The Balance Recovery Confidence (BRC) Scale

ABSTRACT Background Falls efficacy posits an understanding of the perceived ability to prevent and manage falls. There have been no validated self-reported instruments to measure the perceived ability to recover balance in response to destabilizing perturbations. Purpose To develop a scale of balance recovery confidence. Methods Stage one had candidate items generated by 12 community-dwelling adults aged 65 and older using the nominal group technique. Stage two had the scale’s name, instructions, response options, recall period and the items validated for appropriateness with 28 healthcare professionals and 10 older adults using an e-Delphi technique. Stage three had the scale’s psychometric properties evaluated with 84 older adults who had completed self-reported and performance measures. Factor analysis was applied to confirm unidimensionality. The internal structure, reliability and validity of the scale were evaluated using the classical test theory and Rasch measurement theory. Results The 19-item scale was developed and validated with experts’ consensus. The scale is unidimensional with excellent internal structure (Cronbach’s α = 0.975) and test-retest reliability with Intraclass Correlation Coefficient (ICC3,1) = 0.944. Construct validity of the scale was supported by its relationships with the other measures (Activities-specific Balance Confidence scale, Falls Efficacy Scale-International, Late-Life Function and Disability International-Function, handgrip strength dynamometry, 30-second chair stand test, and mini-BESTest). Conclusion The balance recovery confidence scale is a distinct instrument that measures perceived reactive balance recovery. The scale has good psychometric properties and can be used to complement other measurement instruments to help older adults cope with challenges to balance.


Introduction
Falls remain a significant threat to health, independence and quality of life for older adults (Gardiner et al., 2017).The World Health Organization (2021) has identified falls as the second leading cause of death due to unintentional injury.The burden of falls is substantial, ranking it as the 18th leading cause of the age-standardized rate of disability-adjusted life years (James et al., 2020).Falls efficacy rooted in Bandura's self-efficacy theory (Bandura, 1977) can influence individuals' agency to cope with and overcome various falls-related situational conditions (Soh et al., 2021e).Self-efficacy posits an understanding of the different perceived abilities to execute the courses of action required to produce given attainments (Bandura, 2018).
To understand older adults' self-efficacy to tackle a fall threat, obtaining information about the perceived ability to cope with the different demands of falls and falling is imperative (Soh et al., 2021e).A set of fallsrelated self-efficacy expectations namely: balance confidence, balance recovery confidence, safe landing confidence, and post-fall recovery confidence could be characterized through the perceived capabilities needed at the pre-fall, near-fall, fall-landing and completed-fall stages (Soh et al., 2021e).This perspective concurs with Bandura's view that perceived self-efficacy is a differentiated set of self-beliefs linked to distinct realms of functioning (Bandura, 2018).Focusing selfefficacy on a particular domain risks limiting input to what self-efficacy could contribute to the issue of concern (Bandura, 2012).For example, interpreting falls efficacy using measures of balance confidence (i.e. the perceived ability to perform different activities without losing balance) risks overlooking other falls-related self-efficacies such as the perceived ability to arrest a fall, to fall safely, or to get up after a fall.Applying appropriate perceived self-efficacy instruments tailored to a particular domain could potentially serve as useful explanatory and predictive instruments for a more comprehensive understanding of the agency of individuals to cope with falls.
Before commencing this work to develop a scale of balance recovery confidence, two previous studies were performed.The first was a systematic literature review conducted to evaluate the methodological quality of the development, content validity and structural validity of existing falls efficacy-related instruments using the COSMIN method (Soh et al., 2021c).The second was a feasibility study to investigate whether communitydwelling older adults were able to relate to the balance recovery concepts (Soh et al., 2021d).These studies were implemented based on Terwee et al. (2018) recommendation of preliminary work before developing a healthrelated measurement instrument.The most pertinent findings were that many existing falls efficacy instruments lacked sufficient quality or consistency of evidence on content development and validation due to the lack of studies reporting the involvement of the target population (Soh et al., 2021c).Therefore, representatives from the target population need to be involved early in developing a self-reported measurement scale.The review also identified that no existing falls efficacy instruments were purposefully constructed to measure balance recovery confidence.This justified the development of a new instrument for measuring balance recovery confidence.The feasibility study showed that community-dwelling older adults could distinguish nuances of balance concepts, that is, the difference between the ability to recover balance following a loss of balance and the ability to perform activities without losing balance (Soh et al., 2021d).
Many falls efficacy instruments have been designed to measure balance confidence (Soh et al., 2021c).However, balance recovery confidence is conceptually different from balance confidence.Measures of balance confidence, for example, the Activities-specific Balance Confidence scale (Powell and Myers, 1995) aim to measure the perceived ability to perform various activities without losing balance or becoming unsteady.In contrast, balance recovery confidence refers to the perceived reactive ability to recover balance in response to sudden, unpredictable destabilizing perturbations such as a slip, a trip, or a loss of balance from volitional movement (Maki, McIlroy, and Fernie, 2003).Balance recovery confidence refers to the perceived reactive ability to execute change-in-support strategies, such as reach-tograsp or compensatory stepping and overcoming the constraints imposed by the environment such as the availability of handrails or the type of platforms to step on (Maki and McIlroy, 2006).The belief in the capacity to arrest a fall depends on the task the person is performing and their environment (Komisar, McIlroy, and Duncan, 2019) as well as their age-related physiological changes (Alcock, O'Brien, and Vanicek, 2018).Little is known about balance recovery confidence.Do individuals perceive their reactive ability to arrest a fall as similar to their ability to perform activities without losing balance?What is the relationship between balance recovery confidence, balance confidence, and fear of falling?Understanding the role of balance recovery confidence would gain significant headway with the availability of a suitable measurement instrument.
An instrument for balance recovery confidence provides several benefits.First, balance recovery confidence is distinguished from balance confidence based on the theoretical underpinnings of balance control mechanisms (Maki and McIlroy, 1997).Second, a reliable and valid instrument that measures balance recovery confidence could facilitate discussion on recovery performance in near-fall scenarios.Third, balance recovery confidence can supplement other aspects of fall-related self-efficacy to address falls agency in individuals.This study aims to develop and evaluate the psychometric properties of the first self-reported measurement instrument that measures balance recovery confidence: The balance recovery confidence (BRC) scale.

Methods
A three-stage approach was applied to develop the BRC scale (Figure 1).Inclusion criteria for communitydwelling participants were: 1) 65 years and older; and 2) living independently in the community with or without using a walking aid.Exclusion criteria included: 1) presence of clinical observable severe cognitive impairment; 2) need any physical assistance from another person to walk within the home; and 3) unable to provide written consent.Ethical approvals were obtained from ethics committees at Queen Margaret University (QMU) (REF 0197/REF0220) and Singapore Institute of Technology (SIT) (2019129/2020098).Studies were recorded in clinicaltrial.govregistry (NCT04087551/ NCT04577365).All participants provided written informed consent to participate.

Stage one -item generation
An extensive list of relevant items was constructed with representatives of the target population (n = 12).These individuals were considered experts in the level of symptoms, functioning and perceived health and can provide insights regarding the suitability of items to measure the construct of interest (De Vet, Terwee, Mokkink, and Knol, 2011).The participants were recruited using convenience sampling of those who participated in the feasibility study (Soh et al., 2021c).

Design
Two focus group sessions were conducted using the nominal group technique (McMillan, King, and Tully, 2016).Each session involved five phases: Introduction, silent idea generation, round-robin recording of ideas, clarification, ranking, and debriefing.All participants were given two questions: 'What common and everyday activities that older people participate in (at home or outside the home) could result in a near-fall?' and 'How can older people avoid a fall in these near-fall events?' Two facilitators (SS, TT) used a facilitator guide created by the study team to standardize both sessions.Items from the other falls efficacy measures were introduced at the clarification stage by the facilitators if they were not mentioned during the round-robin stage.This was considered acceptable because participants would subsequently rank what they deemed important to measure the construct (McMillan, King, and Tully, 2016).Audio recordings were taken with permission to address any uncertainties arising during analysis.

Analysis
Three researchers (SS, TT, TX) reviewed the items and identified those that fit the performance domain of the BRC scale's conceptual framework (Supplementary Figure 1).The framework was developed based on Bandura's self-efficacy model (Reeve, 2009).Items were then depicted in illustrations to present a detailed description of the precise context.

Stage two -content validation
Content validity is the degree to which the content of an instrument is an adequate reflection of the construct to be measured.In this stage, the preliminary content for the BRC scale was reviewed over two rounds using an e-Delphi Technique (McMillan, King, and Tully, 2016) to meet an acceptable level of content validity through a convergence of opinions and facilitated consensus among experts.The experts comprised healthcare Stage two -content validation: An e-Delphi technique was applied to validate the list of items with a panel of experts comprising healthcare professionals (n = 28) and a new group of community-dwelling older adults aged 65 years and older (n = 10) to meet the consensus of appropriateness using the RAND/UCLA appropriateness scale.Qualitative and quantitative analysis of the content was conducted including name of the instrument, instructions, response options, recall period, and the list of items.The comprehensiveness and face validity of the scale was evaluated.
Stage one -item generation: The Nominal Group Technique was applied to generate a list of items with community-dwelling older adults aged 65 years and older (n = 12).This method was applied to identify a comprehensive list of relevant items for the BRC scale by the target population.The list of items were then illustrated using clear line drawings.
Stage three -Evaluation of psychometric properties: Field testing was done with communitydwelling older adults aged 65 years and older (n = 84).Participants had to complete four selfreported measures: the Activities-specific Balance Confidence scale, the BRC scale, the Falls Efficacy Scale -International, the Late Life Function and Disability Instrument -Function.Participants also had to complete three physical performance tests: handgrip strength dynamometer test, 30-second chair stand test and the Mini-BESTest.A retest of the BRC scale was completed with the Global Perceived Effect scale in a second session arranged between one and two weeks later.Factor analysis demonstrated the unidimensionality.Classical test theory and Rasch measurement theory were applied to evaluate the new scale's internal structure, reliability and validity.professionals (n = 28) and a new group of communitydwelling older adults (n = 10).There is no clear consensus on the ideal size of such a panel (McMillan, King, and Tully, 2016).However having input from different groups of participants with a variety of expertise would help improve the methodological quality of the scale development (Terwee et al., 2018).Healthcare experts were invited if they had at least 3-year working experience in clinical geriatric work or related research.

Design
Experts accessed an online survey through an e-mail link.They then evaluated the appropriateness of the given content to assess balance recovery confidence using the RAND/UCLA appropriateness scale (Fitch et al., 2001) a nine-point Likert scale ranging from 1 (inappropriate) to 9 (appropriate).Appropriateness was defined as the importance, relevance and clarity of measuring balance recovery confidence in communitydwelling older adults.Comments were given in a freetext box.Experts rated the instrument's name, instructions, response options, recall period, the given items' descriptors and illustrations.Two weeks were allocated to complete the survey, with an e-mail reminder sent to those who had not responded after one week.Two additional questions were given in the second round to determine the comprehensiveness and face validity of the content.

Analysis
Three researchers (SS, JL, CW) analyzed the quantitative and qualitative data based on set criteria (Supplementary Table 1 and Supplementary Table 2) using the RAND/ UCLA user's manual (Fitch et al., 2001).

Stage three -evaluation of psychometric properties
Obtaining insights into the structure of the data generated by a newly developed scale and assessing reliability and validity are critical aspects of the scale's development (De Vet, Terwee, Mokkink, and Knol, 2011).Key psychometric properties were prioritized using traditional and modern measurement models (i.e.classical test theory and Rasch measurement theory).

Design
A sample of community-dwelling older adults (n = 84) was recruited using convenience sampling from Singapore Institute of Technology, St Luke's Hospital and word-of-mouth recommendations.Participants had to complete four self-reported questionnaires and three physical performance tests: 1) Late-Life Function and Disability Instrument -Function (LLFDI-F) (Haley et al., 2002); 2) Falls Efficacy Scale -International (FES-I) (Yardley et al., 2005); 3) Activities-specific Balance Confidence (ABC) scale (Powell and Myers, 1995); 4) BRC scale; 5) handgrip strength dynamometry (Durkin, 2014); 6) 30-second chair stand test (Jones, Rikli, and Beam, 1999); and 7) the mini-BESTest (Franchignoni et al., 2010).A second session was scheduled a week later for participants to complete the Global Perceived Effect (GPE) scale (Kamper et al., 2010) and the BRC scale.The GPE was used as an exterior control criterion for the test-retest reliability of the BRC scale.
Self-reported measures LLFDI-F (Haley et al., 2002) was used to evaluate a person's perceived difficulty in performing activities of daily living.There were 32 items with response options of 'none,' 'a little,' 'some,' 'quite a lot,' and 'cannot do.'An additional eight items were given to those who used canes or walkers.The raw scores were transformed into scaled scores (0-100).The closer the score to 100, the higher the ability to carry out the task.
The ABC scale (Powell and Myers, 1995) was used to assess individuals' confidence in performing several progressively more challenging balance and mobility tasks without losing balance.There were 16 questions, with scoring options ranging from 0% (no confidence) to 100% (complete confidence).The mean score was recorded.A higher score depicted a greater degree of confidence in performing activities steadily.
FES-I (Yardley et al., 2005) measured the individual's concerns about falling with basic and more demanding activities.Sixteen questions were answered with a fourgrade scale (1-4) of 'not at all concerned,' 'somewhat concerned,' 'fairly concerned' and 'very concerned.'The total score, which ranged from 16 to 64, was recorded.A higher score reflected a greater level of concern about falling.
The BRC scale assessed an individual's perceived ability to recover balance across several scenarios depicting different perturbations such as a slip, a trip, or volitional movements.Nineteen items were rated using an 11-point scale ranging from 0 (cannot do at all) to 10 (highly certain can do).The total possible score was 190, and the mean score was recorded.A higher score denoted a higher certainty of arresting a fall.

Performance measures
Handgrip strength dynamometry.A hydraulic hand dynamometer (Lafayette instrument) was used to determine isometric maximum handgrip strength.Handheld dynamometry has good test-retest reliability and concurrent validity with upper limb strength in older adults (Legg et al., 2020).Participants had to squeeze the dynamometer, and each hand's highest maximum grip force (kg) was recorded.The standardized protocol recommended by the NIHR Southampton Biomedical Research Center guided the administration (Durkin, 2014).

30-second chair stand test.
The 30-second chair stand test was used to test functional lower extremity strength (Jones, Rikli, and Beam, 1999).This test is a reliable and valid indicator of lower body strength in communitydwelling older adults (Jones, Rikli, and Beam, 1999).Participants had to sit and stand from a chair as many times as possible within 30 seconds.The test was administered by adopting the protocol recommended by Stopping Elderly Accidents, Deaths and Injuries (STEADI) (Centers for Disease Control and Prevention, 2017).The total number of sit-stands achieved was recorded.
Mini-BESTest.The Mini-BESTest 14-item clinical test was used to assess four postural control systems: 1) 'anticipatory postural adjustments' (sit to stand, rise to toes, stand on one leg); 2) 'reactive postural responses' (stepping in four different directions); 3) 'sensory orientation' (stance -eyes open; foam surface -eyes closed; incline -eyes closed); and 4) 'dynamic gait' (gait during change of speed, head turns, pivot turns, obstacles; cognitive 'up and go' with dual-task) (Franchignoni et al., 2010).The total possible score was 28.A higher score depicted a greater level of functional balance.The mini-BESTest is the most comprehensive balance measure for practical applications (Di Carlo et al., 2016).

Sample size
The sample size needed to validate a self-reported measurement instrument has been recommended to be between four and 10 participants per item, or a minimum of 50 participants (De Vet, Terwee, Mokkink, and Knol, 2011).The Rasch measurement theory was used to evaluate the scale's fit with polytomous data and mean square statistics would be relatively independent of the sample size.The minimum number of participants for the first validation study of the 19item BRC scale was set at 76.

Analysis
The R version 4.0.4(https://www.R-project.org) and Winsteps version 4.8.2.0 (https://www.winsteps.com)were used for data analysis.The classical test theory and Rasch measurement theory provided a robust initial assessment of the BRC scale's measurement characteristics at both item and scale levels (Petrillo, Cano, McLeod, and Coon, 2015).Classical test theory assumes that every observed score is a function of an individual's true score plus a random error.Rasch measurement theory works on a stochastic model, ordering persons according to their ability and the items according to their difficulty level.
Dimensionality.Dimensionality refers to establishing the number of meaningful dimensions that can be distinguished in a construct (De Vet, Terwee, Mokkink, and Knol, 2011).The content of the BRC scale was designed to measure the perceived performance to recover balance, and the scale is expected to measure a single dimension.Therefore, the scale's unidimensionality was assessed by confirmatory factor analysis (Prinsen et al., 2018).
Internal structure.The acceptability of the BRC scale was established by the extent to which the scale items were scored, the percentage of missing data for each item and the percentage of people for whom a PROM score can be computed (De Vet, Terwee, Mokkink, and Knol, 2011).An amount of missing item-level data of less than 5% was considered acceptable.The score distribution, including skewness, was presented through item-level response descriptive statistics.
Scale item targeting refers to the extent of congruence between the range of the scale options used and the targeted sample's expression of the latent construct (De Vet, Terwee, Mokkink, and Knol, 2011).Targeting the scale's item was determined by evaluating the extent to which the range of the construct measured by the scale matched the range of that construct in the sample population (e.g. if the scores obtained had spanned the entire range of options).The floor effect (i.e.proportion of the sample at the minimum score) and ceiling effect (i.e.proportion of the sample at the maximum score) should be low defined as < 15% of the sample.The person-item threshold distribution map was assessed to present the relative distribution of items matched to the respondents' latent trait range.
The goodness of fit was measured to reveal any major quality-control problems of the BRC scale.The mean square standardized residual (MNSQ) demonstrated if the items were able to provide productive measurement.The MSSQ score should fall within the 0.5-1.5 range.A score less than 0.5 indicates overfit (i.e. the items would be too predictable relative to the Rasch model) while a score greater than 1.5 indicates too much noise (i.e.randomness).

Construct validity.
Construct validity is defined as the degree to which the scores of a measurement instrument are consistent with hypotheses on its relationship with the scores of other instruments (De Vet, Terwee, Mokkink, and Knol, 2011).The list of hypotheses was made a priori (Table 1) (Soh et al., 2021b).The correlation coefficients between the BRC scale and the other outcome measures were established using Spearman's rho.
Reliability.Reliability is the degree to which the measurement is free from measurement error (De Vet, Terwee, Mokkink, and Knol, 2011).Internal consistency was calculated using Cronbach's coefficient alpha, with an alpha value ≥ 0.70 demonstrating adequate evidence of internal consistency.The testretest reliability of the BRC scale was determined using the intraclass correlation coefficient (ICC 3,1 ) with a 95% confidence interval between the scores obtained over the two sessions.The person separation reliability, an estimate of how well one can differentiate persons on the measured variable was determined by the person separation index (PSI).A PSI > 0.7 was considered an adequate measure of reliability.

Study samples
The demographic characteristics of the participants for the three stages are presented in Table 2.A total of 106 community-dwelling older adults aged between 65 and 84 participated in the study.There was a good gender representation, and most participants had at least secondary education.Twenty-eight healthcare professionals were mainly physiotherapists, occupational therapists and nurses, most of whom were from Singapore and had more than six years of experience.

Stage one -item generation
There was no participant withdrawal in this stage.Ninety-nine items were generated during the 'roundrobin' phase.After the 'clarification' and 'ranking' phases, 56 items were presented.Thirty-two were identified to fit the performance domain of the conceptual self-efficacy framework (Supplementary Figure 1).

Stage two -content validation
Response rates for community-dwelling older adults in both rounds were 100%; for healthcare professionals, they were 70% (first-round) and 79% (second-round).The instrument's name, instructions, response options, recall period and 19 items achieved consensus for appropriateness to assess balance recovery confidence without disagreement.The list was comprehensive, and the scale had face validity (Supplementary Table 3).Figure 2 shows four exemplar items of the BRC scale.The scale is made accessible online at https://doi.org/10.6084/m9.figshare.20279763.v1.

Stage three -evaluation of psychometric properties
The mean age of 84 participants was 71.1 years with a standard deviation (SD) of 4.5, and 59.5% were women.Five participants' data were not used for the test-retest reliability analysis; three participants did not complete the retest of the BRC scale because of Covid-19 restrictions, and two reported a fall before the retest assessment.

Internal structure and construct validity
The distribution of responses was broad across the score categories (Table 3).All response options (0-Table 1.The list of a priori hypotheses used to evaluate the construct validity of the balance recovery confidence (BRC) scale (Soh et al., 2021b).

S/N Hypotheses 1
A moderate positive correlation (0.30-0.59) was expected between the BRC scale and the ABC scale.While balance confidence and balance recovery confidence are unique, they are related constructs of balance control.The ABC scale focuses on the perceived ability to stay balanced during activity performance, and the BRC scale focuses on perceived balance recovery performance.2 A moderate negative correlation (0.30-0.59) was expected between the BRC scale and the FES-I.Balance recovery confidence and fear of falling, while conceptually different, are related, given that people with low balance confidence have a high fear of falling.

3
A moderate positive correlation (0.30-0.59) was expected between the BRC scale and the LLFDI-F.Both instruments measure the perceived physical performance of an individual.The LLFDI-F focuses on the perceived ability to do specific activities, while the BRC scale focuses on perceived balance recovery performance.4 A strong positive correlation (≥ 0.60) was expected between the BRC scale and the HSD test.The HSD test measures handgrip strength.Handgrip strength is necessary for reach-to-grasp maneuvers to recover balance.5 A strong positive correlation (≥ 0.60) was expected between the BRC scale and the CST.CST measures lower limb strength.Lower limb strength is necessary for compensatory stepping maneuvers to recover balance.6 A strong positive correlation (≥ 0.60) was expected between the BRC scale and the MBT.MBT measures the anticipatory and reactive ability for balance and balance recovery.
BRC: balance recovery confidence, ABC: Activities-specific Balance Confidence, FES-I: Falls Efficacy Scale-International, LLFDI-F: Late-Life Function and Disability Instrument-Function, HSD: Handgrip strength dynamometer, CST: 30-second chair stand test, MBT: Mini-BESTest 10) were used in all items but three (Items 3, 10 and 11), using response options between 2 and 10.There were no missing scores.All participants completed the BRC scale.The distribution of the scores had no significant departures from normality (W = 0.97, p = .05),skewness of −0.32, kurtosis of −0.63, and standard error of measurement of 0.22.The BRC scale had a single factor loading for each item (0.727 to 0.921) (Supplementary Table 4).The standardized root mean square residual was 0.057, meeting the criteria of < 0.08.Comparative fit index and Tucker-Lewis Index scores of 0.792 and 0.767 suggested that the content could be further improved in terms of its structural validity for this sampled population.The uniqueness of the items was low (0.15 to 0.47), which implied that the construct of balance recovery confidence could explain the variation in each item.Moderate correlations were found between the BRC scale and the different outcome measures: ABC scale (0.54), LLFDI (0.41), Mini BESTest (0.51), and FES-I (−0.57) (Table 4).Correlations between the BRC scale, handgrip strength dynamometer and 30second chair stand test were weak.The reactive postural control section of Mini-BESTest (section 4-6) showed a slightly stronger congruence to the BRC scale (0.62) than to the ABC scale (0.57).

Targeting of the items
No floor effect was identified through the score distribution table (Table 3).Nine items (Items 3,8,9,10,11,14,17,18,and 19) had a ceiling effect which indicated that these items were easy for the sampled population.Most participants were confident they could recover their balance in those nine scenarios.Based on the Wright distribution map (Supplementary Figure 2), the BRC scale item difficulty level accommodated a wide range of trait abilities in the sample population, from low to high levels of balance recovery confidence.The mean item difficulty level was well structured without being extremely difficult or extremely easy.The modeling fit statistics showed person ability estimate means of 0.86 logits (SD 1.46).This implied that this sample of older adults' confidence in their balance recovery ability was slightly high.The standard deviation of 1.46 logits for the person estimate indicated an adequate spread of person measures.The mean of the infit and outfit mean squares at 1.03 and 1.01, respectively, were harmonious with the Rasch-modeled expectations of 1.This indicated the data fitted well to the probabilistic Rasch model specification.The standardized fit Z values were around zero (infit Z = −0.5;outfit Z = −0.5).The variation of modeled fit scores for persons (infit Z SD = 2.3 and outfit Z SD = 2.3) suggested that most person ability estimates were transformed within the fit statistics.Five items (Items 2,12,14,11,15) were out of the conventionally accepted range of -2 to +2.

Reliability
The BRC scale had excellent internal consistency (α = 0.975).The reliability of the person ability and the item difficulty estimates were high, indicated by 0.93 (person separation = 3.76) and 0.94 (item

Discussion
This study presents the development of the balance recovery confidence (BRC) scale.The BRC scale is designed to measure balance recovery confidence, distinguishing the scale from others measuring balance confidence.The content obtained the consensus to be appropriate for measuring balance recovery confidence.
The key quality criteria of the BRC scale's psychometric properties have been adequately demonstrated to encourage its application in research and clinical practice (De Vet, Terwee, Mokkink, and Knol, 2011).
Representatives from the target population were involved from the beginning of the instrument's development.The intent was to ensure that older adults would suitably use the BRC scale to express their confidence level to recover their balance in the near-fall scenarios.The ideas generated for items to assess balance recovery confidence were noticeably different from those used in the scales of balance confidence (i.e. to assess the confidence to perform different activities without losing balance or becoming unsteady).The clarity of the content constructed for measuring balance recovery confidence was affirmed through content validation.Neither the healthcare professionals nor the community-dwelling older adults expressed significant difficulties in reviewing the content.The diversity of views and the meaningful perspectives offered by the The score distribution reflected that more than 15% of the participants were in the extreme high-end category scored for the item.This indicated a ceiling effect for the items.different groups of experts has been reported in a separate article (Soh et al., 2021a).Recognizing that the self-belief to arrest falls differs from the self-belief in performing activities without losing balance affirms that the roles of different perceived falls-related self-efficacy need further exploration.Examination of the scale structure supported that the BRC scale is unidimensional.The scale showed good internal consistency and excellent test-retest stability.There was the consistency of a priori hypothesized relationships between the BRC scale and three other questionnaires (i.e.ABC scale, FES-I, and LLFDI-F).This supported the theoretical understanding of the BRC scale.However, the inconsistency found in the relationships between the BRC scale and the three performance measures (i.e.handgrip strength dynamometer, 30second chair stand, and Mini BESTest) needed further explanations.First, self-reported and performance measures likely measure different aspects of functioning (Silva, Queiros, Sa-Couto, and Rocha, 2015).Selfreported measures capture the perception of the capability to act, whereas performance measures reflect the actual ability of an individual to perform the action.Previous studies have reported the importance of using different measurement methods to assess functional ability in older adults (Nielsen et al., 2016).Making perceived and actual physical abilities explicit on falls agency is essential as individuals with fatalistic fallsrelated beliefs and behavior would benefit from using different measures (Stevens, Sleet, and Rubenstein, 2018).Another possible explanation could be that simple motor skills cannot adequately reflect perceived reactive balance recovery ability; single-task performance measures may not fully reflect the complexity of realworld situations.
Some key issues that arose during the study deserve to be mentioned.First, the theoretical self-efficacy concept underpinning the BRC scale assumes that individuals consider several factors to cope in a given situation such as: avoiding, adapting, persisting, or necessarily performing in the situation.During field testing, some older adults mentioned that specific scenarios would not happen to them or that they would take appropriate measures to avoid them.All items in the BRC scale are encouraged to be completed because a person's selfreporting of reactive recovery abilities would indicate an understanding of personal capabilities to arrest falls.The score would inform clinicians about their falls prevention behaviors.Second, the BRC scale does not dictate the choice of balance recovery maneuvers used by individuals to achieve a successful recovery performance.A wide range of balance recovery strategies was considered alongside indoor and outdoor falls-initiating events.Different strategies should be explored between clinicians and their patients.It was not possible to detail an infinite number of near-fall scenarios.
This study has some limitations.First, the BRC scale was developed by samples of high-functioning older adults living independently in the community.Most scenarios were challenging and may not be suitable for frail older adults.Second, the sample size of 84 is generally considered small for a rigorous analysis using Rasch measurement theory, and a larger sample size is needed for a more robust analysis.Third, the BRC scale was not studied for responsiveness or its value as a predictive instrument.Studies on these psychometric properties would be needed.The study did not investigate the circumstances and consequences of falls of older adults who fell in the past year.Therefore, previous fall experiences to their balance recovery confidence level could not be examined.Future studies could explore the impact of past fall experiences on balance recovery confidence.

Conclusion
The BRC scale is the first instrument reported in the falls-related literature to measure balance recovery confidence.The steps to develop and validate the content were following best practices.Initial psychometric testing showed encouraging reliability and validity results.Further studies would be needed to demonstrate the BRC scale's potential and psychometric properties fully.

Figure 1 .
Figure 1.Stages to develop the balance recovery confidence (BRC) scale.

Figure 2 .
Figure 2. Four examples of items from the balance recovery confidence (BRC) scale.

Table 2 .
Demographic characteristics of the participants across the three stages of study.

Table 3 .
Distribution of the sample population (n = 84) scoring across the balance recovery confidence (BRC) scale response categories.

Table 4 .
The correlation matrix presents the correlations between the different measurement instruments.