Determinants of COVID-19 vaccine hesitancy in university students and support staff in Pakistan: A machine learning and statistical analysis

Introduction

COVID-19 disease is caused by infection with the severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) and typically presents clinically with respiratory disease, that can be fatal.¹ The SARS-CoV-2 virus outbreak was initiated in Wuhan, China in December 2019 but quickly spread worldwide.^1,2 In Pakistan, two confirmed cases were reported in Karachi on the 26^th of February 2020.³ On 30 January 2020, the WHO declared that the coronavirus outbreak was a public health emergency of international concern and that it was a pandemic on the 11^th of March 2020.⁴ The number of reported cases of COVID-19 disease is over 775.87 million with approaching 7 million deaths.⁵

The development of SARS-CoV-2 antiviral treatments has been an ongoing challenge for scientists.⁶ Although various antibiotics and antivirals have been recommended by the US Food and Drug Administration (FDA) on the basis of their antiviral properties, there is no definitive therapy that has yet been approved that is completely effective against SARS-CoV-2.^7,8

Assuming there is a willingness for uptake, vaccination is one of the preventive measures that can be employed in a pandemic situation as immunization is a highly successful health intervention often able to tackle severe and debilitating diseases within communities.⁹ Eradicating diseases like smallpox, which cost about 100 million USD to eradicate, saves lives and also results in annual savings of roughly 1.35 billion USD.⁹ Likewise, the completion of the polio eradication campaign is anticipated to save lives and an approximately 1.5 billion USD annual cost.^10,11

In early 2021, Pakistan launched its COVID-19 vaccination campaign, approximately one year after the first case. The government regulated the vaccination effort, procuring and distributing seven approved vaccines across Pakistan. Various partners assisted the government by promoting vaccine awareness, facilitating mobile vaccination, improving healthcare facilities, and collecting vaccination data.¹²

There are differences in the rates of vaccination between countries. The analysis of the WHO/UNICEF Joint Reporting Form data from 2015-2017 has cited major causes of vaccine hesitancy among low-income countries as compared to middle-income countries.¹³ According to a survey conducted in 23 countries, the vaccine acceptance ratio is 79.1%, up 5.2% in 2022.¹⁴ However, vaccine hesitancy has increased in eight countries, ranging from the lowest vaccine hesitancy in the UK at 1.0% to the highest in South Africa at 21.1%.¹⁴

In a study conducted in France, 80% of healthcare workers were willing to accept the COVID-19 vaccine, and this willingness grew as vaccination programs became more accessible. However, among those who were hesitant, the primary concern was the fear of adverse events.¹⁵ A cross-sectional study of German healthcare workers revealed an overall COVID-19 vaccination acceptance of 91.7% in February 2021.¹⁶ In a comprehensive search of 33 countries, the level of vaccine take-up differed between low-income and lower-income countries with India having the highest acceptance at 76.7%, and Egypt at the lowest rate of 42.6%.¹⁷

Hence, despite the availability of a COVID-19 vaccine, avoidance and refusal of the vaccine are prevalent and collectively constitute vaccine hesitancy.¹⁸ Vaccine hesitancy was defined by the WHO Strategic Advisory Group of Experts on Immunization as a behavior influenced by several factors and these include issues of confidence, complacency, and convenience.¹³ Opposition to vaccination is not a new concept,¹⁹ and would likely have existed from the time of Edward Jenner and the initiation of inoculation against smallpox in 1796.¹⁹

After the polio outbreak in the 1940s, there were almost 15,000 cases of paralysis per year in the USA.²⁰ However, when the polio vaccination was developed and successfully administered, the paralysis ratio declined simultaneously.²¹ Hence, the number of polio cases fell exponentially.^22,23 That polio remains and is still endemic in Pakistan likely relates to vaccination hesitancy.²³ In 2019, vaccine hesitancy was listed as a top 10 global health threat and the WHO promised to make this a priority for its organization and its partners.²⁴ Hence, the number of polio cases fell in the 1960s and remained less than 10 in the 1970s, and currently, polio has been eradicated from all over the world except for Nigeria, Afghanistan, and Pakistan.

The Global Alliance for Vaccines and Immunizations ((Gavi) Geneva, Switzerland) suggests that for herd immunity against COVID-19 disease, approximately 60% of the population may need to be vaccinated,²⁵ although the numbers for other diseases such as measles require about 95% of a population to be vaccinated, and for polio this is approximately 80%.^25,26

Determinants of vaccine hesitancy according to the WHO ‘3Cs’ model include a lack of confidence (in vaccine safety and efficacy), complacency (low perceived disease risk and low self-efficacy), and convenience (of vaccine availability, affordability, and health literacy, of which mistrust of the benefit of vaccination and low perceived seriousness of COVID-19 disease can be main determinants of vaccine hesitancy.^13,27

Machine learning models serve as powerful tools that offer valuable insights into the medical field; providing an understanding of disease risk, clinical presentations/symptoms and treatment outcomes.^28–30 These models can evaluate potentially hidden patterns within large datasets and have been applied to the complex nature of classification tasks.²⁹ Recent studies have used different machine learning algorithms trained on social media data from platforms such as Twitter to detect anti-vaccination tweets.³¹ Sentiment analysis techniques can be employed to predict vaccine hesitancy, and researchers have developed deep learning models and used transfer learning to identify anti-vaccination tweets.³² COVID-19 vaccine hesitancy and acceptance can be analyzed in diverse datasets and provide an understanding of the factors that influence vaccine hesitancy and acceptance across cohorts by utilizing different machine learning models.^33–35 Random forest (RF) is an ensemble of classification and regression models that can analyze relatively large datasets by predicting outcomes based on the majority vote from trees and has been applied for the consideration of vaccine hesitancy.^32,36 Logistic regression can model the relationship between multiple independent parameters and a dependent categorical variable within the medical field and can also evaluate the relative contribution of each feature to the classification.³⁷

Previous studies have described vaccine hesitancy among the general population of Pakistan.^38,39 However, to date, there is a knowledge gap concerning the drivers for vaccine hesitancy amongst university students and support staff in Pakistan. Hence, the primary goal of this study is to consider the factors influencing COVID-19 vaccine hesitancy in these cohorts and evaluate the impact of gender, age, and education on vaccine hesitancy by utilizing statistical tools. The secondary objective is to identify how accurately machine learning models can classify individuals in the dataset based on their acceptance or hesitancy towards vaccination and extract the key features associated with vaccine hesitancy.

Methods

Study design and setting

A cross-sectional study was conducted from June to November 2021 to evaluate vaccine hesitancy among students and workers of different departments of a public sector university in Lahore, Pakistan. A comprehensive self-administered questionnaire was designed with the help of the Health Belief Model as a theoretical framework,⁴⁰ as per study objectives, to be completed by university students and support staff, followed by a short interview by the lead researcher. The survey included undergraduate students, postgraduate students, technical staff, laboratory attendants, management staff, security workers, and other support staff of the university. The staff from all of these socio-demographic backgrounds were included to assist with the generation of a broader representation of possible trends towards vaccination take-up.

The questionnaire content was considered and approved by the University of the Punjab Institutional Ethics Review Board (IRB), with approval number D/35/FIMS on the 21^st of September, 2021. Written informed consent was obtained from all subjects involved in the study. The participants were provided with detailed information regarding the purpose, procedures, potential risks, and benefits of the study, and their rights to withdraw at any stage without any consequences were emphasized. No participants under the age of 18 were included in the study, and all participants had the opportunity to ask questions and receive answers before giving consent. The use of human participants for the study adhered to the Declaration of Helsinki https://www.wma.net/policies-post/wma-declaration-of-helsinki-ethical-principles-for-medical-research-involving-human-subjects/. Written informed consent was obtained from all participants prior to their involvement in the study. The consent form stated the following:

“I am a Pharm-D student at the University College of Pharmacy, Punjab University. As pharmacy students, we are conducting a project on COVID-19 vaccination. Ethical approval for this study has been granted by the committee of the University College of Pharmacy, Punjab University. The primary objective of our survey is to assess vaccine acceptance and rejection behaviors, along with their potential determinants, through a cross-sectional study among university students and staff. You are invited to participate in this observational study. We assure you that your personal information will remain confidential. If you have any questions, please feel free to ask.” The overall process is shown in Figure 1.

Figure 1. Flowchart of the project methodology.

Results

Socio-demographics

The socio-demographic data from the survey is presented in Table 1. Out of 847 people who participated in the survey, 638 (75%) were students of undergraduate or postgraduate level, 209 (25%) were supportive staff including clerks, management staff, technical staff, lab attendants, shopkeepers, and guards. The overall response rate was 84.7% and the majority of the responders were female students: 582 (68.7%) females and 265 (31.3%) males, since more female participants responded to the survey. Among participants, 701 (82.8%) participants were ≤30 years of age, while 146 (17.2%) were >30 years. In terms of educational status, 711 (84%) participants were university educated (undergraduate or postgraduate programs) while 136 (16%) were non-university educated. According to occupation, 638 (75%) were students and 209 (25%) were supportive staff. Out of the 847 respondents, 363 (43%) were vaccinated, while 484 (57%) remained thus far hesitant. Hence, approximately half of the study population had not yet adopted the COVID-19 vaccine as shown in Table 1.

Table 1. Descriptive studies and bivariate analysis of the socio-demographic parameters associated with COVID-19 vaccine status.

Characteristics	Parameters	Frequency	Percentage	Vaccination status		P-value	Odd Ratio (CI:95%)	Lower value	Upper value
				No (n = 484)	Yes (n = 363)
Gender	Female	582	68.7	329 (38.8%)	253 (29.9%)	0.593	0.923	0.688	1.239
	Male	265	31.3	155 (18.3%)	110 (12.9%)
Age	≤30 years	701	82.8	414 (48.9%)	287 (33.9%)	0.14	1.566	1.095	2.240
	>30 years	146	17.2	70 (8.3%)	76 (9.0%)
Education	Non-university educated	136	16.1	89 (10.5%)	47 (5.6%)	0.033*	1.515	1.033	2.222
	University educated	711	83.9	395 (46.5%)	316 (37.3%)
Occupation	Students	638	75.3	385 (45.5%)	253 (29.9%)	0.01*	1.691	1.235	2.316
	Supportive staff	209	24.7	99 (11.7%)	110 (13.0%)

* p < 0.05.

Among the 582 females, 329 (38.8%) were not yet vaccinated, 253 (29.9%) were fully vaccinated, and for the 265 males, 155 (18.3%) were not vaccinated, while 110 (12.9%) were vaccinated. Comparatively, the hesitancy associated with vaccine take-up was more prevalent in females than in males (p = 0.593) but this did not reach significance. Approximately 49% of the younger cohort (≤30 years) was not yet vaccinated, while approximately 34% had been. Hence, the younger cohort were more resistant towards vaccination when compared to the acceptance from the older (and potentially higher risk) cohort of >30 years of age. Hence, respondents within the 31-65 age range were more accepting of vaccination (p = 0.14).

With regard to the levels of education, approximately 47% of the cohort that were university educated remained unvaccinated, whereas 37% were vaccinated. By comparison, a relatively higher percentage (11%) of the cohort that were non-university educated were not vaccinated compared to approximately 6% that were (p = 0.03).

Approximately 46% of the cohort that were students were not vaccinated whereas 30% were. This was in comparison with support staff in which the proportion of vaccinated was 13%, whereas approximately 12% were not vaccinated (p = 0.01). Hence, occupation has a significant impact on the number of vaccinated individuals and associated vaccine hesitancy within the cohort, as shown in Table 1. The prominent trend amongst supportive staff of the university indicated that being a part of these cohorts had a positive impact on vaccine acceptance (p < 0.05).

Despite the accessibility of the COVID-19 vaccine, 57.1% of the respondents had not been vaccinated (Table 1) and this was due to the following reasons: 13% did not get the vaccine due to a busy routine, 37% of people had trust issues (safety and efficacy concerns) regarding vaccines; approximately 38% were afraid of side effects after vaccination; approximately 7% thought that they may have pre-formed antibodies enough to mitigate their future risk; approximately 2% did not get vaccinated because of pregnancy, and approximately 3% of the cohort had other co-morbidities (such as diabetes, hypertension, and allergic issues etc) and needed further consultation from their physicians, as shown in Figure 2.

Figure 2. Reasons for not getting vaccinated against COVID-19 disease.

Correlation analysis

Figure 3 indicates the correlation of vaccine acceptance (dependent variable) with demographics and beliefs of an individual (independent variables), in which “Q3: Do you think you need to boost your immunity?” is positively correlated with vaccine acceptance at 94%. Similarly, Q1 “Do you have any trust issues (safety and efficacy concerns) regarding COVID-19 vaccines?” and “Q2: Are you afraid of side effects after vaccination?” negatively influence vaccine acceptance at 59% and 38%, respectively. In terms of demographics, only education showed an 11% correlation with vaccine acceptance as compared to occupation, gender and age variables are shown in Figure 3.

Figure 3. Correlation plot between all independent variables and the dependent variable, vaccine acceptance.

Key variables were further evaluated by using a Structure Model Equation to consider the relationship between vaccine hesitancy and an individual’s beliefs. Several latent variables, such as trust in healthcare providers, perceived risk and efficacy and observed variables, social influence (future risk estimation) and past vaccination status represent the indicators to evaluate vaccine hesitancy (Figure 4). The arrows in Figure 4 indicate the relationship of these variables to vaccine hesitancy and ‘e’ refers to the error terms and these are represented as small ovals attached to the variables. Results indicated that trust issues and safety and efficacy concerns had a 0.88 (88%) and 0.57 (57%) positive impact on hesitancy, respectively. The future risk of COVID-19 disease and past vaccination status were 0.67 (67%) and 0.13 (13%) negative impact factors on vaccine hesitancy, respectively.

Figure 4. A Structural Equation Model (SEM) diagram depicting factors affecting vaccine hesitancy.

The beliefs, concerns of vaccine hesitancy, and past vaccination status were considered and 535 (63.2%) responders had safety and efficacy concerns regarding the vaccine, and 458 (54.1%) of the cohort were concerned with side effects of the vaccine. Furthermore, 460 (54.3%) of responders believed in their natural immunity and 160 (19%) people had not yet been vaccinated. The details of trust issues, safety and efficacy concerns, future risk perception, and past vaccination status are covered in Table 2.

Table 2. Beliefs of participants towards complacency, confidence in the COVID-19 vaccine, and its influence on vaccine hesitancy.

Parameters	Questions	Responses		P value	Odds Ratio	95% (CI)
		No	Yes			lower	higher
Trust issues	Do you have any trust issues (safety and efficacy concerns) regarding COVID-19 vaccines?	312 (36.8%)	535 (63.2%)	<0.01*	0.060	0.042	0.086
Safety and efficacy concerns	Are you afraid of side effects after vaccination?	389 (45.9%)	458 (54.1%)	<0.01*	0.199	0.149	0.268
Future risk estimation	Do you think you need to boost your immunity?	460 (54.3%)	387 (45.7%)	<0.01*	16.125	10.945	23.756
Past vaccination status	Have you got vaccinated before?	160 (18.9%)	687 (81.1%)	<0.01*	1.841	1.276	2.656

* Denotes a significant result with a p value <0.05.

Machine learning models

We evaluated the performance of our machine learning models in classifying individuals based on their acceptance or hesitancy towards vaccination. The accuracy score of random forest was 96% and logistic regression was 97%. To validate this further, precision, recall and F1-score were considered. Precision measures the proportion of true positive prediction (vaccine hesitant people) among all positive predictions made by the model. The precision of both models was high (>90%). Recall (sensitivity) is the proportion of true positive prediction among all actual positive values. These values were close to 1 indicative that both models can identify the majority of vaccine hesitant and non-hesitant people in the dataset. The F1-score provided a balance between the two metrics, precision and recall. Both models had high F1-scores for both classes, consistent with a good balance between precision and recall. Overall, a comparative analysis of both models indicates that logistic regression can perform better than random forest in classifying the individual based on vaccine hesitancy profile in terms of precision and accuracy as shown in Table 3.

Table 3. Evaluation parameters for the machine learning models.

Parameters	Outcomes	Machine learning models
		Random Forest	Logistic Regression
Accuracy score		0.96	0.97
Precision (Sensitivity)	0 (Vaccine acceptance)	0.99	1
	1 (Vaccine hesitant)	0.93	0.93
Recall	0 (Vaccine acceptance)	0.95	0.95
	1 (Vaccine hesitant)	0.99	1
F1-Score	0 (Vaccine acceptance)	0.96	0.97
	1 (Vaccine hesitant)	0.96	0.96

The AUC-ROC for the random forest model is 0.98 and 0.99 for the logistic regression model, as shown in Figure 5.

Figure 5. AUC-ROC plot of random forest (left panel) and logistic regression (right panel).

Confusion matrix

Both the random forest and logistic regression models correctly predicted non-hesitancy (actual value 0) 96% of the time (true negatives) but incorrectly predicted hesitancy (actual value 1) 5% of the time (false negatives) in both cases. For individuals who are hesitant (actual value 1), the models correctly predicted hesitancy 68% of the time (true positives), but the logistic regression model achieved a slightly higher accuracy rate of 69%. Additionally, both models incorrectly predicted non-hesitancy 1% of the time (false positives), but the logistic regression model did not make any incorrect predictions of non-hesitancy as shown in Figure 6.

Figure 6. Confusion matrix of random forest (left panel) and logistic regression (right panel) models.

Features importance

In logistic regression, feature extraction involves assigning weights to important features to calculate the probability of the predicted outcomes. These weights indicate the impact and relationship of these variables to the predicted outcome. If future risk estimation is higher in people, it indicates more vaccine hesitancy, and similarly, the occupation and education of the individual also influence vaccine hesitancy. However, feature importance predicts that a person with trust issues regarding vaccine safety and efficacy, and who is afraid of side effects, may be more likely to be vaccine hesitant, as shown in Figure 7.

Figure 7. Important variables of hesitancy plot using machine learning model.

Discussion

Immunization is a useful strategy to limit the spread and health impact of COVID-19 disease. However, a lack of engagement with a vaccine program is still a problem in some countries such as Pakistan. A major obstacle to vaccine acceptance and the establishment of a herd immunity threshold is a lack of confidence in vaccines, particularly mistrust regarding the safety and effectiveness of the COVID-19 vaccine.⁴¹ Hence, this study aimed to determine the significance of demographic characteristics (gender, age, education, and occupation) in relation to vaccine acceptance using a Chi-square test. The study was designed to determine major contributing factors towards vaccine hesitancy by statistical and machine learning models. The four factors studied in this regard were gender, age, level of education, and occupation. The results revealed that gender and age did not significantly impact vaccine acceptance, while education and occupation were significantly associated (p < 0.05) with vaccine acceptance. Furthermore, individual vaccine hesitancy was considered within the context of the WHO ‘3Cs’ determinants of Confidence, Complacency, and Convenience.¹³ By employing machine learning algorithms on our data, we were able to predict vaccine hesitancy and acceptance with 96% and 97% accuracy using random forest and logistic regression analyses, respectively. Our study revealed that low perceived disease risk without vaccination, mistrusted information, the fear of side effects, and occupation and education were the strong predictors of vaccine hesitancy by statistical and machine learning driven data.

Previous studies have also identified vaccine hesitancy in high-income countries with high accuracy, achieving 82% sensitivity and 79–82% specificity using different machine learning models. A predictive model of vaccine hesitancy in America showed that gradient boosting and a random forest algorithm can correctly classify a vaccine acceptant individual with 97% accuracy, and a vaccine hesitant individual with 72% accuracy, and suggested that the lack of interest in vaccination is the key predictor when using the Scientific Advisory Group on Emergencies (SAGE) vaccine hesitancy parameters.³⁴ Zhou et al. (2015) developed a supervised classifier to identify anti-vaccine tweets using a random sample that utilized a support vector machine (SVM) method with 88.6% accuracy.⁴²

A systematic search covering low- and lower-income countries suggests that the average vaccine acceptance ratio was 58.5% with the highest acceptance in India (76.7%) and the lowest in Egypt (42.6%).¹⁷ A similar survey of lower-income countries revealed that Burkina Faso and Pakistan have the lowest acceptance rates at 66.5%,⁴³ with Bangladesh similar with 66.6% of people accepting vaccination.⁴⁴ Our study, albeit with a smaller sample size, had an acceptance rate of 43% which is similar to the Egyptian ratio.^17,43

Early in the pandemic, the vulnerability of older people to fatal COVID-19 disease was publicized due to a potentially weakened immune system and the presence of other comorbidities.⁴⁵ Herein, the older age group (>30 years of age) did have a higher vaccination take-up than the younger group (<30 years), although this difference did not have a significant impact on vaccine hesitancy (p = 0.14) (Table 1). Vaccine hesitancy trends in both genders were similar and the influence of gender on vaccine hesitancy was not statistically significant (p = 0.593). In contrast, education level had a significant impact on vaccination status (p = 0.033). A cross-sectional study on vaccination hesitancy in Egypt reported that, despite having relevant medical education, a number of students refused to get vaccinated due to concerns about side effects and misconceptions about vaccines.⁴⁶ Similarly, vaccine hesitancy behavior was observed among Chinese medical students.⁴⁷ In our studies, despite students having different educational backgrounds, their opposition to getting vaccinated remains consistent. However, there is a higher willingness among supportive staff to receive the vaccination (p = 0.033) as detailed in Table 1.

The second most prominent fear expressed in our findings was safety and efficacy concerns related to COVID-19 vaccines. Previous studies have reported that conspiracy beliefs and misinformation are linked to vaccine hesitancy and these can be propagated by various streams of media such as social-, electronic- and print-media.⁴⁸ Individuals are more likely to get vaccinated if they receive vaccination information from official health sources rather than unofficial ones.⁴⁹ Public perception about vaccination can be influenced by healthcare professionals and people planning to get vaccinated by trusted doctors for information regarding COVID-19.⁵⁰

Another determinant of vaccine hesitancy is confidence in the vaccine that relates to its safety and efficacy. Approximately 38% of our study participants were afraid of the side effects of vaccination, which can be a key discriminating variable for past vaccination refusal.⁵¹ Furthermore, mistrust in the benefits of vaccines also influences vaccine confidence.⁵²

Another contributing factor to vaccine reluctance was complacency and a decreased sense of the severity of the COVID-19 virus. This finding indicates the need for a strategic plan of public health messaging targeting that addresses this issue. Our study findings indicate that past vaccination refusal is a key parameter in COVID-19 vaccination hesitancy (Table 2). Hence, past vaccination history can also contribute to vaccine hesitancy, with younger people’s previous vaccination history indicative of whether they will be less resistant to new vaccines. People who previously had positive attitudes toward vaccination practices had more confidence in immunization against COVID-19 disease, and this is consistent with other studies conducted in the USA and Italy which showed that previous uptake of influenza vaccine was a predictor of COVID-19 vaccination intent.^53,54 In addition, individuals who had never received any vaccines in their lifetime displayed a higher degree of vaccine hesitancy (Table 2).

A number of suggestions and recommendations could be adopted to try and mitigate some of the concerns that relate to vaccine hesitancy (Table 2). A counselling facility at a pharmacy or healthcare center could promote contact with a health professional to assist with questions and concerns about the safety and efficacy of vaccines. Educational campaigns at college and university levels will also help to provide authentic and relevant information to students and consideration could be given to check and monitor their exposure to anti-vaccination propaganda that can be spread on social media. Vaccine hesitancy may be better addressed by the continuous monitoring of vaccine safety, addressing public concerns through research, providing compensation for adverse events, and introducing specific training for healthcare professionals to counter anti-vaccination misinformation.⁵⁵

A spread of information by teachers, motivational speakers, religious scholars and celebrities could provide validation of appropriate and established facts regarding vaccination since vaccine take-up by others can influence vaccination intention.⁵⁶ Printed sources and media advertisements should be meaningful and appropriate. Countries such as Israel (and the UK) have successfully used public awareness campaigns to promote vaccination and combat vaccine hesitancy.⁵⁷ A campaign of parental education may improve vaccination in children and young adults and public confidence will improve with post-vaccination follow-up and publicity regarding pharmacovigilance. A collaboration with employers to organize workplace vaccination drives, making it convenient for employees to get vaccinated during working hours may be fruitful as this approach has been successful for influenza vaccination.⁵⁸ Ultimately, with an increase in the number of vaccinated people in Pakistan, closer and potentially more rapid herd immunity could be achieved and the mortality and morbidity arising from SARS-CoV-2 infection will likely decline.

Conclusions

This study considered the hesitancy to the COVID-19 vaccine in a Pakistani cohort and how this may be impacted by socio-demographics. By utilizing both conventional statistical methods and machine learning models, this research revealed that trust issues, safety and efficacy concerns, future risk perception, and education and occupation are contributing factors and key drivers of vaccine hesitancy. These findings suggest that possible interventional educational programs and counselling by healthcare providers regarding the safety and efficacy of vaccines may represent possible approaches to mitigate vaccine hesitancy.

Author’s contributions

Conceptualization, Methodology: FR; Data curation, Writing- Original draft preparation: AN and MHS; Visualization, Investigation: FR; Supervision: FR; Software: AN; Validation: FR and FKH; Writing-Reviewing and Editing: FR and WGC. All authors have read and agreed to the published version of the manuscript.

Ethics and consent

The questionnaire content was considered and approved by the University of the Punjab Institutional Ethics Review Board (IRB), with approval number D/35/FIMS on the 21^st of September, 2021. Written informed consent was obtained from all subjects involved in the study. The participants were provided with detailed information regarding the purpose, procedures, potential risks, and benefits of the study, and their rights to withdraw at any stage without any consequences were emphasized. No participants under the age of 18 were included in the study, and all participants had the opportunity to ask questions and receive answers before giving consent. The use of human participants for the study adhered to the Declaration of Helsinki https://www.wma.net/policies-post/wma-declaration-of-helsinki-ethical-principles-for-medical-research-involving-human-subjects/.

Written informed consent was obtained from all participants prior to their involvement in the study. The consent form stated the following: “I am a Pharm-D student at the University College of Pharmacy, Punjab University. As pharmacy students, we are conducting a project on COVID-19 vaccination. Ethical approval for this study has been granted by the committee of the University College of Pharmacy, Punjab University. The primary objective of our survey is to assess vaccine acceptance and rejection behaviors, along with their potential determinants, through a cross-sectional study among university students and staff. You are invited to participate in this observational study. We assure you that your personal information will remain confidential. If you have any questions, please feel free to ask.”