Translation and transcultural validation of the Dutch Hospital for Special Surgery Paediatric Activity Brief Scale (HSS Pedi-FABS)

Background: There is a need for a validated simple Dutch paediatric activity scale. The purpose was to translate and transculturally validate the Dutch Hospital for Special Surgery Paediatric Functional Activity Brief Scale (HSS Pedi-FABS) questionnaire in healthy children and adolescents. Methods: The original HSS Pedi-FABS was translated forward and backward and was transculturally adapted after performing a pilot study among children and professionals. The nal version of the Dutch HSS Pedi-FABS was validated in healthy children and adolescents aged 10 to 18 years old. Children who had any condition or injury limiting their normal physical activity were excluded. The interval between the rst questionnaire T0 (HSS Pedi-FABS, Physical Activity Questionnaire for children or adolescents (PAQ-C/A) and Tegner activity scale) and the second questionnaire T1 (HSS Pedi-FABS) was two weeks. Construct validity, interpretability and reliability were evaluated. Content validity was evaluated through cognitive interviews among a smaller group of children and through a questionnaire among professionals. Results: To evaluate content validity, 9 children and adolescents were interviewed, and 30 professionals were consulted. Content validity among professionals showed a relevance of less than 85% for most items on construct. However, content validity among children was good with a 92% score for item relevance. Readability was scored at a reading level of 11- to 12-year-olds. The validation group consisted of 110 healthy children and adolescents (mean age of 13.9 years ± 2.6). Construct validity was considered good as 8 out of 10 hypotheses were conrmed. The Dutch HSS Pedi-FABS showed no oor or ceiling effect. Analysis of the internal consistency in the validation group resulted in a Cronbach’s alpha of 0.82. Test-retest reliability was evaluated among 69 children and adolescents and revealed an Intraclass Correlation Coecient (ICC) of 0.76. Conclusion: The Dutch HSS Pedi-FABS showed good psychometric properties in a healthy Dutch paediatric and adolescent population. Limitations of the current Dutch HSS Pedi-FABS are content validity on construct of items reported by professionals. Future research is desirable to establish psychometric properties in children and adolescents with

subjective measures. Objective measures such as accelerometers and heart-rate monitors provide highly reproducible and accurate data on physical activity but are often rather expensive, time-consuming, and may require technical expertise. [6] Self-reported measures, such as questionnaires, are often used to assess physical activity in children and adolescents because of their advantages, such as low costs, minimal participant burden, and easy administration. However, problems may arise with the length of the questionnaire, understanding the questions, or accurately recalling physical activity especially in a young target population. [7] Multiple self-reported activity scales already exist in the orthopaedic eld. [8,9] However, they either focus on disability, are too long and time consuming, are not applicable to children, or are joint or activity speci c. [8] Moreover, a recent review by Hidding et al. [10] argued that there is a lack of physical activity questionnaires with excellent validity and reliability. [10] In 2013, the Hospital for Special Surgery Paediatric Functional Activity Brief Scale (HSS Pedi-FABS) was developed to assess the physical activity level in children and adolescents aged 10 to 18 years old. [8] The HSS Pedi-FABS is a simple, validated paediatric activity scale, which may be useful to evaluate physical activity level as a prognostic factor in clinical outcome research. [11] It has excellent scale reliability, robust construct validity, and shows no oor or ceiling effects. [8] Besides, the HSS Pedi-FABS has recently been recommended by the 2018 International Olympic Committee (IOC) consensus statement and will be used as activity-rating scale in the Paediatric Anterior Cruciate Ligament Monitoring Initiative (PAMI). [12] This European initiative launched by the European Society for Sports Traumatology, Knee Surgery and Arthroscopy (ESSKA) aims to create a pan-European system to collect and analyse data to provide stronger scienti c evidence in paediatric ACL injury treatment. [12,13,14] Yet, the HSS Pedi-FABS is currently only available in English and Italian. [8,13] Therefore, it is crucial to transculturally validate the Dutch HSS Pedi-FABS. It is hypothesized that the Dutch HSS Pedi-FABS has adequate psychometric properties in a healthy paediatric and adolescent population, comparable to the psychometric properties of the original HSS Pedi-FABS. [8] Methods Translation procedure Translation of the original HSS Pedi-FABS was performed using a forward-backward translation procedure. [15] The HSS Pedi-FABS was translated from English into Dutch by two native Dutch speakers.
Translations were compared, discrepancies between them were discussed and a preliminary version of the Dutch HSS Pedi-FABS was established. Subsequently, this preliminary version was translated back into English by two independent English native speakers who were unfamiliar with the original questionnaire. The translated version was compared to the original version of the HSS Pedi-FABS to check for similar item content. Differences and inconsistencies were discussed and adjustments were made to form the pre-nal version. This pre-nal version was evaluated as a pilot among children and professionals for cross-cultural adaptation which resulted in minor adjustments. The developer of the original HSS Pedi-FABS was consulted to discuss cross-cultural adaptations. [8] Finally, the Department of Patient Communication at the Máxima Medical Centre evaluated this pre-nal version. Some linguistic adjustments were made, and the nal version of the Dutch HSS Pedi-FABS was established (Appendix 1).

Participants
The study population consisted of the content validity group and the validation group. The content validity group consisted of two subgroups: target population and professionals. Participants of the content validity target population were recruited at a sports club and through a personal network.
Physically active children aged 10 to 18 years old were included in this group. Professionals from relevant disciplines were recruited from 4 Dutch teaching hospitals: Máxima Medical Centre Eindhoven/Veldhoven, VieCuri Hospital Venlo, Maastricht University Medical Centre and Erasmus University Medical Centre Rotterdam. The validation group was recruited through primary and secondary schools in the Netherlands and at the out-patient department of the paediatric orthopaedic clinic at the Máxima Medical Centre and Erasmus University Medical Centre. Children or adolescents aged 10 to 18 years were included in this group. Children who had any condition or injury limiting their normal physical activity were excluded. Study procedure Figure 1 shows the study procedure of the translation and validation of the Dutch HSS Pedi-FABS. After translation, content validity was assessed through cognitive interviews in participants representing the target population and through questionnaires in professionals. [16] Interpretability and construct validity were evaluated within the validation group. Participants received an information letter together with a set of questionnaires at school (T0). If they completed the baseline questionnaires, they were asked to ll out the Dutch HSS Pedi-FABS again by email or post two weeks later (T1) and answer the anchor question "Did your level of physical activity change since you completed the previous questionnaires (± 2 weeks ago)?". Responses of participants reporting stable level of activity were used to assess reliability of the Dutch HSS Pedi-FABS.

Questionnaires
At baseline (T0), children and adolescents within the validation group completed the HSS Pedi-FABS, Physical Activity Questionnaire -Children or Adolescent (PAQ-C or PAQ-A), Tegner activity scale, and questions concerning age, school, weekly participation in sports and level of competition.
The HSS Pedi-FABS questionnaire is an 8-item metric to assess physical activity in children and adolescents between 10 and 18 years old. [8] The questionnaire consists of 6 Likert-based items regarding frequency of activities, one item on the level of sports and one item on supervision. [8,13] Scores range from 0 to 30, with a higher score indicating more physical activity. [8,13] To date, the PAQ-C and PAQ-A are the only validated Dutch questionnaires to assess physical activity in children and adolescents. [10,17]. The PAQ-C was originally designed for children aged 8 to 14 years and consists of 10 questions. [17] The PAQ-A was designed for adolescents aged 14 to 20 years and consists of 9 questions. [17] Both PAQ questionnaires contain one question in the form of a checklist of common sports and activities which are scored on frequency of participation. [17] Subsequently, the mean is calculated which results in the total score ranging from 1 (low activity) to 5 (high activity). [17] Although the Tegner activity scale is not validated in the paediatric population, it is often used as an activity scale for children and adolescents. [18] The Tegner activity scale is a 1-item activity scale ranging from 0 (no activity) to 10 (high activity) and is widely used as an activity rating system for a variety of knee disorders. [19,20] Readability Readability of the Dutch HSS Pedi-FABS was assessed with the Dutch version of the Flesch reading ease (FRE) test. [21] A score of 0 re ects academic language while a score of 100 re ects the reading level of children in 4 th grade (age 9-10 years). The aim was to attain a readability score between 60 and 80 points, which re ects the reading level of children aged 11 to 13 years old.

Content validity
Target population Relevance, comprehensiveness and comprehensibility were evaluated through cognitive interviews. [22] All interviews were conducted by one researcher (TK) and were audio-recorded for transcription. A semistructured interview guide was used containing prede ned probe questions which addressed comprehension of the instructions, items, recall period, and response options. [23] All items of the HSS Pedi-FABS were also rated on relevance, and participants were asked to suggest potential missing concepts. Interviews were transcribed verbatim and analysed by one researcher (TK) using ATLAS.ti version 8.4 (Scienti c Software Development GmbH, Berlin, Germany). Cognitive codes were applied using the Problem Classi cation Coding Scheme (CCS), which consists of ve categories: comprehension and communication; memory retrieval; judgement and evaluation; response selection; and other (see Appendix 2). [24] A sixth category containing four codes was added to determine relevance.

Professionals
Professionals from relevant disciplines (orthopaedic surgeons, residents in orthopaedic surgery, physiotherapists, sports physicians, rehabilitation physicians, and trauma surgeons) rated the relevance and comprehensiveness of the Dutch HSS Pedi-FABS. A questionnaire was used to evaluate the relevance of each item for both the target population and the construct; the response options and recall period were rated on appropriateness. Comprehensiveness was addressed by asking whether important concepts or items were missing in the questionnaire.
Items were considered relevant for the target population or construct if at least 85% of professionals rated them as relevant. Content validity of the Dutch HSS-Pedi FABS was rated as su cient if at least 85% of the items were considered relevant by both professionals and participants. [16] Interpretability Interpretability was assessed by examining the distribution of HSS Pedi-FABS scores at T0, including the mean and standard deviation (SD). Moreover, oor and ceiling effects were evaluated and considered present if more than 15% of the participants scored either the lowest or highest score possible. [8,16,25] A positive rating of interpretability of the HSS Pedi-FABS was given if oor and ceiling effects were absent.

Construct validity
Hypothesis testing was used to assess construct validity; criterion validity was not evaluated since no gold standard is available for questionnaires on physical activity. Hypotheses were de ned about the relationship between the HSS Pedi-FABS and outcome measures which measure either the same or a different construct (convergent or discriminative validity, respectively) ( Table 3). These hypotheses were formulated by a panel consisting of experts and based on the study of Fabricant et al. [8] To evaluate convergent validity, a correlation of r ≥ 0.50 was expected between the HSS Pedi-FABS scores and a) the PAQ scores (total), b) PAQ-C scores, c) PAQ-A scores, d) the Tegner activity score, e) self-reported hours of weekly participation in sports, and f) the level of competition, all assessed at baseline (T0). Age and BMI were expected not to correlate (<0.30) with scores of the HSS Pedi-FABS which re ects divergent validity. Correlations of the hypotheses con rming convergent validity should be at least 0.1 higher than the correlations that indicate discriminative validity. [22] The latter was operationalized as two hypotheses: hypotheses conforming convergent validity should be at least 0.1 higher than age (hypothesis 9) and BMI (hypothesis 10). Construct validity of the HSS Pedi-FABS was considered good if at least 75% of the prede ned hypotheses were con rmed. [16] Reliability Internal consistency, test-retest reliability and measurement error were evaluated as measurement properties of reliability. [16]. All participants within the validation group were included for analysis of internal consistency. All participants who completed the baseline questionnaires were invited to complete the Dutch HSS Pedi-FABS a second time two weeks (T1) after completion of T0. Only participants who reported no change in their activity pattern during the interval period were included in the test-retest analysis.

Comparison between HSS Pedi-FABS versions
The psychometric properties of the original, the Italian and Dutch HSS Pedi-FABS were compared.

Statistical analysis
Statistical analyses were performed with IBM SPSS Statistics 24. Descriptive statistics were used to describe baseline characteristics of the participants. The aim was to include at least 7 participants and construct validity, interpretability, and reliability, at least 100 participants needed to be included in the validation group. [22] Shapiro-Wilk test was used as test for normality of the baseline characteristics and HSS Pedi-FABS outcomes in the validation group. [26] Spearman rank correlations were calculated to assess construct validity. To determine internal consistency, Cronbach's alpha was calculated. Test-retest reliability was evaluated by means of a two-way random effects model of Intraclass Correlation Coe cient (ICC) in absolute agreement. Cronbach's alpha and ICC coe cients of 0.70 or higher are considered to re ect good reliability. [16] The Standard Error of Measurement (SEM) was calculated as SEM = SD * √ (1 -reliability), where the ICC re ects reliability. [16,27] The Smallest Detectable Change (SDC) was de ned as 1.96 * √2 * SEM. [28] The signi cance level was set at 5% for all statistical analyses.

Results
The content validity population consisted of 9 participants. A total of 132 children and adolescents were included for the validation study, of which 22 participants reported a condition or injury limiting their normal physical activity and were excluded for analysis. In Table 1, the characteristics of the content validity population and validation population are shown. 39.1% of the children received help from parents for completing the questionnaires. "BMI = body mass index"; "h = hours"; "No. = number"; "SD = standard deviation"; "y = years".

Readability
The readability level was estimated at 71 which corresponds to a readability level of 11-to 12-year-old children.

Target population
The interviews (n = 9) yielded 32 different codes, 28 as de ned in the CCS and 4 extra codes to evaluate the relevance of each item (Appendix 2). In total, 127 times a code was applied; 54 times this was a relevance code and 73 times a code from the CCS ( Table 2). Considering the relevance of the items, 92.3% (50/54) was considered relevant by the participants of which 69% (37/54) was evaluated as highly relevant; in three cases (5.6%) it was unclear whether the item was considered relevant and only once an item was indicated as not relevant (1.9%). Over half of the applied codes from the CCS (45/73) were classi ed in the comprehension and communication category ( Table 2). The items that were most often regarded as complex or vague were cutting and pivoting. Sometimes, participants struggled with the difference between endurance and duration. Furthermore, a few participants had di culty estimating the frequency of the requested item.

Professionals
The 30 professionals consisted of 9 orthopaedic surgeons, 2 trauma surgeons, 1 sports physician, 13 residents and 5 physiotherapists. Figure 2 shows the relevance according to the professionals for each item with regard to the target population and construct. Item 5 and 6 were considered relevant by less than 85% of the professionals for both the target population and construct. Item 4, 7 and 8 also achieved a relevance score of less than 85% for the construct. The recall period of one month was considered "good" by 67% of the professionals, 20% found that the recall period was too long and 13% that it was too short. Almost half of all professionals suggested that additional items were necessary to measure physical activity. The most frequently suggested additional items were cycling to school (n=4), other physical activities such as playing outside or physical education (n=4) and injuries (n=3).
Overall, content validity among the target population was considered good and among professionals acceptable for relevance of the target population but insu cient for relevance of the construct.

Construct validity
Spearman rank correlations were calculated to evaluate the prede ned hypotheses for the construct validity, as shown in Table 4. Except for the PAQ-C and weekly participation in sports, the hypotheses regarding convergent validity were con rmed. Both hypotheses on discriminative validity were also con rmed. Furthermore, all correlations evaluating convergent validity were at least 0.1 higher than the correlation of age or BMI with the HSS Pedi-FABS. Eight out of the ten (80%) hypotheses were con rmed, indicating a good construct validity. "BMI = body mass index"; "HSS Pedi-FABS = Hospital for Special Surgery Paediatric Functional Activity Scale"; PAQ-A/C = Physical Activity Questionnaire -Adolescent/Children".

Reliability
Analysis of the internal consistency of the HSS Pedi-FABS in 110 children, resulted in a Cronbach's alpha of 0.82, indicating a good internal consistency.
Of the 110 children that were included at baseline, 89 children responded at follow-up. For analysis, 14 children were excluded because of altered activity patterns and six children due to technical errors at T1. The mean interval between baseline and follow-up was 19 days ± 7.2. The mean HSS Pedi-FABS score at follow-up (n = 69) was 17.7 ± 5.8. Test-retest reliability of the Dutch HSS Pedi-FABS was su cient with ICC = 0.76 (p < .001). The SEM was calculated at 2.8 points and the SDC at 7.9 points, on a scale from 0 to 30.

Comparisons between the HSS Pedi-FABS versions
Several differences in psychometric properties were found between the original English, Italian and Dutch versions (Table 5). Compared to the English and Italian version, the Dutch HSS Pedi-FABS showed a lower, but acceptable, test-retest reliability. [8,13] Compared to the English and Italian HSS Pedi-FABS, the percentage of children scoring the lowest and highest possible scores, indicating oor-or ceiling effect, was lower. [8,13]  "HSS Pedi-FABS = Hospital for Special Surgery Paediatric Functional Activity Scale"; "ICC = Intraclass Correlation Coefficient"; "N/A = Not Applicable"; "PAQ-A/C = Physical Activity Questionnaire -Adolescent/Children"; "SDC = Smallest Detectable Change"; "SEM = Standard Error of Measurement"; "-" stands for not assessed.

Discussion
The most important ndings of this study are that the Dutch HSS Pedi-FABS has a good internal consistency, acceptable test-retest reliability, good construct validity and a positive interpretability rating in a Dutch population of healthy paediatric and adolescent participants.
Although the overall interpretation of the psychometric properties was similar, certain differences were found among the HSS Pedi-FABS versions, of which the test-retest reliability and oor and ceiling effects were the most important. [8,13] These differences may be caused by the differences in interval between T0 and T1, the inclusion criteria and population characteristics. For example, the differences in oor and ceiling effect could be caused by the differences in inclusion criteria. [8,13] The English HSS Pedi-FABS study included athletically active adolescents, the current study healthy children and adolescents and the Italian study children with knee pathologies. [8,13] Therefore, it might be expected that the mean score of the English HSS Pedi-FABS study is higher than the Dutch and the Italian, but also that the scores are distributed in the higher score ranges. [8,13] As the Italian study included children with knee pathologies, the mean score was lower and had a signi cant oor effect. [13] The mean score of the Dutch HSS Pedi-FABS was more similar to the mean score of the HSS Pedi-FABS in a study on normative data, although a higher oor effect was found in that study. [11] In contrast to the positive ratings on the outcomes of construct validity, interpretability and reliability, content validity showed different results. Content validity among the target population was considered to be good. However, content validity reported by professionals was acceptable for the relevance for the target population, but insu cient for the construct. The questionnaire did not reach 85% relevance on most items for the construct. Besides, 47% of the professionals suggested an additional item to measure physical activity. As multiple issues appeared already in the pre-nal version during the pilot study among professionals, several transcultural adaptations were made to solve these issues and the original author was consulted. It was decided to maintain the original form and content of the HSS Pedi-FABS. No additional items were therefore added nor was the content changed.
Compared to other paediatric activity scales, the HSS Pedi-FABS has multiple advantages. [8] The HSS Pedi-FABS is a short and simple scale compared to other questionnaires, that potentially minimizes questionnaire fatigue and increases compliance. [8] Also, the HSS Pedi-FABS is a general measurement of physical activity and not speci ed on sports or joints, which provides a potential for broader application in clinical outcomes research. [8,11] In previous studies, the HSS Pedi-FABS has shown to capture changes in physical activity due to recent injury more likely than the Marx Activity Scale, to have more correlations with an athlete's participation in sports than the Tegner activity scale and to be reliable as patient reported outcome measure (PROM) captured electronically as on paper. [29,30,31] This study had certain limitations. Criterion validity could not be assessed, as there was no "gold standard" for questionnaires on physical activity. [16] Furthermore, a potential source of bias is the "proxy problem", as 39% of the children received help from parents in completing the questionnaire. [32,33] Selfreports of children are not equal to reports by proxy-respondents and a parents' report can therefore not be substituted for the child's report. [32,33] The readability level, however, was estimated to correspond to a readability level of 11-to 12-year-old children. [21] As children and adolescents aged 10 to 18 years were included, it seemed to be a rather high percentage of children receiving help from parents in completing the questionnaire. For children who experienced problems in the comprehensibility and who were not able to complete the questionnaire properly, parental assistance might be desirable. This was also advised on the instruction form of the questionnaire. However, as comprehensibility in the current study in this population did not seem to be an issue, it is doubtful whether parental help was necessary and this "proxy problem" might be a source of bias. [32,33] For future use, the Dutch HSS Pedi-FABS also has a limitation. The current study included healthy children and adolescents without a condition or injury limiting their normal physical activity.  Study procedure. "HSS Pedi-FABS = Hospital for Special Surgery Paediatric Functional Activity Brief Scale"; "PAQ-C/A = Physical Activity Questionnaire -Children or Adolescents".

Figure 2
Relevance for the target population (left) and the construct (right) rated by professionals (n = 30)

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.