Study design
Between 9th August 2021 and 23rd August 2021, we conducted an online, descriptive cross-sectional study using quantitative technique.
Study setting
This study was conducted in Uganda. There are currently 53 universities in Uganda turning out over 40,000 graduates annually [17]. However, only 12 universities offer HP courses with an estimated population of 10,000 students. These include Makerere University (Mak), Mbarara University of Science and Technology (MUST), Busitema University (BU), Gulu University (GU), Kabale University, Kampala International University (KIU), King Ceasor University, Islamic University in Uganda (IUIU), Uganda Christian University (UCU), Soroti University, Muni University and Lira University (LU).
Target population
All university students male and female, 18 years or older, from first to fifth years of study offering health care programs at any of the above 12 medical schools in Uganda. Programs that participated include Bachelors in Medicine and surgery (MBChB), Bachelor in Pharmacy (BPHAR), Bachelor in Nursing/ Midwifery (BSN/MW), Bachelor in Dental surgery (BDS), Bachelor in Public health/ Environmental health (BPH), Bachelor in medical Radiography (BMR), Bachelors of science in Anesthesia among others.
Sample size
The sample size was calculated using Epi Info StatCalc by the Center for Disease Control (CDC) for population surveys. With an estimated population size of 10,000 health care students in Uganda at expected frequency of 50% and confidence limits of 4.0%, the estimated sample size at 95% confidence interval was 566 students.
Sampling procedure
We used convenience sampling method, whereby those who were able to access the link to the online survey questionnaire sent out were involved. We identified all relevant WhatsApp, telegram and email groups of HP students through a coordinator at each university and continuously shared the invitation link to the online questionnaire.
Study Variables
Dependent variables included questions on perceptions, attitudes and practices concerning herbal medicine use in the treatment of COVID-19. The independent variables included age, sex, rural or urban setting of growing up, religion, year of study, program and University of study.
Data collection tool.
The questionnaire used was adopted from two different pre-validated questionnaires used by two related studies [18-19] and were modified to suit our study setting and objectives. The tool consisted of four sections;
Section I: Had 7 questions which we used to assess for the HP students’ socio-demographic factors including age, sex, religion, growing up setting, academic year, program and University of study.
Section II: 6 questions were used to assess HP students’ perceptions regarding HM use in COVID-19 treatment by indicating YES, NO, or IAM NOT SURE against a given statement. The questions were adopted from Alotiby et al [18].
Section III: Had 6 questions, each scored on a 5 Likert scale ranging from strongly agree, agree, neutral, disagree, and strongly disagree. It was used to assess students’ attitude towards HM use in COVID-19 treatment. The questions were adopted from Samara et al [19].
Section IV: Had 10 questions used to assess the students’ practices and use of HM in COVID-19 treatment. The questions were adopted from Samara et al [19] and some were added by the investigators.
Quality Control
Questions were designed in simple English words for effective understanding by the participants. Soft copies of the questionnaires were designed with checks to allow valid and complete entries only. The questionnaire was pretested among 15 veterinary medicine students and all required changes were made before sending it out to the final participants.
Data management and analysis.
Fully completed questionnaires were extracted from Google Forms and exported to Microsoft Excel 2016 for cleaning and coding. The cleaned data was exported to STATA 15 for analysis. Numerical data was summarized as means and standard deviations (SD) or median and interquartile range (IQR). Categorical data was summarized as frequencies and proportions. Blooms cut-off of 80% was used to determine whether or not a participant had good perception, positive attitude. Perception was assessed using 6 questions and each correct answer scored 1 point. Attitudes were assessed using a 5-point Likert-item questions with responses including strongly disagree, disagree, neutral, agree and strongly agree and the scores were 1-5 for the respective responses. The total score was 30, and a score of ≥24 (≥80%) was considered a positive attitude. Associations between independent variables and dependent variables were assessed using Chi-square and Fischer’s exact test. Binary logistic regression analysis model was constructed. We accounted for important confounders in the model. Data was presented as adjusted odds ratio (aOR) and 95% confidence interval (95% CI). All data analyses were two-tailed and p<0.05 was considered statistically significant.