Study design and setting
An institutional-based cross-sectional study was conducted in randomly selected primary hospitals of Northwest Ethiopia, located Northwest of Addis Ababa. The study period was from March 1 to May 30, 2019. Essential information was obtained through the administration of structured questionnaire-based interviews. The study area had one comprehensive specialized referral and teaching hospital, one private general hospital in Gondar town, ten primary hospitals, and a number of health centers. The study participants were recruited from the selected primary hospitals located in the towns of Addis Zemen, Debark, Wogera, Kolladiba, Chilga, and Metema.
Source population, Sample size determination and Sampling procedure
The source population was all diabetic patients who had been treated with insulin in the study area. The study population included those diabetic patients who were using insulin as their primary therapy or as additional therapy and visited those hospitals during the study period. Patients or patient caregivers who were 18 years of age and above were included in the study. Patients should be on insulin treatment for at least one month and refilling insulin prescriptions at one of the hospitals. Patients who did not consent to participate in the study or those who were seriously ill, unable to hear or speak, physically disabled, having dementia or cognitive impairment and difficulty of getting consent were excluded from the study. A convenient sampling technique was used to collect data.
Data collection instruments, procedure and management
The data collection format was initially prepared in English. It was then translated to the local language (Amharic) and back-translated to English to ensure proper meaning. Trained pharmacy professionals, under investigators’ daily supervision, collected the data. The questionnaire focused on socio-demographic and related information, experiences, practices and knowledge of insulin storage and handling techniques, and an observational checklist of patients’ skills related to self-insulin administration. Both the practice and the knowledge questions contain 14 items. Practices were measured by Likert scale type (Never=1; Sometimes=2; Often/Usually =3; Always=4) and graded as poor, fair and good for scores of <50% (<28), 51-75% (29-42) and >75% (>42) out of 56 points, respectively. The knowledge of respondents’ was measured with dichotomous outcomes as “right” who answered the question correctly and “wrong” who answered the question incorrectly. Finally, all the responses were summed up to an overall score and categorized into three levels such as adequate, moderately adequate, and inadequate knowledge levels. Scores of >75% (>10.5), 51-75% (8-10.5), and <50% (<7) out of 14 points were said to be adequate, moderately adequate, and inadequate, respectively [19]. The respondents’ skills were measured through a checklists of five observational (demonstration) techniques related to insulin self-administration procedures. Based on American diabetic association of insulin administration, the checklist was marked as correct, incorrect, and skipped and given scores of 2, 1, and 0, respectively [20, 21]. If a patient or a caregiver correctly performed all the critical steps, the observation (demonstration) was considered as correct; if any of the critical steps was missed or performed incorrectly, the demonstration was considered as incorrect; and it is considered as skipped if any of the steps was jumped.
Data entry, analysis, and interpretation
The data was entered and analyzed using statistical package for social sciences (SPSS), version 22.0.[22]. Frequency, percentages and median were used to describe the variables in univariate analysis while Chi-square, Mann-Whitney U and Kruskal-Wallis tests were used to describe and test the statistical significance of variables in the bivariate analysis. Pairwise multiple comparisons were done for those groups who had significant knowledge and practice median differences on the Kruskal-Wallis H test. The knowledge and practice data were transformed into categorical values. The knowledge levels were classified as inadequate, moderately adequate and adequate knowledge; and practice levels were expressed as poor, fair, and good practices. P-value < 0.05 and 95% confidence interval (CI) were used as cut-off points for determining the statistical significance of associations among different variables. Pearson’s correlation coefficient test was done to assess the degree of correlation between the patients’ knowledge, and their practice levels in their insulin handling techniques and injection practices.
Data quality control
Before any data collection was performed, a pretest was done on 15 patients from one of the randomly selected primary hospitals. These patients were not included in the final data analysis. Important amendments were made and formatted as needed. The data accuracy and completeness were consistently checked by using double entry and any mistakes and omissions were corrected.
Informed consent and Confidentiality
Before any of the activities, written informed consent was sought from each of the participants. The voluntariness of the participation and the aims of the study were explained clearly. The participants were given the right to interrupt at any time if they do not want to continue. The respondents were interviewed and observed while doing a demonstration privately and independently. All information was kept confidentially by giving serial numbers.