Assessing health literacy in the eastern and middle-eastern cultures

An expert panel comprised of a purposive sample of 11 individuals, representing the fields of medical education (n = 2), public health (n = 3), and clinical research (n = 1) as well as local community members (n = 5), was convened to examine relevant domains in the assessment of health literacy in Eastern and Middle Eastern settings, with the explicit intent of minimizing the influence of social desirability by the survey participants in this cultural context. The experts were fluent in English and at least one other local language. Delphi technique, a group consensus gathering methodology, was used to reach agreement among the panel members [13]. After a comprehensive review of the literature pertaining to health literacy short screening tools, the content areas for inclusion in the questionnaire were primarily drawn from the Health Literacy Screening Brief Questions [14] to generate a conceptual framework relevant for Eastern and Middle Eastern cultures (Fig. 1).

A focus group (N = 10), comprised of a convenience sample of students (n = 2), patients (n = 3), healthcare providers (n = 1) and members of the local community (n = 4), was conducted to provide feedback on the content and readability ease of the survey questions proposed by the expert panel [15]. Participants included both men and women between the ages of 18–50 years, who were able to understand, read and write English and at least 1 other local language. Participants especially from the community who were unable to consent were excluded. The Content Validity Index (CVI), a rating that the content is valid and consistent with the desired outcomes, was applied [16]. Accordingly, each member of the focus group was independently advised to score each item of the survey generated by the expert panel between 0–100 points based on relevance and importance to the local culture and setting, with higher scores representing greater relevance and importance of the survey item. Items obtaining a mean score of 80 % ormore were accepted as survey questions [16].

Since the populations of the Middle East and Asia are predominantly non-native English speaking, a readability test was then applied to all resulting survey questions. The Flesch-Kincaid scale was chosen because of its convenience for computerized use, excellent reproducibility and high correlation with other established readability scales [17]. The Flesch-Kincaid reading ease determines comprehension difficulty of written text on a scale of 0–100, with higher scores indicating that the material is easier to read. The Flesch-Kincaid grade level formula translates the score to a grade level, indicating the number of years of education generally required to understand a given text [17]. Question difficulty was included in the item development to ensure that the final items could distinguish between low, moderate and high levels of health literacy [18].

The revised version, containing the survey questions resulting from the focus group feedback, was used for the field pre-test. The pre-test was conducted using a purposeful sampling method, with the goal of including an equal distribution of participants with respect to age, gender and formal education. Patients from outpatient clinics from 2 major hospitals in the emirate of Abu Dhabi and 2 city hospitals in Mumbai, India were recruited for the study. Given the fact the study was conducted as the government-owned public hospitals in the UAE and India, a cross-section of urban and rural population participated in the study. Pilot field pre-testing involved face-to-face interviews between patient participants fromAbu Dhabi and Mumbai and the multilingual physician researchers. All respondents provided verbal consent prior to participation. A 5-point likert scale was used to assess participant responses [14, 18]. The response options included “1 = never, 2 = rarely, 3 = sometimes, 4 = most of the time and 5 = always” and the mean score for all item responses ranged from a minimumof 13 to a maximum of 65 for each participant. The scores distinguished between three levels of health literacy: 1–26 for “inadequate,”27-39 for “marginal” and 40–65 for “adequate” [19]. The researchers also used an electronic tablet device to display pictures of hospital signs, appointment slips, medical insurance forms, informed consent forms, English and local language newspaper reports, and local currencies for the purpose of identification and to assess comprehension.

The final questionnaire EMAHL13 was translated from the original English version by certified health translators into the most prevalent languages in the UAE, namely Arabic, Hindi, Urdu, Tagalog and Malayalam. The surveys were then back-translated to English, in order to ensure that the structure, content and intent of the survey-items did not alter during translation. Although physician-researchers conducted the face-to-face interviews with the participants, the survey-items were translated into local languages to optimize participant comprehension.

Data was analysed using SPSS Statistical Software Version 20 (SPSS Inc. Chicago, USA) [20]. The domains were identified using factor analysis [20]. The widely used Principal Component Analysis (PCA) was used as the extraction method to undertake factor analysis; and Varimax rotation was used to rotate the factors to better fit the data [20, 21]. Convergent validity to assess if the survey items converged to measure a construct was also conducted using the Correlation coefficient matrix method [20]. The percentage of total variance by each factor was calculated and pattern matrix was used to identify the domains. Kiaser-Meyer-Olkin (KMO) sampling adequacy and Bartlett’s tests (to assess the strength of the relationship among the variables) were also applied to the EMAHL13 [20, 21] . The reliability of the inventory and its subscales were tabulated using Cronbach’s alpha [20].

Characteristics of age, gender, nationalities, education and professional status of the members of the expert panel were equally distributed. Approximately half (6/11) of the expert panel members were women. Forty-five percent (5/11) of the panel was composed of community members, 27 % were public health professionals, and 9 % were healthcare providers. The average age of the participants was 37 ± 3.9 years, with 45 % having education levels at the high school level or below. The percentage of Asians and Arabs was almost equally distributed at 40 % each (data not shown).

The expert panel accepted 3 of the 5 functional domains from Chew et. al. Health Literacy Screening Brief Questions [14], specifically navigating the health care system, filling-out medical forms, and reading and understanding patient information materials. A fourth domain, differentiating and demonstrating medications and publication, consisted of non-health related items, such as the ability to read newspapers. Fifteen items were adapted from the 3 domains of the Health Literacy Screening Brief Questions, an additional 4 questions were derived from non-health related and medication domains, culminating in a 19-item culturally relevant health literacy survey (Table 1). Items from both the reading and understanding of patient materials domain (31.5 %) and medication instruction /non-health related domain (31.5 %) constituted the majority of the total items (Table 1). This was followed by 21 % of items from the filling-out medical forms domain. Only 16 % of the items were selected from the navigating through the health system domain (Table 1).

Sixty-percent of focus group participants were women, with an average age of 32.6 ± 3.8 (mean ± standard deviation). The average age of the male focus group participants was 37.8 ± 7.3. Sixty percent of the participants had a baccalaureate degree and were employed (data not shown). The focus group session provided feedback on the structure, content and clarity of the survey questionnaire. The questions were formulated as “you can” taking cues from the common local communication pattern in Eastern and Middle Eastern societies. The focus group participants reviewed the 19 semi-structured, open-ended items proposed by the expert panel and chose to include or exclude survey items based on their relevance and importance to the local population. Table 1 indicates the content validity index score for relevance and importance of the local setting. Survey items that obtained average scores less than 80 % were thereby removed from the questionnaire. Removed items included completing health insurance forms (66.5 %) and patient registration forms (64.1 %), reading pharmacy prescriptions (61.7 %) and medical supply store location (67.7 %), understanding consumer product information (70.7 %), and understanding diagnosis instructions (66.7 %). Structural revisions to avoid repetitive or duplication of the survey questions were also addressed by the focus group. The Flesch-Kincaid readability ease (FRES) testindicated a score of 61 and Flesch-Kincaid Grade Level (FGLS) of 7. It has been demonstrated that a reading score of 60 indicates standard readability ease [17].

The fieldtest participants were primarily women in both India (315/450, 70 %) and the UAE (272/450, 60 %). Compared to 66 % in the UAE, 80 % of the participants in India were employed. Quality assessment of the survey included survey length, completion time, understanding demographics, layout and scoring ease. There were no significant differences between the quality assessments of the survey instrument by the participants from either country (Fig. 2). The items related to the patient registration form, consumer product advertisement, medical store, pharmacy prescription, health claim form and diagnosis instructions were removed from the questionnaire in order to avoid duplication, as well as for their limited ability to directly assess health literacy (Table 2). The final health literacy construct (EMAHL13) is shown in Table 2. The average time required to complete the EMAHL13 survey was 11 min for participants from India and 13 min by UAE participants.

Validity and Reliability of the construct: The consistency of the health literacy scale (EMAHL13) was measured using the principal component analysis, obtaining a Kaiser-Meyer-Olkin value of 0.92, reaching statistical significance. Bartlett’s test of sphericity was significant (0.000). The principal component analysis revealed the presence of a component with an eigen value of 11.1, explaining 85.2 % of total variance, correlation coeeficient indicating relationship between the items are shown in Table 3. Cronbach’s α reliability assessment of each of the four domains ranged between 0.77 and 0.99 (Table 3) . The overall Cronbach’s α reliability score for all 13 items for the final construct was high at 0.85.