Study design

A cross-sectional study design was employed to collect data from travelers embarking on international travel through the airport.

Study setting

This study was conducted at Murtala Muhammed International Airport (MMIA) in Nigeria. Nigeria’s premier international air gateway. MMIA is located in Lagos and was commissioned in 1978 [30]. According to the Nigerian Bureau of Statistics, 3,202,837 passengers traveled through the international airport in 2019 [31]. The Port Health Services, a government agency located within the airport arena, provides pre-travel health vaccination services. The activities at the airport are overseen by the Federal Airport Authority of Nigeria (FAAN).

Recruitment process

This study was planned for January 2020, but data collection was carried out from October to November 2020 after easing the COVID-19 lockdown in Nigeria. Data collection took place daily at the departure gates of MMIA based on a pre-determined and Institutional Review Boards (IRB) approved study protocol.

Participants – inclusion and exclusion criteria and sample size

We included travelers aged 18 years and above and excluded travelers using MMIA for transfer or transit to other countries. A brief screening survey was conducted to identify eligible participants. The sample size was determined using Leslie Fisher’s formula for prevalence studies. The assumed proportion of YF vaccination was 76% from a previous study, and the level of precision was set at 0.04 [32]. The computed sample size was increased by 10% to compensate for non-responses. This gave a total sample size of 486.

Participants recruitment: sampling technique and selection of study participantsStage 1: selection of travelers’ destinations based on WHO regions

Four out of the six WHO regions were included in this study [33]. The four areas were African, American, European, and Eastern Mediterranean because Nigeria does not have direct flights to the remaining two regions (i.e. South-East Asian and Western Pacific Regions).

Stage 2: proportional allocation of travelers to WHO regions using a stratified sampling method

Based on the 2019 Winter flight schedule received from the Commercial Travel Section of the Operation Unit of MMIA, the expected monthly traveler volume was estimated to be 223,704. This was proportionally allocated to 4 WHO regions to ensure the representativeness of travelers and travel destinations. The African region, comprising 113,316, was further stratified into the West African, East African, Central African, and Southern African regions.

Stage 3: selection of study participants using simple random sampling

A simple random sampling method was used to select study participants per region. The samples were randomly selected using electronically generated random numbers. The data collection was completed in four weeks using a daily data collection plan to meet the data target for each WHO region. Questionnaires were administered at the departure gates of each flight. If a selected traveler declines to participate in the study, the traveler representing the following randomly selected number was approached to participate.

Data collection

A semi-structured questionnaire, adapted from the International Health Travel Questionnaires [7, 13] and questionnaires from previous studies, was used to collect relevant data from study participants (supplementary document 1). The questionnaire was pre-tested at the Nnamdi Azikiwe International Airport, Abuja, Nigeria.

Variables

The independent variables of interest were educational level, nationality, Cholera endemicity at the destination, YF endemicity at the destination, traveler’s awareness of pre-travel vaccination, and inspection of vaccination cards during the previous trip(s). Their covariates are listed as a footnote under Table 1.

Table 1 Multivariable logistic regression analysis of the factors associated with pre-travel health practices among respondents

The dependent variable is pre-travel health practices (a composite score of the practice of pre-travel consultation/advice, pre-travel vaccination status, and pre-travel preventive measures). Pre-travel consultation or advice was defined as information obtained from a health professional [22]. Pre-travel consultation was categorized as taken and not taken; vaccination was categorized as vaccinated and not vaccinated based on reported vaccination for YF among those traveling to endemic countries. Self-reported YF vaccination was used as a proxy for vaccination because it was readily accessible at the Nigerian Port Health Services compared to the Cholera vaccine. To validate the vaccination status, we inspected the vaccination cards of travelers. Pre-travel preventive measures were categorized as taken and not taken among travelers going to YF endemic and Plague endemic regions. The preventive measures assessed were actions to prevent (a) the bite of infected vector fleas for Plague and mosquitoes for YF, (b) contact with infectious bodily fluids or contaminated materials, and (c) the inhalation of respiratory droplets/small particles from a patient with pneumonic Plague.

The composite score of pre-travel health practices ranged from 0 to 3. Having taken pre-travel consultation/advice was scored 1 while not taken was scored 0, pre-travel vaccination was scored 1 while non-vaccination was scored 0, and having taken pre-travel preventive measures was scored 1 while non-practice was scored 0. Those with a score of 3 were categorized as having good practices, while those with scores of 2 or less were categorized as having poor practices. We ensured no overlap of the scores to prevent misclassification of travelers.

Statistical analysis

Data analysis was done using IBM SPSS Statistics 25.0. Descriptive statistics were presented in frequency tables, while pre-travel health practices were visually represented using a pie chart. A binary logistic regression analysis was employed to examine the association between the independent and dependent variables of interest and pre-travel health practices. Furthermore, a multivariable analysis was conducted to reduce the influence of potentially confounding variables, controlling for relevant covariates (Table 1). Before fitting the model, variables with cell counts less than 5 were recategorized. Educational level was recategorized into 3 levels: secondary/high school or less, undergraduate or College, and Graduate (Masters/Ph.D./Professional). The undergraduate or College and Graduate (Masters/Ph.D./Professional) were also referred to as higher educational levels under the discussion.

We used a directed acyclic graph (DAG) to identify possible confounders of our selected exposures and controlled for them in the multivariable analysis. The significance level was set at 5%, and the odds ratio was presented with a 95% confidence interval.