Several population-based intervention trials have assessed the contribution of vitamin A deficiency (VAD) to child mortality [1–8]. Independent meta-analyses of these trials have indicated that in areas where VAD is prevalent, child mortality is reduced by 23–34 % following vitamin A interventions [9–12]. This significant improvement in child survival has been largely attributed to a reduction in mortality from measles, severe diarrhoea, dysentery, and possibly falciparum malaria [7, 13–15]. A recent Cochrane review has concluded that vitamin A supplementation (VAS) should be given to all children at risk of deficiency, particularly in low and middle income countries [16, 17]. In light of this compelling evidence, WHO recommends that in settings where VAD is a public health problem, children 6–59 months old be provided vitamin A supplements to reduce child morbidity and mortality [18].

In India, VAD has long been recognized as a public health problem [19–21]. Surveys carried out between 2002 and 2005 found that 62 % of preschool-age children had sub-clinical VAD (serum retinol < 20 μg/dl). These surveys indicated that the prevalence of more severe forms of VAD was three times higher among children from schedule caste (SC) or schedule tribe (ST) households [22, 23]. Furthermore, India’s 2006 National Family Health Survey indicated that the proportion of SC/ST children 6–59 months who had received vitamin A supplements in the 6 months preceding the survey was a mere 17 % [24].

Since 2006, the Government of India and India’s State Governments have made a concerted effort to increase the coverage of the VAS programme by strengthening the implementation of bi-annual VAS rounds as part of the national VAS programme, managed by the Ministry of Health and Family Welfare. The programme aims at delivering preventive VAS to children 6–59 months old. For programmatic convenience, the first VAS dose (100,000 IU) is administered with the measles vaccination at ~9 months, while the subsequent eight doses (200,000 IU each) are administered every 6 months up to the age of 59 months using bi-annual VAS rounds as the main delivery platform [25]. The delivery of vitamin A supplements is ensured by the Auxiliary Nurse Midwife (ANM) of the Health Department, supported by the Accredited Social Health Activist (ASHA), the frontline worker of the National Rural Health Mission (NRHM), and the Anganwadi Worker, the frontline worker of Integrated Child Development Services (ICDS) programme, India’s flagship programmes for child health, nutrition and development.

In 2006, the states of Bihar, Chhattisgarh, Jharkhand, Madhya Pradesh, Odisha, Rajasthan and Uttar Pradesh—with 52 % of India’s under-five population (i.e. 59.4 million)—were home to 74 % of the under-five deaths in India, with a combined under-five mortality rate of 84 deaths for every 1000 live births (63 in India) [24, 26, 27]. Furthermore, 32.9 % of the combined population in these seven states survived below a minimum standard of living, with poverty levels as high as 47.3 % in the population quintile with the highest proportion of SC/ST households [28]. The objective of this analysis is to assess whether India’s national VAS programme is reaching the districts with a higher concentration of children who belong to SC/ST households in the seven states of India with the largest burden of under-five mortality.

We used the latest publicly available data by the Office of the Registrar General and Census Commissioner of India to determine the population size, population composition (including the proportion of SC/ST households), and the number of children 6–59 months (SC, ST, and non-SC/ST) in each of the seven states and 255 districts included in our analysis [29].

Data collection and collation procedures took into account the number of children 6–11 months old who received VAS through the routine immunization programme (measles vaccination at ~9 months) and the number of children 12–59 months old who received VAS through the bi-annual rounds. The number of children 6–59 months old who benefitted from the VAS programme in each district was determined using standardized bottom-up data collection and collation approach with data flowing up from the VAS supplementation sites in the village to the block, from the block to the district, and from the district to the state in each given calendar semester (semester 1: Jan 1–June 30; semester 2: Jul 1–Dec 31).

Following international recommendations, VAS coverage was defined as the proportion of eligible children who received at least one VAS dose in a given calendar year while full VAS coverage was defined as the proportion of eligible children who received two VAS doses in a given calendar year. For any given district and calendar year, VAS coverage was computed as that of the semester with the highest VAS coverage whereas full VAS coverage was computed as that of the semester with the lowest VAS coverage, thus assuming that all children who benefitted from the VAS programme in the semester with the lowest VAS coverage also did in the semester with the highest VAS coverage [30].

STATA12 (StataCorp. 2009; Stata: Release 12. Statistical Software. College Station, TX: StataCorp LP) was used for all data analyses. Ethical approval was not sought as we analyse anonymous data that cannot be linked to individual children, caregivers, or household identity.

In 2006, the highest SC/ST quintile reported the highest full VAS coverage (52.6 %) while the two lower SC/ST quintiles reported the lowest full VAS coverage (~37.0 %). Between 2006 and 2011, the full VAS coverage increased in all SC/ST quintiles. This increase was inversely correlated with the SC/ST concentration quintile: highest (42.5 percentage points) in the lowest SC/ST quintile and lowest (8.6 percentage points) in the highest SC/ST quintile (Table 1). However, the mean full VAS coverage over the 6-year period (2006–2011) was positively correlated with the SC/ST concentration quintile: highest (62.5 %) in the highest SC/ST quintile and lowest (47.9 %) in the lowest SC/ST quintile.

Additionally, we analysed the VAS coverage and full VAS coverage by ST concentration by dividing the 255 districts into five quintiles: the lowest quintile comprising the 20 % districts with the lowest concentration of ST households and the highest quintile comprising the 20 % districts with the highest concentration of ST households (Table 2). In 2006, the lowest ST quintile reported the lowest full VAS coverage (9.5 %) while the remaining ST quintiles reported full VAS coverage ranging between 47.3 % and 60.8 %. Between 2006 and 2011, the full VAS coverage increased in all ST quintiles. The highest increase (≥37.6 percentage points) was observed in the two lowest ST quintiles. However, the mean full VAS coverage over the 6-year period (2006–2011) was lowest (31.3 %) in the lowest ST quintile while it was highest (>64.5 %) in the two highest ST quintiles (Table 2; Fig. 1).

		ST concentration quintile
	Year	Lowest	Lower	Middle	Higher	Highest
VAS coverage	2006	26.5	55.9	77.3	71.5	70.0
	2007	35.1	84.5	90.6	92.1	90.2
	2008	61.7	96.3	97.0	103.7	99.1
	2009	86.9	101.5	93.8	98.7	98.4
	2010	61.0	93.9	95.8	99.0	98.6
	2011	81.9	99.7	100.4	93.5	84.9
	2006–11	58.6	88.7	92.5	93.0	90.2
Full VAS coverage	2006	9.5	47.3	60.8	58.7	48.6
	2007	23.2	16.3	39.9	67.3	62.6
	2008	32.1	76.7	68.7	71.6	55.3
	2009	75.5	93.1	64.1	74.9	73.1
	2010	0.0	13.1	46.1	81.0	85.1
	2011	48.2	84.9	69.9	72.6	62.5
	2006–11	31.3	55.3	58.2	70.9	64.5

Note: Coverage figures are indicated as percentages

Between 2006 and 2011, India’s national VAS programme experienced a significant expansion in the seven states with the highest burden of child mortality. The increase in full VAS coverage was observed in all SC/ST quintiles. Importantly, the 2006–11 mean full VAS coverage was highest in the district quintiles with the highest concentration of SC/ST households. A recent review indicates that the critical success factors of the VAS programme in Bihar and Odisha—two of the seven states included in our analysis—have been: strong leadership by the state government, close coordination between the departments of Health and Family Welfare and Women and Child Development, effective district-level planning prior to each biannual VAS round, a stable procurement and distribution mechanism, appropriate training and supervision of staff, and intensive social mobilization and communication [31].

Besides documenting the significant expansion of the VAS programme in India’s seven high burden states, our analysis highlights substantial differences in full VAS coverage between districts with a high concentration of SC households and districts with a high concentration of ST households. The 6-year period mean full VAS coverage was highest in the district quintiles with the highest concentration of ST households but was lowest in the district quintiles with the highest concentration of SC households. This is potentially the result of geographic targeting/emphasis by the VAS programme as ST households tend to cluster in the same geographic area due to their ancestral link to the land where they live. SC households, on the contrary, tend to be geographically mixed with non-SC households and therefore are less likely to benefit from geographical targeting by public health programmes.

Despite impressive increases in full VAS coverage—including in districts with a high concentration of socially disadvantaged households—a large a proportion of children are not fully covered, particularly among SC and ST children, who are potentially among the most vulnerable to VAD and its consequences. Furthermore, as full VAS coverage increases across all SC/ST concentration quintiles, the SC and ST children who are not covered by the programme are increasingly evenly distributed across SC/ST quintiles. Thus, in addition to efforts to scale up and/or consolidate the VAS programme in the states where VAS coverage is low or erratic, states need to give particular attention to expanding the coverage of the programme in sub-district-level units (i.e. blocks and villages) with higher concentrations of SC/ST children, keeping in mind that SC children are less likely to benefit from geographical targeting/emphasis and therefore may require more innovative programming approaches.

It will also be crucial that the state VAS programmes be part of an integrated and sustained assault on VAD that encompasses improved child feeding and increased vitamin A intakes through the life cycle. Studies have documented that 42 % of Indian children are not fed vitamin A-rich foods regularly and that most children do not even meet 50 % of the recommended vitamin A requirements [22–24, 32, 33]. Importantly, studies have shown that Indian children who are not fed vitamin A-rich foods tend to be younger and are less likely to have received vitamin A supplementation in the previous 6 months [34]. This situation underscores: (1) the importance of breast milk as a major source of preformed retinol for children in the first 2 years of life; (2) the importance of vitamin A-rich complementary foods for children 6–23 months old; and (3) the need to universalize the coverage of the VAS programme to enhance children’s vitamin A intake, particularly among the most vulnerable children: the youngest and the socially excluded.

Indian states have achieved significant progress in expanding the coverage of the VAS programme. However, a large proportion of children are not benefitting from this child survival intervention, particularly among SC/ST children. These children are potentially among the most vulnerable to VAD and its consequences. India’s national VAS programme needs to be strengthened in sub-district-level units (i.e. blocks and villages) with higher concentrations of SC/ST children, with particular emphasis on SC children. To accelerate progress in equitable programme coverage, it will be important to strengthen monitoring and reporting mechanisms so that state-level programme planners and managers have access to real-time and accurate information on programme performance by village, block, and district.

In addition, as India approaches the 2015 deadline to achieve the Millennium Development Goals (MDG) targets for child survival, it will be important to reassess the prevalence of clinical and sub-clinical VAD in preschool age children (with appropriate geographic and socio-economic data disaggregation). This will address recent concerns about the extent and severity of VAD in India and the relevance of the national VAS programme [35, 36] and will provide the evidence base to design India’s way forward post 2015.