This study was conducted in three Health Districts (HDs)—Lukelenge, Muya, and Diulu—located in Mbujimayi, the provincial capital of Kasai-Oriental in central DRC ( Figure 1 ).

HDs are responsible for implementing health strategies defined by the National Ministry of Health (MoH) and overseen by the Provincial Health Department (PHD). Each Health District is designed to meet the health needs of a population ranging from 150,000 to 200,000 residents. COVID-19 vaccination commenced in Kinshasa on April 19, 2021, after the arrival of 1.8 million doses of COVID-19 vaccine ChAdOx1-S (AstraZeneca®/Covishield) [14] [19] . Due to slow initial uptake, around 1.44 million doses were at risk of expiring by June and July. To prevent wastage, these doses were redistributed to other African countries [14] [17] . Subsequently, additional vaccines were procured from various donors and initiatives [14] [19] . As shown in Figure 2 , mass vaccination campaigns in Kasai-Oriental underwent four phases between December 2021 and November 2023, each lasting 7 to 16 days (total, 45 days). Further data regarding the number of vaccines administered during these campaigns would provide valuable insights, but they were not accessible upon request from authorities at the time of conducting the survey.

This figure situates the survey on the chronology schematizing the main events linked to vaccination against COVID-19 in the population studied. It appears that in total, the population had benefited from 45 days of mass vaccination.

We conducted a cross-sectional household-based survey in the study area from May 12^th to 18^th, 2023, during the fourth phase of COVID-19 vaccination in Mbujimayi. Then, during the survey, a single individual aged 18 and above was recruited within each targeted household on a “first come, first serve” basis. Written informed consent was obtained from each participant before their inclusion, ensuring adherence to ethical standards. Individuals unable to participate due to medical or availability reasons were excluded from the survey to maintain data integrity. Sample size calculations for each HD considered population size and estimated vaccination coverage. With a 50% assumed vaccination coverage [24] , a 5% margin of error, 95% confidence interval, and 90% response rate, aiming for a minimum of 422 participants per HD. A minimum of 1266 participants were thus targeted by the field survey.

Ten investigators proficient in Ciluba, the local language, underwent training to administer the survey. The survey units were selected using a multi-stage random sampling technique. A total of three zones, each comprising an equivalent number of wards (19 health wards), were randomly selected from the ten health zones that constitute the town of Mbujimayi. Within each zone, six health areas were randomly selected to form the sample. These neighbourhoods were delineated and subdivided into parcel blocks using satellite images obtained via Google Maps ( https://maps.google.com/ ). Three plot blocks were then randomly selected within each neighbourhood. Finally, within each plot, one adult per household was interviewed using a systematic random sampling procedure with a sampling strategy. They conducted in-depth interviews with individual participants using a standardized semi-structured questionnaire (Additional File 2). This questionnaire was adapted from the International Citizen Project on Coronavirus Disease 2019 (ICPcovid, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7932542/ ; accessed on 10 February 2021). Prior to its implementation, the questionnaire was field-tested via a pre-survey. It was developed using the open-source KoboToolbox platform ( https://www.kobotoolbox.org/ ) as a mobile web application. It gathered information on participants’ sociodemographic characteristics (e.g. age, gender, marital status, education level, religion, and occupation), vaccination status, as well as personal attitude toward COVID-19 and its vaccines. Specifically, it included inquiries about participants’ understanding of COVID-19, their vaccination history or intentions, satisfaction with their vaccination status, reasons for vaccine refusal, and perceptions of vaccine effectiveness or safety.

In this study, participants’ personal opinions on COVID-19 were categorized as “correct” if they expressed the view that “COVID-19 is a serious illness requiring protection” in response to the question regarding their perception of COVID-19. Any divergent opinion was classified as a misconception surrounding the disease. Similarly, participants’ personal opinions on the COVID-19 vaccination were considered “correct” if they affirmed the effectiveness, usefulness, and safety of the vaccines.

We operationally defined COVID-19 vaccination non-compliance as not receiving any vaccine dose recommended by national health policies. To assess this, we explored the number of vaccine doses each participant received. Additionally, we delineated vaccine hesitancy from refusal and access limitations. Vaccine hesitancy encompassed a range of behaviors from delayed acceptance to reluctance, without reaching outright refusal, despite the availability of vaccination services. Vaccine refusal refers to the active decision to decline vaccination, regardless of vaccination service availability. In contrast, access limitation involves barriers preventing individuals or communities from obtaining vaccines despite their willingness to be vaccinated.

Data analysis was conducted using R software version 4.3.0 (The R Development Core Team, R Foundation for Statistical Computing, Vienna, Austria, 2019). Absolute and relative frequencies summarized qualitative variables, while medians with corresponding Interquartile Ranges (IQRs) described quantitative characteristics. Qualitative variables were compared between groups using the Chi-square test or Fisher’s exact test, whereas quantitative variables underwent comparison using parametric tests (e.g. ANOVA test or t-test) or nonparametric tests (Kruskal-Wallis’s rank sum test or Wilcoxon rank sum test with correction for continuity), after confirming normality and homogeneity of variances with Shapiro-Wilk’s and Levene’s tests, respectively. Univariate and multivariate hierarchical Logistic Regression Models (GLMs) assessed potential predictive factors’ effect on anti-SARS-CoV-2 vaccine non-compliance. The final model was chosen based on the Akaike’s “An Information Criterion” (AIC) values through a stepwise algorithm incorporating both “backward” and “forward” methods. A significance level of p < 0.05 was considered for interpreting all statistical tests.

Figure 3 and Table 1 provide an overview of the key sociodemographic characteristics and clinical profiles of the individuals enrolled in this study. Of the initially recruited 1500 individuals, 1496 were successfully included, resulting in a response rate of 99.7%. Notably, 38.4% (n = 575), 31.0% (n = 464), and 30.6% (n = 457) of the participants hailed from the Diulu, Muya, and Lukelenge health districts, respectively. The median age across all participants stood at 33 years (interquartile range, IQR: 23.3), with a predominant age group of 20 to 40 years. Most participants were female (58.8%), irrespective of their health district of residence. Moreover, participants reported being mostly married (62.0%), adhering to Christianity as their religion (83.8%), and attaining a secondary level of education (55.0%). Approximately one-third of participants (32.2%) identified themselves as unemployed. Notably, all participants met the survey’s inclusion criteria by self-reporting apparent good health. Additionally, most of them (91.6%, n = 1354) claimed to have not undergone testing for COVID-19, expressing confidence in not being infected since the pandemic’s onset ( Table S2 ). However, 70.9% of respondents (n = 1,030) admitted having experienced symptoms consistent with a possible COVID-19, as defined by the World Health Organization (WHO), over a two-week period, primarily including fever (54.8%) and headache (48.1%).

This figure represents the main characteristics of the study participants, including their distribution by age (Panel (A)), gender (Panel (B)), symptoms mentioned (Panel (C)), and COVID-19 testing history (Panel (D)).

Note: *IQR: Interquartile Range.

This study delved into participants’ perceptions regarding COVID-19 and its vaccines, as shown in Table 2 . Surprisingly, only 35.2% (n = 526) considered COVID-19 as a natural disease posing a significant health threat. Conversely, nearly half held diverse opinions, considering COVID-19 as an artificially created disease to harm people, a disease exclusively affecting certain racial groups, or even as a nonexistent or spiritually induced illness. A notable proportion of them (17.8%; n = 266) refrained from expressing an opinion on COVID-19. Regarding vaccines, merely 19.8% (n = 296) believed in their efficacy against COVID-19, while the majority (77.5%) expressed skepticism, deeming them either ineffective, unsafe, or both.

Participants were asked to evaluate their compliance with government-imposed non-pharmacological interventions during the pandemic ( Table S3 ). Only 12.7% reportedly adhered to social distancing guidelines of maintaining a 1.5-meter distance in public, and merely 5.7% declared consistently wearing face masks when leaving their home. Additionally, just 24.8% reported regularly coughing or sneezing into their elbows as recommended. Hand hygiene practices were reportedly also suboptimal, with only 19.9% of respondents declaring often washing hands (i.e. several times on any occasion) and 28.6% frequently using hand sanitizer (i.e. several times on any occasion).

Nearly half of participants remained unvaccinated against COVID-19, resulting in an estimated vaccination coverage of 49.1% [95% CI: 47.5; 52.6]. Among the vaccinated individuals (n = 727), 71.8% (n = 522) had received only one vaccine dose, while merely 8.0% (n = 58) had received three or more doses, as shown in Table 3 . Primary reasons for vaccination hesitancy or refusal were fear of vaccines and their side effects (38.2%; n = 217) as well as general distrust of vaccines (21.7%; n = 123). The vast majority of those hesitating or refusing the booster dose could not articulate a specific reason for their stance (79.5%).

The distribution of vaccine refusal, hesitancy, and access barriers among the study participants is depicted in Figure 4 . Out of 756 unvaccinated individuals, a significant majority (67.3%) expressed no clear intention of future vaccination, displaying hesitancy (30.4%) or outright refusal (36.9%). Conversely, 32.7% of this cohort expressed willingness to be vaccinated but reported not having access to vaccines. Moreover, among those who had received one or two doses, only 43.7% (n = 318) expressed intent to receive a booster dose, while 23.2% (n = 169) were hesitant, and 33.0% (n = 240) outright refused it.

These Sankey diagrams visualize networks of individuals displaying refusal, hesitancy, and access barriers to COVID-19 vaccines (Panel (A)) or vaccine booster doses (Panel (B)) within health districts.

In Tables S4-S6 , we assessed the characteristics of participants exhibiting varying attitudes towards COVID-19 vaccination. Participants who outright refused vaccination and booster doses were notably clustered in the Lukelenga health district (OR = 1.5; p = 0.003), in the Diulu health district (OR = 1.7; p < 0.001) among males (OR = 1.3; p = 0.033) and people who were still single (OR = 1.4; p = 0.011). They often harbored misconceptions regarding the nature of COVID-19 (OR = 2.2; p < 0.001) while expressing confidence in their “innate” protection against the virus (OR = 1.5; p < 0.001) and holding skepticism towards the usefulness of vaccines (OR = 0.5; p < 0.001). Conversely, individuals displaying hesitancy towards vaccination and boosters were more frequently identified as residents of the Lukelenge health district (OR = 2.0; p < 0.001). They reported no conviction in their “innate” protection against COVID-19 illness (OR = 4; p < 0.001). While expressing trust in the safety of vaccines (OR = 2.7; p < 0.001), as those who refused vaccines, they were also skeptic of the usefulness of vaccines (OR = 2; p < 0.001). Additionally, this group was more likely to be unemployed (OR = 1.5; p = 0.000). Finally, participants facing limited access to vaccination predominantly hailed from the Muya health district (OR = 2.7; p < 0.001). They exhibited no confidence in their natural protection against COVID-19 (OR = 1.6; p = 0.016).

We employed hierarchical logistic regression models to explore the determinants of COVID-19 vaccine non-observance, with geographic location treated as a random effect. Our final multivariate model yielded a marginal R² of 0.43, indicative of substantial explanatory capability. Key predictors independently associated with vaccine non-adherence encompassed the health district of residency, age, marital status, occupation, and attitudes towards COVID-19 and its vaccines, as illustrated in Figure 5 and Figure S1 .

People residing in Diulu and Muya health districts exhibited significantly increased odds of vaccine non-observance compared to those in the Lukalenge health district, with aORs of 3.94 [95% CI: 2.60 - 6.06; p < 0.001] and 3.46 [95% CI: 2.25 - 5.38; p < 0.001], respectively. Single participants exhibited 1.7 times higher odds of non-vaccination compared to married individuals [aOR = 1.66; 95% CI: 1.23 - 2.23; p = 0.002]. Unemployment of participants was linked to 1.6 times higher odds of non-vaccination relative to employment or schooling [aOR = 1.61; 95% CI: 1.18 - 2.21; p = 0.003]. Likewise, participants holding misconception around COVID-19 and its vaccines respectively experienced a 2.1-fold increase [aOR = 2.09; 95% CI: 1.64 - 2.66; p < 0.001] and a 3.3-fold increase [aOR = 3.27; 95% CI: 2.4 - 4.5; p < 0.001] in the likelihood of non-vaccination, compared to those with accurate perceptions. Conversely, each incremental year in age was associated with approximately a 2% decrease in the odds of vaccine non-observance [aOR = 0.98; 95% CI: 0.97 - 0.99; p = 0.002].

The Forest plot presented here elucidates the impact of pivotal factors on the odds of COVID-19 vaccine non-observance within the framework of the final multivariate model. Each horizontal line within the plot depicts the effect of an individual factor, represented as an adjusted Odd Ratio (aOR), delineated by a corresponding box along with its associated 95% confidence interval. Factors positioned to intersect the vertical line set at aOR = 1 indicate negligible influence on non-compliance with COVID-19 vaccination. Conversely, factors positioned away from this line, either above or below, significantly decrease or increase the likelihood of non-compliance, respectively.

The study reveals alarming trends in COVID-19 vaccination uptake, indicating a coverage rate of only 49.1% within the studied population, with many individuals receiving only a single dose. Two years after mass vaccination efforts began, this low uptake deviates from targets set by health ministries and WHO recommendations [8] [14] . This situation could have extended the population’s vulnerability to the virus, allowing for ongoing transmission and potentially leading to the emergence of new viral strains [25] . Additionally, it might have imposed extra pressure on the healthcare system and the local economy [14] . The vaccination coverage in the study population likely reflects the broader landscape nationwide, supported by evidence from online surveys and reports indicating approximately 50% coverage across the country [24] [26] . This finding highlights the need for targeted interventions to overcome vaccination barriers and boost community-wide vaccine acceptance in the study population and nationwide.

Refusal, hesitancy, and access limitations likely played varying but significant roles in the low vaccination rates observed in the study population. It is noteworthy that individuals expressing vaccination refusal or hesitancy, as well as those facing access limitations, exhibited distinct profiles, providing valuable insights for developing targeted strategies to address low vaccination uptake [27] [28] . Vaccine access limitations could indeed be a consequence of global access inequities worsened by the COVID-19 pandemic [29] . Additionally, locally organized brief vaccination campaigns, spanning a total of 45 days divided into 4 phases, may have contributed to lower uptake rates in the study population. Besides population behaviors, challenges hindering vaccination efforts in many resource-limited regions likely encompass inadequate healthcare infrastructure, funding constraints, limited public awareness, and suboptimal government policies [30] [31] . Consequently, the observed low vaccine coverage may be attributed to both vaccine scarcity and inadequacies in vaccination campaign implementation. These findings underscore the importance of optimizing vaccine distribution strategies and enhancing accessibility to achieve broader coverage. Extending the vaccination campaign duration and its integration have the potential to enhance the overall vaccination coverage.

Conversely, more than half of unvaccinated individuals expressed hesitancy or outright refusal of vaccination. Likewise, among those who had received one or two vaccine doses, the majority were hesitant about receiving a booster dose. This reveals a concerning level of reluctance towards COVID-19 vaccination in the population. Consistently, before the initiation of COVID-19 vaccination in the country, surveys conducted in the capital city indicated that the majority of people were not willing to receive the vaccine [32] . Among African countries, the DRC displayed the lowest willingness for COVID-19 vaccination [33] . Respondents’ reasons for vaccine reluctance, including a general distrust of vaccines, suggested that this issue may have longstanding roots predating the COVID-19 pandemic. Indeed, there are well-documented historical instances of vaccine hesitancy in the country, such as the mass boycott of the polio vaccine in the 1990s [34] . Effective education campaigns are needed to improve vaccine acceptance and readiness for future health crises that may require vaccination efforts.

Besides, fear of vaccine Adverse Events (AEs) could have also played role in vaccine hesitancy or refusal. While concerns about AEs have been commonly raised with both initial vaccination and boosters across different populations, approved COVID-19 vaccines have demonstrated safety and effectiveness on a global scale [35] - [40] . Reports of AEs such as stroke and myocarditis may have fueled negative campaigns against mass vaccination during the pandemic [38] [41] - [44] . Nonetheless, the benefits of COVID-19 vaccination appeared to outweigh the risks of AEs, thereby reinforcing their recommendation where available [27] [36] [44] - [46] . Additional reasons provided by respondents for not getting vaccinated highlight the significant role of misinformation. Therefore, the spread of misinformation, especially through social networks, and the politicization of scientific knowledge observed during the pandemic likely posed a threat to vaccine acceptance across the country [47] [48] . Moreover, the widespread perception that Africa was less at risk from COVID-19 may have contributed to reinforcing refusals and hesitation, widening the gap between vaccination efforts and vaccination coverage [49] . These findings highlight the need for accurate, evidence-based information, which should be readily accessible to the population. Moreover, there is a pressing need for the implementation of robust policies to further disseminate knowledge concerning the AEs, effectiveness, and safety of vaccines [50] - [52] .

Factors contributing to non-compliance with COVID-19 vaccines have been extensively examined, revealing a range of predictors, including sociodemographic factors, professional conditions, individual health, perceptions of vaccines, social phenomena, and information received [53] [54] . These factors vary across surveys, necessitating tailored investigations for specific populations [53] - [56] .

During the current survey, significant spatial disparities in vaccine uptake were observed across the study area, indicating the need to consider geographic clustering when modeling non-vaccination factors [57] . Residents in certain HDs showed higher odds of vaccine non-compliance, highlighting the importance of equitable spatial distribution of vaccination services. Globally, geographic factors have been pivotal in determining the distribution and accessibility of COVID-19 vaccines, resulting in varying vaccination rates across regions [58] . Urban areas have generally seen more efficient vaccine rollout due to superior health infrastructure and logistical capabilities, whereas rural and suburban areas face challenges in vaccine accessibility and administration [10] . International collaborations have been initiated and have played a crucial role in addressing disparities in vaccine coverage within Africa [59] . However, local-scale spatial variations, such as those observed in this survey, have received less attention in vaccination policies. The Ministry of Health (MoH) established 19 vaccination sites simultaneously, each with similar campaign strategies to serve estimated populations of 479,459, 448,745, or 351,838 inhabitants in Diulu, Muya, or Lukelenge [60] .

Age is inversely correlated with vaccine non-observance, with older individuals displaying higher vaccination rates. The influence of age on COVID-19 vaccination has varied across Africa, with some studies reporting higher acceptance among younger age groups [61] - [65] . The DRC’s vaccination policies may have influenced the age distribution of vaccinated individuals, initially prioritizing older populations, individuals with comorbidities, and healthcare workers [14] [66] . Over time, eligibility expanded to younger age groups [14] . However, children under 12 and pregnant women remained ineligible, with limited access for those aged 12 to 18. Despite lacking evidence on age groups’ roles in disease dynamics, policymakers should have adjusted target populations based on global vaccine safety and efficacy data [40] . Misinformation on social media and perceived lack of severity of COVID-19 among young people may have contributed to hesitancy or refusal to vaccinate.

Marital status independently influenced vaccine observance, with single individuals having nearly twice the odds of non-observance compared to married counterparts. Inconsistent findings in the literature emphasize the importance of population-specific strategies to enhance vaccine acceptance [53] - [55] . While marital status may not directly impact COVID-19 vaccine compliance, household dynamics and caregiving responsibilities could shape perceptions of vaccine necessity. Married individuals in our study area often reside in multigenerational households, prioritizing vaccination due to caregiving duties, potentially affecting vaccine uptake. Further research is needed in this regard. Additionally, unemployed individuals had 1.61 odds of non-observance compared to their employed counterparts. Occupations with frequent public interaction may heighten perceived exposure risk, prompting vaccination prioritization. Nevertheless, the relationship between profession and vaccination warrants nuanced examination, notably considering healthcare workers’ hesitancy despite high exposure risks [67] [68] .

Finally, misconceptions about COVID-19 and its vaccines significantly magnified vaccine non-uptake rates 2 to 3 times in the study population. As reported elsewhere [69] - [71] , misinformation, often intertwined with specific religious beliefs, could have fueled distrust in COVID-19 vaccination efforts. Targeted public health campaigns are vital to address and correct these misconceptions, fostering confidence in vaccination programs and enhancing pandemic preparedness [71] [72] . Messages should be tailored to reflect sociodemographic characteristics and cultural norms, ensuring accessibility and relevance to diverse populations [73] . Transparency about vaccine development, approval, and dissemination, along with providing information on health policy-making processes and scientific evidence, can help dispel myths and enhance overall vaccine acceptance [73] - [75] . Vaccination campaigns should prioritize equitable access for high-risk individuals and settings to ensure community-wide protection against COVID-19 [76] .

This study had several limitations. First, focusing solely on recipients’ perspectives may have overlooked systemic issues like supply chain weaknesses. Second, reliance on self-reports could have introduced biases, and age restrictions might have overestimated coverage. Third, findings might not apply to the entire Congolese population due to specific collection sites, and the cross-sectional design could have limited capturing evolving dynamics. Further research in diverse regions is warranted for broader insights. However, given the deceleration of vaccination campaigns, significant changes in coverage post-survey are unlikely.