A cross-sectional survey was conducted between December 2015 and June 2016 among caregivers of children aged 12–59 months in all of Lebanon except the district of Nabatieh and was designed to provide district-based vaccine coverage estimates .
The study sample included resident children in Lebanon, irrespective of their nationality. Sample size calculations were made at the district level, assuming a conservative vaccination coverage of 50%, a desired precision of ±5%, a probability of achieving that precision of 0.95 and a design effect of 2. This led to a required sample of 390 children per district, that is, a total sample of 10,140 children from 26 districts.
Following the World Health Organization (WHO) cluster evaluation survey methodology, 26 clusters were randomly selected in each district proportionally to the population estimates obtained from the Central Administration of Statistics, which are based on a population census of 2009, and the United Nations High Commissioner for Refugees [10,11,12]. The clusters that were likely to be sampled more than once were assigned a fixed number of starting points based on how often they would be selected with certainty. Fifteen children were recruited from each cluster, with each child being selected from a different household.
Households were identified by randomly selecting a landmark from a list of landmarks in each cluster identified prior to the study with the support of local authorities. From the selected landmark, a direction was chosen by spinning a pen or another sharp object to approach the first household. Subsequent households were visited according to proximity, selecting the nearest household to continue data collection. If the end of a street was reached, the neighboring street was chosen following a clockwise approach. Participants were recruited until the total number of 15 was reached within each cluster. In the event that more than one eligible child was found in the same household, each child’s name was written on a slip of paper and the participating child was chosen at random. If the parent or legal guardian was absent during the field visit, the household was revisited at least twice. The same applied to empty houses where fieldworkers were able to inquire from neighbors that an eligible child should be hosted. Facility traceback to find documented evidence of vaccination was piloted and was found not to be feasible.
Sixty data collectors worked on the ground to collect data throughout the districts. All fieldworkers underwent intensive training on field practices, interview techniques and ethical considerations, as well as pilot test activity. A survey questionnaire (Additional file 1) consisting of closed and open-ended questions about demographic information and the child’s immunization status was developed, reviewed and approved by the MoPH, the Lebanon country office of the WHO and the United Nations Children’s Fund (UNICEF). The tool was forward- and back-translated from English to Arabic in order to ensure consistency and pilot tested in the cadasters of Dekweneh and Nabaa in Lebanon. Data collection was performed using paper questionnaires and the electronic KoBoCollect application on tablets. Epidata software was used to enter any nonelectronically collected information. Pictures of the child’s (one or more) immunization card(s) available in the house were also taken. Further, when a child was known to have missed the vaccination, caregivers were asked about reasons why their child had not been vaccinated for each type of vaccine separately.
A response rate of 94.3%, accounting for all surveyed cases irrespective of their nationality, led to a total of 9560 children; however, for this study, we excluded 245 (2.6%) children who were not Lebanese or Syrians living in the communities. Data were analyzed using Stata software, version 14. Descriptive analyses were presented as proportions and means with standard deviations where appropriate. National and district level vaccination coverage estimates and 95% confidence intervals (CI) were calculated based on the Taylor Series Linearization method to retrieve results for the entire sample. National estimates took into account the sampling design (stratum, district and governorate-specific weight). Vaccination cards were used for validating received vaccinations, and coverage rates of children with a well-documented vaccination card were presented. If the vaccination card was incomplete or missing, the recall of caregivers was considered to assess the child’s vaccination status. Dropout rates were calculated as the difference in coverage between the first and third doses for each of the following: polio, DTP, HepB and Hib; the first and second doses for MCV; and the first dose of DTP and first dose of MCV. A multivariable logistic regression analysis of completed vaccination coverage, also accounting for the sampling design, was performed for each vaccine separately. Significance was considered at a p-value < 0.05 following a t-distribution.
Before each interview, oral informed consent was obtained from the child’s caretaker. Written consent was not obtained as this is not a common practice for this type of studies in Lebanon, given the low levels of literacy among certain populations and the non-sensitive nature of the information obtained. All participants were informed of their completely free choice of participation and the strict application of confidentiality to any of the participants’ disclosed information. Prior to the start of any interview, participants received a brief but thorough explanation of the scope and aim of the survey. The Institutional Review Board at Sagesse University approved the study. Only the study team handled the database and pictures to ensure data confidentiality.