This study has revealed that a change in the delivery strategy of interventions in a health system can have important effects on coverage without compromising equity. Our data has shown that vitamin A supplementation coverage in children between 1 and 2 years of age increased from 13% in 1999 to 76% in 2002. National coverage figures from the 1999 Tanzania Reproductive and Child Heath Survey (TRCHS) were similar (14%)  to what we observed in study districts. However more recent national coverage estimates from the 2004/2005 DHS, which includes around 7,000 children , are rather lower at (46%), than we found in our 2002 survey. These differences could be due to geographic variations in coverage or to a drop between 2002 and 2004. In contrast, 85% and 90% coverage at national level have been recently reported by two other sources (Ndossi, G.D. 2004 and Helen Keller International, 2004), the second of which was a population based assessment in 12,000 children from 21 regions of mainland Tanzania. The most likely explanation for the discrepancies between the 2004/05 DHS and HKI surveys, both of which cover the entire country, are a combination of survey timing and differences in questionnaires wording and implementation. The HKI survey was conducted during August 2004, and had a short recall period, as it was little more than a month after the June 2004 VAS campaign. The DHS was conducted from October 2004 to January 2005. Further, the HKI survey question on VAS had a specific point reference of June 2004 whereas the DHS question referred to a 6 month period prior to the survey (giving the scope to capture VAS provided through EPI+ as well as disease targeted approaches, which would have increased coverage estimates). Interview techniques, whether or not mothers were actually shown the capsules, and whether the capsules shown were of the right colour, could also help to explain the differences observed (Mugyabuso, report in preparation). Sampling methodology also differed between the surveys.
We found no evidence of socio-economic inequities in vitamin A supplementation in either 1999 or 2002, despite the major increase in coverage, unlike what has been reported elsewhere . By socio-economic inequity in VAS we mean the trend in coverage level across socio-economic quintiles. New health interventions are often embraced by the richest first and followed gradually by the poorest . So that one would expect equity to get worse as coverage increases before it can get better. Here we found no evidence of increased inequity despite a major increase in coverage and this highlights the need for devising ways to maintain these achievements . We did not find any evidence of gender or district differentials in VAS coverage. Our results on the lack evidence of inequity in gender are similar to what was reported by Bishai and colleagues . A negative association between vitamin A supplementation and maternal knowledge of child health danger signs was rather unexpected and think it is likely to be a chance finding than a real difference.
Vitamin A supplementation campaigns in Tanzania have been concerted efforts by the Government and stakeholders from the health sector. To maximize coverage, activities were strategically integrated into commemoration of the Day of the African Child and World AIDS Day events. During implementation, partnerships were forged between the Presidents Office for Regional Administration and Local Government (PORALG), Ministry of Health and the Tanzania Food and Nutrition Centre (TFNC). Funding support came primarily from UNICEF and vitamin A supplies from Canadian International Development Agency (CIDA). Other agencies and Non Governmental Organizations such as USAID and Plan International also joined these efforts, albeit in later rounds.
This study had two potential limitations. First, we depended largely on the mothers' or caretakers' accounts of the vitamin A supplementation status of their children. We also checked the child's MCH card or notebook for information about vitamin A supplementation but this had limited use as the campaign staff were trained not to record the vitamin A dose on the children's health cards. Secondly, neither the 1999 nor the 2002 household surveys were designed solely to evaluate the change in delivery strategy or policy, but were rather designed to measure coverage of various interventions for children under five, vitamin A supplementation being one, as recommended by UNICEF and WHO . Thirdly, the increase in vitamin A supplementation coverage may have been due in part to other factors and not to the change in the delivery strategy above. However, we are not aware of any factors that could be responsible for the dramatic increase in coverage despite a comprehensive investigation of contextual factors [22, 23]
Although the bi-annual campaign-based strategy for vitamin A delivery resulted in a major increase in coverage, it remains important that this approach sustains universal coverage while not undermining routine health programs .
One among the many obstacles for health systems in low-income countries is fragmented health information systems. Reliable and timely health information is an essential foundation for public health action. Such systems are vital for informing decision-makers to enable them to identify problems and needs, track progress, evaluate the impact of interventions, and make evidence-based decisions on health policy and effect policy change. Routine health programs may be weak due to poorly motivated and often overburdened health staffs, whose job includes reporting from multiple data collection systems and who suffer and who suffer from a lack of supervision. The VAS coverage rates that we have observed so far in Tanzania need to be sustained and monitored regularly if we are to reach the MDGs in 2015. The MCH "Road to Health" cards offer a practical opportunity to capture information on interventions that children actually receive through routine systems or campaigns.
During the mass distribution campaigns, vitamin A supplementation was not recorded on the health card. If campaigns are going to continue, monitoring of progress would be greatly simplified if planners ensure that information for each individual child is recorded. To our knowledge, vaccines or vitamin A supplementation campaigns that are outside routine EPI are not generally registered on children's health cards.
Coverage figures of the campaigns since 2001 have been recorded on tally sheets and subsequently summary sheets that were later collated by Tanzania Food and Nutrition Centre. The problem with this approach is the potential to over-estimate coverage because of lack of proper denominators, or inflated numerators due to multiple doses to the same children. Documenting any extra contacts such as National Immunization Days (NID) or other related health events on health cards not only offers a useful opportunity to reliably estimate coverage figures, but also creates a less demanding and more integrated health information system. In our study, we used information provided by mothers as well as that written on health cards to estimate coverage figures. This source of information is also open to error but is likely to be a conservative estimate, and has the advantage of being household-based .
Social mobilization campaigns are good for catch-up coverage over a short period of time. This coverage has been sustained for at least three years. In spite of the campaigns, national VAS coverage may have dropped as low as 46% in 2004 . Given the evidence on the cost per death averted in Tanzania  and the role vitamin A can play in reducing child mortality, allocation of resources required to implement such child health interventions needs to be maintained.