An institutional-based cross-sectional study was conducted at eight higher education institutions found in Adama City, which is situated 100 km southeast of the capital of Ethiopia, Addis Ababa. Of the total population, ~86,069 are young adults attending higher education institutions. In the city administration, there are eight higher education institutions, four universities (Adama Science and Technology University, Harambe University, Rift Valley University, and Unity University), and four colleges (Adama TVET College, Adama Comprehensive Hospital and College of Health Sciences, Oromia Police college, and Adama General Hospital and Medical college). The study was conducted between 01 February and 30 March 2022.

The source population comprised students enrolled in higher education institutions within the city during the study period, and the study population encompassed all students from these institutions who provided consent to participate in the study. Individuals who refused to participate were excluded from the study.

The sample size was determined using EPI Info version 7 statistical software, employing the double population proportion formula, with COVID-19 vaccine distrust as an independent variable (8). The considerations included a 95% confidence level, 80% power, a 1:1 ratio, and an additional 10% contingency to account for non-response rates, finally resulting in the recruitment of 530 study participants. Four higher educational institutions were selected through simple random sampling. Study participants were selected from those who were attending class during the data collection period from February to March 2022 using a systematic random sampling technique. The sampling interval (K) was calculated by dividing the total number of students who were attending classes in the higher education facility during the data collection period by the sample size. A number between one and K was taken as a random start, and every Kth value was selected as a sample unit. The sample size was allocated for each higher education facility proportionally, which is 132 for three institutions and 134 for one institution.

Dependent variable: COVID-19 Vaccine hesitancy

Independent variables: Socio demographic factors (Age, gender, socioeconomic status, Level of education, geographic area of origin, ethnic group, and Religion).

Contextual influences- communication and media environment, anti- or pro-vaccination lobbies, historical influences, politics/policies and perception of the pharmaceutical industry.

Individual and group influence related factors- knowledge, health system and healthcare providers trust, personal experience and risk/benefit (perceived, heuristic).

A half-day training session was conducted for both data collectors and supervisors, focusing on elucidating the study's objectives and significance, instructing on the adept collection of pertinent information, detailing the procedures of data collection techniques, and thoroughly covering the questionnaire's contents. The questionnaire underwent a pre-testing phase involving a representative sample of 27 students (5% of the sample size) from Admas University, and this institution was not included during the major data collection for the research. A necessary modification was made to the questionnaire based on the nature of the identified gaps. Data were entered into Epi-data version 7.2.1 and cleaned and explored for outliers, missed values, and any inconsistencies.

After checking, coding, and entering Epi data version 3.1 and validating and comparing it to the original data, corrective measures were taken accordingly. Data were exported to the Statistical Package for Social Science [SPSS] Version-27 software for analysis. Through SPSS's transform function, variables were computed and recorded. The study's outcome variable, COVID-19 vaccine hesitancy, was measured by asking participants the following questions: “Will you get vaccinated if you get COVID-19 vaccine?”, for which the answer was dichotomized into “Yes” or “No” alternatives. Students who responded “Yes” were recorded as COVID-19 vaccine acceptance, while those who responded “No” were coded as COVID-19 vaccine hesitancy.

A descriptive analysis utilizing frequency tables, proportions with a 95% confidence interval, and summary measures was conducted. To choose candidate variables for multivariable analysis, a bivariable logistic regression analysis was applied. Based on the presumption of selection criteria, variables with a p-value of < 0.25 were taken as candidates for the final multivariable analysis model. The multi-collinearity test was carried out to observe the linear correlation among independent variables by using standard error. The Hosmer-Lemeshow goodness of fitness test was used to assess the appropriateness of the model; the result was negligible (p = 0.65), indicating that the model was fitted. Using adjusted odds ratio (AOR) with 95% confidence intervals (95% CI), the strength of associations between the dependent and independent variables was finally evaluated. A p-value of < 0.05 was used to indicate the significance of the association.

The median age of those 530 higher education students who were selected for the study was 22 years (IQR = 2). Only 17 (3.2%) of the selected students were 30 years of age and above. A greater proportion of the student population, that is, 377 (71.1%), were male, and most of the study participants, 485 (91.5%), were single. The majority of the students belong to the Oromo and Amhara ethnic groups, and 377 (71.1%) students were urban residents. More than half of the participants (52.6%) were Orthodox Christian by religion, and 471 (88.9%) students were attending their bachelor's degree in their respective higher education institutions. Most of them, 294 (55.5%), reported that they came from low-monthly income families, followed by 161 of them (30.4%) from middle-monthly income families (Table 1).

The majority of the study participants, 334 (63%), trust health professionals for information regarding the COVID-19 vaccine, while social media is the least trusted source, with 209 (39.4%) students indicating their trust in it for information about the COVID-19 vaccine. Among all the study participants, 270 (50.9%) students re-considered the choice to get vaccinated after hearing or reading reports about the COVID-19 vaccine on social media. A large proportion of students, 345 (65.1%), agree with some groups or leaders that do not support COVID-19 vaccination for different reasons. Whereas, leaders (religious, political, teachers, and healthcare workers) supporting COVID-19 vaccines for young adults in their respective communities are 52.3% (Table 2).

COVID-19 vaccine hesitancy was measured by asking participants, “Will you get vaccinated if you get COVID-19 vaccine?”, for which the answer was dichotomized into “Yes” or “No”. The study showed that the prevalence of coronavirus vaccine hesitancy among students in higher education institutions in Adama City administration is 47.5% (Figure 1).

A total of 8 variables, including residence in rural areas [COR = 1.632 (95% 1.117, 2.382)]; agreeing with leaders and groups that do not support COVID-19 vaccination [COR = 3.814 (2.585, 5.626)]; a previous history of bad reaction to the vaccine [COR = 1.740 (1.225, 2.470)]; coming from a community whose leaders support COVID-19 vaccination for young adults [COR = 0.388 (0.273, 0.550)]; thinking it is important to get vaccinated to protect those that cannot be vaccinated for different reasons [COR = 0.166 (0.114, 0.241]; believing COVID-19 vaccines are safe [COR = 0.131 (0.072, 0.238)]; trusting their government in making the best decision in the interest of its people [COR = 0.391 (0.269, 0.567)]; and thinking the government is pushed by lobbyists or vaccine makers to recommend certain vaccines [COR = 1.045 (0.732, 1.491)], were found to have a statistically significant association with coronavirus hesitancy.

To control confounding and find independent factors of coronavirus vaccine hesitancy, multivariable analysis was used. Variables with a p-value of < 0.25 in the bi-variable logistic regression analysis were candidates for multivariable analysis. Out of those associated variables in the bi-variable analysis, the odds of vaccine hesitancy were 2.4 times higher in students who were from rural areas compared to their urban counterparts (AOR = 2.398, 95% CI: 1.476–3.896). The odds of vaccine hesitancy were 2.3 times higher in students who agree with leaders and groups that do not support COVID-19 vaccination compared to those who support the vaccine (AOR = 2.292, 95% CI: 1.418–3.704). Vaccine hesitancy was 40.2% less likely in students who came from a community whose leaders supported COVID-19 vaccination for young adults compared to those whose leaders did not support it (AOR = 0.598, 95% CI: 0.381–0.940). The odds of vaccine hesitancy were 65.7% less likely in those who believed COVID-19 vaccines were safe compared to those who did not (AOR = 0.343, 95% CI: 0.168–0.701) (Table 3).