Transcultural adaptation and validation of the QUALEFFO-41 Questionnaire for Hebrew-Speaking Israeli women with postmenopausal osteoporosis with and without vertebral fractures

The translation process employed professional translators with relevant research backgrounds. A physiotherapist specializing in osteoporosis and a translator unassociated with the field, fluent in English and Hebrew, performed the initial translation. The questionnaire was then back-translated into English by another therapist experienced in osteoporosis and a translator fluent in Hebrew and English. After discussing and reaching a consensus, the translators determined the final Hebrew version of the questionnaire. This version was then pretested on a small group of 15 women from the target population to ensure appropriateness and comprehensibility.

Our objective was to obtain a sample size exceeding 100, which, according to Consensus-based Standards for the Selection of Health Measurement Instruments (COSMIN) standards, is deemed “very good” for evaluating construct validity, measurement error, reliability, and internal consistency. These standards rate a sample size of 50–99 as “adequate,” 30–49 as “doubtful,” and anything less than 30 as “inadequate” [17]. In the context of our study, a sample of 100 should include both osteoporotic patients with vertebral fractures and osteoporotic control subjects without fractures.

The research was conducted within the Continuing Care and Home Hospitalization Unit at Clalit Health Services, Jerusalem District. Participants were recruited from seven primary care clinics in the Jerusalem District of Clalit Health Services (Ramot Eshkol, Mekor Baruch, Kiryat Yovel, Beit HaKerem, Maoz, German Colony, and Metudela) and the Ramot Eshkol Physiotherapy Institute.

Recruitment occurred through three channels: (1) direct referral by physiotherapists who identified eligible patients during treatment, (2) response to study advertisements posted in clinics and physiotherapy institutes, and (3) database screening of Clalit Health Services records following approval from the Data Extraction Committee. For the test–retest reliability assessment, 20 participants (10 from each group) were randomly selected to complete the QUALEFFO-41 questionnaire twice at a 2-week interval, either at their homes or at the clinic. The questionnaires were self-administered and completed outside of treatment sessions.

Inclusion criteria for the study were only Jewish (Ashkenazi or Sephardic), postmenopausal women aged 50–90 years diagnosed with osteoporosis. Those with at least one spinal fracture and no history of osteoporotic fractures in other areas were categorized into the OVF group, while those without spinal fractures comprised the OPC group. Exclusion criteria included women in an acute condition within a month following a spinal fracture, women with dementia, those unable to read Hebrew, as well as those diagnosed with secondary osteoporosis due to factors such as drug use, metabolic bone disease, end-stage renal failure, heart failure, or tumors. Additionally, women with other pathological conditions causing functional limitations or chronic back pain due to various diagnoses, such as disc herniations, spondylolysis, spondylolisthesis, and spondyloarthropathy, were excluded.

Data were analyzed using SPSS version 24. A p-value of 0.05 or less was considered a priori to indicate statistical significance for all research outcomes. Descriptive statistics (means, standard deviations, and percentages) were employed to analyze demographic characteristics. Chi-square and T-tests were used to compare categorical and continuous variables between the study and control groups. The same descriptive statistics were used to compare questionnaire scores within each group, including the floor and ceiling effects. The latter measures the proportion of subjects with the lowest and highest possible scores, respectively, with high values indicating difficulty in differentiating subjects or assessing changes after an intervention.

Reliability was tested via internal consistency, assessing item homogeneity within the test, and the test–retest method, assessing tool stability through repetition under identical conditions. Internal consistency was measured using Cronbach’s alpha, with a value of 0.70 or greater deemed satisfactory, while the ICC test was used to assess test–retest reliability.

For validity, discriminant and convergent validity tests were conducted using Pearson correlation for the QUALEFFO-41 questionnaire and Spearman correlation for the EQ-5D questionnaire. Concurrent validity was assessed using the EQ-5D questionnaire as the gold standard, examining the correlation between similar sections of both questionnaires. The degree of association was evaluated using a Spearman correlation.

The sensitivity of the questionnaires was measured via the area under the receiver operating characteristic (ROC) curve, which reflects the sensitivity and specificity of the questionnaire. We considered an area under the curve (AUC) of 0.7 or higher as marginally adequate in terms of sensitivity. The cutoff score for the questionnaire was calculated using the Youden index. We considered correlation coefficient (CC) values as follows: 0–0.20 indicates a poor correlation, 0.21–0.40 as medium, 0.41–0.60 as good, 0.61–0.80 as very good, and 0.81–1 as excellent. For comparing the discriminative abilities of the QUALEFFO-41 and EQ-5D questionnaires, we calculated the statistical significance of the difference between their respective areas under the curve (AUC). The comparison was performed using the formula Z = (AUC1 − AUC2)/√ (SE1² + SE2²), where AUC1 and AUC2 represent the respective AUC values, and SE1 and SE2 represent their standard errors. The resulting Z-score was then converted to a p-value using the standard normal distribution, with p < 0.05 considered statistically significant. This method allows for direct comparison of the diagnostic accuracy between the two questionnaires while accounting for their respective standard errors.

The questionnaire was translated into Hebrew and back-translated into English per the methodological protocol. Adjustments were made to accommodate the Israeli population, including alterations in measurement units and culturally relevant activities (e.g., going to synagogue). The International Osteoporosis Foundation approved the English version after two revisions. A pilot trial involving 15 women affirmed its comprehensibility and relevance, and no further changes were made.

The study involved 155 postmenopausal women aged 50–90, of which 49 were in the OVF group. The average age of this group was 76.84 ± 8.13 years. The OPC group, with an average age of 70.42 ± 8.08 years, comprised 106 women. Five participants withdrew due to specific back pain from a new disc herniation or failure to comply with the inclusion criteria due to osteopenia. The two groups had no significant differences regarding origin, marital status, or work type. Demographic details are presented in Table 1.

The QUALEFFO-41 and EQ-5D questionnaires reflected a poorer quality of life among the study group compared to the control group. The QUALEFFO-41 questionnaire showed a ceiling effect in pain and basic daily activities for both groups, but no significant floor effect. Detailed scores and distributions can be found in Table 2.

The Cronbach’s alpha coefficient for the QUALEFFO-41 questionnaire was α = 0.88 for both groups, indicating good internal consistency, except in the OPC group’s mental, leisure, and social activity domains)α = 0.67(. The EQ-5D questionnaire showed lower consistency, with α = 0.77 in the OVF group and α = 0.70 in the OPC group. A repeat test showed good to excellent correlations in most areas for both groups (Table 3).

The construct validity of the QUALEFFO-41 questionnaire was assessed using discriminant and convergent validity, with correlations evaluated using Pearson’s correlation coefficient. In the study group, 37 out of 41 questions (90%) showed good to excellent correlations, with the remaining four questions belonging to the mental domain. The correlation index ranged from 0.89 to 0.79 for pain, 0.89 to 0.71 for basic daily activities, 0.83 to 0.73 for housework, 0.83 to 0.62 for mobility, 0.85 to 0.59 for leisure and social activities, 0.92 to 0.78 for general health, and 0.78 to 0.3 for mental functioning.

Regarding discriminant validity, most QUALEFFO-41 questions (89%) showed higher correlations with their domain than with other domains, except for one question in the mental domain. For convergent validity, most questions showed a satisfactory relationship with expected variables, namely basic daily activities, household tasks, mobility, leisure and social activities, general health condition, and mental functioning (67%), except for three questions in the mental domain. In the control group, 40 questions (97%) exhibited good to excellent correlations, with the remaining questions belonging to the leisure domain. The correlation index ranged from 0.88 to 0.74 for pain, 0.83 to 0.64 for basic daily activities, 0.92 to 0.75 for housework, 0.83 to 0.48 for mobility, 0.70 to 0.22 for leisure and social activities, 0.87 to 0.72 for general health, and 0.73 to 0.41 for mental functioning. Regarding discriminant validity, most questions showed higher correlations with their domain compared to other domains (57%), except for four questions in the leisure and social activity domain. For convergent validity, most questions demonstrated satisfactory correlations (86%), except for one in the leisure and social activity domain. These findings are presented in Table 4.

The ROC curve analysis showed sufficient sensitivity of both QUALEFFO-41 and EQ-5D questionnaires in diagnosing women with osteoporosis and fractures. The area under the curve (AUC) was 0.767 (0.043) for QUALEFFO-41 and 0.756 (0.041) for EQ-5D. Comparative analysis of the discriminative abilities between these questionnaires revealed no statistically significant difference (difference = 0.011, Z = 0.185, p = 0.853), suggesting both questionnaires demonstrate comparable effectiveness in discriminating between women with and without vertebral fractures. The domains showed varying levels of predictive ability, with physical function demonstrating good discrimination (AUC = 0.75 [0.043] for QUALEFFO-41 and 0.77 [0.041] for EQ-5D), while the pain domains showed lower discrimination (AUC = 0.63 [0.050] for QUALEFFO-41 and 0.68 [0.05] for EQ-5D). Detailed scores for each domain are in Table 5, and the ROC curves for the QUALEFFO-41 and EQ-5D are available in Figs. 1 and 2.

Bild vergrößern

Bild vergrößern