Water System Publications
distribution system and diarrheal disease transmission: a case study in Uzbekistan
Semenza J, Roberts L, Henderson A, Bogan J, Rubin C
Deteriorating water treatment facilities and distribution systems pose a significant
public health threat. particularly in republics of the former Soviet Union. Interventions
to decrease the disease burden associated with these water systems range from
upgrading distribution networks to installing reverse osmosis technology. To provide
insight into this decision process, we conducted a randomized intervention study
to provide epidemiologic data for water policy decisions in Nukus, Uzbekistan,
where drinking water quality is suboptimal. We interviewed residents of 240 households,
120 with and 120 without access to municipal piped water. Residents of 62 households
without piped water were trained to chlorinate their drinking water at home in
a narrow-necked water container with a spout. All study subjects (1583 individuals)
were monitored biweekly for self-reported diarrheal illness over a period of 9.5
weeks. The home chlorination intervention group had the lowest diarrheal rate
(28.8/1,000 subjects/month) despite lack of access to piped water in their homes.
Compared with the two groups that did not receive the intervention this rate was
one-sixth that of the group with no piped water (179.2/1.000 subjects/month) and
one-third that of the households with piped water (75.5/1.000 subjects/month).
More than 30% of the households with piped water lacked detectable levels of chlorine
residues in their drinking water, despite two-stage chlorination of the source
water, and were at increased risk of diarrhea. Forty-two percent of these municipal
users reported that water pressure had been intermittent within the previous two
days. The dramatic reduction in diarrheal rates in the home-chlorination intervention
group indicates that a large proportion of diarrheal diseases in Nukus are water-home.
The home-chlorination group had less diarrhea than the group with piped water,
implicating the distribution system as a source of disease transmission. Taken
together, these epidemiologic data would support the hypothesis that diarrhea
in the piped water group could be attributed to cross-contamination between the
municipal water supply and sewer, due to leaky pipes and lack of water pressure.
Relatively inexpensive steps, including chlorination, maintaining water pressure,
and properly maintaining the distribution system, rather than reverse osmosis
technology, should reduce diarrheal rates.
quality of drinking water is closely associated with human health, and providing
safe drinking water is a major public health priority. However, there are various
methods of improving the quality of drinking water, and public health officials
often must decide which methods are most appropriate for a given situation. We
conducted a randomized intervention trial in Nukus, Uzbekistan to provide epidemiologic
data to assist such a policy decision.
is located 250 km south of the Aral Sea, in one of the newly independent states
of the Former Soviet Union. Poor water quality and water shortages pose a considerable
public health threat to the 200,000 residents of Nukus, 20% of whom lack access
to piped water. As a result of decades of extensive irrigation of cotton fields,
water and land in Uzbekistan has become salinized and the availability of potable
water has decreased. At the time of our study, the U.S. Agency for International
Development had upgraded a two-stage chlorination system for the municipal water
supply, but no disease monitoring had been conducted in Nukus to verify the effectiveness
of this program.
goal of our randomized intervention study was to determine water-home disease
incidence through active surveillance. Intervention households lacking piped water
were supplied with equipment for home-chlorination of drinking water. Self-reported
incidence of diairhea was compared in both a group with and a group without access
to piped water.
approach allowed the identification of possible routes of disease transmission
and enabled different options for water treatment to be prioritized, such as upgrading
the existing water plant, improving management of the existing distribution
system, revising the distribution system technology, or investing in reverse osmosis
In June 1996,
scientists from the U.S. Centers for Disease Control and Prevention (CDC), in
collaboration with local health officials, identified neighborhoods in Nukus that
did not receive piped water from the municipal water supply. We constructed a
detailed map of the six largest neighborhoods without piped water and estimated
the number of houses. From these areas, we randomly selected 12 index homes, each
of which would be part of a cluster of 10 households. Selection criteria required
that a child less than five years of age lived in the house. The number of clusters
in each neighborhood was assigned systematically proportional to the population.
Because this project was a program evaluation and not human subjects research,
IRB review was not required because the study did not fall under the human subjects
regulations (45 CFR 46). Nevertheless, informed consent was obtained in the study.
Trained interviewers began by interviewing an adult member at the index home;
upon completion they turned left when leaving the house, skipped the next house,
and proceeded until one adult member of each of 10 households had been successfully
interviewed. The same sampling method was used to select another 12 clusters (of
10 households, each with a child) with piped water on the premises (Table 1).
The index houses in the city were selected systematically, proportional to population
from medical charts at the eight polyclinics serving Nukus, where all members
of the population are registered.
1: Selected demographic characteristics of participants by study group, Nukus,
HHs with piped water
130 HHs with no intervention
HHs without piped water
HHs with intervention
HHs with no miervennon
<nve years old
monthly income in US
$ (range) ||76|
of meal or fish consumed by HH
in last week (range) |
of taxi by interviewee last month (range)
garden size in m2 (range)
years lived in current home (range)
HH = household
assess water consumption, sanitary conditions, socio-economic status, and health
status, we developed a questionnaire that was translated into Russian and back-translated
into English for content verification. The questionnaire was field-tested and
revised by six interviewers who were native to Nukus and fluent in Russian as
well as the regional language of Karakalpak.
computer-generated random integers, we divided the 120 households without piped
water into two groups: an intervention and a nonimervention group. A 1.5% chlorine
stock solution and a narrow-necked water container with a spout were provided
to the intervention group. Members of the intervention households were taught
to add chlorine solution to each newly collected container of water. Because water
was dispensed through a tap on the container, neither hands nor utensils could
be immersed in the chlorinated water. thereby preventing recontamination. Members
of households in the intervention group also received hygiene education and were
asked to obtain all of their drinking water only from the provided container and
to wash their fruits and vegetables only with this chlorinated water.
Water samples collected
from selected water sources and households underwent microbiologic analysis for
fecal coliform. A sterile 250-ml polypropylene container was used for water collection.
The water was stewed at 4°C and analyzed within 24 hr. Duplicate samples of
100 ml were filtered through a membrane and incubated at 45°C on liquid m-TEC
medium (Difco Laboratories, Detroit, MI) for 18 hr1. The average of
the duplicate colony counts was taken if the counts ranged between 0 and 200.
A count exceeding 200 colonies per 100 ml was recorded as 200, and less than one
visited 240 households in the study (120 without piped water and 120 with piped
water) twice a week for 9.5 weeks. Study participants did not know the specific
day or time at which the interviewers would return. In a brief
interview using standardized questions, interviewers recorded occurrences of diarrhea
and dysentery reported by an adult member of the household as well as the age
and sex of the affected individuals. Diarrhea was defined as three or more liquid
stools over a 24-hr period. People with diarrhea had to report having had no diarrhea
at one of the twice-a-week visits before they could be counted as experiencing
another episode. Self-reported dysentery was defined as a diarrheal episode with
visible blood in the stool.
regularly monitored compliance in the intervention group by testing the water
in the containers for both free and combined chlorine levels with the N,N, diethyl-p-phenylene
diamine sulfate (DPD) colorimetric method, and by checking the remaining volume
of the concentrated chlorine solution. Questionnaire data were double-entered
into Epi-lnfo (CDC, Atlanta, GA) for analysis. Diarrheal surveillance was documented
and verified by comparing original questionnaires to an Excel (Microsoft. Redmond.
calculated the diarrhea rate for each household, while controlling for intra-household
correlation. The household rate was defined as the number of episodes of diarrhea
in a household over the study time period divided by the number of people in that
family. The mean household rate was compared with each cohort using a t-test with
each household having equal weight. We used a Taylor series approximation to calculate
confidence intervals (CIs). Relative risks (RRs) wore calculated by dividing the
average diarrheal rates for we cohort by that for the referem cohort.
We used logistic regression
for the univariate analysis of 50 potential risk factors for diarrhea for all
ages per household. Because of possible intra-household correlation, we again
used the household as the unit of analysis. The outcome of each household was
defined as 0 for no diarrhea cases, and as 1 if there were one or more cases.
All models controlled for the number of subjects in the household. Varibles
that were significant in the crude, univariate analysis on the 0.1 level were
entered into a stepwise logistic-regression model.2 The final multivariate
model contained the variables that were meaningful predictors (P = 0.1) of diarrhea.
A total of 818
females and 765 males living in the 240 homes were included in the study. Of these
participants, 344 were children less than five years of age (Table 1) and of these
344 children, 115 (71 boys and 44 girls) were breast-fed during the study period.
Boiling and settling the drinking wafer prior to consumption were the most common
water treatments used in all three groups. Prior to the intervention, none of
those interviewed reported filtering or chlorinating their water. The origin of
source water for the different groups is listed in Table 1. Of those households
with piped water on the premises, three also had a well and one also obtained
water from vendors. All wells tested (n =7) were found to be contain inated with
coliforms (mean as 54 colonies/100 mi). Comparison of coliform density of water
samples taken from household water containers in the intervention group (n = 15)
and the nonintervention (n = 11) group with no piped water showed that they were
not statistically different (comparison of no intervention versus intervention
by the Wilcoxon rank sum test: P = 0.97; mean = 52 versus 47 colonies/100 ml.
median = 40 versus 47 colonies/100 ml. range = 0 to > 200 versus 2 to >
200 colonies/100 ml).
investigation of chlorine levels in all of the 120 study households with piped
water revealed that at the beginning of July 1996, water in 45 (38%) of the houses
lacked any detectable levels of free or bound chlorine as measured by the DPD
colorimetric method. The water in 32 (27%) of the houses lacked detectable levels
when retested at the end of July 1996.
and socioeconomic status. Interviewers reported visible feces inside the latrine
or in the garbage can in 39 (17%) of the homes, and no toilet paper of other anal
cleaning material in the latrine or toilets in 101 homes (43%). In subjective
evaluations of the cleanliness of children, the household, and water containers,
interviewers assessed that households in the areas with no access to piped water
were in generally worse sanitary condition than those in the central city with
piped water. This difference is also reflected in the contrast between socioeconomic
status as classified by total family income or meat consumption (Table 1). However,
interviewers found no dramatic difference in the number of years spent in the
current home between residents with piped water and those without piped water:
interviewees with access to piped water had resided in their current home for
11.2 years on average compared with 8.2 years among those without piped water
surveillance. Active diarrheal surveillance over 9.5 weeks revealed a mean
monthly diarrheal rate of 75.5/1.000 among individuals with piped water on their
premises and 179.2/1.000 among those without piped water on the premises (Figure
1). A similar comparison between the intervention group with home chlorination
(rate = 28.9/ 1,000 individuals/month) and the group without home chlorination
showed that home chlonnation reduced diarrhea by 85%
(RR = 0.15, 95% CI = 0.07-0.31) (Table 2). Individuals living in households with
access to piped water but without a detectable chlorine residual were at increased
risk compared to households with detectable chlorine residuals (RR = 1.6. 95%
CI = 0.7-3.7).
rates and relative risks for children less than five years of age are shown in
Tables 3 and 4. Over the course of the study, three infants less than one year
of age were reported by a family member to have died of diarrhea. One of the deceased
was from a home with piped water, and two lived in houses in the nonintervention
group in the area with no piped water. In each of the cases a household member
reported that the deceased infant had bloody stools: however, independent confirmation
from a health care provider could not be obtained.
the obvious advantages of clean drinking water, residents expressed little willingness
to pay for 20 liters of clean drinking water (mean value per 20 liters = 7.2 Sum
or $0.20) or a container for home chlorination (mean value per container = 12.6
Sum or $0.30). However, the home chlorination procedure was readily accepted by
this population based on a compliance rate of 73% for detectable chlorine residuals
in the water at the time of the visit.
multivariate analyses with non-age-specific diarrhea within a household as the
outcome and correcting for family size, water source and intervention status,
we found the following variables (Table 5) to be risk factors: a water source
farther than 200 meters from the residence compared with a source closer than
200 meters (OR = 2.2. 95% CI = 0.9-4.9). 20-40 liters of water stored in the house
compared with less than 20 liters stored (OR = 2.2. 95% CI = 1.1-4.1), and more
than 40 liters of water stored in the house (OR = 2.0, 95% CI. = 0.9-4.4). Use
of toilet paper after defecating was found to be protective (OR = 0.7. 95% CI
Our active diarrheal
surveillance, in combination with the randomized intervention study, revealed
that poor water quality contributes to a substantial diarrheal burden for the
population of Nukus that consumes municipal water. This conclusion is based on
the following three findings. 1) Among people living in homes without piped water,
those who were randomized to treat their water with chlorine had an 85% reduction
in diarrheal rates, compared with those who did not chlorinate their water (Figure
1). This implies that a large fraction of diarrheal pathogens in Nukus are spread
through water and that other routes of transmission play only a minor role. It
is unlikely that chlorination of drinking water would affect disease transmission
due to poor hygiene, lack of sanitation, inter-household or person-to-person transmission,
or contaminated foods to the degree observed in the intervention group. 2) People
in the home-chlorination intervention group had a 62% reduction in diarrheal rates
compared with those living in areas with access to piped water, despite less sanitary
conditions and lower income in the areas without access to piped water (Table
1 and Figure 1). This suggests that some, if not most, of the diarrhea among those
with piped water is associated with the public water supply. 3) In approximately
one-third of the households with piped water, the water did not contain adequate
levels of chlorine, and 42% of the interviewees from households with piped water
reported frequent loss of water pressure. As a result, leaky sewer lines or unsanitary
ground water could cross-contaminate the drinking water pipes, which in turn would
lead to both fecal spread and depletion of chlorine residuals. An indication of
the hazard posed by the Nukus public water system is illustrated by the fact that
people in houses without a chlorine residual in their piped water experienced
60% more cases of diarrhea than did those with a chlorine residual in their piped
water. This finding is not statistically significant and should only be taken
as supporting evidence.
1. Diarrhea rates in Nukus, Uzbekistan by chlorination status. June-August
1996. *No data for area with piped water.
three sets of findings all indicate that contaminated drinking water is a major
cause of diarrheal diseases in Nukus. Because the water supply is chlorinated
twice at the plant, it is likely that the distribution system is a primary source
of diarrheal pathogens. Increasing the amount of water available would help maintain
pressure in the pipes and decrease
cross-contamination. Availability can be increased by minimizing water loss through
proper management of the distribution system. Under current conditions, water
that passes through the two-step clilorination process is believed to become recontaminated
in the distribution system, and the quality of the drinking water is therefore
suboptimal. Likewise, highly purified water from a reverse osmosis plant can only
reach consumers through an intact distribution system with no cross-contamination,
a situation that our study indicates does not currently exist in Nukus. Based
on the results of this study, Nukus health officials improved management of water
distribution by locating and repairing leaky pipes. Furthermore, reverse osmosis
was abandoned as a treatment option and water quantity will be increased in the
2: Comparison of diarrheal burden among three study groups over a 9.5-week
period using the household (HH) as a unit of analysis*
95% CI = apporoximate 95% confidence interval
rates among children less than five years of age by study group
|Total no. of Children||
4: Comparison of diarrheal disease rates among children less than five years
of age in the 3 study groups, from June 26 to August 17, 1990*
chlorination vs. no home chlorination (intervention vs. control)|
chlorination vs. piped water|
home chlorinaiion vs. piped water|
of risk factors with diarrhea for all ages in household in multivariate analysis*
to water source =<200 meters (ref.) |
to water source >200 meters |
after defecating: other (ref.) |
after defecating: paper|
of water stored in home < 20 (ref.)|
of waier stored in home >20<=40 |
of water stored in home >40 |
CI =confidence interval; ref. = reference.
of diarrheal diseases is an important public health strategy for reducing morbidity
and mortality, paricularly among children less than five years of age. In enteric
disease outbreaks, sources of infection have been traced to a variety of origins
such as surface water, household water containers, foods and drinks, and poor
sanitary conditions. Contaminated municipal water supplies have been implicated
in disease transmission in diarrheal outbreaks around the world.6,8-13
However, few studies to date have evaluated the burden and routes of transmission
of enteric diseases in relatively developed countries with suboptimal sanitary
infrastructure, such as the newly independent Central Asian States. Our analysis
of disease transmission in Nukus reveals that in this case the endemic disease
burden is caused primarily by contaminated public water supply rather than by
poor sanitation, poor hygiene practices, or other causes.
remediating the municipal water supply it is also important to extend piped water
services to houses in the periphery of Nukus, which disproportionately suffer
from diarrheal diseases. Implementation of water treatment with delivery systems
as well as sewage treatment plants are obviously desirable measures, but are often
hampered by the fact that they are expensive, time-consuming, and laborious to
construct. In fact, the lack of piped water in these outlying neighborhoods of
Nukus is not a recent occurrence, as reflected by the number of years occupants
lived in their homes, and indicates an ongoing developmental problem. A short-term
alternative involves the use of a narrow-necked water container with spout and
chemical disinfectant such as that used in this and other studies.14-16
Our data suggest that this simple combination of water disinfection and a physical
barrier to recontamination, if used appropriately and supplemented with hygiene
education, can be an affordable, effective short-term means of reducing diarrheal
rates in those areas where potable water is currently not available. Our study
not only documents the feasibility and effectiveness of this method but also establishes
the behavioral and cultural willingness of our study population to adopt such
an intervention, provided the containers are made available. As part of a training
course in environmental epidemiology, local health officials participated in this
study; the goal was to illustrate a study design appropriate for local circumstances,
relying on limited infrastructure and resources but nevertheless powerful in its
the study, we monitored the durability of the narrow-necked, flexible, plastic
water containers used by the intervetion group. Because of leaks and punctures,
19 of the 62 containers had to be replaced, indicating a half-life of 4.9 months
for this type of container. This suggests the need for more durable water containers,
and the effectiveness of locally produced aluminum containers will be assessed
in nonintervention households in the coming months.
intervention group, in addition to having access to disinfected water, may have
changed their sanitary habits during the course of the study, and such behavior
modification may have contributed to their reduced diarrheal rates. Besides being
instructed to use chlorinated water for drinking purposes, intervention households
were also asked to wash their fruits and vegetables with the treated water. This
guidance, together with the fact that the intervention group received a valuable
container, may have led to improved hygienic behavior among members of this group.
However, we do not believe that this is the case for the following reasons: hygiene
practices were observed to be at a consistently high level in all three study
groups, especially with regard to hand washing and food washing; our hygienic
message was brief and not reinforced: there is no reason to believe that prior
differences in hygiene practices existed in the different groups; and during the
intervention all three groups received equal numbers of visits and similar amounts
of attention. The purpose of our message was to emphasize that this water was
not only for drinking but also for food preparation. It is possible that members
of me intervention group were more likely to wash their food because they knew
their water was treated. In this case, an increased tendency to wash food should
be viewed as an indirect benefit of chlorination that would likely occur elsewhere
in this population when safe domestic water becomes available.
inhabitants of Nukus have a considerable burden of diarrheal diseases regardless
of their water source, compared with locations where drinking water meets current
microbiologic standards.12 Individuals with access to piped water at
their residence experience fewer episodes of diarrheal disease than those without
access to piped water. Nonetheless, drinking home-chlorinated water results in
lower diarrheal incidence than drinking the piped water provided in Nukus. This
is probably due to cross-contamination between water lines and unsanitary ground
water which results in a depletion of chlorine levels in the piped water. Thus,
providing water with adequate chlorination levels to all residents should be the
top priority of Nukus health officials. In fact,
this study resulted in better system management with a focus on maintaining pressure
throughout the distribution system, accompanied by monitoring to ensure that pressure
remains stable. Until the distribution system can deliver uncontaminated water
to households in Nukus, reverse osmosis treatment has no technical, financial
or public health basis and has indeed been abandoned.
We are particularly grateful to Drs. G. Heitbaev and A. Mustanna of the Sanitary-Epidemiologic
Services (SES), Nukus for support of the field investigation. We also thank Paul
Dryer (EPT), Bruce Ross and Dr. Terence Chorba (CDC, Almaty, Kazakhkstan), and
Maureen Blyler (CDC, Atlanta). Finally, many thanks are given to the people of
Nukus, who were warm and open, and endured both the hardship of life and the taste
of chlorine with a smile.
support: This study was funded by the United States Agency for International
Development (USAID). the Environmental Policy and Technology Project (EPT), and
CDC. The work was facilitated by U'SAID/Central Asian Republics and undertaken
jointly by the SES of Uzbekistan, SES Karakalpakistian, EPT, and CDC.
1. Mates A. Shatter
M. 1989. Membrane nitration differemation of
E. coli from coliforms in the examination of water. J Appi Baciennl
SAS Institute. 1992. SAS/STAT Software: Changes and Enhancements.
Release 6.07. Technical Report P-229. Cary, NC:
Mujica OJ, Quick RE, Palacios AM, Beingolea LB, Vargas R, Moreno
D, Barret TJ, Bean NH, Seminario L, Tauxe RV, 1994.
Epidemic cholera in the Amazon: the role of produce in
disease risk and prevention. J Infect Dis 169: 1381-1384.
Feachem RG, 1984. Interventions for the control of diarrhoeal diseases
among young children: promotion of personal and domestic
hygiene. Bull World Health Organ 62: 467-476.
RE, Thompson BL, Zuniga A, Dominguez G, De Brizuelu
EL, De Palma O, Almeida S, Valencia A, Ries AA, Bean
NH, Blake PA, 1995. Epidemic cholera in rural El Salvador: risk factors in a region
covered by a cholera prevention campaign. Epidemiol Infect 114: 249-255.
6. Black RE, Lopez
de Romana G, Brown KH, Bravo N, Bazalar OG, Kanashiro HC, 1989. Incidence and
etiology of infantile diarrhea and major routes of transmission in Huascar Peru.
Am J Epidemiol 129: 785-799.
Black RE, Brown KH, Becker S, Alim ARMA, Merson MH, 1982. Contamination of weaning
foods and transmission of enterotoxigenic escherichia coli diarrhoea in children
in rural Bangladesh. Trans R Soc Trop Med Hyg 76: 259-264
Swerdlow DL, Mintz ED, Rodriguez M, Tejada E, Ogampo C, Espejo L, Greene KD, Saldana
W, Seminano L, Tauxe RV, Wells JG, Bean NH, Ries AA, Pollack M, Vertiz B, Blake
PA, 1992. Waterborne transmission of epidemic cholera in Trujillo, Peru: lessons
for a continent at risk. Lancet 340: 28-32.
Ries AA, Vugia DJ, Beingolea L, Palacios AM, Vesquez E, Wells JG, Baca NG, Swerdlow
DL, Pollack M, Bean NH, Seminario L, Tauxe RV, 1992. Cholera in Piura, Peru: a
modern urban epidemic. J Infea Dis 166: 1429-1433.
Sacks JJ, Lieb S, Baldy LM, Berta S, Patton CM, White MC, Bigler WJ, Witte JJ,
1986. Epidemic campylobacterlosis associated with a community water supply. Am
J Public Health 76: 424-428.
Esrey SA, Feachem RG, Hughes JM, 1985. Interventions for the control of diarrhoeal
diseases among young children: improving water supplies and excreta disposal facilities.
Bull World Health Organ 63: 757-772.
Payment P, Richardson L, Siemiatycki J, Dewar R, Edwardes M, Franco E, 1991. A
randomized trial 10 evaluate the risk of gastrointerstinal disease due to consumption
of drinking water meeting current microbiological standards. Am J Public Health
Cardenas V, Saad C, Varona M, Linero M, 1993. Waterborne cholera
in Riohacha, Columbia, 1992. Bull Pan Am Health Organ
Mintz ED, Reiff FM, Tauxe RV, 1995. Safe water treatment and storage
in the home: a practical new strategy to prevent waterborne
disease. JAMA 273: 948-953.
Quick RE, Venczel LV, Gonzalez O, Mintz ED, Highsmith AK, Espada A, Damaini E,
Bean NH, De Hannover EH, Tauxe RV, 1996. Narrow-mouthed water storage vessels
and in situ chlorinauon in a Bolivian community: a simple method to improve drinking
water quality. Am J Trop Med Hyg 54: 511-516.
Deb BC, Sircar BK, Sengupta PG, De SP, Mondal SK, Gupta DN, Sana NC, Ghosh S,
Mitra U, Pal SC, 1986. Studies on interventions 10 prevent eltor cholera transmission
in urban slums. Bull World Health Organ 64: 127-131.
J, Roberts L, Henderson A, Bogan J, Rubin C. Water distribution system and diarrheal
disease transmission: a case study in Uzbekistan. American Journal of Tropical
Medicine and Hygiene 1998; 59: 941-6.