Methods We undertook a fi eld study of 30 households in a peri-urban neighbourhood adjacent to the city of Mzuzu,
Malawi. We used a random number generator to assign households to one of three water purifi cation interventions:
boiling, Waterguard chlorine solution (PSI/Malawi, Blantyre, Malawi) , or Tulip table-top ceramic fi lter. We analysed
samples taken from a drinking-water storage container in each household at baseline and 2 weeks after baseline, and
recorded total coliform and Escherichia coli count s. Participants completed an initial structured questionnaire about
water sources used, sanitation, health, water consumption patterns, and socioeconomic variables in their household.
Follow-up questionnaires at 1 week and 2 weeks after the intervention, focused on use and acceptance of the assigned
treatment modality. We used WHO health risk standards, and correlated drinking water quality with the questionnaire
responses. We used Fisher’s exact test for data analysis.