| چکيده
|
:
|
Introduction: In the context of the COVID-19 pandemic, vaccination has been identified as one of the most effective tools for controlling disease transmission and reducing mortality. However, public vaccine acceptance, despite widespread availability, remains a significant challenge in many countries, including Iran. This study aims to identify the most critical factors influencing COVID-19 vaccine uptake/non-uptake in the general population of Iran.Material & Methods: This study utilized data from the National COVID-19 Vaccine Acceptance Survey conducted in Iran between 2021–2022. The relationships between demographic characteristics, perceived risk, perceived severity, perceived susceptibility, awareness, attitude, trust, worry, and accessibility with vaccine uptake were evaluated. Participants were categorized into two groups based on their vaccination status: individuals who had received at least one dose of the vaccine and those who had not received any doses. Random Forest (RF) and Logistic Regression (LR) algorithms were applied to identify factors associated with vaccine acceptance or refusal. Given the imbalance in the outcome data, a weighting approach was utilized. Furthermore, to facilitate the interpretation of the coefficients, all continuous variables were standardized before entering the model.Feature importance in the RF model was assessed using the Gini and permutation importance methods. In the LR model, backward selection was applied to identify key variables and reduce the number of predictors. The models' performance was assessed and compared using metrics such as the area under the receiver operating characteristic curve (AUROC), sensitivity, specificity, positive predictive value, negative predictive value, and accuracy. After selecting the best-performing model, the most important variables associated with COVID-19 vaccine acceptance were identified. The analyses were performed using Python and SPSS software, version 26, and a significance level of less than 0.05 was considered.Results: Among 3,777 participants, 1,743 (46.02%) were women, and the mean age of respondents was 42.87 ± 12.38 years. The findings showed that 3,484 participants (92.4%) had received at least one dose of the COVID-19 vaccine. No significant gender differences were observed in vaccine acceptance. Comparing the performance metrics of the models demonstrated that both RF and LR methods performed well in predicting vaccine uptake/non-uptake, but notable differences existed between them. According to the result, LR outperformed RF in all performance indices for classifying individuals based on vaccine uptake. LR achieved a sensitivity of 75.31% and an accuracy of 82.6%, compared to 71.60% sensitivity and 81.08% accuracy for RF. The higher AUC of LR (0.87) compared to RF (0.84) reflects better group discrimination and a superior balance between sensitivity and false-positive rate. According to the RF model, the most influential variables included: the necessity of vaccination for returning to normal life, the role of vaccination in preventing disease onset, willingness to receive booster doses if required, vaccination to prevent disease transmission, the severity and mortality of COVID-19, vaccination as a collective action to curb disease spread, mandatory vaccine certification for workplaces, universities, and schools, concerns about immediate post-vaccination side effects, and age.The results of the multivariable LR showed that several variables, including age (OR = 0.61, 95% CI: 0.51-0.72), perceived risk of COVID-19 (OR = 0.65, 95% CI: 0.53-0.79), belief in the effectiveness of vaccination in preventing the transmission of COVID-19 to others (OR = 0.47, 95% CI: 0.40-0.55), belief in the severity and lethality of COVID-19 and the requirement for vaccination (OR = 0.51, 95% CI: 0.43-0.60), perceiving oneself as susceptible to COVID-19 (OR = 0.43, 95% CI: 0.38-0.57), awareness of the temporary nature of the side effects of the COVID-19 vaccine (OR = 0.74, 95% CI: 0.66-0.84), and awareness of the effectiveness of the COVID-19 vaccine in preventing infection and reducing mortality (OR = 0.85, 95% CI: 0.76-0.97) were directly associated with the odds of not receiving the vaccine.In contrast, the calculation of the potential benefits and risks of the vaccine by individuals (OR = 1.15, 95% CI: 1.01-1.31) and a history of chronic illness (OR = 1.29, 95% CI: 1.11-1.51) showed an inverse relationship with the odds of not receiving the vaccine.Conclusions: The findings of this study highlight the critical role of demographic, social, and psychological factors in vaccination decision-making. The research emphasizes the importance of a comprehensive consideration of these factors alongside technical aspects in designing vaccination programs. Enhancing transparent communication and boosting public trust can reduce social resistance and improve vaccination coverage. These findings provide valuable insights for designing health policies and targeted interventions at the national level, as well as for managing similar crises in the future.Keywords: Vaccination refusal, COVID-19, Random Forest, Logistic Regression, Iran
|