Factors contributing to low uptake of vaccination in underserved communities: a mixed-method diagnostic design in Southwest region of Cameroon

¹Rural Doctors, Buea, Cameroon, ²Regional Delegation for Public Health, Southwest, Buea, Cameroon, ³Faculty of Health Sciences, University of Buea, Buea, Cameroon

^&Corresponding author
Edgar Mandeng Ma Linwa, Rural Doctors, Buea, Cameroon

Introduction    

In 2024, World Health Organization (WHO) and United Nations International Children's Emergency Fund (UNICEF) reported that 89% of infants globally (115 million) received at least one diphtheria, tetanus and pertussis (DTP) vaccine dose and 85% (109 million) completed all three, representing gains of 171,000 and 1 million children compared to 2023 [1]. Yet, 20 million missed at least one DTP vaccine dose, including 14.3 million zero-dose children, which is 4 million above the 2024 Immunization Agenda 2030 target and 1.4 million more than in 2019, highlighting the urgent need to address vaccine hesitancy and access barriers in underserved communities [1]. In Cameroon, full vaccination coverage for children aged 12-23 months grew from 40% in 1991 to 52% in 2018 but fell 1% from 2011, with ZD children doubling from 5% to 10% in that period, while WHO/UNICEF National Immunization Coverage (WUENIC) data show DTP1 coverage remained stable at 75% and DTP3 coverage slightly increased from 67% in 2019 to 68% in 2022, despite COVID-19-related urban declines [2,3]. In Yaoundé (Cameroon), 37% of children aged 12-59 months were fully immunized, while 16% had never received any vaccine; living in wealthier households, having a mother with primary or higher education, and living with either parent significantly reduced the likelihood of being zero-dose [4]. In 2021, the highest rates of ZD children were reported in the Southwest (31% of children) and Northwest (24%) regions of Cameroon, two regions affected by sociopolitical unrest [3]. It is likely that the burden and factors linked to vaccine hesitancy and zero-dose vaccination in rural, conflict-affected areas differ markedly from those in stable regions like Yaoundé. For example, in Ethiopia, being unmarried [Odds Ratio (OR)=1.8], older child age (24-35 months) (OR=3.7), and greater distance to health facilities (OR=1.4), were additional factors linked to ZD prevalence [5]. Other factors fuelling ZD prevalence in sub-Saharan Africa include no formal paternal education, limited decision-making power in mothers, financial barriers, lack of media access, low community literacy and reduced country-level health expenditure [6]. Limited data on ZDand vaccine hesitancy drivers in conflict zones like Cameroon's Southwest region highlight the need for mixed-method research to understand complex phenomena and inform health policymakers, aligning with Newman et al. (2003) research purposes of understanding phenomena and informing constituencies [7]. The objectives are threefold: i) describe ZD prevalence and correlates; ii) explore barriers and facilitators to vaccination; and iii) explain relationships between socio-cultural factors and hesitancy.

Methods     

This study adhered to the guidelines for Conducting and Reporting Mixed Research in the Field of Counselling and Beyond [8]

Study design: the study utilised a fully mixed concurrent equal-status convergent parallel design collecting and analysing quantitative (cross-sectional survey) and qualitative (FGDs) data simultaneously during the same phase, with equal emphasis to address the research questions [9]. This design will enable triangulation of quantitative statistical associations (e.g., prevalence) with qualitative narratives (e.g. barriers) for corroboration, validation, and a comprehensive understanding of low uptake in low-performing settings. The convergent design was chosen due to limited data collection time, equal value of both data types, and the research team´s expertise in quantitative and qualitative methods. A pragmatic paradigm guided the study, integrating diverse data under a unified framework focusing on practical solutions rather than absolute truths, judging truth by its consequences, and supporting mixed methods despite philosophical inconsistencies [10].

Research questions: the study was guided by mixed research questions: i)What is the prevalence of ZD children in the Southwest region of Cameroon, and what demographic and access factors are associated? ii)What are parents´ perceptions and experiences of vaccination barriers? iii)How do quantitative associations align with qualitative barriers (e.g., trust, myths) to explain low uptake?

Sampling procedures

Quantitative phase: the quantitative phase employed multistage sampling. First, a purposive selection of the two health districts (Tombel and Eyumojock) with the lowest reported DTP3 coverage in the Southwest Region (39% and 35%, respectively) was made based on unpublished 2023 EPI data. Within these districts, health areas and communities were purposively selected to represent six predefined community types: remote/rural, conflict-affected, religious groups, displaced populations, urban slums, and ethnic minorities. The sample size for the cross-sectional survey was calculated to provide a precise estimate of the zero-dose (ZD) prevalence, the study´s primary quantitative outcome. Based on the most recent available regional estimate of 31% ZD prevalence in the Southwest Region in 2021 [3], a minimum sample size was determined using the single proportion formula:

Qualitative phase: the qualitative phase employed purposive criterion-based sampling to engage community stakeholders such as parents, community health workers, and community leaders. Two FGDs (8-10 per group) were held, one in Eyumojock and one in Tombel, ensuring gender balance and community type representativeness, to generate information-rich perspectives, aligning with recommendations [11,12]. FGDs were chosen over individual interviews for their cultural fit in communal settings, fostering authentic dialogue and group synergy that reveal social norms and collective barriers [11,13]. Thematic analysis followed the process outlined by Braun and Clarke (2006) [14]. Limitations, such as potential group thinking and silent voices, were addressed through skilled moderation [12]. Findings are presented narratively with illustrative quotes; frequencies are not reported for FGD data to avoid inappropriate quantification of perspectives from a small, purposive sample. Full questionnaire and interview guide can be found in supplementary files. These sampling schemes ensured analytic generalisations and case-to-case transfer, rather than mere statistical representativeness [15].

Data collection: mixed data collection strategies included structured surveys and semi-structured FGDs [16]. Questionnaires were adapted from a WHO-standard demographic and health survey module [17], and grounded in the Health Belief Model (HBM) [18]. The sample questionnaire used can be found in Annex 1. It assessed ZD vaccination status, demographics (age, education, income, ethnicity, religion, distance to health facility), household composition, knowledge (3 items), attitudes (4 items), perceived susceptibility/severity (2 items), benefits/barriers (4 items), and cues to action/self-efficacy (2 items). We also used the vaccine acceptance scale (20-item validated scale, α=0.97 historical [19], 0.47 current study). The questionnaires were administered via Kobo Toolbox by 15 trained enumerators in community settings, achieving a 100% completion rate. Focused group discussions used a piloted guide probing vaccination experiences, community beliefs, influencers, and barriers. They were conducted in pidgin English (60 minutes, audio-recorded) in community halls. Visual aids (flipcharts) and skilled moderation ensured engagement. Quality was ensured through enumerator training, pilot testing, and audio verification by an independent researcher.

Data analysis: data analysis integrated the convergent parallel design and followed Onwuegbuzie and Teddlie´s (2003) seven-stage mixed-methods framework, which includes: a) data reduction; quantitative (descriptives, Chi-square for sociodemographics, HBM, vaccine acceptance; univariable and multivariable logistic regression logistic regression, R v4.2, odds ratio) and qualitative (thematic coding of FGD beliefs, barriers, NVivo 12); b) data display (tables, matrices); c) data transformation (quantitating themes e.g., myth frequency); d) data correlation (e.g., acceptance scores vs. themes); e) data consolidation (joint displays); f) data comparison (predictors vs. themes); g) data integration (meta inferences linking quantitative predictors and qualitative barriers to inform tailored interventions). Univariable logistic regression was performed for all potential predictors. Variables with p < 0.200 in univariable analysis or theoretical importance were considered for the initial multivariable model. A backward stepwise selection procedure was used. Model performance was assessed using the likelihood ratio test, classification accuracy, and pseudo-R² measures. Variables such as hearing and visual impairment were excluded due to concerns about measurement validity. Backward stepwise selection was applied, iteratively removing non-significant variables until the best-fitting model was achieved. Zero-dose status was assessed via VS13 ('Has any of your children aged less than 2 years received any type of vaccination?’).

Data validation and trustworthiness: quantitative validity was ensured by addressing internal threats such as selection bias via multistage sampling across the six most vulnerable community types. External validity was also ensured through context-specific generalizability to conflict-affected settings. A key limitation was the low internal consistency (Cronbach´s ?=0.47) of the multi-item vaccine acceptance scale, indicating poor reliability. Consequently, this scale was not used as a composite measure, and findings related to "vaccine acceptance" are based on the single global willingness item (VS14) [9,20]. Qualitative trustworthiness was achieved through thick descriptions of vaccination experiences, beliefs, and barriers using NVivo 12. Mixed-methods legitimation included: a) sample integration (parallel quantitative and qualitative samples targeting sociodemographics, HBM constructs, and community perspectives); b) weakness minimization (triangulation of survey predictors with FGD themes); and c) multiple validities (quantitative statistical rigor, qualitative credibility) [21]. Recall bias was mitigated through enumerator training and audio verification. Mixed criteria ensured design suitability, implementation fidelity, and integrative efficacy for zero-dose prevalence and barriers [22].

Operational definition of terms: i) zero-dose: children under two years who received no routine vaccinations, as reported by parents in survey item VS13 (Has any of your children aged less than 2 years received any type of vaccination?); ii) vaccine acceptance: parental willingness to vaccinate their child under two for some or all routine vaccines, measured by survey item VS14 (Would you consider getting your child vaccinated against common vaccine-preventable diseases like diphtheria, tetanus, pertussis- hepatitis B and Haemophilus influenzae type b?”); ii) self-efficacy/confidence: parents´ belief in their ability to ensure their child´s vaccination, assessed by survey item SS15 through a 6 point Likert scale; iv) Remote community: areas with limited health facility access (e.g., >30-minute walk), identified by geographic isolation in Tombel and Eyumojock; v) Conflict-affected Ccmmunity: areas impacted by ongoing violence or insecurity, that have disrupted local health services and mobility; vi) religious Community: groups defined by shared religious beliefs influencing health decisions, and uncategorisable in other community types; vii) displaced community: populations forced to relocate due to conflict or other crises, per parental self-report in the survey; viii) urban slums: informal urban settlements with poor infrastructure and limited health access, identified by residential characteristics in the survey; ix) ethnic minorities: groups with distinct cultural or linguistic identities, per parental self-report in the survey.

Ethical considerations: the study received approval from the Institutional Review Board of the Faculty of Health Sciences (N°2024/2572-08/UB/SG/IRB/FHS of 28 August 2024). Informed consent was obtained, confidentiality ensured, and no incentives were provided. The report adheres to validated reporting guidelines [8].

Results     

Quantitative analysis: a total of 424 parents were surveyed. The prevalence of zero-dose vaccination was 9.2% (39/424) and did not differ significantly across community types (χ² 4.59, p=0.468). Strikingly, stated willingness to vaccinate was very high overall (98.1%) and among parents of ZD children (89.7%). The final multivariable logistic regression model (n=423) included three predictors (Table 1). The model fit was significant (Likelihood Ratio χ²=14.40, df=3, p=0.002) but explained limited variance (Nagelkerke R²=0.07). The significant predictors were: i) lower self-efficacy/confidence in ensuring child vaccination (protective effect of high confidence: aOR = 0.50, 95% CI: 0.25-1.00; p = 0.049); ii) disagreement that vaccines are a major advancement for humanity (aOR = 2.35, 95% CI: 1.08-5.10; p = 0.031); iii) perceived seriousness of vaccine-preventable diseases was marginally associated (aOR = 1.86, 95% CI: 0.93-3.72; p = 0.079).

Qualitative analysis: two focus group discussions (FGDs) with parents in Eyumojock (n=9) and Tombel (n=8), identified barriers and facilitators to vaccination through the lens of the Health Belief Model (HBM). In Eyumojock, participants perceived high susceptibility to and severity of vaccine-preventable diseases, valuing vaccines for protection (EHA1, perceived benefits). However, some participants reported mixed feelings, with fear of injection pain reducing perceived benefits (EHA4, perceived barriers). Misinformation, such as vaccines causing population control (EHA5), heightened perceived barriers, alongside access issues like insecurity and stockouts (EHA3, EHA6). Community chiefs and religious leaders were key cues to action (EHA3, EHA4), with many participants suggesting incentives and education to enhance self-efficacy (EHA1, EHA4). In Tombel, participants endorsed vaccine efficacy (THA06, perceived benefits), but some feared poor administration (THA03, perceived barriers), and noted Apostolic Faith resistance (THA03, perceived barriers). Misinformation about vaccine harm was prevalent (THA02), undermining perceived benefits. Health workers and chiefs were trusted cues to action (THA03, THA04), with some participants recommending training and formal communication to boost self-efficacy (THA04, THA06). Across districts, misinformation and access barriers dominated, with Tombel emphasizing administration fears and Eyumojock highlighting insecurity. Full thematic analyses with thick descriptions for Eyumojock and Tombel can be found in Annex 1.

Meta-inferences: Table 2 presents the joint display of key meta-inferences and indicate; i) convergence on an access-acceptance paradox: high quantitative willingness (98%) converges with qualitative recognition of vaccine benefits, yet qualitative access barriers explain the quantitative link between low self-efficacy and ZD status; ii) convergence on misinformation: quantitative disagreement with vaccines as an advancement aligns with qualitative myths, reducing perceived benefit. iii) divergence on severity: qualitative discussions acknowledged disease severity, but this did not emerge as a strong differentiating factor for ZD status in the quantitative model, suggesting other factors are more salient.

Discussion     

This mixed-methods study in Cameroon´s conflict-affected SWR reveals a critical paradox: extremely high parental willingness to vaccinate (98.1%) coexists with a persistent, though lower-than-expected, zero-dose prevalence of 9.2%. This finding challenges the framing of the issue purely as "vaccine hesitancy" and redirects focus to structural and delivery system failures as the primary drivers of zero-dose status in this context. The significantly lower ZD prevalence found (9.2% vs. the 31% regional estimate for 2021) requires careful interpretation. It may reflect genuine progress due to intensified local microplanning and outreach campaigns in these districts [3,23]. However, methodological factors, including potential sampling bias towards more accessible households within insecure areas and the reliance on unverified parental recall for ZD status, may also contribute to an underestimate. The true prevalence likely lies between these figures, but the dramatic difference underscores the potential impact of focused community engagement. Our regression model, while identifying significant cognitive predictors (low perceived benefit, low self-efficacy), had critical limitations. Its inability to correctly identify any ZD child (0% sensitivity) strongly illustrates that, unmeasured, predominantly structural variables, such as stockouts, episodic insecurity, and variable service delivery, may be the principal determinants. This aligns perfectly with our qualitative findings where these access barriers dominated narratives. The quantitative link between low self-efficacy and ZD status likely captures parents' realistic assessment of these systemic obstacles, not a psychological deficit. The high willingness to vaccinate, even among most ZD parents, is a crucial asset. It suggests demand is not the bottleneck. This aligns with studies emphasizing systemic barriers in fragile settings [24]. and contrasts with studies from more stable urban areas where hesitancy is more prominent [4]. Our qualitative data on trusted leaders (chiefs, pastors) as key influencers corroborates evidence from Ethiopia and elsewhere on the importance of community networks [5].

The 9.2% zero-dose prevalence aligns with WHO/UNICEF national estimates for Cameroon (10% in 2018), but is notably lower than the 31% reported for the Southwest region in 2021, likely reflecting localized interventions such as microplanning (including district-level mapping and community engagement to target zero-dose children), the Handshake Model integrating vaccination with community services, and geospatial catch-up campaigns targeting missed communities [3,23]. High vaccine acceptance suggests awareness of vaccine benefits, corroborated by qualitative themes valuing disease prevention. However, misinformation, particularly myths about population control, undermines uptake, though agreement with vaccine protection indicates that most ZDV parents believe in vaccine efficacy, suggesting potential for targeted education. Low self-confidence may stem from logistical barriers (e.g., stockouts, insecurity in Eyumojock), reducing parents´ perceived ability to vaccinate. Tombel´s focus on administration fears reflects distrust in health worker competence, a concern amplified in conflict zones with strained health systems [24,25]. Eyumojock´s emphasis on insecurity highlights how sociopolitical unrest disrupts access [24].

Limitations include the cross-sectional design, preventing causal inference; the poor reliability of the vaccine acceptance scale, confining acceptance analysis to a single item; potential social desirability bias in stated willingness; and possible under-sampling of the most inaccessible, security-compromised households, potentially biasing the ZD prevalence estimate downwards. The use of parent-reported, unverified vaccination status is a notable source of potential misclassification. For policy and practice, the primary implication is to shift resources from persuasive communication campaigns, which our data suggest are less urgently needed, toward solving logistical and access problems. Investments should prioritize: i) predictable, resilient vaccine supply chains to eliminate stockouts; ii) reaching the last mile through funded mobile clinics and community health worker networks; formally engaging and training trusted community leaders as vaccine advocates to counter misinformation and mobilize communities. Future research should employ longitudinal designs and integrate verified vaccination data with real-time metrics of access (e.g., stockout frequency, security incidents) to better model the dynamics of zero-dose in conflict settings.

Conclusion     

In conflict-affected Southwest Cameroon, the barrier to childhood vaccination may not be parental refusal but systemic failure. High willingness to vaccinate may be hindered by stockouts, distance, insecurity, and misinformation. This demands a strategic shift from persuasion to access: prioritize reliable supply chains, mobile outreach to remote communities, and formal engagement of trusted local leaders to counter myths. These actions are critical to convert high acceptance into coverage and achieve immunization equity.

Competing interests     

The authors declare no competing interests.

Authors' contributions     

Study concept and design: Edgar Mandeng Ma Linwa, Pamela Oben Besong, Michael Ngenge Budzi and Nelson Sontsa Njedock. Data collection: Jamin Ghangha, Ismail Shifu, Martha Ngoe, Kibu Odette Dzemo and Neris Mabou Nfor. Analysis and interpretation of data: Edgar Mandeng Ma Linwa and Samuel Nkengfua. Manuscript writing: Edgar Mandeng Ma Linwa. Final approval of manuscript: All authors. Dickson-Shey Nsagha supervised the study. Edgar Mandeng Ma Linwa and Samuel Nkengfua had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. All authors have read and approved the final vresion of this manuscript.

Acknowledgments     

We thank all participants who accepted to participate in this study.

Tables     

Table 1: factors associated with zero-dose vaccination status (n=424)

Table 2: meta-Inferred joint display of qualitative and quantitative data

Annexes     

Annex 1: supplementary materials (PDF - 440KB )

References