Joe Brown and Mark Sobsey
University of North Carolina School of Public Health
Department of Environmental Sciences and Engineering
Submitted to UNICEF – Cambodia, 5 May 2006
This study is an independent follow-up assessment of two large-scale implementations
of the household-scale ceramic water filteration after 2 and 4 years in use.
Approximately 1000 household filters were introduced by Resources Development
International (RDI) in Kandal Province from December 2003 and 1000+ filters by
International Development Enterprises (IDE) in Kampong Chhnang and Pursat provinces
from July 2002. The American Red Cross, CIDA, AusAID, UNICEF, and the World Bank
Development Marketplace Programme have supplied support to these two NGOs for
various parts of the production and distribution cycle of the filters.
In October 2003, IDE completed a field study of the ceramic water filtration devices after one year in use,
yielding promising results. The study used bacterial analyses of water samples and user
surveys to measure the performance, acceptance and use of ceramic water filtration devices in 12 rural villages.
The field study also assessed health improvements, time savings, and expense savings.
In August 2005, RDI completed a similar internal study for the filter distribution in Kandal
province, although findings from this assessment have not yet been released. The
present study follows up on these previous assessments and represents an independent
appraisal of the performance of the ceramic water filtration projects undertaken by IDE and RDI. It is
hoped that the findings produced will aid in assessing the water quality and health
impacts of the ceramic water filtration interventions to date and yield useful information on the
sustainability of the filters as implemented.
The study was carried out in two parts:
(1), a cross-sectional study of households
that originally received filters to determine uptake and use rates and associated factors;
(2), a nested longitudinal prospective cohort study of 80 households using filters and
80 control households to determine the microbiological effectiveness and health impacts
of the filters in household use. We measured (i) the continued use of the filters over
time as the proportion of filters still in use since introduction, and identified factors
potentially associated with filter uptake and long term use; (ii), the microbiological
effectiveness in situ of the filters still being used, as determined by the log10 reduction
values of the indicator bacterium E. coli; and (iii), the health impacts of the filters as
determined by a prospective cohort study using data on diarrheal disease prevalence
proportions among filter users versus non-users. We also collected a variety of other
survey data intended to elucidate successes and challenges facing the long-term
sustainability of this intervention in Cambodia. Stratified analyses, logistic regression,
and log-risk regression with Poisson extension of generalized estimating equations
(GEE) were employed in analysis of cross-sectional and longitudinal data to determine
factors associated with long term filter use and effectiveness of filters currently in use.
Major findings are that (i), the rate of filter disuse was approximately 2% per
month after implementation, due largely to breakages; (ii), controlling for time since
implementation, continued filter use over time was most closely positively associated
with related water, sanitation, and hygiene practices in the home, cash investment in the
technology by the household, and use of surface water as a primary drinking water
source; (iii), the filters reduced E. coli/100ml counts by a mean 95.1% in treated versus
untreated household water, although demonstrated filter field performance in some
cases exceeded 99.99%; (iv), microbiological effectiveness of the filters was not
observed to be closely related to time in use; (v), the filters can be highly effective
against microbial indicator organisms but may be subject to recontamination, probably
during regular cleaning; and (vi), the filters were associated with an estimated 46%
reduction in diarrhea in filter users versus non users (RR: 0.54, 95% CI 0.41-0.71).