Development and field testing of low-cost, quantal microbial assays with volunteer reporting as scalable means of drinking water safety estimation.

ABSTRACT

AIMS: To evaluate a low-cost water quality test for at-scale drinking water safety estimation in rural India. METHODS AND RESULTS: Within a longitudinal study to characterize variability in household drinking water safety in rural Maharashtra, we piloted a low-cost presence-absence (LCPA) microbial test designed to be used by volunteer residents in rural areas. In comparing the LCPA results with standard laboratory methods for enumeration of Escherichia coli, we found that LCPA tests using modified mTec media were highly sensitive in detecting drinking water of moderate risk (88% of tests were positive at E. coli counts of 11-100 CFU per 100 ml) and high risk (96% of tests were positive at E. coli counts of 101 + CFU per 100 ml). The LCPA tests demonstrated low specificity for E. coli specifically, due to concurrent detection of Klebsiella 38% of LCPA tests were positive even when E. coli was not detected in a 100 ml sample by membrane filtration, suggesting the test would be conservative in risk estimation. We also found that 47% of participants in rural villages in India were willing to conduct tests and return results after a brief training, with 45% of active participants sending their water testing results via short message service. CONCLUSIONS: Given their low cost (~US$0.50 as piloted) and open-source format, such tests may provide a compelling alternative to standard methods for rapid water quality assessments, especially in resource-limited settings. SIGNIFICANCE AND IMPACT OF THE STUDY The lack of availability of water quality data constrains efforts to monitor, evaluate and improve the safety of water and sanitation infrastructure in underserved settings. Current water testing methods are not scalable because of laboratory and cost constraints. Our findings indicate the LCPA or similar low-cost microbial tests could be useful in rapid water safety estimation, including via crowdsourcing.