A statistical analysis of the pros and cons of field kits to guide
well-switching in arsenic prone regions
Abstract
Field kits for testing the level of a toxicant in the environment are
inherently less accurate than a laboratory instrument. Using a specific
example, we argue here that kit measurements still have a key role to
play when the spatial distribution of a toxicant is very heterogeneous.
The context is provided by the groundwater arsenic problem in
Bangladesh. We combine here two data sets, a blanket survey of 6595
wells over a 25 km2 based on laboratory measurements
and 900 paired kit and laboratory measurements from the same area. We
explore different hypothetical mitigation scenarios based on actual data
that rely on households with a high-arsenic well switching to a nearby
low-arsenic well. We show that the decline in average exposure to
arsenic from relying on kit rather than laboratory data is modest in
relation to the logistical and financial challenge of delivering
exclusively laboratory data. Our analysis indicates that the 50 ug/L
threshold used in Bangladesh to distinguish safe and unsafe wells,
rather than the WHO guideline of 10 ug/L, is close to optimal in terms
of average exposure reduction. We also show, however, that providing kit
data at the maximum possible resolution rather than merely classifying
wells as unsafe or safe would be even better. These findings are
relevant as the government of Bangladesh is about to launch a new
blanket testing campaign of millions of wells using field kits.