Study design and setting

This community-based cross-sectional study was conducted in the Buea Health District (BHD) from August 2023 to March 2024. Buea, the regional capital of the South West Region of Cameroon, has an estimated population of 177,297 inhabitants and is the home to the largest English-speaking university in the country, the University of Buea. The BHD covers a total surface area of 870 km² and consists of seven health areas: Bokwango, Bova, Buea Road, Buea Town, Molyko, Muea, and Tole.

Participants

We included parents of children aged 9–18 years who were residents of the BHD and consented to participate in the study. We excluded individuals visiting the BHD and those who withdrew consent during the study.

Sample size and sampling

We estimated that a minimum of 943 participants would be needed to detect an odds ratio (OR) of 1.5 with a power of 80% and a two-sided alpha of 0.05. The proportion (0.31) of parental HPV hesitancy was obtained from a study conducted in Northwest Ethiopia²⁰.

We used a two-stage sampling method to recruit participants. In the first stage, we randomly selected four of the seven health areas in the BHD: Bokwango, Bueatown, Molyko, and Muea. In the second stage, parents were recruited by convenient sampling from community centres (e.g., marketplaces, schools, hospitals, meeting houses, places of worship) and their homes. Participants were selected using convenience sampling because it requires significantly fewer financial and human resources compared to random sampling²¹. The number of participants recruited per health area was allocated proportionally to the population size of each health area, as estimated by the South West Regional Delegation of Public Health.

Data collection

Five individuals were selected from the study areas and trained in interviewing participants and data recording to minimise the risk of social desirability and interviewer bias during data collection. Data were collected using a modified version of the WHO’s standardised questionnaire (version 2.0) designed to assess COVID-19 vaccine hesitancy in adults²². The COVID-19 questionnaire consisted of seven sections: sociodemographic factors, medical history, history of vaccination, perceived risk of COVID-19, confidence in the vaccine, willingness to receive vaccine, and social influences on vaccination decisions. The questionnaire for the present study retained most items from the original WHO questionnaire but excluded questions not relevant to the present study, including perceived stigma related to the participants getting HPV and freedom of participant to meet family and friends without infecting them.

The questionnaire for the following study was structured as follows:

• Sociodemographic factors: age, gender, marital status (single, married, cohabiting, divorced, widow/widower), highest level of education, religion, employment status, monthly household income (in FCFA), and number of children.

• Medical history: History of chronic illness (yes or no).

• History of vaccination: Receipt of recommended and other vaccines for children (yes or no) and parental vaccination (yes or no).

• HPV-related awareness: Awareness of HPV (yes or no), HPV vaccine (yes, no, or not sure), cervical cancer (yes or no), or HPV as a cause of cervical cancer (yes or no).

• Perceived risk of cervical cancer: Perceived risk of cervical cancer as serious disease (yes or no).

• Perceived vaccine safety: Perceived safety of HPV vaccine for children (safe, not safe, or not sure) and concerns about severe vaccine adverse effects (yes, no, or not sure).

• Social influence: Participants think their religious leaders will support children receiving the HPV vaccine (yes, no, not sure); participants have heard any bad news regarding the HPV vaccine (yes or no).

• Trust: Trust in pharmaceutical companies manufacturing vaccines (yes, no, or somewhat) and trust in the MOH (yes, no, or somewhat).

• Willingness to vaccinate child: Participants were asked whether their children had received the HPV vaccine (yes, no, or not sure) and, if not, whether they intended to vaccinate them in the future (yes, no, or not sure).

The adapted questionnaire was then pretested among 20 randomly selected parents living in the Tole health area who were not included in the study. The questionnaire was administered via interviews in English and took 15 min on average to complete.

Outcomes and definitions

The primary outcome of this study was parental HPV vaccine hesitancy, defined as parents’ unwillingness to accept the HPV vaccine for their female children. Parental HPV vaccine hesitancy was assessed using the question, “Has your child taken the HPV vaccine?” and the responses were “Yes,” “No,” and “Not sure.” Participants who responded “No” or “Not sure” to the previous question were asked a follow-up question: “Do you intend to give them the vaccine?”; a “No” or “Not sure” response depicted HPV vaccine hesitancy.

The secondary outcomes were HPV infection and cervical cancer awareness. Cervical cancer awareness was assessed by asking, “Have you heard of cervical cancer?” and HPV awareness was assessed using the question “Have you heard of HPV?”. The responses were either “Yes” or “No.” Participants who responded “No” were considered unaware.

Ethical approval and considerations

Ethical approvalwas obtained from the Institutional Review Board of the University of Buea (2023/1949-02), followed by administrative approval from the South West Regional Delegation of Public Health. This study was conducted in accordance with the principles of the Declaration of Helsinki. Participation in the survey was voluntary, and participants’ autonomy and anonymity were assured. All participants provided informed consent before enrolment into this study.

Statistical analysis

All analyses were performed using R programming software (version 4.3.1, 2023, The R Foundation for Statistical Computing, Vienna, Austria). Categorical variables were summarised using frequency and percentage, while quantitative variables were summarised using mean and standard deviation (SD).

Multivariable logistic regression generated adjusted ORs and 95% confidence intervals (CIs) for factors associated with lack of awareness of HPV infection and cervical cancer after adjusting for age, age-squared, sex, health area, marital status, education, employment status, monthly household income, and history of chronic disease (objective 1). To investigate the factors associated with parents’ hesitancy towards HPV vaccination for their children (objective 2), we additionally adjusted for the number of children and the number of children squared as covariates.

The responses of participants recruited from one health area might be more similar than those of participants from other health areas, introducing clustering into the data. Clustering underestimates standard errors from models such as generalised linear models and inflates type I error rates, although it does not affect estimates of the regression coefficients²³. To account for clustering by health area, the “vcovCL” function from the sandwish package was used for clustered covariance estimation and the “coeftest” function from the lmtest package was used to generate robust standard errors and p-values.

For analyses of ordered categorical variables (like level of education), the Floating Absolute Risk (FAR) method was used to calculate the variance of the log OR for each category, including the reference category²³. This group-specific variance reflects the amount of information in that category and was used to calculate the group-specific CI. This method enables pairwise comparisons of the odds of an outcome variable (e.g., vaccine hesitancy) across any two categories rather than restricting comparisons between other categories to an arbitrarily defined reference category²³. The group-specific FARs and their 95% CIs were then plotted to visualise the shape of associations. The group-specific 95% CIs are narrower than classical CIs and were, therefore, only used for plotting; robust CIs were used when reporting the odds in one group compared to the reference category²³.

For analysis of the lack of awareness of HPV and cervical cancer, two-tailed p-values less than 0.05 were considered statistically significant. For analysis of vaccine hesitancy, two-tailed p-values were corrected using Bonferroni’s method by dividing the conventional threshold (0.05) by the number of exposures investigated (0.05/19 = 0.0026). Therefore, two-tailed p-values less than the Bonferroni threshold of 0.0026 were considered statistically significant. There were no missing data in the dataset used for this study.