this summary from percent score reported in the Wright et al., study to scores that are out of 96 so that they are on the same metric as the adult CB&M scores in this summary). While responsiveness to change was not evaluated as a part of the pediatric reliability study, the CB&M had been used previously by members of the reliability study team in a program evaluation context. Review of the scores of 17 youth with ABI showed a CB&M mean change score of 22.6 points [SD = 13.0] from admission (mean score = 62.5) to discharge (mean score = 85.1)(Wright et al, 2007). This magnitude of change was well beyond the MDC90 for the CB&M, suggesting that one can be confident that the observed change in scores in the program evaluation sample represented actual improvement in the children at follow-up. In comparison, the follow-up sample changed by only 6.8 points on the GMFM, and the follow-up mean score was within 3 points of the maximum score of 100% on the GMFM indicating lack of scope for detection of further change.

Distribution of Scores
Adult
The CB&M was developed for the higher-functioning individual with TBI. Despite the high levels of independence with functional mobility in the study samples, a wide range of CB&M total scores has been demonstrated across studies. For example, in a study of 32 patients with TBI, the majority of whom were walking independently without an aid and with maximal walking velocities within age-related healthy values (1.71 ±0.45 m/sec), the mean CB&M scores ranged from 62.1±18.3 out of a total of 96 (range 23-90) (Howe et al, 2006). The CB&M has not demonstrated a ceiling effect and as referenced below has also been able to differentiate between balance performance in healthy populations from decade to decade.

Children and Youth
Wright et al., (2010) also found a wide range of CB&M scores for the 32 participants in their reliability study. The mean CB&M scores varied from 64.7±17.5 out of a total of 96 (range 23.4-100). A ceiling effect was not evident. The inclusion criteria for the pediatric reliability study was that the child be able to stand on one leg independently (right or left) for at least 3 seconds and stand independently for 10 seconds (as measured within the Gross Motor Function Measure (Russell et al., 1988).

Healthy Age-Referenced Values (Zbarsky et al 2010)
The interpretation of scores of functional outcome measures can be enhanced by an understanding of the performance of individuals without disability across the life span. This may be especially true when measuring performance of patient populations that are at the high end of functional abilities. To establish age-related reference values for the CB&M, 121 healthy participants aged 20-79 completed the CB&M. Means, standard deviations and 95% confidence intervals were calculated and are referenced in the table below. The confidence intervals listed can be used to determine if a patient falls within or below the healthy range for his/her age group. This is helpful in identifying the presence and degree of balance impairment. The values for the 70+ age group should be used with caution due to the small sample size.

Table 1. Community Balance and Mobility Scale (CB&M): Healthy Age-Related Reference Values

Age Group	N	Mean*	SD	95% CI
20-29	24	88.71	3.53	87.22-90.20
30-39	27	86.33	5.78	84.05-88.62
40-49	23	84.35	4.03	82.61-86.10
50-59**	26	77.43	6.55	74.99-79.88
60-69**	17	64.94	8.22	60.72-69.17
70-79**	4	49.75	6.95	38.70-60.80

*CB&M scored out of 96
**Significant difference from the group in the previous decade