The Science of Diabetes Self-Management and Care2023, Vol. 49(6) 449 –461© The Author(s) 2023Article reuse guidelines:sagepub.com/journals-permissionsDOI: 10.1177/26350106231207349journals.sagepub.com/home/tde
AbstractPurpose: The purpose of this study was to validate the Literacy Assessment for Diabetes (LAD), the Diabetes Numeracy Test (DNT), and the Simplified Diabetes Knowledge Test (DKT) in the Arabic language and context.
Methods: Three hundred eighty four, ≥18-year-old patients with type 1, type 2, or gestational diabetes mellitus were recruited from 3 endocrinology clinics in the United Arab Emirates. Exploratory factor analysis using principal component was performed. Achieved scores were compared using Pearson bivariate correlation.
Results: All 60 LAD items loaded on 1 factor accounting for 66.7% of the variance, with internal consistency α = .991. Average score = 68.7%. Nineteen out of 26 items were retained on the DNT and grouped into 4 factors, prescription reading and directions, proper dose coverage, nutrition, and insulin, with good internal consistency (α = .721). Average score = 73.2%. All 20 DKT items loaded on 3 factors accounting for 41.2% of the variance, causes and consequences of the high blood sugar level, prevention of the disease, and misconceptions about diabetes self-management, with good internal consistency (α = .799). Average score = 71.9%. A moderate and significant correlation between the DKT and DNT (r = .56, P < .001) was observed.
Conclusions: Three tools to assess diabetes literacy, numeracy, and knowledge were psychometrically tested to establish their validity and reliability in the Arabic language and context. The tools could be used to assess patient skills and competence in navigating the health care system and managing their diabetes.
Health literacy has been defined as “the degree to which individuals have the capacity to obtain, process, and understand basic health information and services needed to make appropriate health decisions.”1 Low health literacy was found to be linked to poor patient disease knowledge, insufficient self-management skills, less adherence to disease treatment, high risk of hospitalization, and higher health care costs.2-4 Research has shown that patients with low health literacy may not fully understand medical information or instruction.5 Findings about available health and disease information presented to patients show that the information is developed at a higher level than what patients could understand, which could affect the effectiveness of the information.6,7 Numeracy is part of health literacy and has been defined as “the degree to which individuals have the capacity to access, process, interpret, communicate, and act on numerical, quantitative, graphical, biostatistical, and probabilistic health information needed to make effective health decisions.”8 Numeracy affects perceptions of health risks and benefits, which could impact an individual’s decision-making, health behaviors, and outcomes.8 For example, poor numeracy skills in a global disease like diabetes9 could lead to difficulty understanding glucose readings, calculating carbohydrate consumption, adjusting medications, and performing self-management activities.6 Consequently, health literacy, including numeracy, should be assessed to identify those at low levels, which might be a significant barrier to patients’ understanding of their health needs and to their achievement of optimum health outcomes. Furthermore, assessing individual health literacy and numeracy is vital before interventional and educational health programs are designed so they are tailored to patients’ specific needs.
Similarly, knowledge of patients with diabetes in matters such as the effect of diet on diabetes control or hypoglycemia symptoms and their treatment is essential for effective self-management behaviors and better glycemic control.5 Diabetes knowledge deficits are a global problem that are associated with higher A1C and negative metabolic control among patients with diabetes.10,11 In addition, poor diabetes-related knowledge was associated with less awareness about diabetes complications and medication nonadherence.10,12 Therefore, knowledge of patients with diabetes needs to be assessed for educational and interventional programs to be effective and to avoid wasting effort and resources.10 Tailoring educational material to suit patients’ varying needs of education is recommended to achieve a more personalized and comprehensive approach in an individual patient diabetes management.6
Several instruments have been developed to assess health literacy, numeracy, and knowledge of diabetes. Three widely used instruments are the Literacy Assessment of Diabetes (LAD),13 the Diabetes Numeracy Test (DNT),14 and the Simplified Diabetes Knowledge Test (DKT).15 The LAD was designed to measure an individual’s ability to pronounce 60 words that are frequently used during clinic visits or when diabetes self-care instructions are given.13 The 60 words are arranged on 3 separate lists in ascending difficulty, considering the length, number of syllables, and frequency of use of the word. The DNT, on the other hand, was developed to measure patients’ numeracy required to perform daily self-management skills in diabetes.14 The original DNT consisted of 43 items covering 5 self-management domains: nutrition, exercise, glucose monitoring, oral medications, and insulin use. The Simplified DKT consisted of a 20-item tool developed to assess general knowledge of diabetes among adults with diabetes (types 1 and 2), including knowledge on diet, blood glucose control, exercise, medications, complications of diabetes, and insulin use. A “true/false/I don’t know” scale was adopted in the latest version of the tool compared to the original multiple-choice version, and the total score equaled the sum of all responses that a participant answered correctly.15
Because the current prevalence of diabetes is estimated at around 9.2% in adults who live in the Middle East and North Africa region,9 it is important to make these tools available to assess literacy and knowledge among this population. Limited published studies have undertaken an assessment of diabetes literacy in Arab countries;7 the original LAD and DNT have not been culturally adapted or validated for use in the Arabic language and context. Furthermore, although some research has been conducted on the translation and cultural adaptation of the various versions of the DKT,10-12 the Simplified DKT has not been psychometrically tested in this context. To address these shortcomings, recent work on the tools was carried out to translate and culturally adapt them to the Arabic language and context16 following the International society for Pharmaeconomics and Outcomes Research (ISPOR) guidelines for translation and cultural adaptation of patient-reported outcomes.17 Nevertheless, these culturally adapted tools have not undergone validation and psychometric evaluation. Therefore, the aim of this study was to continue the work on these culturally adapted tools (LAD, DNT, and Simplified DKT) by providing further validation and evaluation of their psychometric properties in a sample of Arabic-speaking adults with diabetes in the United Arab Emirates (UAE). This will help establish the suitability and reliability of the tools for use in the clinical and research settings.
Institutional review board approval was obtained before the conduct of the study (Reference No. F-H-18-03-02).
In following the ISPOR guidelines for translation and cultural adaptation of patient-reported outcomes,17 the tools were initially forward translated to Arabic by 2 independent professional translators, which led to producing 2 Arabic translations of the tools; the 2 translations were reconciled by the research team to produce 1 version. A third translator translated the reconciled Arabic version back to English. This new English version was compared to the original tool to resolve any differences. Cognitive interviews were conducted on a sample of 10 participants to establish the face and content validity of the tools and helped to further refine and finalize them. All members of the research team were bilingual of Arabic and English, hence, they worked together to revise all versions of the instruments by checking meanings and refining minor differences.16 The following sections illustrate how the outcomes from the translation and cultural adaptation process were used in this study.
As an outcome of the translation and cultural adaptation of the LAD, all items were retained in the translation but were rearranged to accommodate alphabet-based listing traditionally used in in Arabic language as opposed to the original listing of items in ascending difficulty.16 The adapted LAD was used in this study, where the same instructions of the original English LAD addressing participants were used to coach participants: “Would you please read aloud as many words as you can from the lists? Start with the first word on List 1. When you come to a word you cannot read, try your best or say ‘Pass’ and go on to the next word.” If the participant took longer than 5 seconds to read a word, they were asked to skip that word and move to the next one on the list. As per the original instructions, the LAD was scored by marking a plus (+) sign next to each correctly pronounced word and a minus (−) sign next to each mispronounced or not attempted word. Total scores represented the sum of all correctly pronounced words; these were then used to assign a reading grade level based on a conversion table present on the scoring sheet within the tool. Scores between 0 and 20 were considered fourth-grade level or below, scores between 21 and 40 were considered fifth- through ninth-grade levels, and scores between 41 and 60 were considered higher than ninth-grade level.13
The translation and cultural adaptation of the DNT led to reduction of the number of items to 26 from the original 43 covering the 5 diabetes care domains, eliminating items that were of similar mathematical concepts within each individual domain, new items added to include newly marketed medications not included in the original version, and images of medications added within the tool to help participants match the medications with their names.16 Assessments of numeracy in the DNT covered general mathematical skills of addition, subtraction, multiplication, division, fractions, decimals, numerical hierarchy, and multistep calculations.14 Items were scored as binary outcomes (correct or incorrect), and the scores were calculated as the sum of all correct answers for each individual participant. In the culturally adapted version, 3 items focused on nutrition such as label interpretation and carbohydrate counting, 1 item related to exercise (highlighting insulin dosing adjustments required based on physical activity), 1 item covered blood glucose monitoring, 15 items assessed patient handling of oral medication use and refill, and last, 5 items focused on insulin use and dosing adjustments. The original general mathematical skills were retained in the adapted version.
Culture adaptation of items in the translated DKT required changing some terms and replacing others to preserve the intended meaning in the Arabic context, but all 20 items were retained.16 The Simplified DKT tool contained 20 items in a true/false format, 18 of which pertained to general knowledge of diabetes, and the remaining 2 pertained to insulin use. Total scores were calculated as the percentage of correct responses achieved by the participants. The tool was considered easy to read because its reading level was set at the sixth grade.
According to Krejcie and Morgan,18 384 completed questionnaires were needed to have an estimate of precision at the 95% confidence level with α = .05. In the literature, there are varying opinions on the appropriate sample size for factor analysis; some researchers argue that 300 cases are appropriate, whereas others prefer to use a minimum ratio of 5 to 10 participants per questionnaire item to be analyzed, up to a total of 300.19 The 400 cases retained for analysis in this study met both sample size criteria.
Participants who were 18 years and older, diagnosed with type 1 or type 2 diabetes mellitus or gestational diabetes, and able to speak Arabic were included in the study. Patients who had mental illness, signs of cognitive impairment, or vision or hearing loss were excluded. Permission to recruit participants was obtained in the endocrinology clinics in 2 medical centers and at 1 tertiary care teaching hospital specialized in providing advanced diabetes care in the UAE. Potential participants were approached while waiting for their medical appointments in the endocrinology clinics, informed of the nature of the study, and invited to participate. An explanatory statement was given to them detailing the aim of the study and the type of the data to be collected. Agreeing participants provided their written consent and completed the questionnaires using printed copies of the tools with the help of 2 trained research assistants, who provided instruction on completing the LAD, DNT, and DKT tools without giving any interpretation or explanation. Alternatively, if a participant elected, they were interviewed at a place and a time of their preference outside of their medical appointments. The research assistants also noted the time each participant took to complete the tools.
Other data collected included participants’ sociodemographic characteristics, such as age, gender, nationality, marital status, the highest level of attained education, annual earned income, and diabetes type.
Descriptive statistics were used to describe participants’ characteristics, their diabetes-related information, and LAD, DNT, and DKT scores. Prior to conducting factor analysis, the suitability of the data for factor analysis was assessed in terms of adequacy of the sample size. Principal component analysis was used as a method of exploratory factor analysis20 with varimax rotation of factors that had eigenvalues higher than 1. Items with factor loadings 0.40 or greater were considered “significant,” and loadings of 0.50 or greater were considered “very significant.” To retain an item on a scale, the factor loading of the item should be higher than 0.20 with no higher loading on another factor.21,22 Reliability was assessed using Cronbach’s alpha to measure the internal consistency of the factor, with preferred values between .70 and .90.23 Factor scores were computed by adding the scores for the individual items constituting the factor and dividing by the number of items. For the descriptive results of each instrument, the mean and standard deviation of the proportions of correct answers were calculated.
Additionally, convergent validity was assessed with the hypothesis that a positive correlation would be expected between literacy and numeracy, literacy and knowledge, and numeracy and knowledge in diabetes. Thus, Pearson bivariate correlation coefficient was used to measure possible correlations between scores on the LAD, DNT, and DKT. Analyses were conducted using SPSS V.28.0 software package (IBM Corp). The statistical significance was defined at .05.
A convenience sample of 400 participants was recruited at a response rate of 83%, of whom 60.8% were females. The mean age of the participants was 40.9 (SD = 15.1) years. Most of the participants were married (68.0%); originally from Syria (23.0%), Jordan (14.2%), or Egypt (13.3%); and had a college degree (64.3%; Table 1). More than half of the participants had type 2 diabetes (56.8%), followed by type 1 diabetes (34.5%) and gestational diabetes (8.8%).
The scree plot is a common method used to help decision-making about the number of factors to retain in an exploratory factor analysis. It is a graphic representation of the number of factors proposed. To analyze it, we used the “elbow method” (ie, the “elbow” of the graph suggests retaining all components before the curve flattens out). The analysis of the scree plot indicated a possible organization of the LAD items in more than 1 component (Supplemental Figure 1a). However, the exploratory factor analysis revealed that the majority of the items had higher factor loadings in the first dimension, suggesting the presence of 1 factor (Table 2), which accounted for 66.7% of the variance. The Kaiser-Meyer-Olkin test for sampling adequacy was 0.964, and Bartlett’s test of sphericity was very significant (χ2 = 47 968.112, df = 1170, P < .001), showing that the LAD data were adequate for factor analysis.
The LAD test showed excellent internal consistency (α = .991).
The LAD required 7 minutes, on average, to complete. Correct participant responses on LAD items are presented in Table 2. The average participant performance score was 41.2 (SD = 21.9; maximum = 60, percentage average = 68.7%). At most, half of the participants recognized words such as cholesterol, glycogen, nephropathy, ketones, ketoacidosis, and triglycerides; only 60% recognized the word hypoglycemia. The majority of participants, 248 (62%), were at or above ninth-grade level; 57 (14.2%) were between fifth-grade and ninth-grade levels; and 95 (23.8%) were at or below fourth-grade level according to achieved scores on the LAD.
The factor analysis extracted 4 factors with an initial cumulative explained variance of 39.2% (Supplemental Figure 1b). Items with low loadings (<0.4) and those that were not theoretically meaningful to the factor were removed, leaving 19 out of 26 items to be retained. Varimax rotation showed strong loadings (>0.40) of each item on 1 factor (Table 3), except for Items 11 and 20, which were acceptable (≈0.3). The research team elected to keep the 2 items due to their theoretical fit with other items within the factor. The scale with 19 items explained 47.8% of the variance. Accordingly, the 19 items were grouped into 4 new factors: prescription reading and directions (Items 7, 12, 13, 15, and 17), proper dose coverage (Items 5, 6, 8-11, 18, and 20), nutrition (Items 1-3), and insulin (Items 21-23), still covering the initially proposed 5 self-management domains of nutrition, exercise, glucose monitoring, oral medications, and insulin use; Table 3).14 The Kaiser-Meyer-Olkin measure of sampling adequacy for factor analysis was 0.712, and Bartlett’s test of sphericity was very significant (χ2 = 2549.164, df = 325, P < .001), indicating that the DNT data were suitable for factor analysis.
DNT showed a good internal consistency for the total scale (α = .721). Cronbach’s alphas for the factors varied between α = .444 (Factor 4) and α = .739 (Factor 1; Table 3).
On average, participants completed the DNT tool in 25 minutes. The average participant performance score on the newly developed DNT was 13.9 (SD = 3.2; maximum = 19, average percentage = 73.2%). The proportion of correct participant responses on DNT items are presented in Table 3. Participant performance scores on DNT factors and the various mathematical skills (organized according to the original scale) are presented in Table 4.
The exploratory factor analysis of the DKT showed an unforced 5-factor structure, with eigenvalues higher than 1, explaining 52.2% of the variance. However, only 1 item (Item 8) saturated on 1 of the suggested factors, and this solution revealed not to be theoretically meaningful. Therefore, we tested the solutions of forcing 4-factor and 3-factor structures. On consideration of the scree plot (Supplemental Figure 1c) and the theoretical criterion, the organization of the items in 3 factors was considered most acceptable to the data, accounting for a cumulative variance of 41.2%. As is shown in Table 5, the factor structure for the 3 factors was clear; 17 out of the 20 items revealed strong factor loadings higher than 0.4, and all item loadings ranged from 0.332 to 0.791. Therefore, the items were aggregated on the factor on which they presented higher loading. The theoretical criterion was used in cases where factor loadings were similar on 2 factors. Thus, the authors opted to include Item 14 in Factor 1 and Item 18 in Factor 2. The factors were labeled after careful analysis of the content of the included items by the research team. Factor 1 included 9 items (2, 6, 11, 12, 14-16, 19, and 20) and assessed knowledge on the causes and consequences of the high blood sugar level, Factor 2 evaluated knowledge on prevention of the disease and included 6 items (1, 8, 9, 13, 17, and 18), and Factor 3 assessed misconceptions about diabetes self-management and was composed of 5 items (3, 4, 5, 7, and 10). Sampling adequacy, measured by the Kaiser-Meyer-Olkin, was 0.844, and Bartlett’s test of sphericity was very significant (χ2 = 1877.669, df = 190, P < .001), demonstrating that the DKT data were adequately suited for factor analysis.
The internal consistency of the total scale was good (α = .799), which was comparable to that of the original tool (α = .71) The reliability of the factors varied from α = .547 in Factor 2 to α = .793 in Factor 1 (Table 5).
The time to complete the DKT was between 10 minutes and 15 minutes. The average participant performance score on the DKT was 14.3 (SD = 3.6; maximum = 20). Participants replied correctly to the majority of the DKT items (71.9% average score), varying from 26.8% for Item 7, about knowledge on feet health (“Wearing shoes a size bigger than usual helps prevent foot ulcers”), to 97.8% for Item 17, about knowledge on prevention of diabetes (“Having regular check-ups with your doctor can help identify the early signs of diabetes complications”). On average, participants presented higher proportion of correct answers in the factor prevention of the disease (80.7%), followed by the factors on misconceptions about diabetes self-management (69.2%) and the causes and consequences of the high blood sugar level (67.5%).
The bivariate Pearson correlation test revealed a moderate, positive, and significant correlation between the DKT and the DNT (r = .56, P < .001), thus more knowledge on diabetes was correlated with better diabetes numeracy skills (Table 6). No significant correlations were observed between the LAD and the DNT or DKT scores.
The current study provided psychometric evaluations of the Arabic version of 3 tools (LAD, DNT, and DKT) to assess literacy, numeracy, and knowledge about diabetes. Previously, the rigorous translation and cultural adaptation process undertaken to develop the tools was described.17 The psychometric analysis in this study showed that the Arabic versions of LAD, DNT, and DKT were valid and reliable instruments to be used during medical encounters in Arabic-speaking communities with diabetes. Assessing literacy, numeracy, and knowledge related to a disease is an essential yet often a neglected element of health care provision.24 A large body of evidence shows that patients with lower literacy, numeracy, and knowledge about their disease experience poorer health outcomes.25,26 The psychometric validation of the tools should enhance the capacity and resources for researchers and health care professionals to accurately assess health literacy, numeracy skills, and knowledge level of concerned individuals, which should enable tailoring health care delivery to people with diabetes. Within this context, several diabetes health organizations highlighted that relying on a generic and broad treatment plan is ineffective.27
In literacy assessment, LAD showed an excellent internal consistency (α = .991), higher than that of the original instrument (α = .860),13 and all items had higher factor loadings, which testified to the scale’s reliability. The participants correctly recognized more than half of the words presented (M = 41.2, SD = 21.9). However, this value seemed lower than what has been reported in similar studies where high levels of diabetes literacy reading were found.28,29 The LAD contained a mix of lay and medical vocabulary and was designed intentionally with half of the items being lower than the fourth-grade level.13 Although most of the participants (85%) in the sample had an educational level of at least 10th grade, 14.2% were positioned between fifth and ninth grades, and 23.8% were positioned lower than or equal to fourth-grade level. This result indicated that more than one-third of the participants would still have difficulty recognizing common vocabulary or language used during their medical appointments because most medical literature is written at higher levels than sixth grade.30 This was obvious, for example, because many participants could not recognize keywords associated with diabetes, such as cholesterol, triglycerides, or hypoglycemia. Hence, they would be eligible to receive low-level instruction including explanation of these specific words. Like the sample in the original tool (average age was 41), our sample was young (a mean age of 40), but unlike the original tool, it mostly constituted of females. These results seemed to be contrary to what could be expected considering that higher literacy levels have been reported among females.29 Concerning age, the literature reported lower levels of health literacy more commonly found among older individuals.31 Further application of the tool should study participant-specific factors that may be associated with literacy levels. The tool is valuable for use in the clinical setting because it is quick and nonstigmatizing. Patients should be reassured that they are not expected to know all the words in the tool but it would be most useful if they knew them and that after taking the test, they would be given written or audio instruction that is suitable for them based on the assessment.13
Concerning numeracy, the original DNT comprised 43 items;14 26 items were retained after the previous translation and cultural adaptation process,17 and those were tested here. Based on the exploratory factor analysis, a shorter tool with 19 items grouped into 4 factors was proposed. It is important to note that the retained items preserved all the mathematical and numerical skills assessed by the original instrument,14 allowing to infer the different skills included in numeracy assessments. Moreover, the diabetes care domains in the original tool were also retained after factor analysis, although organized into 4 factors. The suggested regrouping of items after factor analysis was helpful in 2 ways: data reduction of items into 4 factors leading to a more focused analysis of the findings and more stable results in future use of the instrument because the items have been psychometrically grouped.
The 19-item DNT revealed a Cronbach’s internal consistency coefficient of .721, which was considered good reliability, although lower than the original instrument (α = .900). For the individual factors, Cronbach’s alpha for each of the extracted factors ranged from acceptable (Cronbach’s values did not go above .7) to good (Cronbach’s values ≥.7). Kehoe20 suggested that values lower than .7 were satisfactory for short tests, indicating that the items within each factor were internally consistent. Other validity tests included the corrected item-scale correlations, which were all significant, indicating homogeneity of the items within the factors. In addition, there were differences in the mean scores in different mathematical skills, which indicated the ability of the current tool to distinguish the levels of numeracy. In principle, the DNT retained the items that discriminated against diabetes-related numeracy skills for a diabetes educator or clinician to focus on during discussions with the patient.
The mean score in the DNT was reasonable (M = 13.9, SD = 3.2; average percentage of correct responses = 73.2%) and slightly higher than the scores found in the original instrument (61%),14 possibly due to the relatively highly educated sample in this study. As reported earlier,14 this research recognized several likely problems for patients while navigating health care for diabetes, including nutrition-related areas such as food label interpretation and carbohydrate calculations, date of when to refill medications, understanding of insulin amounts on blood glucose level, and understanding of medication titration instructions. It is important to note that as diabetes progresses, the complexity of the management regimen for a particular patient may also progress; therefore, patients need continuing education and support to adequately treat their diabetes. A diabetes educator or clinician may use the DNT to help target their education and support materials to the individual need.
The psychometric analysis of the DKT revealed a robust instrument that kept the 20 items from the cultural adaptation, with good internal consistency of the scale (α = .784), like that of the original instrument (α = .71).15 The original instrument proposed an organization in 2 factors: general knowledge of diabetes and insulin use. The exploratory factor analysis in this study proposed an organization of the items in 3 factors, which was believed to provide more detailed information about different dimensions of knowledge of diabetes (causes and consequences of the high blood sugar level, prevention of the disease, and misconceptions about self-management). This proposal allowed to identify areas in this sample where more information is needed, specifically the causes and consequences of high blood sugar levels. Participants replied correctly to most of the DKT items (71.8% correct answers in the total sample) in comparison to 65% average score on the original version of the tool. These results were consistent with those reported in another validation study in the USA32 but higher than a previous study conducted in the UAE that showed 55% correct responses using the Michigan Diabetes Knowledge Test, a tool that was structured as questions with single correct answers where 3 or 4 answer options were given.33 The higher level of knowledge found in this study may be explained by the younger age and the higher educational level of the sample, typically associated with better knowledge of diabetes.34 Here, only 2.5% of the participants never attended school, compared to 23% of the participants with no schooling in that study.33 Additionally, a true/false answer format (the DKT) has been associated with better participant performance in assessments of knowledge of diabetes.15 This simplified version of the original tool reduced the difficulty associated with multiple-choice formats found in the revised version of the Diabetes Knowledge Scale15 and in its assessment of patient knowledge of diabetes, would consequently help in the design of targeted interventions that fit the needs of individual patients and possibly communities with similar characteristics. Because knowledge about diabetes and self-care activities evolve continually, there may be a need to further update this tool in the future to capture new information relevant to diabetes education.
As hypothesized, convergent validity of knowledge of diabetes was confirmed with higher numeracy skills; hence, assessing 1 domain of diabetes care could inform the likelihood of mastery or difficulty with the other. However, as observed in previous studies,35 no correlations were found between the LAD scores and the DNT or DKT scores; this indicated that literacy and numeracy were different constructs, and thus, they should be assessed and intervened separately.
Three tools of diabetes literacy, numeracy, and knowledge have been psychometrically tested in an Arabicspeaking population and have offered further validation to the originally developed tools by testing them in another cultural context than the one they were originally developed. Future application of the tools should focus on patient-specific factors that may be associated with levels of literacy, numeracy, and knowledge of diabetes. This should further indicate how participants would perform on these measures on the one hand and guide efforts to either modify or address these patient factors.
The data used in the psychometric evaluation of the tools depended on self-report by the participants, hence there might have been elements of bias or forgetfulness that were not accounted for. This study employed convenience sampling in recruitment of participants, which could indicate that those who did not participate in the study might have had different literacy and numeracy skills and knowledge than those that participated. Most participants in this study were women who were college graduates; this could have affected the achieved scores. However, these issues were minimized by the large, diverse sample of several represented nationalities and the wide spectrum of age groups.
Three tools to assess diabetes literacy, numeracy, and knowledge have been psychometrically tested to establish their validity and reliability in the Arabic language and context using standard methods of factor analysis. All items were retained in the LAD and loaded on 1 factor. Nineteen items (out of 26) on the DNT retained all mathematical skills and diabetes care domains in the original tool. All 20 items were retained on the DKT but distributed over a higher number of factors in comparison to the original tool. Diabetes literacy, numeracy, and knowledge scores were suboptimal in key areas of diabetes care despite high participant education, indicating these were separate constructs that needed to be tested regardless of literacy. Numeracy was correlated with knowledge of diabetes but not with literacy. The tools will provide valuable contribution to the assessment of diabetes literacy, numeracy, and knowledge in Arabic-speaking countries, and the methods used to validate them offer further strengthening of the originally developed tools.
Acknowledgments go to the physicians and staff at the Ajman Specialty Hospital, Emirates European Hospital, and Thumbay Hospital for allowing access to the patients and for facilitating the interviews with the participants. A special thank you goes to the participants for volunteering their time and sharing their personal information for the sake of advancing knowledge and research. The authors would also like to thank Ms Shrooq Mahameed and Sidra Shamsuddin for their dedicated work in participant recruitment and data collection and handling. Lastly, the support of the Research and Deanship of Graduate Studies at Ajman University for their generous support of this study is very much appreciated.
The authors declare that there is no conflict of interest.
This work was supported by the Deanship of Research and Graduate Studies at the institution (Grant No. CRG-A-2018-PH-01).
Sanah Hasan https://orcid.org/0000-0002-3043-0322
Supplemental material is available online with this article.
From Department of Clinical Sciences, Center of Medical and Bio-Allied Health Sciences, College of Pharmacy and Health Sciences, Ajman University, Ajman, United Arab Emirates (Dr Hasan); College of Pharmacy, University of Sharjah, Sharjah, United Arab Emirates (Dr Alzubaidi); Research Institute for Medical and Health Sciences, University of Sharjah, Sharjah, United Arab Emirates (Dr Samorinha); and College of Humanities and Sciences, University of Science & Technology of Fujairah, Fujairah, United Arab Emirates (Dr Al Radhaideh).
Corresponding Author:Sanah Hasan, Department of Clinical Sciences, College of Pharmacy and Health Sciences, Center of Medical and Bio-Allied Health Sciences, Ajman University, Al Juf Ajman, Ajman, 346, United Arab Emirates.Email: s.hasan@ajman.ac.ae
