A community based, cross-sectional study was conducted from November 2016 to February 2017 in Baghdad city, Iraq. Baghdad city has two large areas named Al-Kharkh and Al-Rusafa. Three districts were selected by random cluster sampling method from these areas. The community pharmacies where the undergraduate students underwent training in these districts were used to invite the participants. Systematic samples were randomly selected. A structured interview included collection of the socio-demographic and translated OSES-A data. Each participant was interviewed individually by the researcher or a trained 5^th year undergraduate student after obtaining written or verbal informed consent. Some of participants gave only verbal informed consent because they considered it impolite behavior given that they had already given verbal consent for participation. The study protocol and ethical approval (including verbal informed consent) were approved by the Scientific Committee of Al-Rafidain University Collage, Baghdad, Iraq.

A recommendation suggested that at least 5 subjects per item are needed to evaluate the reliability and validity of a questionnaire [19]. The original OSES consist of 12 questions; therefore 60 participants were needed for the purpose of validation. However, for factor analysis, it is preferable to use 300 subjects [20]. Moreover, with this number of participants, it would be possible to discriminate between high and low correlations in measuring correlations [21]. Only 400 participants were accepted to be involved in this study, however, 45 of them were ineligible due to incomplete responses. Therefore, only 355 participants were selected for this study. Thirty participants of them were randomly selected for the pilot study in face validation step of the translated questionnaire and not included in the final analysis. Moreover, 25 participants from the sample population were randomly selected for test- retest within 1–2 weeks.

All participants completed the structured questionnaires including OSES-A. The original OSES is in English language and designed to assess self-efficacy of behaviors toward OP [22, 23]. The OSES is a twelve items rated by individual on a 100 mm visual analogue scale to assess the confidence in performing osteoporosis preventive behaviors. The OSES has two subscales namely: OSES-Exercise and OSES-Calcium. The possible total score range from 0 to 1200 with each subscale range score from 0 to 600. A cut-off point (858) was used to categorize the osteoporosis self-efficacy scores into two levels: low and high OSES-A levels [24].

Beside OSES-A, unpublished but valid and reliable Osteoporosis Knowledge Tool (OKT-A) and Osteoporosis health belief Scale (OHBS-A) Arabic version tools, were administered before OSES-A, respectively, according to the developer instructions. Also, the original OKT and OHBS are in English language and designed to assess OP knowledge and health beliefs about developing osteoporosis, respectively [23, 25]. The OKT is 24 multiple-choice items regarding risk factors and its prevention. From 0 to 24 is the possible score range and the highest value indicate the highest level knowledge score. While, the OHBS consist of 42- likert type scale (1 = strongly disagree, 5 = strongly agree). The total possible score range from 42 to 210. Cut-off points (14 and 169) were used to categorize OKT-A and OHBS-A scores into two levels: low and high, respectively [26, 27].

Forward–backward-forward translation method was used to translate the questionnaire from English into Arabic according to translation international guidelines including forward translation, reconciliation, reverse translation, debriefing [28-30]. The translation process was conducted by two independent, expert translators. Thereafter, an expert panel of eight clinical pharmacists and the researcher reviewed the Arabic version for reconciliation. Then, back-translation of the reconciled version was carried out by another two independent expert translators. Subsequently, discussions between the expert panel, translators, and the researcher were held to resolve any inconsistencies and check the capacity of the items to measure the construct that it proposes to measure then a final version was decided (face validity process) [31].

Finally, a pilot study was conducted by distributing the questionnaire to 30 participants and the questionnaire was modified according to their feedback after discussions between the expert panel, and the researcher. Those 30 participants were excluded from the final study outcome and analysis.

Exploratory Factor Analysis (EFA) was examined to found the factor structures of OSES-A. A principal axis factoring method for extraction with direct Oblimin (Oblique) rotations was used for EFA. The criteria for EFA were: factor loading greater than 0.40, Kaiser–Meyer–Olkin (KMO) value (> 0.5), Bartlett’s test of sphericity (significant level < 0.05). The number of factors retain were depend on: Kaiser’s criterion (eigenvalue ≥ 1.0) and theoretical meaning of the rotated factors [32].

Reliability with a minimum acceptable criterion above 0.5 was applied to measure the consistency of a measurement item [33]. The internal consistency was evaluated using Cronbach’s alpha and corrected item total correlations between the scales and their corresponding items (correlation of < 0.20 is considered poor). Pearson’s correlation coefficient was used to evaluate test–retest reliability [19].

The statistical analysis of the validation processes included assessing construct validity and reliability (Cronbach’s alpha and test-retest). Descriptive statistics, percentages, and frequencies were used as appropriate. The chi square (χ²) test was employed for categorical variables to find any association, whereas for continuous data, Mann–Whitney U and Kruskal–Wallis tests were used to evaluate the differences between the groups when required. In addition, logistic regression analysis using backward method was used to identify the factors affecting OSES-A. Predictive Analytics Software (PASW) version 19.0 was used to analyze data in this study and significance level was set at P value <0.05.

The age of the participants was a range between 18 to 87 years with an average of 41.82±12.452 years. Nearly 46% of respondents were male. About 79% of the respondents had educational level more than 12 years. About 23% of the respondents were single and around 38% had monthly income less than 500,000 Iraq Dinar (IQD; 1 US dollar is equivalent to 1,250 IQD). By employing the recommended scoring method, the mean scores (M±SD) of the OSES-A was 658.43±222.014 which considered low.

Table 1 shows the distribution of the two levels of osteoporosis self-efficacy and the demographic data results. Only 16.67% of the study population was found to have high OSES-A level. The results showed significant differences between the following independent variables in relation to total OSES-A scores: age, gender, marital status, family history of osteoporosis, self-reported osteoporosis and osteoporosis diagnosis or screening. In addition, there were significant associations between ages, gender, and self-reported osteoporosis with OSES-A levels (Table 1). Furthermore, the results revealed low self-efficacy in all dimensions (less than 60%) with the lowest value appeared in the exercise subscales (51.03%).

Correlations were performed to determine the relationship between total OKT-A and OHBS-A with total OSES-A. There were positive correlations between total OSES-A score with total OKT-A score, r=0.274, and OHBS-A, r=0.238 (all Ps<0.01). The binary logistic regression revealed that OKT-A (categorical) and OHBS-A (categorical), age (continuous) and gender were predictors for total OSES-A (Table 4) and the model explain about 84.70% of the dependent variable.