Sir

Your Editorial “Millennium development holes” (Nature 446, 347; doi:10.1038/446347a 2007) is timely in highlighting the complexities of monitoring the ambitious development goals on which the world is focused. As you note, the uneven quality of data means that our confidence in reported or predicted achievements varies by country and by indicator. Your call for more investment in evidence-based approaches is welcome, but the solutions are more complicated than this.

A significant handicap in a country's efforts to evaluate interventions is the requirement, by multiple agencies, to monitor multiple indicators for multiple internationally led projects. Its scarce resources can be undermined by the creation of parallel reporting structures, by demands for overlapping surveys for different purposes, and by financial support that is skewed to meet the donor's needs to report internationally. This results in an ever-widening gap between national capacity and international expectations, and the influx of more international experts to fill the 'holes'.

Agencies and donors want national estimates of Millennium Development Goal indicators in order to make international comparisons and to monitor their investments. Because the underlying country data are often weak, agencies develop predictions or estimates to fill data gaps. But countries need more than national averages; they require intimate knowledge of changing disparities in indicators of access and outcomes, by administrative areas and between socio-economic groups. This knowledge can be gleaned only from empirical data collected through information systems that are reliable at sub-national level.

Additional investment is required, but it should be directed to supporting countries' efforts to strengthen their own statistical systems in order to produce the evidence they need nationally and sub-nationally. Until international agencies agree to provide coordinated support, the disparities will remain unnoticed and the 'holes' unfilled.