Study design and site
This study is a cross-sectional survey using an interview questionnaire investigating the prevalence of HbA1c self-awareness among Saudi patients with type 2 diabetes, the accuracy of reported HbA1c, their association with glycemic control, and the factors that might be related to glycemic control. The study was conducted in the outpatient diabetes clinic at tertiary hospitals in Riyadh, Qassim, and Jeddah.
Eligibility criteria
The inclusion criteria were: either male or female patients between the ages of 18–75 years, who had been diagnosed with type 2 diabetes mellitus for at least 1 year, and who have regular follow up visits for diabetes care. We excluded any patients younger than 18 years and older than 75 years, those with a past history of bariatric surgery, past history of renal insufficiency, with a creatinine level greater than 1.5 mg/dl, with known underlying illness (e.g., malignancy or hemoglobinopathies), that received a blood transfusion within the past 30 days, or being pregnant. Also, the participants were excluded if, for any reason, they did not have the ability to communicate verbally.
Patient enrolment
Patients were approached by using a simple time random interval technique (17). Three different days every week were chosen. During the early morning of each of these days, we visited the outpatient diabetes clinics and obtained the sample frame of all patients who met our inclusion criteria. A random sample was then selected from the sample frame. The informed consent form was taken from the patients before the administration of the questionnaire, and all patient’s information was handled with strict confidentiality. Trained data collectors were assigned to collect the data at each center, and all of them followed the same process. The data was collected using the questionnaire through face-to-face interviews from March to April 2018.
Instrument development
Data for this study were collected using a questionnaire through an interview. The interview questionnaire was developed based on an extensive literature review and experts’ opinions (18-21). The questionnaire was composed of two main sections. The first section assessed the patient’s background information (age, gender, nationality, highest educational level, location, monthly income, current occupation, marital status, smoking status, duration of diabetes, and type of diabetes treatment). The second part addressed HbA1c (medication compliance, frequency of hospital visits, patient awareness of HbA1c, and education about the test and its target by a healthcare provider). The most recent HbA1c, weight, and height data were gathered by the data collector from the patient’s file.
The patient’s self-awareness of HbA1c was assessed based on a score of four questions. Patients were asked if they have heard about or are aware of the term HbA1c. Those who answered yes proceeded to answer three other questions on HbA1c, including what does HbA1c mean, their target HbA1c goal, and whether they can correctly identify their current HbA1c value. Participants were categorized as having good HbA1c self-awareness if they could answer 3 out of 4 questions on HbA1c correctly. While those who did not hear about the test before or scored less than 3 out of 4 were categorized as not having a good awareness of the test (19). The reported HbA1c was considered accurate if within ± 0.5% of the recorded HbA1c. Glycemic control was defined as having a measure of HbA1c of less than 7%.
The questionnaire was initially developed in English. Two accredited translators were assigned to translate it to Arabic and then back-translated it to English, both the original and final English versions were reviewed, and any disagreements between them were solved by the principal author and the translators (22, 23). To check the validity and reliability of the final Arabic version, a pilot study was done that included 60 patients with type 2 diabetes. The piloted participants were not included in the study.
Sample size calculation
Using the following single proportion formula, assuming that alpha is equal to 1.96 for a 95% confidence level, the precision is equal to 4%, and a prevalence of 50%, the calculated sample size was 600 subjects.
where: Z = Z value, p = the prevalence, and d = precision.
Statistical analyses
All analyses were conducted using SAS Version 9.3 (SAS Institute, Inc, Cary, NC) (24, 25). Frequencies and percentages were calculated for all categorical variables used in the analysis. Means and standard deviations were calculated for continuous variables. A p-value of less than 0.05 was considered statistically significant for all analyses.
Mean self-reported HbA1c with the corresponding standard deviation was calculated for patients who had reported their HbA1c. Additionally, mean laboratory-generated HbA1c and corresponding standard deviations were calculated for all patients. Levene’s test was used to examine whether the variance in the mean laboratory-generated HbA1c values was different between those who had reported their HbA1c and those who had not. We also calculated the mean difference between these two groups and corresponding 95% confidence intervals for these mean differences. Finally, a Peta-square was calculated to determine whether these mean values were significantly different. We performed the same analysis comparing mean reported, and actual HbA1c values among patients who had accurately and inaccurately reported their HbA1c.
An interclass correlation coefficient (ICC) was calculated to determine the relationship between actual and reported HbA1c among patients who had reported their HbA1c.
T-tests were performed to determine whether the mean values for several continuous variables were different among those who had good HbA1c self-awareness and those who did not. We also used chi-square tests to determine whether the distribution of different categorical variables was different between these two groups. We performed the same comparisons between those who had good self- awareness and knew their HbA1c goal and those who had good self-awareness but did not know their HbA1c goals.
A chi-square test was also performed to determine whether the distribution of different categorical variables was different among those with glycemic control and without glycemic control.
Finally, logistic regression was performed to determine whether the odds of having glycemic control differed significantly according to HbA1c self-awareness, socio-demographic, and clinical characteristics as covariates. We ran univariate models to examine these characteristics separately, and a multivariate model examining all variables simultaneously.