Inequitable coverage of vitamin A supplementation in Nigeria and implications for childhood blindness

The data used were collected from the Nigeria Demographic and Health Survey (NDHS) 2013 and the published literature on population-based childhood blindness surveys in Nigeria. The Demographic and Health Surveys in Nigeria are implemented by the National Population Commission, with developmental assistance from partners such as the United Nations Population Fund (UNFPA) and the United Kingdom Department for International Development (DFID).

A steering committee is responsible for coordination and oversight and consists of high-level stakeholders from the National Bureau of Statistics, the Ministry of Health and the National Population Commission. The goal of the NDHS is to generate data to assist policy makers make informed policy choices and develop strategies for improving population health outcomes [16].

The NDHS is a nationally representative source of data on demographic and health statistics and collects information on child health characteristics, including maternal health factors, from a nationally representative sample of households. Sample selection was done by two -stage cluster sampling- first by selecting clusters from a list of enumeration areas, next by systematically selecting households from a list in the selected clusters using the 2006 census frame.

Data collection was by questionnaire. For the females, only women aged 15–49 were selected. Information sought included socio-demographic characteristics - age, religion, education, literacy, marriage, health characteristics, fertility preferences, antenatal, delivery, and postnatal care; breastfeeding and infant feeding practices, vitamin A supplementation, child immunisation and childhood illnesses. For mothers with children 6–59 months of age, data collected on relevant immunisations and vitamin A supplementation were based on mother’s recall and immunisation card where available. In addition, the household wealth index was calculated as a score of weighted household assets (such as ownership of vehicles or household facilities) and national level wealth quintiles were obtained. The response rate of women 15–49 years in the 2013 NDHS was 97.6% [16]. The detailed methodology can be found elsewhere [16].

Data sources for the childhood blindness surveys were from three population based childhood blindness surveys in three geopolitical zones in Nigeria; Northwest [15], South east [13] South-south [14] zones. These three studies used the key informant methodology to identify children suspected of having visual impairment, to determine population-based estimates and the causes of childhood blindness [17].

The Children’s Data Recode File component of the data from the Nigeria Demographic and Health Survey (NDHS) for 2013 was retrieved as the database for this study. After reviewing the detailed data coding, further data cleaning was performed by the researchers. Data for 24,327 children aged 6–59 months were included in this analysis.

For the NDHS survey, VAS coverage was defined as the percentage of children 6–59 months who received VAS within the 6 months preceding the survey [16].

Study factors comprised a range of socioeconomic, and individual factors, and their inclusion in the present analysis was based on findings from similar previously published studies in Sub-Saharan African countries like Guinea [18] Ethiopia [19] and a UNICEF summary of global VAS coverage [4] where an association was found between these variables and VAS coverage. These include age of the index child, urban/rural residence, geopolitical zone, mothers’ educational attainment (categorised as no schooling, primary or secondary and higher education), and household wealth index. The household wealth index was used to categorise households into wealth quintiles (as ‘poorest’, ‘poorer’, ‘middle’, ‘richer’ or ‘richest’) [16].

Data were analysed using STATA V.12.1 (Statcorp, Texas). To allow for the cluster sampling survey design used in the NDHS, appropriate weights were used to restore the representativeness of the sample and analyses were performed using “svy” commands in STATA.

The main outcome measure was the proportion of children (6–59 months) who received at least one dose of Vitamin A in the 6 months preceding the survey. To explore the effects of the independent variables on VAS coverage, bivariate regression models were fitted separately for the 2013 NDHS data set. Subsequently, a multiple regression model was developed to assess the effects of variables found to be significant at p ≤ 0.05. These included socioeconomic, maternal education and regional variables. Models were restricted to children aged 6–59 months, living with respondents (eligible women aged 15–49 years). Both crude and adjusted odds ratios (OR) are presented with 95% confidence intervals. All tests were two-sided, and variables with p < 0.05 in the final multivariable model were considered as independent factors.

For the population-based childhood blindness surveys, causes of blindness were classified as corneal blindness and other causes. Odds ratios were computed to determine the odds of corneal blindness in each geopolitical region. Tests of significance were set at the 5% level.

The odds of a child in the South-south zone receiving VAS were almost two and a half times that of a child in the northwest zone. This zonal inequity has also been recorded in two previous NDHS in Nigeria [28, 29]. Wide intra country disparities in VAS have been reported in several countries [18‐20]. In northern Nigeria which is predominantly muslim, cultural and religious beliefs may affect the uptake of health interventions, especially for maternal and child health [22, 30, 31]. It has been reported that issues pertaining to fathers’ approval significantly affect the uptake of immunisation or VAS in the northwest [22] and northeast [32] regions in Nigeria. However, the contribution of fathers’ approval to poor uptake of health intervention in the northern geopolitical zones needs to be interpreted with caution as the same may apply in other areas of the country where similar studies were not done. It was previously suggested that supply- side disparities in distribution of health facilities was a contributory factor for the health inequities between northern and southern Nigeria. However, recent research suggests that this is not the case and residents in the north have more physical access to vaccination centres than their southern counterparts suggesting that any disparities are likely due to socio-cultural differences [33]. This suggests that strategies to improve the uptake of health interventions should include targeted advocacy to community and religious leaders and male heads of households. A qualitative study on factors affecting the uptake of immunisation in Nigeria highlights the importance of sensitizing religious leaders especially in northern Nigeria [34].

There is a growing body of literature that suggests that mothers’ education and socioeconomic status correlate directly with improved health outcomes and our study was no exception. In our study, the odds of the child of a mother with higher education receiving VAS was over three times that of the child of a mother without education. This is similar to results from Ghana [24] and India [20]. This may be because maternal education confers more awareness of the benefits of VAS. Research from Mali shows no disadvantage in VAS coverage for the children of uneducated mothers. The decentralised and context specific approach to disseminating health related information implemented in Mali may have reduced the inequity in access to health information [35]. Also, in a study in Sierra Leone, while data on socioeconomic and regional inequities are unavailable, high VAS coverage was achieved within clusters ranging from 86.7 to 97.8%. This was made possible by extensive and context specific social mobilisation [23]. This suggests that health promotion messages that are readily understandable by women without formal education need to be developed and scaled up.

The Africa Vaccination Week (AVW) is an initiative devised by the WHO AFRO to promote vaccination and ensure equity and access to its benefits. Several countries have used the Africa Vaccination Week to boost coverage of maternal and childhood health interventions, including VAS [36]. Nigeria’s equivalent of the (AVW) is the bi-annual Maternal New-born and Child Health Week (MNCHW) which started in 2010 and targets children under 5 years of age, pregnant women and women of child-bearing age with specific interventions. However, a recent evaluation of the initiative reveals that the current model of the MNCHW has failed to reach the most marginalized and there is no evidence that the MNCHW has significantly contributed to coverage of essential MNCHW interventions in Nigeria [8]. It has been suggested that failure of MNCHWs to reach the most marginalised populations in Nigeria is the because the bulk of interventions is facility-based and not community-based, and may not address the barriers of distance, opportunity costs and security [8].

This is reflected in our study which shows a low VAS coverage and high coverage disparities.

However, the NSPAN recommends that VAS be delivered not only in health facilities but also in community structures and campaigns/outreaches [26]. This has the potential to increase VAS coverage.

In addition, we can learn from the successful strategies of other African countries that have achieved high and equitable VAS coverage [18, 35] Such strategies include: developing context specific VAS strategies that are culturally and socially acceptable to the specific target audience [35], and maximizing the potential of the African Vaccination week by meticulous planning, coordination, supervision training and social mobilisation to ensure universal coverage for maternal and child health interventions [23].

Unless stakeholders in Nigeria develop and implement feasible strategies that target underserved and marginalised populations, achieving VAS coverage targets will be high on rhetoric and low on delivery.

In many LMICs, corneal blindness from VAD and measles, which used to be the commonest cause of childhood blindness, is decreasing due to measles immunisation and VAS [37, 38]. In the southeast and south-south zones, corneal blindness was responsible for less than a sixth of all childhood blindness but in the northwest zone with the lowest VAS coverage, over half of children were blind from corneal scarring. Corneal blindness from VAD or measles is a reflection of individual (care giver), community and systemic lapses; there is need for care givers and community members to be sensitised about the importance of child health interventions and the consequences of non-compliance; it is crucial for health workers to understand the importance of VAS and measles immunisation [39]; finally, it is imperative that health systems deliver effective strategies to resolve both demand and supply side constraints to providing universal coverage with child health interventions.

In interpreting the results of this study and other NDHS studies, a few methodological limitations and strengths should be taken into consideration. The major strengths of the study are its national representativeness, large sample size and the availability of individual and household level factors which could influence VAS coverage. A major limitation is that the data on VAS available for analysis did not indicate whether data came from observation of children’s immunization cards, or from recall by mothers. The latter may be subject to recall bias depending on the interval between VAS and the NDHS survey.

Finally, analyses of the childhood blindness data should be interpreted with caution as these are population-based estimates derived from studies with small sample sizes and the data may not be normally distributed. In addition, the key informant method has limitations, and may not identify all children who are blind in a given population. However, due the challenges in the epidemiological investigation of relatively rare diseases like childhood blindness, this method has been used to provide population-based estimates of rare diseases and disability [17].