This cross-sectional study complied with the requirements of the Declaration of Helsinki,21 and the Ethics Committee of Federal University of Vales do Jequitinhonha and Mucuri approved the study protocol (73359517.3.0000.5108). We defined the measurement properties of HAGOS-Br according to the recommendations of the COSMIM checklist and previous studies.14,22–25

The translation and cross-cultural adaptation of the HAGOS-Br was conducted following the guidelines for cross-cultural adaptation of self-administered questionnaires26 after obtaining the consent of the authors of the original questionnaire.9

Initially, 2 bilingual translators worked independently to translate the HAGOS from Danish into Brazilian Portuguese. The separate translations were later compared and consolidated into a consensus preliminary initial version of the HAGOS-Br. Two native Danish translators performed the back-translation, independently, then, compared their work and reached consensus on the back-translated Danish version of the HAGOS. A specialist committee audited all of the translations, compared them, and discussed with the translators to resolve any discrepancies and to develop a pre-final version of the HAGOS to be tested in Brazil, titled HAGOS-Br (http://www.koos.nu/).

Subsequently, this pre-final version was submitted for assessment to 25 Specialist Musculoskeletal and Sports Physical Therapist from all regions of Brazil. This committee analyzed the clarity and clinical applicability of the pre-final version of the questionnaire, as well as its adequacy to the socio-cultural and educational reality of the region. Therefore, through a Delphi study, with questions answered on a Likert scale, a level of agreement of at least 80%, defined as the cutoff point, was reached.

After obtaining consensus among the professionals, we tested the pre-final version in a self-administered manner on 25 university students who, in addition to answering the questionnaire, were asked to indicate possible difficulties in understanding some term and/or answer options. A level of understanding of 85% was defined as the cut-off point. After we obtained a proper level of comprehension of all items of the questionnaire in the pre-testing stage, the HAGOS-Br questionnaire was finalized (Supplemental Online Material).

Participants were divided into two groups: asymptomatic group (AG) and symptomatic group (SG). Participants were considered eligible for the study if they were 18–55 years of age, were physically active (at least 2.5 h per week), and agreed to sign the informed consent form. In addition, the AG was composed of asymptomatic individuals who participated in sports with high-demand for the hip and groin region; and the SG was composed of individuals who reported hip and/or groin pain with an intensity equal or greater that 1 point on a 0–10 Numeric Pain Rating Scale (NPRS) at rest or during/after sports participation/physical activity that persisted more than 6 weeks.8,9,15,27 By including both symptomatic and asymptomatic individuals, the chances of achieving a data spread across the entire HAGOS 0–100 points range was optimized. As HAGOS is used in sporting populations for both screening, prevention, and treatment it is important that usage of the whole scales(s) is considered and included.28–30 Potential participants with self-reported limiting comorbidities31 or not able to complete the questionnaires were excluded. The flow chart of the inclusion procedure of participants is presented in the Supplemental Online Material.

Data collection was performed on two different occasions with an interval period of 4–7 days. First, the participants were classified regarding hip and/or groin pain.8,9 Next, the HAGOS-Br was self-administered after a brief explanation about the structure of the questionnaire. Finally, the Brazilian version of the Short Form-36 items (SF-36)32 and Lower Extremity Functional Scale (LEFS)33 were also completed by the participants.

The HAGOS includes 37 items, evaluated in 6 subscales: Symptoms (7 items); Pain (10 items); ADL (5 items); Sport/Rec (8 items); PA (2 items) and QOL (5 items). Standardized answer options are given (5 Likert boxes) and each question gets a score from 0 to 4, where 0 indicates no problem. The six scores are calculated as the sum of the items included, ex. (100−(totalscoresubscale)×100maximumsubscalescore). Raw scores are then transformed to a 0–100 scale, with zero representing extreme hip/groin problems and 100 representing no hip/groin problems. Scores between 0 and 100 represent the percentage of total possible score achieved. An aggregate score is not calculated because it is regarded desirable to analyze and interpret the different dimensions separately.9

SF-36 is a generic instrument of easy comprehension and application that aims to evaluate QOL. It is presented as a multidimensional questionnaire consisting of 36 items in eight subscales: physical functioning, physical role, bodily pain, general health, vitality, social functioning, emotional role, and mental health. Each subscale is scored separately, with values ranging from 0 to 100, with 0 indicates the worst condition and 100 the best condition.32

The LEFS aims to assess the functional level of individuals with lower limb injuries. This scale categorically evaluates 20 items about functions related to routine activities. The score of each item is given through a 5-point Likert scale, ranging from 0 to 4. The final score of the scale is given by the sum of the score of each item, generating a result ranging from 0 to 80, which should be transformed in a percentage value from 0% to 100%, with 0 indicating severe impairment and 100% no functional impairment. For the Brazilian version of the LEFS, differences above 11 points are considered clinically important.33,34

To measure average pain intensity experienced at rest and during/after sports participation/physical activity, an 11-point NPRS was also used.15,35

The sample size was based on previous studies,9,16,36,37 and consistent with the literature, which recommended to include at least 50 participants.14,24 Only symptomatic participants were used for test-retest reliability, measurement error, internal consistency, structural validity, and interpretability analysis (n = 51, while both groups were used for construct validity(n = 103). Descriptive statistics were used for demographic variables and scores on the questionnaires. Data are presented as mean ± standard deviation (SD). Statistical analysis was performed with IBM SPSS Statistics Version 22 (IBM Inc., Armonk, NY, USA), using a significance level of 5%. The statistical software AMOS version 22.0 (IBM Inc., Armonk, NY, USA) was used to carry out the confirmatory factor analysis.

Test-retest reliability is the extent to which scores for the same patients are unchanged for repeated measurements over time. To evaluate this property, we use only stable individuals of the symptomatic group. Participants who did not present clinical changes in LEFS were considered stable and included in the test–retest reliability analysis.14,22 Test-retest reliability was assessed using the intraclass correlation coefficient (ICC2,1), and were calculated for each subscale using a two-way random effects model, type absolute agreement. ICC values of 0.70 or higher indicate high test-retest reliability.22,38

The measurement error (systematic and random error of a patient's score that is not attributed to true changes in the construct to be measured) was calculated using the SEM (SEM = SD × √1-ICC), which was converted to the SDC at an individual level with the 95% confidence interval (SDC95 = 1.96 × √2 × SEM).14,24,39,40

The internal consistency is the extent of interrelatedness among the items of the questionnaire, thus measuring the same construct, to verify the homogeneity of the questionnaire.22,24

For internal consistency analysis, we calculated Cronbach's α values for each subscale of the HAGOS-Br. Values between 0.70 and 0.95 are considered to indicate good internal consistency.24,38

The structural validity is the extent to which the scores of an instrument are an adequate reflection of the dimensionality of the construct to be measured.22 To assess the structural validity the confirmatory factor analysis (CFA) was performed by applying the maximum likelihood estimation method, available in AMOS software. The method was conducted to test the six-factor structures identified in the exploratory factor analysis of the original HAGOS.9 The goodness of fit for the competing models was evaluated through the following fit indices41: the chi-square test, the goodness of fit index (GFI), the root mean square error of approximation (RMSEA), the normed fit index (NFI), the Tucker–Lewis index (TLI), and the comparative fit index (CFI). The criteria used to determine a good model fit were a nonsignificant chi-square. For the GFI, NFI, TLI, and CFI, values above 0.90 indicate an adequate fit, and values above 0.95 indicate a good to very good fit and RMSEA < 0.08.42,43 The magnitudes of factor loadings > 0.3 were considered acceptable and statistically significant (p < 0.05).

Because there is no gold standard, validity was expressed in terms of construct validity, which is related to the score on a patient-reported outcome (PRO) instrument consistent with a priori hypotheses, based on the assumption that the PRO instrument measures the target construct.14,22

Construct validity of the HAGOS-Br subscales, SF-36 subscales, LEFS, and NPRS were investigated by Spearman's correlation coefficient. We defined strong correlations as r ≥ 0.7 (or -0.7 in case of an inverse relationship), moderate correlations as 0.5 < r <0.7 (or -0.5 < r < -0.7), and weak correlations as r ≤ 0.5 (or -0.5).24

Because the HAGOS is better designed to measure physical health than mental and/or social function, we expected the highest correlations between HAGOS-Br subscales and the SF-36 subscales of physical function, physical role due to physical health problems, and bodily pain (convergent validity) and lower correlations between the HAGOS-Br subscales and the SF-36 subscales of general health, vitality, social functioning, emotional role due to emotional problems, and mental health (divergent validity). Furthermore, we expected higher correlations between the HAGOS-Br subscales and LEFS (convergent validity); and between HAGOS-Br subscale of pain and NPRS at rest or during/after sports participation/physical activity (convergent validity). A priori hypotheses were formulated (Table 1), and good construct validity was based on meeting the criteria in at least 75% (42/56) of the indicated hypotheses.24

Interpretability is the extent with which it is possible to assign qualitative meaning to an instrument's quantitative scores or change in scores.22 This includes the distribution of total scores, change in scores, and floor and ceiling effects. Floor and ceiling effects were defined as being present if more than 15% of patients reported lowest (0) or highest (100) possible scores, and were assessed by calculating the relative frequency of the maximum and minimum values on the HAGOS-Br.24

The HAGOS-Br subscale scores for both test occasions, as well as the corresponding ICCs for absolute agreement, for the 51 participants in the symptomatic group are presented in Table 3. For all comparisons analyzed, ICCs were consistent, and higher than 0.70.

The SEM and the SDC at an individual level with the 95% confidence interval of HAGOS-Br are presented in Table 3.

The CFA revealed a Cronbach's Alpha for the HAGOS-Br subscales ranging from 0.86 to 0.96 (Table 4).

The CFA showed that the six constructs detected in the original HAGOS were confirmed, with the constructs presenting correlations from 0.68 to 0.96 (Supplemental Online Material). We tested for the six-factor structure with correlations between factors, however, we obtained poor indexes of model adjustment (chi-square= 1746.57, p < 0.001, GFI= 0.549, RMSE= 0.134, NFI= 0.654, TLI= 0.791, CFI= 0.741). We therefore checked the structure of each scale separately. For all HAGOS-Br scales, acceptable fit indexes were shown (Table 4). For RMSEA, four scales showed acceptable goodness-of-fit (<0.08). All questions showed acceptable factor loadings (>0.3) (Table 4).

Table 5 shows the Spearman correlations between the HAGOS-Br subscales, SF-36 subscales, LEFS questionnaire, and between the HAGOS-Br subscale of pain and NPRS at rest or during/after sports. A correlation between the HAGOS-Br and the SF-36 subscales was found, from the lowest to the highest, in mental health, emotional role, vitality, social functioning, physical role, bodily pain, and physical function. There was a strong correlation between the HAGOS-Br subscales and the LEFS questionnaire, as well as for the HAGOS-Br subscale of pain and NPRS both at rest or during/after sports. We confirmed 43 of the a priori hypotheses, formulated resulting in 76.7% agreement.

Overall, no ceiling or floor effects were observed for the HAGOS-Br subscales.