Via The Lancet, a brief and disturbing report: New global estimates of malaria deaths. I've put in a link to the first cited study; for the rest, click through and see the References.
Christopher Murray and colleagues' systematic analysis of 1980—2010 global malaria mortality figures reveals alarming deficiencies in current data collection, reporting systems, and analysis and interpretation of surveillance data.
The huge differences between Murray and colleagues' mortality figures for malaria and those of WHO now call into question surveillance data for tuberculosis, HIV/AIDS, and non-communicable diseases (NCDs).
Global cause-specific morbidity and mortality statistics are presently based on a compilation of data from clinical records, death certificates, annual governmental returns, and verbal autopsy studies, all of which are known to be grossly inaccurate. Record keeping and reporting systems from most developing countries remain poor. Verbal autopsy studies are few, subjective, inaccurate, often misleading, and yield limited information.
The most reliable method for obtaining cause-specific mortality rates is through autopsy examination of the cadaver. Autopsy rates have declined substantially and actual causes of death in most sub-Saharan African and Asian countries are unknown. Autopsies for research purposes have been few, but have yielded invaluable information about tuberculosis, HIV/AIDS, cancers, and other missed preventable disorders.
An enormous amount of funding and resources is invested by donors and governments in interventions to reduce mortality from malaria, tuberculosis, HIV/AIDS, and NCDs, and yet there are no accurate means of assessing the effectiveness of these interventions. It was only through acquisition of an accurate evidence base from autopsy studies of African children that WHO, government, and donor attention was focused on childhood tuberculosis. Funding agencies and governments need to consider making major investments into pathology services to establish routine autopsies in developing countries.