Study design
A population-based household survey was conducted to assess vaccination coverage and its characteristics in the Colombian Amazon department among children above 6 months and < 8 years old. Age was restricted because only children < 8 years. old were exposed to the vaccination schedule under study.
Study population
The municipalities included in the survey were Leticia, Puerto Nariño, and Tarapacá. Leticia is the capital of the department of the Amazon with an estimated population of 39,667 inhabitants with 37 % living in rural areas. Puerto Nariño is the second most populated municipality with 7574 inhabitants, 73.3 % living in the rural area. Tarapacá is a rural area with small human settlements scattered along the banks of the Putumayo and the Cothue rivers with a population of 3992 inhabitants. Leticia and Puerto Nariño were chosen because they have the largest population in the department. Tarapacá was selected as representative of the most isolated rural areas where a substantial proportion of the Amazon population lived.
Sample size
The main goal of this study was to assess the prevalence of HBV infection and HBsAg carriage among children living in rural areas of the Amazon. Sample size was calculated to assess an HBsAg carriage prevalence of 1 % within a confidence limit of 0.5 and 95 % of confidence level. The number needed to fulfill those assumptions was estimated to be ~1300 children. At the end, more than 900 children < 8 years were available for the analysis on vaccine coverage. That sample size would be enough to estimate a vaccination coverage of 80 % with confidence limits of 8 %.
Sampling criteria
For Leticia and Puerto Nariño, only rural areas settled along the Amazon river (pop < 8 years ≈ 2700 according to EPI field census) were included in the study because HBsAg carriage was considerably higher and vaccination was lower than in urban settlements or rural areas placed elsewhere. [6–8]. Most people belong to aboriginal tribes and live under the poverty line with low access to health and sanitation services. In Leticia 12 of 21 villages were included; in Puerto Nariño 17 villages out of 20 were included and in Tarapacá, 8 communities out of 13 were included. Six communities were excluded because of a very low population (only one family) or language constraints. One community (YAGUAS) rejected to participate in the study.
Community and tribal leaders in every municipality were invited to a meeting, conducted by the researchers, where they were briefed about the following issues: objectives of the study, importance of hepatitis B infection for their communities, relevance of the vaccination program for hepatitis B control, and the relevance of monitoring infection prevalence periodically.
Once a community approved to participate in the study, field workers started household visits in order to find out the information on vaccination status of every child who match the inclusion criteria within the household. All children born from 2004 onwards, present at the time of the visit and whose parents consented to participate in the study were included. A household was defined as all those sleeping under the same roof regularly. The field work team was composed of bacteriologists, nurses, and a field coordinator. The survey was conducted between July 2011 and March 2012.
For Leticia and Puerto Nariño participants were selected from villages settled along the banks of the Amazon River and the Loretoyaco River. It was restricted to those settlements because they tend to have lower coverages and higher HBV prevalence than areas accessible by road [6].
Selection of participants
In the villages selected for the study every household was visited and parents invited to participate in the study. All eligible children found in any given household were included in the study if their caretakers approved to participate. Households were visited starting from the farthest to the nearest to the center of the community.
Hepatitis B vaccination data
Vaccination dates were recorded from vaccination cards. When vaccination cards were not available or difficult to read the Expanded Program on Immunization (EPI) databases were consulted. These databases are updated by the EPI nurses and contain vaccination data from every child living in rural areas. Recall data was not used. A timely birth dose should be given in the first 24 h after delivery as recommended by the WHO [1]; however in our study, vaccination cards show dates of birth and administration of monovalent HBV dose but lack data on both, exact hour of birth or hour of vaccine administration. Therefore, a timely birth dose was defined as a monovalent dose of the hepatitis B vaccine administered on the same date of delivery or the day after. A timely vaccinated child was described as having a timely monovalent dose plus 3 doses of pentavalent vaccine given at 2, 4 and 6 months with at least 30 days between each dose.
The research protocol was approved by the medical ethics committee of the school of medicine, Universidad Nacional de Colombia, October 8, 2010. All participants were informed of the purpose and procedures of the study and all the parents or legal guardians gave their signed consent for their children to participate.
Independent variables
The following variables were collected : 1) vaccination status, 2) mother’s education level, 3) level of wealth measured by ownership of boat or electric appliances, 4) overcrowding (defined as more than 5 people per room), 5) mother’s ethnic group, 5) children sex, 7) site of birth (home, health care facility other), 8) person attending birth, 9) child’s health insurance regime for which there were three options: belonging to the Contributive regime or to the Subsidized regime. In Contributive regime workers in the formal sector pay a monthly sum and have access to a wider health care set of services. The Subsidized regime is designed for informal workers who pay a lower sum and are entitled to a more restrictive health care portfolio. It should be emphasized that hepatitis B vaccine is provided freely by the Colombian M o H and every child has the right to be vaccinated regardless his/her insurance scheme.
Statistical analysis
Information was entered into databases using EpiInfo 3.5.3 ®, and statistical analysis were performed in SPSS 19® and Stata 11.1 ® software. Coverage with the birth dose, coverage with a timely birth dose and mean time to the monovalent hepatitis B vaccine dose were described by birth cohort. An inverse Kaplan-Meier curve was used to describe coverage and timeliness of vaccination with the monovalent birth dose. To identify predictors of receiving a timely birth dose of monovalent hepatitis B vaccine, bivariate analysis were conducted relating it to the following independent variables: municipality, mother’s education level, household wealth, crowding, ethnic group, child sex, site of birth, and child’s health insurance. Odds Ratios (OR), 95 % confidence intervals, and p values were estimated to assess the strength and statistical significance of those associations. All independent variables found related to timely monovalent dose (p < 0.1) were included in a multivariate model (unconditional logistic regression) using a stepwise approach.
Children were selected using villages as the sampling unit, so children in a specific village tended to be similar, but different, in a range of characteristics, compared to children in a different village. This “clustering effect” may lead to underestimate the magnitude of the variance of the OR estimates. Hence, a mixed-effects logistic regression analysis was performed to control for that cluster effect. The model was built using the same strategies described for the unconditional logistic regression. Results from both models were compared and, if no important difference was observed, only results from the simple model was presented. Timely vaccination with pentavalent was assessed in a similar way.
To assess the direction and magnitude of potential selection biases introduced by children without vaccination card, additional analysis were conducted. Children without a valid vaccination card, whose vaccination record could not be recovered from EPI local registries, were classified as no vaccinated and were included in the analysis. Then, changes in ORs values were noted and appraised if they occur. To assess the influence of potential information bias, an additional analysis was conducted classifying as unvaccinated the 30 children without vaccination card.