Cross-cultural adaptation and psychometric properties of the Brazilian-Portuguese version of the Falls Behavioral (FaB-Brazil) Scale

Abstract Purpose To develop and cross-culturally adapt a Brazilian-Portuguese version of the Falls Behavioral Scale (FaB-Brazil) and to verify its psychometric properties. Material and Methods The translation and cross-cultural adaptation process of the scale followed standard guidelines. The FaB-Brazil scale was applied to 93 community-dwelling older people. Cronbach’s alpha was calculated to evaluate internal consistency and the intraclass correlation coefficient (ICC) to evaluate interrater and test–retest reliability. The standard error of measurement (SEM), minimal detectable change (MDC), ceiling and floor effects, convergent and discriminative validity were evaluated. A significance level of .05 was set for statistical analyses. Results Internal consistency was moderate (α = 0.73). An excellent inter-rater (ICC = 0.93; p < 0.001) and a good test–retest (ICC = 0.79; p < 0.001) reliability were found. The SEM was 0.27 and MDC was 0.53. Neither ceiling nor floor effects were found. Convergent validity was established by the positive correlations between the FaB-Brazil scale, age, and functional mobility, and by the negative correlations between the FaB-Brazil scale and balance confidence, community mobility and EuroQol-5D (p < 0.05). No significant differences were found between males and females and between non-fallers and fallers. Conclusions Our results offer evidence for the reliability and validity of the FaB-Brazil scale for community-dwelling older people. Implications for Rehabilitation Fall-related behaviors should be part of the fall risk assessment of community-dwelling older people. The Brazilian-Portuguese version of the Falls Behavioral Scale (FaB-Brazil) is reliable and valid for assessing fall-related behaviors in community-dwelling older people. The FaB-Brazil scale may be used to raise awareness about potential fall hazards and to guide fall prevention programs.


Introduction
Falls are common among older people and a major public health problem that is related to increased health care costs and risk of disability, nursing home admissions, hospitalization, and mortality [1,2].Around 30% of people over 65 years fall each year, with many falling recurrently [1].Several contributing risk factors are associated with falls, including intrinsic (e.g., history of fall, age, sex, medications, muscle weakness, impaired balance and mobility, and fear of falling) [1,2], extrinsic (e.g., slippery floors, uneven surfaces, and poor lighting) [1,2], and behavioral (e.g., reduced attention when negotiating the environment, moving about and doing things at a speed beyond the person's capabilities, not scanning ahead when walking, being distracted, overexertion, and not noticing environmental hazards) [3].
In this context of a dynamic relationship between the environment and behavior, Clemson et al. developed the Falls Behavioral (FaB) Scale, an assessment tool to evaluate everyday behaviors and actions that increase the risk or offer protection from falling in older people [4,5].It has been shown to be reliable and valid in this population and is comprised of 30 items which evaluate fall prevention behaviors in 10 dimensions as follows: cognitive adaptations, protective mobility, avoidance, awareness, pace, practical strategies, displacing activities, being observant, changes in level, and getting to the phone [4].
Since fall prevention programs require a multifactorial approach [6], it is important not only to assess intrinsic and extrinsic risk factors, but also behavioral ones, specially fall-related risk-taking behaviors which are potentially modifiable.The FaB scale has been translated into Turkish [7], French Canadian [8], and European Spanish [9], with all versions showing good psychometric properties.However, the FaB scale is not available in Brazilian-Portuguese. Therefore, the aims of this study were to develop and cross-culturally adapt a Brazilian-Portuguese version of the FaB scale (FaB-Brazil) and to verify its psychometric properties.

Materials and methods
This cross-sectional study was carried out in two phases.In phase 1, the translation and cross-cultural adaptation process of the FaB scale was performed with the permission of Dr Lindy Clemson, first author of the scale [4].In phase 2, the analysis of measurement properties was performed by applying the FaB-Brazil scale to older people.The study was approved by the research ethics committee of the Roberto Santos General Hospital under the study number CAAE: 84229318.7.0000.5028.All participants provided written informed consent prior to assessment.

Translation and cross-cultural adaptation process
The translation and cross-cultural adaptation process of the FaB scale followed standard guidelines [10].The original FaB scale [4,11] was translated to Brazilian-Portuguese independently by three bilingual translators whose mother tongue was Brazilian-Portuguese.The translators had different background: a physiotherapist and a neurologist who were aware of the concepts being examined in the scale, and a naive translator who was not aware of the concepts being examined and had no medical background.All translations were discussed by the translators and a synthesis of these translations was produced to reach a consensus version.Next, two back translations of the consensus version were produced by two professional translators whose mother language was English.They were not aware of the concepts explored and had no medical background.
All versions of the FaB (the original version, three Brazilian-Portuguese versions, the Brazilian-Portuguese consensus version, and two backward translations English versions) were examined by an expert committee comprised of health professionals from different regions of Brazil, including five physiotherapists (including one with methodological background and one working in the field of geriatrics with expertise in the area of falls), one psychologist working in the field of geriatrics and one neurologist with epidemiological background and expertise in the area of falls.Members of the committee were asked to examine the equivalence between the English and Brazilian-Portuguese version in the following areas: semantic equivalence (meaning of the words, multiple meanings to a given item, and grammatical difficulties in the translation), idiomatic equivalence (meaning of colloquialisms and terms), experiential equivalence (replacement of tasks or situations that are not experienced in the Brazilian culture) and conceptual equivalence (conceptual meaning of the words between cultures) [10].Consensus should be reached on all items and, if necessary, an item should be reworded in order to reach consensus on discrepancies.The prefinal version was thus reached and then tested in fifteen participants from the target setting.

Participants
Participants were recruited from the Universidade Aberta � a Terceira Idade (UATI)/Bahia State University (UNEB), Salvador, Bahia, Brazil between November 2018 and December 2018.Eligibility criteria included people aged 60 years and older, able to walk without assistance of another person, with or without an assistive device.Participants were excluded if they had cognitive impairment based on Mini-Mental State Examination (MMSE) scores with cutoff points adjusted to level of education (no education, 13; <8 years of education, 18; �8 years of education, 26) [12], neurological conditions, severe visual disturbance or hearing impairment, vestibular dysfunction, or comorbidities that would affect locomotion or balance.

Assessment
Demographic and clinical data, including age, sex, marital status, living status, years of education, history of falls (none vs any) and frequency of falls (none; one; two or more) in the previous 12 months, use of an assistive device and the extent of community mobility (number of times left home in the past monthnever; once or twice in past month; once a week; two or three times a week; four or more times a week) were recorded.
Participants were assessed with the FaB-Brazil scale (Supplementary Appendix).In line with the original scale [4,11], the questionnaire includes 30 items describing everyday behaviors and actions related to falling and participants are required to indicate how often they perform each activity.Scores on each item range from 1 (never) to 4 (always) with an option to answer "does not apply" on 14 items (4, 5, 6, 9, 10, 13, 14, 18, 19, 25, 26, 28, 29, and 30).Higher scores reflect more protective behaviors and five items (7, 8, 9, 10, and 19) need to be recoded to reflect this [4,5,11].Item 23 has been previously reworded (I go out on windy days to I avoid going out on windy, icy or wet days) [5] and therefore there is no need to be recoded.The total score for each individual is calculated as a mean of all items, excluding any of the specific 14 items answered as "does not apply." A mean score was calculated for each item, the total score and each dimension (i.e., cognitive adaptations (7 items; 6, 12,13,21,24,25,26) [11] was partially translated and adapted and is available in the Supplementary Appendix.
Participants were also assessed with the modified Barthel Index to evaluate performance in ten activities of daily living, such as bathing, dressing, and walking on a level surface [13].Scores on each item range from 1 (unable) to 5 (independent) and total scores of 10 indicate total dependence, 11-30 indicate severe dependence, 31-45 indicate moderate dependence, 46-49 indicate slight dependence, and 50 indicate independence [13].The International Physical Activity Questionnaire (IPAQ) short form was used as a self-report measure of physical activity level and the types of activities assessed were walking, moderate-intensity activities and vigorous-intensity activities [14,15].Physical activity was categorized as high, moderate, or low and results are also reported in metabolic equivalent minutes/week [15].
The Activities-specific Balance Confidence (ABC) scale was used to assess self-perceived balance confidence during the performance of daily activities, such as reach on tip toes, sweep the floor and step onto or off an escalator, without losing balance or becoming unsteady.This scale is related to the construct of fear of falling and scores range from 0% (no confidence) to 100% (full confidence) [16].The EuroQoL-5D (EQ-5D) was used to assess health-related quality of life and comprises five dimensions (mobility, self-care, usual activities, pain/discomfort, and anxiety/ depression).Each of the five dimensions is divided into three different severity levels, i.e., level 1 indicating no problems, level 2 indicating some problems, and level 3 indicating extreme problems.These labels have no arithmetic properties and should not be used to derive a summary score [17].An EQ-5D index score was also calculated with scores ranging from 0 to 1 (best healthrelated quality of life) [17,18].
The Timed Up & Go test (TUG) reflects functional mobility and requires participants to stand up from an armchair, walk forward for 3 m, turn around, walk back to the chair, and sit down; results are recorded in seconds [19].The TUG was performed twice, with the second trial recorded as the test result and participants wearing their regular footwear and using their customary walking aids.

Procedures
All assessments were performed on the day on which each participant was recruited and administered in the order described above.Participants were allowed to rest as needed at any time during the evaluation.Following the initial assessment, only the FaB-Brazil scale was readministered on the same day by another rater in order to verify inter-rater reliability.Two weeks later, one of the original raters readministered the FaB-Brazil scale in order to verify test-retest reliability.

Statistical analysis
Data were analyzed using IBM SPSS version 21 (IBM Corp, Armonk, New York).Histograms indicated that the FaB-Brazil scale showed approximate normal distribution and that all other measures were not normally distributed.Descriptive statistics were calculated for demographic and clinical variables and data were described as mean (standard deviation [SD]) or median (interquartile range [IQR]).Cronbach's alpha was calculated to evaluate internal consistency of the FaB-Brazil scale.A Cronbach's alpha coefficient of �0.80 was interpreted as good, �0.70 to <0.80 as moderate, and <0.70 as low [20].
An intraclass correlation coefficient (ICC; absolute agreement, two-way mixed effects model) was calculated to evaluate interrater and test-retest reliability in a subset of participants.ICC values of �0.90 were interpreted as excellent, �0.75 to <0.90 as good, �0.50 to <0.75 as moderate, and <0.50 as poor reliability [21].There was no missing data in the FaB-Brazil total score.
The standard error of measurement (SEM) of the FaB-Brazil scale was calculated using the following formula: SEM ¼ SD baseline � square root of (2 � [1 À r xx ]) where r xx ¼ test-retest reliability [22].The minimal detectable change (MDC), which means the lower boundary of a potentially meaningful change [22], was calculated using the following formula: MDC ¼ Z-score level of confi- dence � SEM, considering a confidence interval (CI) of 95%, where the z-score ¼ 1.96 (the value on a standard normal curve associated with a 95% CI) [23,24].
Ceiling and floor effects were considered to be present if more than 15% of participants reached the minimum and maximum scores on the FaB-Brazil scale, respectively [25].
Convergent validity between the FaB-Brazil scale and age, history of falls (none vs any) and frequency of falls (none; one; two or more) in the previous 12 months, the extent of community mobility, modified Barthel Index, IPAQ, ABC, EQ-5D and TUG were evaluated using Spearman's rank correlation coefficient.The correlation between the FaB-Brazil scale and its dimensions was also tested using Spearman's rank correlation coefficient.A coefficient of 0.90 to 1.00 was interpreted as indicating very strong correlation, 0.70 to 0.89 as strong, 0.40 to 0.69 as moderate, and 0.10 to 0.39 as weak [26].
Participants were classified as non-fallers (no falls in the past year) or fallers (� 1 fall in the past year).In order to verify discriminative validity, comparisons of the FaB-Brazil scale mean scores between male and female and fallers and non-fallers were performed using student t test.A significance level of .05 was set for all statistical tests.

Translation and cross-cultural adaptation process
Initially, some changes were made by the translators on two items in order to keep idiomatic equivalence.Item 15 "I notice spills on the floor" was replaced by "I notice when the floor is wet" and item 23 "I avoid going out on windy, icy or wet days" was replaced by "I avoid going out on windy or wet days".All members of the expert committee agreed with both changes.Additionally, the expert committee suggested to reword item 19 "When wearing bifocals or trifocals I misjudge a step or do not see a change in floor level" and it was replaced by "When wearing bifocal or multifocal spectacles I misjudge a step or do not see a change in floor level."Also, considering the culture differences and the level of education across the different Brazilian regions, the expert committee suggested to replace some words in Brazilian Portuguese in order to reach experiential equivalence on the following items: 2 "I do things at a slower pace," 3 "I talk with someone I know about things I do that might help prevent a fall," 4 "I bend over to reach something only if I have a firm handhold," 7 "I hurry when I do things," 9 "To reach something up high I use the nearest chair, or whatever furniture is handy, to climb on," 11 "I get help when I need to change a light bulb," 12 "I get help when I need to reach something very high," 13 "When I am feeling ill I take special care of how I get up from a chair and move around," 14 "When I am getting down from a ladder or step stool I think about the bottom rung/step," 17 "I have made changes at home to make the lighting better," and 21 "When I walk outdoors I look ahead for potential hazards."Based on testing the prefinal version on 15 participants, no further changes to the FaB-Brazil scale were deemed necessary.

Participant characteristics
One hundred fourteen individuals were assessed for eligibility; four were excluded (vestibular dysfunction, n ¼ 2; stroke, n ¼ 1; severe knee osteoarthritis, n ¼ 1) and 17 declined.Therefore, 93 individuals, aged 69 years (IQR 64-73.5)and median Barthel index 50 points (IQR 50-50), were enrolled in this study (Table 1).Most participants were women (73%; 68/93), left home four or more times a week in past week (83%; 77/93), did not live alone (82%; 76/93) and reported low (39%; 36/93) or moderate (40%; 37/93) levels of physical activity.Twenty-six (28%; 26/93) individuals had at least one fall in the past year and only two (2%; 2/93) walked with an assistive device.The FaB-Brazil scale had a mean score of 2.81 (SD 0.42).Table 2 shows the mean score of each item and each dimension of the FaB-Brazil scale.Dimensions with the greatest protective behaviors were changes in level, being observant, and awareness, while pace, displacing activities and protective mobility were the dimensions with the greatest risky behaviors.

Internal consistency
The FaB-Brazil scale had a Cronbach alpha internal consistency of 0.73.

Inter-rater reliability
Thirty participants were assessed to determine the inter-rater reliability of the FaB-Brazil scale.The first assessment had a total mean score of 2.76 (SD 0.50) and the second assessment of 2.71 (SD 0.42).The ICC for the total mean scores between the first and the second assessment was 0.93 (95% CI 0.86-0.97;p < 0.001).N/A 2.57 (1.21) a These items need to be recoded to reflect more protective behaviors.b Some items have an option to answer "does not apply" and the total score for each individual is calculated as a mean of all items scored 1-4.FaB-Brazil: Brazilian-Portuguese version of the Falls Behavioral scale; N/A: not applicable.

Test-retest reliability
Thirty participants were assessed to determine the test-retest reliability of the FaB-Brazil scale.The first assessment had a total mean score of 2.69 (SD 0.34) and the retest of 2.73 (SD 0.38) (Figure 1).The ICC for the total mean scores between the first assessment and the retest was 0.79 (95% CI 0.57-0.90;p < 0.001).

Measurement error
The SEM was 0.27 and the MDC was 0.53.

Ceiling and floor effects
No participants reached the minimum and maximum scores on the FaB-Brazil scale.Therefore, neither ceiling nor floor effects were found in the FaB-Brazil scale.

Convergent validity
Significant positive weak correlations were found between the FaB-Brazil scale and age, TUG and the mobility, self-care and pain/ discomfort dimensions of the EQ-5D.Significant negative weak correlations were found between the FaB-Brazil scale and community mobility (number of times left home in past month) and EQ-5D index, while a significant negative moderate correlation was found between the FaB-Brazil scale and ABC scale.The variables years of education, frequency of falls (none; one; two or more) in the previous 12 months, history of falls (none vs any) in the previous 12 months, the usual activities and anxiety/depression dimensions of the EQ-5D, MMSE, Barthel, level of physical activity and MET-minutes/week were not significantly correlated with the total mean score of the FaB-Brazil scale (Table 3).
The dimensions cognitive adaptations, protective mobility, and avoidance had significant positive strong correlations with the total mean score of the FaB-Brazil scale, whereas practical strategies, displacing activities, and being observant had significant positive moderate correlations.The dimensions awareness and pace had a significant positive weak correlation with the total mean score of the FaB-Brazil scale, while the dimensions changes in level and getting to the phone were not significantly correlated with the total mean score of the FaB-Brazil scale.

Discussion
The aims of this study were to develop and cross-culturally adapt the FaB-Brazil scale and to verify its psychometric properties.Our results offer evidence for the reliability and validity of the FaB-Brazil scale for the assessment of habitual behaviors and actions related to falling in community-dwelling older people.As expected, internal consistency of the FaB-Brazil scale was moderate, similar to the most recently published European Spanish version [9] but in contrast with the original English [4], French Canadian [8], and Turkish [7] versions, which presented good internal consistency.We found excellent interrater reliability, which has not been previously reported.Additionally, we found good test-retest reliability, although it was found to be excellent in the European Spanish [9], the original English [4], and French Canadian [8] versions.These differences in both the internal consistency and test-retest reliability of the FaB scale are likely attributable to differences in methodologies and samples.Clemson et al. [4] used the original version of the FaB scale, where I adjust the lighting at home to suit my eyesight and I go out on windy days (items 17 and 23, respectively), have subsequently been modified in the current version of the scale to I have made changes at home to make the lighting better and I avoid going out on windy, icy or wet days, respectively [5].Also, Filiatrault et al. [8] set a 1-week interval between test-retest interviews and Alonso-Casado et al. [9] included participants living in geriatric or private homes and required participants to complete two versions of the FaB scale at least 30 minutes apart.It is also possible that there were variations in scoring procedures used by authors.
In relation to the construct validity, our results indicate that those who report more protective behaviors related to falling are less confident in their balance, in line with previous studies using the FaB scale [8,9].This may be explained because some items of the FaB scale comprise situations that challenge balance and require balance confidence to be performed without some form of adaptation, such as I hold onto a handrail when I climb stairs; I get help when I need to reach something very high; when I stand up I pause to get my balance; I bend over to reach something only if I have a firm handhold.Additionally, reduced balance confidence has been shown to be a risk factor for falls in older adults [27].Since balance confidence and fear of falling are related [28], it is important to note that carefulness has been recognised as a protective strategy and some older people who experience fear of falling adapt by moving more carefully [29].
The association between fall-related behavior and both functional and community mobility suggest that those who adopt more protective behaviors have impaired mobility, activity limitations and participation restrictions, in line with previous findings [4].Our results are strengthened by the association between protective behaviors on the FaB-Brazil scale and poorer scores on the mobility and self-care dimensions of the EQ-5D.Although the TUG test is primarily a measure of functional mobility [19], it also reflects balance, walking ability, and fall risk [30].Therefore, we conclude that those with more protective behaviors related to falling had reduced balance and gait capabilities.
Our results indicate that the older the person, the more protective behaviors adopted, similar to other findings [4].Increasing age is a well-known risk factor for falls [1] and its relationship with all dimensions of the FaB scale has been shown by Clemson et al. [4], suggesting that older people make behavioral changes to prevent a fall.Considering that more protective behaviors were related to reduced self-perceived balance confidence, impaired mobility, and decreased times left home, it was not surprising to find that greater protective behaviors were associated with poorer health-related quality of life, in line with previous findings [9].Interestingly, our results also suggest that pain/discomfort is associated with greater protective behaviors.These findings support previous research showing that pain contributes to balance [31,32] and gait impairment [33], and mobility limitations [31,32,34] in older adults.
Our findings of no association between the FaB-Brazil scores and sex and fall status are in line with those found by Filiatrault et al. [8].However, our results are in contrast with other findings and these differences may be due to the older group and larger sample size in Clemson et al. [4] compared to the current sample.Future research should explore fall-related behaviors among fallers and non-fallers considering both past falls and falls prospectively identified in order to investigate whether older people who have experienced a fall adopt more protective behaviors than those who have not a history of falls and whether the adoption of risky behaviors predicts future falls.
Different dimensions contribute to understanding older people's risk-taking behaviors that may increase the risk of falling or attitudes that may protect from falling while performing daily activities.In line with the findings of the development study [4], the dimensions cognitive adaptations (behaviors associated with reflection, intention and planning while performing indoor and outdoor activities), protective mobility (negotiating the environment in a supportive or protective way), and avoidance (avoidance of risky situations) were most strongly correlated with the overall FaB-Brazil score, highlighting their importance to fall prevention.Changes in level (strategies to cope with changes in support surface level), being observant, and awareness (both dimensions include behaviors related to noticing hazards) [4] were the dimensions with the greatest protective behaviors related to falling, suggesting that participants in this study adopted more protective behaviors while performing activities in the community and tasks that challenge balance.On the other hand, pace (avoiding hurrying and turning quickly), displacing activities (avoiding activities that cause displacement, i.e., going out on windy or wet days) and protective mobility (behaviors such as pausing after standing up or taking extra care when doing everyday tasks while unwell) [4] were less commonly adopted, showing that participants were less concerned about slowing down and adopting protective mobility strategies.
In spite of the finding that the minimum amount of change necessary for the mean score of the FaB-Brazil scale to be considered a true change in the period between evaluations [24] is 0.53, it is not possible to compare our data with other authors since they have reported the MDC of the FaB scale based on points instead of the mean score [9].Additionally, given that neither ceiling nor floor effects were present in the FaB-Brazil scale, it is suggested that no extreme items are missing in the lower or upper end of the scale, not limiting content validity [25].
We followed standard guidelines to translate and cross-cultural adapt the FaB scale [10].However, we acknowledge that the assessment of community-dwelling older people with no cognitive impairment based on the MMSE may limit the generalizability of our findings.In addition, our analyses with respect to falls data were limited since we did not collect fall frequency data greater than two falls.Future studies measuring behavior change following intervention for fall prevention [35] and investigating the relationship between risk-taking behaviors and measures of balance, gait and fear of falling in this population are warranted.Also, responsiveness needs to be determined.
In conclusion, the FaB-Brazil scale is a reliable and valid tool for measuring protective behaviors that reduce the risk of falls in community-dwelling older adults.Therefore, it may be used to raise awareness about the subtle everyday habitual or intentional behaviors of older people, which may be protective or risky with respect to falling in order to guide personalised fall prevention.

Figure 1 .
Figure 1.Scatterplot of the total FaB-Brazil mean scores at first assessment on the x-axis and at retest on the y-axis, with the 45 � line reflecting perfect agreement, i.e., y ¼ x.Note that the datapoints are evenly distributed above and below the 45 � line, indicating an absence of bias.

Table 1 .
Demographic and clinical characteristics of the study sample.Data are reported as mean (SD), median (interquartile range) where data were ordinal or not normally distributed, or n (%) for categorical predictors.

Table 2 .
Mean (SD) of each item of the FaB-Brazil scale and its dimensions (n ¼ 93).