Last fall, I blogged about a seemingly alarming new research findings published in the Lancet by Dhingra and co-authors that claimed that the estimates of mortality from malaria in India were being underreported by a factor of 10. I like many others, asked how could the WHO’s estimates be so wrong?
Five months later, it seems that this controversy is still buzzing away (mosquito pun intended). The Lancet recently published a series of correspondences have seriously questioned the validity of these newly published estimates and have left me scratching my head once again.
I was particularly troubled by some of the arguments made by Naman Shah and co-authors (you may know Naman from his excellent malaria blog topnamen). Apparently a small validation exercise had been carried out in India to validate the use of the verbal autopsy methods used in this study but the validation exercise found large discrepancies between what was being counted as a malaria death using verbal autopsy methods and what was being clinically validated in hospitals. I saw the same data presented last week at the GHME conference in Seattle which makes me believe that this is a really big faux pas in the health metrics world.
So while I am no expert on these things, I am surprised that if one is using a new method and that there is data available from a validation exercise that shows large discrepancies that these limitations would not become apparent during the peer review process – in particular given how much controversy there is about the use of verbal autopsy methods in general and in particular regarding the use of them in other geographic contexts for malaria.
It is great to have new methods to track and monitor things that have been challenging to measure in the past, but as we all know, newer is not always better, and I believe that new methods need to demonstrate their superiority before we take them at face value. This might be one case where we might need a bit more convincing.Share on Facebook