Rates of coverage and determinants of complete vaccination of children in rural areas of Burkina Faso (1998-2003)

Study population

Our study population was children aged 12-23 months at the time of the survey [26]. In all, the study involved 805 children (898 after weighting) distributed among 44 rural health districts, for the year 1998, and 1,360 children (1,461 after weighting) distributed among 48 rural health districts, for the year 2003. During the DHS data analysis, to ensure representativeness of the sample, a weighting was applied that took into account the method of sampling [9, 10]. An administrative reform increased the number of districts in 2003.

Sources of data

Three data sources were used.

1. The Demographic and Health Surveys (DHS) of 1998-1999 and 2003 [9, 10]: the sample, drawn from a random sampling stratified to two levels, is representative at both the national level and the level of residential areas (urban/rural). Burkina is divided into communes, which are basic territorial communities organized into sectors or villages. There are two types of communes, urban and rural. The rural commune is a collection of villages that has a combined population of at least 5,000 inhabitants and whose economic activities generate budgetary resources of at least 5 million CFA francs [27]. The data were collected through direct interviews in people's homes, and those related to vaccination were gathered from vaccination cards or mothers' statements in the absence of written documentation.

2. The Statistical Yearbooks of 1997 and 2002 [28, 29]: These are prepared annually from reports and information gathered from all the health facilities and administrative structures in the country's healthcare system. They provide information on the country's health facilities coverage (structures and staffing) and health services utilization.

3. To better understand the evolution of the factors that might explain complete vaccination of children, we carried out individual interviews with decision-makers at the central and regional levels, as well as with field workers in Burkina's healthcare system. These were semi-directed, face-to-face interviews, recorded by the first author (DS) and transcribed under his supervision. We interviewed nine decision-makers at the central level and four at the regional level, three field workers, and five representatives of non-governmental organizations.

Ethical approval was granted by the Comité d'éthique pour la recherche en santé of Burkina Faso (agreement N° 2007-056 from November 8, 2007).

Variables

The dependent variable is the child's vaccination status, which is dichotomous: the child is either completely vaccinated or not. Completely vaccinated children are those who have received the BCG, measles, and yellow fever vaccines as well as the three doses of DTP and oral polio vaccines during their first year of life, according to either their vaccination cards or their mothers' statement.

Statistical analyses

Comparison of factors associated with complete vaccination of children, determined by logistical regression in 1998 and 2003

In 1998 a child from a well-off household was 1.7 times more likely to be completely vaccinated compared to one from a poor household, but in 2003 the standard of living did not appear to have an effect on the probability of being completely vaccinated. In 2003, having fewer children in the household was associated with a higher probability of being vaccinated. This factor was not associated with complete vaccination in 1998. In addition, the positive relationship observed in 1998 between complete vaccination of children and continuity from prenatal care to institutional delivery appeared even more marked in 2003.

Systematic variation of complete vaccination in 1998 and in 2003

Variance analysis of a factor with random effects showed systematic variation in full vaccination both in 1998 (χ² = 11.262 to one degree of freedom (DF) with p = 0.00079) and in 2003 (χ² = 9.167 to one DF with p = 0.0024642). However, the proportion of this variation that is attributable to differences between health districts is five times less in 2003 (7.5%) than was seen in 1998 (37.14%).

Comparison of factors associated with complete vaccination of children, determined by multilevel analysis in 1998 and 2003 (final model)

In both 1998 and 2003 the effects of the level 1 variables did not vary by health district and the models constructed are random intercept models.

At level 1, continuity from prenatal care to assisted delivery remains associated with vaccination status in both 1998 and 2003. In 1998 the standard of living was associated with vaccination status while in 2003 the number of other children aged five or under became significantly linked with complete vaccination.

At level 2, in 1998 each one-unit (i.e., 1%) increase in the proportion of educated women in a health district increased children's chances of being completely vaccinated by a factor of 1.14; in 2003 the impact of the proportion of educated women in a health district was not significant. In both 1998 and 2003, no interaction was found among the variables of level 1, nor between variables of levels 1 and 2 (Table 2). Variables related to district resources (mean number of inhabitants per vaccinator staff) and to accessibility (average catchment area, mean number of inhabitants per CSPS) had no significant impact on complete vaccination in either 1998 or 2003.

Comparison of complete vaccination coverage of children aged 12-23 months in 1998 and 2003 and key events for EPI

Even if complete vaccination coverage has not yet reached the desired level within this context of multiple vaccination-promotion initiatives, it did increase significantly in 2003 (41.2%) in comparison with 1998 (25.9%), with a chi-square of 60.39 to one degree of freedom and p < 0.000001.

Increase in vaccination coverage

Reduction in vaccination coverage disparities

Differences between health districts in relation to complete vaccination of children tended to lessen between 1998 and 2003. In fact, even though the healthcare system was decentralized after the creation of health districts in 1994 [33], with district management teams being given a certain amount of autonomy, the central authority at the Ministry of Health retained a major role in decision-making. Among the domains kept at that level were the supply of vaccines and consumables, funding, equipment for the cold chain and rolling stock, and, to a lesser extent, the regulation and distribution of development partners whenever possible, as was revealed by the various actors in the individual interviews. It must be noted that vaccination is funded within a normative framework, based on action plans developed by the health districts. The functioning of this regulatory framework for funding vaccination activities in health districts is recognized and accepted by all actors in the healthcare system (central, regional and district levels, NGOs). This is illustrated by the statements of various resource persons. One NGO representative explained: "At the beginning of the year, we inform the health districts and the regional health administrations which activities will be funded. This way, people know what activities to put into their action plans to get them funded." A manager at the district level spoke along the same lines: "Our action plan is based on a planning framework (directives) from the central authority. These directives tell us that this year, for vaccination, here's how much you can put in as activities that are in accordance with national policy. Then it is adopted by the board of the health district." He added that, "the funding agencies, through a concerted action framework, are already in agreement with the central authority, and when they come to the resource allocation session, there isn't much discussion."

This involvement of the central government has the benefit of making the districts comparable in terms of funding and supplies of material resources and, to a lesser extent, in terms of organization. On the other hand, decisions taken at the central government level, by means of the planning framework sent to the health districts and the specifications imposed by development partners regarding activities they are prepared to fund, leave little room for each district to resolve specific local problems, which will not be included in the health districts' action plans because they are not targeted by funding agencies. Over time, this situation can discourage actors at the operational level from developing any initiatives. For example, if there is a delay in receiving funding, actors in the field have no alternative but to adjust their action plans, even if they would be able to pre-finance the activity concerned, as is recounted by one actor at the regional level: "The Ministry of Health no longer allows pre-financing, which means the districts have to revise their action plans. This affects the morale of the workers and also, indirectly, vaccination, and especially its quality." Another case in point is the fight against meningitis; a person at the central level told: "It was recommended that the regions and health districts should be able to include, in their action plans, a certain number of activities related to fighting meningitis epidemics. These epidemics had become so frequent that they were listed as priority issues in all the districts. Unfortunately, this still hasn't been done, but we continue to ask for these activities to be included in the health districts' action plans so that they can benefit from some funding sessions." In fact, the resolution of problems that are specific to each district in general, and to the area of each CSPS in particular, is the responsibility of the health centres' management committees; as was said by a regional administrator, "the operational funding of vaccination is local." It seems reasonable to assume that the remaining differences between health districts could be explained, in part, by the community's involvement in these districts' health centres and by whether or not the management committee, which in principle reflects the community, functions well.

Poverty

Both in 1998 and in 2003, our results confirm that poverty and resource constraints are an impediment to complete vaccination of children in rural areas [34]. Independently of poverty, in 2003, we found a link between the number of children aged five and under in the household and the probability of being vaccinated. It may be that in five years the situation has become more worrisome [35], such that despite vaccination being free [20], households have become more careful about the use of family resources, which affects the priority given to children and increasingly takes into account the opportunity costs related to wait times, missed appointments, and absence from work in the fields and in the market. Continued poverty would motivate households to ration resources more carefully as the number of children grows [36].

Experience of healthcare services utilization

The experience of healthcare (maternity) services utilization by the mother appears to be the most stable and strongest explanatory factor for complete vaccination of children. Mothers' loyalty to vaccination services depends on the quality of that experience. In fact, it was this concept of loyalty that led to using the DTP3 rate to assess the performance of vaccination services and thus their capacity to retain their users [37].

Education

In this study, factors related to education did not appear to play a fundamental role in complete vaccination of children. In 1998, the community's level of education appeared to be an explanatory factor for complete vaccination in children, but this was no longer the case in 2003. Our observation differs from what has been reported in other contexts [38–42]. This reflects an increasing popularization of vaccination in a population that is becoming more organized into associations that deal with specific issues and, at the same time, become useful channels of communication. In addition, vaccination's benefits and positive effects are widely recognized. People attribute the drop in cases of measles and poliomyelitis to vaccination, and when they speak, for example, of poliomyelitis, they refer to cases of paralysis in the past, noting that these have become rare thanks to the drops that children receive. The following statements from resource persons illustrate this. According to a decision maker at the central level, "Before, when you entered many homes, you saw poliomyelitic children dragging along, and this was frequent. But the disabled persons we encounter today are of a certain age, they are no longer the very youngest ones; that has become the exception. Measles was very common. In our region, there was an adage that if your child had not yet had measles, he was not yours. It was even said that, if you hadn't had measles by the time you died, you would get it in your tomb--all to say, measles was inevitable. But we have seen that with vaccination, it is avoidable, and its impact in terms of morbidity and mortality has greatly diminished. All these observations have been made in recent years." An administrator at the regional level said, "I had both pertussis and measles, because I wasn't vaccinated. But today our children don't know what pertussis is, and measles has become sporadic." In such a context, it seems reasonable that a mother's ability to decide for herself to use curative medical services creates a favourable situation for vaccination, which, moreover, has been declared a free service [20].