The sample consisted of 279 participants, both sexes aged 18–60 years, divided into 178 with non-specific CLBP and 101 asymptomatic controls. COSMIN suggests the rule of thumb of at least 5 times the number of items and ≥100 as a suitable sample size for structural validity analysis.31 As the FreBAQ includes 9 items, a sample of at least 100 participants is suitable. Participants were enrolled through social media, posters in clinical and non-clinical settings, and radio program calls.

For the CLBP group, eligibility was established using the following criteria: (1) a medical diagnosis of low back pain and pain duration equal or longer than 3 months, with pain located between T12 and the gluteal fold25 (2) back pain intensity on the numerical pain rating scale (NPRS) at the time of assessment ≥;326 (3) disability score ≥14% on the Oswestry Disability Index (ODI);27 and (4) Portuguese writing and reading skills. Participants were excluded if they had: (1) signs of serious conditions (red flags); (2) specific low back pain with major neural issues (e.g., active radiculopathy, symptomatic stenosis, spondylolisthesis); (3) neurological, psychiatric, or rheumatological disorders; (4) prior spinal surgery; (5) pregnancy; (6) vestibular disorders; (7) uncontrolled diabetes; (8) persistent neurological deficits (e.g., paresis, hemiparesis); (9) a Mini-Cog test score below 3; or (10) poor comprehension on the Cloze Test (score under 57%). Both the Mini-Cog and Cloze tests are further described in Supplementary file 1.

Participants assigned to the control group were required not to report low back pain symptoms. Individuals who exhibited cognitive changes as determined by the Mini-Cog test, or who had current pain or a history of back or lower extremity pain that restricted activity for more than five consecutive days within the past 12 months,16 were excluded from the study.

All study procedures followed the COnsensus-based Standards for the selection of health Measurement Instruments (COSMIN) recommendations.30,31 The study was conducted in two distinct phases. During the initial phase, the FreBAQ underwent cross-cultural adaptation, translation, and preliminary testing with a cohort of 28 participants diagnosed with CLBP, from March to September 2017. The second stage consisted of assessing the questionnaire's measurement properties (structural validity, internal consistency, measurement error, test-retest reliability, and hypothesis testing for construct validity) through a longitudinal observational case-control study involving 103 participants with CLBP and 101 controls, conducted at the Federal University of São Carlos (UFSCar). This second stage was conducted from February 2024 to November 2024, and participants attended the Research Laboratory of Movement and Pain for two sessions (test and retest assessments).

The study received ethical approval (CAAE 62,797,816.5.0000.0064 and CAAE: 73,402,523.7.0000.5504). The original version author authorized the FreBAQ translation and validation into Brazilian Portuguese. All participants signed the informed consent form and were previously informed about the procedures to be performed in the study, including their benefits and risks, in accordance with National Health Council resolution (CNS) 466/2012.

The FreBAQ is a self-administered questionnaire comprising nine items designed to evaluate back awareness. Each question is rated according to the following classifications: never (0), rarely (1), once in a while (2), often (3), and always (4). The score is obtained by summing the individual scores (0–36). The higher the score, the greater the distortion of back awareness. The measurement properties tested in the original version showed suitable internal consistency (Cronbach’s α = 0.77) and moderate reliability (Intraclass Correlation Coefficient [ICC] =0.65).18

The Mini-Cog28 and Cloze29 tests were used to assess cognitive status and reading comprehension, respectively. They are further described in the Supplementary file 1.

For the assessment of the hypothesis testing for construct validity, the following questionnaires were also administered to the participants: NPRS,32 Central Sensitization Inventory (CSI),33 ODI,34 Roland Morris Disability Questionnaire (RMDQ),35 Tampa Scale for Kinesiophobia (TSK),36 Pain Catastrophizing Scale (PCS),37 Patient Health Questionnaire-9 (PHQ-9),38 Generalized Anxiety Disorder (GAD-7),39 Pain Self-Efficacy Questionnaire (PSEQ),40 Fear-Avoidance Components Scale (FACS),41 Fear of Daily Activities Questionnaire (FDAQ-BR)41 and Pain Vigilance and Awareness Questionnaire (PVAQ).42 In the supplementary file, the reader can find a detailed description of the scales and questionnaires used in the study (Supplementary File 1). All the questionnaires used in the current study were translated into Brazilian Portuguese and showed acceptable measurement properties.

Translation and cross-cultural adaptation procedure of FreBAQ to Brazilian Portuguese and Pre-test Phase

The FreBAQ questionnaire was translated according to COSMIN guidelines.43,44 The five-step translation process is detailed in Supplementary file 2.

Structural validity measures the degree to which an instrument’s scores represent an adequate reflection of the dimensionality of the construct to be measured.34 Internal consistency assesses the degree of the interrelationship between scale items.30,34

Reliability assesses the degree to which repeated measurements at different time points yield similar responses from clinically stable individuals.46 Reliability was assessed in two subsamples of individuals with CLBP: 103 participants were classified as stable across both test and retest assessments based on reported pain intensity at the time of evaluation, while 109 participants were considered stable across both test and retest assessments based on reported pain intensity during the previous seven days. The sample sizes exceeded 100 participants, aligning with COSMIN risk of bias for systematic review of measurement properties.31

Only participants considered stable were included, i.e., those with a difference in pain intensity between assessments (test and retest) lower than or equal to 2 on the NPRS.47 The value of two points is recognized in the literature as minimal important change (MIC) for pain intensity (NPRS) in CLBP, and it has been reported as a parameter to distinguish patients who showed a clinically meaningful improvement/worsening from stable patients, when comparing before and after estimates for individual patients.47 In addition, the context was the same for test-retest assessments. Measurement error refers to the systematic and random errors in an individual's score that are not attributable to true changes in the construct being measured.48

Hypothesis testing for construct validity is the degree to which the scores of an instrument are consistent with previously established hypotheses, as reflected in relationships with other instruments (convergent validity) or differences between relevant groups (discriminant validity).49,50 To assess convergent validity, the FreBAQ-Br score was compared with scores from the following instruments: NPRS,18,51,52 RMDQ,54 ODI,53 PCS,18,53 TSK,51 FACS,51 FDAQ,51 GAD-7,54 PHQ-9,18 PVAQ,53 CSI,53 and duration of low back pain.19,55 Based on established research, hypotheses regarding correlations between questionnaire scores were developed to evaluate convergent validity (Supplementary file 3). Discriminant validity was assessed by comparing FreBAQ-Br scores between individuals with CLBP and asymptomatic participants. Based on existing literature, it was anticipated that individuals with CLBP would demonstrate higher FreBAQ-Br scores than asymptomatic controls.18,53

In total, 16 hypotheses were tested for convergent validity and one hypothesis for discriminant validity of the FreBAQ-Br. COSMIN quality criteria for good measurement properties recommend that at least 75% (13 out of 17) of the hypotheses raised a priori (for each scale or subscale) be confirmed to rate hypothesis testing for construct validity as acceptable.48

The analyses were performed using the SPSS statistical package for Windows and IBM SPSS, version 22. All variables were described using means and standard deviations. Differences between groups were expressed as mean differences (MD) with 95% confidence intervals (CI). Categorical variables were described using absolute and relative frequencies.

Structural validity of the FreBAQ-Br was assessed using Confirmatory Factor Analysis (CFA) (IBM® SPSS® Amos™). Maximum Likelihood with 1000 bootstrap resamples was used.56 Goodness-of-fit (GF) for each factor structure was assessed using several descriptive criteria: (1) Root-Mean-Square Error of Approximation (RMSEA), (2) Comparative Fit Index (CFI), (3) Standardized Root Mean Square Residual (RMSR), and (4) Tucker-Lewis Index (TLI).57,58 The acceptability (sufficient rating) of structural validity was analysed based on the following indices: RMSEA < 0.06, goodness-of-fit indices (CFI and TLI > 0.95), and RMSR < 0.08.31,57 The magnitudes of factor loadings of 0.3 or more were considered adequate for each scale item.31

Internal consistency was analyzed using Cronbach’s α with acceptable results (sufficient) when α≥0.70 for each unidimensional scale or subscale.31 Pearson’s Correlation coefficient was used to assess hypothesis testing for construct validity.31

Test-retest reliability was calculated using ICC (ICC2.1, two-way random mixed-effects, agreement). ICC>0.7 was considered acceptable (sufficient).31 The standard error of measurement (SEM) was analyzed using the following formula:59 SEM = SD x √(1 - ICC), in which SD = standard deviation. The SDC is considered a distribution-based measure and was calculated as follows: SDC95 = 1.96 × √2 × SEM.

During the translation stage, a discrepancy was observed in the translation of Q6 of the FreBAQ (“I can't perceive the exact outline of my back”) among the translators. Translator#1 proposed the following writing for the question: “Não consigo perceber o contorno exato das minhas costas”, while translator#2 suggested: “Eu não consigo perceber o exato alinhamento da minha coluna”. The expert committee agreed on the following translation for the Q6: “Eu não consigo perceber os contornos exatos das minhas costas”. The questions with lower comprehension scores in the pre-testing were Q1 (“My back feels as though it is not part of the rest of my body”) (mean score=8.04) and Q6 (I can’t perceive the exact outline of my back) (mean score of 7.71) (where 0 = “I did not understand at all” and 10 = “I understood everything”) (Supplementary file 4). The negative nature of Q1 and Q6 may explain the slightly lower comprehension score. Detailed information on the sample and the pre-testing results is provided in Supplementary file 4. For each questionnaire item, fewer than 20% of participants reported comparable difficulties; therefore, no modifications were implemented during the pre-testing phase.45

Table 4 (Supplementary file 4) illustrates the data obtained in the pre-testing stage of the study. The mean comprehension score for title/instructions/response options of the FreBAQ ranged from 7.39 to 9.64 (Table 4 – Supplementary file 4). In addition, the mean comprehension score for each FreBAQ question ranged from 7.7 to 8.9. Therefore, the translation process was considered successful, as the Brazilian Portuguese version of the FreBAQ was well understood by most participants. The complete version of the FreBAQ-Br is presented in Fig. 1.

Fig. 1.

Complete version of the Fremantle Back Awareness Questionnaire – Brazilian Portuguese version (FreBAQ-Br).

The FreBAQ‑Br demonstrated acceptable goodness‑of‑fit indices for structural validity, meeting the criteria for a sufficient rating. The model showed excellent fit, with CFI = 0.97, TLI = 0.96, RMSEA = 0.05 (90% CI: 0.00, 0.09), and RMSR = 0.04, supporting a unidimensional factor structure. As illustrated in Fig. 2, all item factor loadings exceeded the recommended threshold of 0.30.

Fig. 2.

Structural validity path diagram of the Fremantle Back Awareness Questionnaire Brazilian Portuguese version (FreBAQ-Br). Q = questions. e = error.

The FreBAQ showed adequate (sufficient) internal consistency (Cronbach’s α = 0.84). Reliability and Measurement error were, respectively, ICC = 0.81 (95% CI = 0.73, 0.87) and SDC = 3.33 (Table 2). Mean scores (SD) for the FreBAQ in CLBP on the first day of assessment (test) and at the retest were, respectively: 14.06 (SD=8.25) and 13.76 (SD=7.81). Thus, the construct's stability was confirmed, as no significant difference was observed in FreBAQ-Br scores between test and retest assessments.

For convergent validity, correlations between the FreBAQ-Br and comparator instruments confirmed 15 of 16 predefined hypotheses (Table 3). For discriminant validity, individuals with CLBP reported higher FreBAQ-Br scores than asymptomatic controls, with a MD=11.28 points (95% CI= 9.5, 13.07), supporting the a priori hypothesis (Table 3). In total, 16/17 hypotheses were supported (94.11%) (Table 3).

The study has some limitations: 1) Responsiveness was not assessed in our study. As a result, it cannot be concluded that the FreBAQ-Br score can identify changes over time. FreBAQ responsiveness has not been reported in the literature, warranting further research. 2) The MIC is important to determine if the changes in a condition are clinically relevant. We encourage future studies with a longitudinal design to determine MIC. 3) The pre-testing sample differs slightly from the sample recruited for analysis of the measurement properties. Nevetheless, the findings from the questionnaire comprehension index indicate that the tool is straightforward to understand. 4) The Mini-Cog test is appropriate for evaluating cognitive status in older adults, but its effectiveness in identifying cognitive impairment in adults has not yet been documented in the literature. We recommend that future studies investigate the accuracy of the Mini-Cog test to assess cognitive status in younger samples. 5) Compared to the CLBP group, the pain-free control group had higher socioeconomic status and educational levels. Nevertheless, the mean FreBAQ score for pain-free participants found in earlier studies (2.4)20 closely matches our own results (2.7), indicating that our data are consistent with previous research.