Thirty low-income Kenyan households using turbid river and relatively cleaner rain water participated in a 6 month in-home Biosand filter (BSF) performance study comprised of surveys and monthly monitoring of BSF influent and effluent water for turbidity and fecal coliforms (FC). River–river (influent–effluent) sample pairs (n = 155; 90% of observations) resulted in average BSF instantaneous FC and turbidity removals of 1.41 log10 (96.1%) and 32.5%, respectively. Cumulative distributions of influent and effluent quality demonstrated unambiguous improvement of river water but rain water improvement was limited and less reliable. Filter performance varied considerably within and across units. A hierarchical set of hypothesized factors affecting filter bacterial performance variability was assessed. BSF effluent FCs were positively correlated with influent (flush water) FCs and influent and effluent turbidity, and negatively correlated with turbidity applied to-date and days since maintenance. Interrupted use and moving the BSF negatively impacted effluent quality. Households with children age 6–10 collecting BSF filtered drinking water, or with more members, had higher effluent FCs. BSFs fed only river water performed better, on average, than mixed-source filters. Implications for BSF implementation in developing countries are discussed, including aqueous chemistry aspects of performance.

Supplementary data