Study design, population and study site
This household survey was conducted among sexually active women aged 18–45 years, from the general population in two local government areas (LGA) in Ibadan. Two contiguous peri-urban communities were selected in Akinyele LGA: (Moniya [population: 53,000] and Sasa [population: 25,000]). Mokola [population: 77,000], a high-density urban community in Ibadan North LGA was the second study site. These study sites were selected because they are heterogenous in culture, religion and socio-economic/demographic profile. Each community is comprised of smaller enumeration areas (EA)—this is a small, compact area, carved out of a bigger locality or a group of localities with well-defined and identifiable boundaries. It usually has a unique name and code for identification. EAs represent the primary sampling unit during census.
The study excluded young adolescents (< 18 years), pregnant or nursing mothers, women not resident at the study sites and those that declined participation including refusal for biological sample collection. Assuming an alpha of 0.05 and a design effect of 2 owing to the clustered sampling design, a sample size of 300 was calculated to be able to estimate the prevalence of HPV and determine associated risk factors in each of the four anatomic sites with adequate precision among women. The research assistants, nurses and laboratory staff had an intensive three-week training and participated in community engagement before the study commenced.
Study procedures
Sampling and enrolment of study participants
Eligible people were selected through a two-stage sampling technique. The first stage sampling involved random selection of four EAs each at Mokola and Moniya, and one EA at Sasa by probability proportional to size using the 2006 National Population Commission Census lists of the EAs [15]. Each house was assigned a unique identification number. House listing of women aged 18–45 years were done by female research assistants. The house listing data from each of the study sites were stratified into two age groups, young adults (18–24 years) and older adults (25–45 years), to serve as a sampling frame. The second stage sampling involved a systematic random sampling of eligible women from each sampling frame till the desired sample size of 310 was achieved.
Eligible participants were randomly enrolled at home by female research assistants who explained the study objectives and provided potential participants with an information leaflet in English or in a local language (Yoruba/Hausa/Igbo). Individuals that agreed to participate were given appointment to be seen in a local primary healthcare centre. Potential participants had reminder phone calls at 72 h and 24 h before and a short message service on the morning of their clinic appointment.
Clinic visit, interview, sample collections and follow-up
At the clinic, participants signed a written or witnessed informed consent following a repeat explanation of study procedures, including collection of biological samples. A face-to-face interview was conducted in English or in Yoruba for participants that could not understand English. The interview covered socio-demographics, sexual behaviours (e.g. vaginal, oral and anal sex) and hygiene practices, intravaginal practices, alcohol, smoking and stimulant use, previous STI and awareness about HPV vaccine. After the interview, a female nurse collected blood for a rapid diagnostic HIV test and other biological samples. Samples from participants who did not wish to know their HIV were tested by an RDT in the laboratory. This HIV testing protocol was based on the National Guidelines for HIV Prevention Treatment and Care, Federal Ministry of Health, Nigeria [16]. HIV positive participants were referred for management following national protocols.
Samples were also collected by a nurse from the mouth, cervix, vulvar and anal cavity in separate sample bottles. Briefly, an oral sample was collected using a 30 s oral rinse and gargle method with 10mls of Scope mouth wash (Procter & Gamble©). The nurse demonstrated the rinse and gargle procedure to a participant prior to sample collection. The participant sample was then collected into a 10 ml labelled sample bottle. and placed immediately into a cold box (2–4 °C). For the vulvar sample, the tip of a Dacron swab was used to rub the introitus on either side of the vaginal orifice without touching the urethral orifice. The cervical sample was then collected by inserting a sterile Cusco speculum into the vagina to expose the cervix. The tip of a new Dacron swab was inserted into the cervical os and gently rotated 360 degrees with care being taken to avoid trauma to the cervix and potential bleeding before removing it. An anal sample was collected with the participant in a left lateral position. A Dacron swab was inserted into the anal canal (about 5–6 cm beyond the anal verge) and rotated 360 degrees with gentle pressure around the anal verge before removal. Each of the samples collected with swabs were placed in separate 2 ml cryotubes that were labelled with a barcoded sticker prior to being placed into a cold box (2–4 °C). All samples were subsequently stored in a − 80 °C freezer at the Institute of Advanced Medical Research and Training, College of Medicine, University of Ibadan, Nigeria until shipment to Catalan Institute of Oncology, Spain. Health related incentives (a bar of soap, toothbrush and paste), a soft drink and biscuit and 1000 Naira (2.8GBP) to cover transportation.
HPV DNA sample analysis
HPV genotyping of the samples was performed at the Catalan Institute of Oncology, Spain using the Anyplex™ II HPV28 assay (Seegene, Seoul, South Korea). This assay detects 28 HPV genotypes, including HR-HPV (HPV-16, -18, -31, -33, -35, -39, -45, -51, -52, -56, -58, -59 and -68), LR-HPV (HPV-6, -11, -40, -42, -43, -44, -53, -54 and -70) and possibly carcinogenic genotypes (HPV-26, -61, -66, -69, -73 and -82).
DNA extraction from the mouthwash was performed using the Maxwell® 16 LEV Blood DNA kit (Promega Corp., Madison, WI, USA). Briefly, mouthwash liquid samples were spin at 12.000xg for 20 min. After discarding supernatant, cell pellet was eluted in 300ul of lysis buffer + 30ul Proteinase K and DNA extraction performed following the manufacture’s instructions. DNA was extracted from cervical, vulvar and anal dry swabs using the Maxwell 16 Buccal swab LEV DNA Purification kit (Promega Corp., Madison, WI, USA). Briefly, dry swabs were eluted in 300 ul of lysis buffer + 30 ul Proteinase K, incubated for 20 min. at 56 °C and spin at max. speed for 2 min. DNA extraction from eluted samples was performed following the manufacture’s instructions.
The Anyplex™ II HPV28 detection test was performed as recommended by the manufacturer. The data recording and interpretation were automated with Seegene viewer software (Seegene, Seoul, South Korea). Beta (β)-globin was used as the internal control to demonstrate the presence of human DNA in each sample and to determine invalid samples. However, any sample that was positive for HPV but negative for Beta (β) globin was considered as valid.
Data management
Data were double entered into REDCap software (Vanderbilt University, Nashville Tennessee, USA). Thereafter, the raw data were exported in CSV format and saved. The data were imported into STATA 16.0 (Stata 2019. Statistical Software: Release 16. College Station, TX: StataCorp LLC) software for analysis.
The participants’ selected descriptive variables were summarised using frequencies and proportions for the categorical variables and mean and standard deviations for the continuous variable. The primary outcome was prevalence of any HPV infection. The prevalence of HR-HPV and LR-HPV infection and groups classified according to the 2009 IARC epidemiological oncogenic classification (Groups 1, 2a, 2b and 3) for each of the anatomical sites were also calculated with their 95% confidence intervals. The trends of association between each classification of HPV infection and the age group of participants were calculated with ANOVA.
A conceptual framework for the risk factor analysis of any HPV infection was developed (Additional file 1: Fig. S1). Associations with any HPV infection were explored to determine independent risk factors for infection. Any HPV infection was treated as a binary outcome; each anatomical site—oral, cervical, vulva and anal cavity—was analysed separately. Logistic regression was applied to obtain crude estimates for the association between any HPV infection and potential risk factors. Adjusted estimates were obtained using a hierarchical modelling technique. Age group and study sites were included in the adjusted estimates a priori. Level 1 sociodemographic variables included ethnicity, religion, highest educational level, ever had Qur’anic education, current occupation, monthly income and current marital status. Each variable was added one by one to the model that included age group and study site. P values were obtained by likelihood ratio tests. Any variable that met a p value ≤ 0.10 was included in the adjusted model. All level 1 variables were adjusted.
Level 2 behavioural variables included age at first vaginal sex, age difference between first vaginal sex partner and the participant, lifetime number of vaginal sex partners, ever cleansed vagina, condom use during last vaginal sex, ever had oral sex (given or received), history of transactional sex, ever had mutual masturbation, history of female genital mutilation, alcohol use, illicit drug use, ever had STIs and ever heard of HPV. Each level 2 variable was added one by one to a model that included level 1 variables that met a p value cut off of ≤ 0.1 after adjusting the ‘core variables’. Any level 2 variable that met a p value cut off of ≤ 0.1 was included with the level 1 core variables in the level 2 adjusted model. Level 3 biological variables were laboratory detection of concurrent HPV infection from the other three anatomical sites apart from the outcome measure. For example, if the outcome measure was to determine risk factor for any cervical HPV, concomitant detection in vulvar, anal and oral sites were included as explanatory variables. Each level 3 variable was added one by one to a model that included level 1 ‘core variables’ and level 2 factors that met a p value cut off of ≤ 0.1. Any level 3 variable that met a p value cut off of ≤ 0.1 was included with the level 1 and level 2 ‘core variables’ in the level 3 adjusted model. This strategy allowed the effects of variables at each level of the framework to be assessed, adjusted for more distal variables.
The concordance of HPV between oral, cervical, vulvar and anal samples in individual participant was defined as the presence of the same type of virus across the four sites. The proportion of concordance for specific HPV type was calculated as the number of each HPV type in all the four sites, any of three and any of two anatomical sites.