Download:
pdf |
pdfAttachment D
CDC Water and Health Study: Pilot for an Epidemiologic Study of
Low Pressure Events in Drinking Water Distribution Systems (WRF #4390)
Pilot Evaluation Report
August 2014
Julia Gargano, PhD ‐ Principal Investigator; Epidemiology lead
Vincent Hill, PhD, PE ‐ Co‐investigator; Laboratory lead
Elizabeth Adam, MPH ‐ Study Coordinator
Chandra Schneeberger ‐ Environmental Microbiologist
Table of Contents
EXECUTIVE SUMMARY ................................................................................................................................................................. 1
INTRODUCTION ............................................................................................................................................................................ 6
Rationale for National Study of Health Effects Associated with Distribution System Low Pressure Events (LPEs) ................ 6
Goal and Aims for the National Study ..................................................................................................................................... 7
Epidemiological Study Design .................................................................................................................................................. 7
PLANNING AND PREPARATION FOR PILOT STUDY ...................................................................................................................... 9
Pilot Study Goals and Objectives ............................................................................................................................................. 9
Preparations for Pilot Launch .................................................................................................................................................. 9
Establishment of Pilot Study Procedures and Roles .............................................................................................................. 10
Selection of LPE and Non‐LPE Areas ................................................................................................................................. 11
Water Sample Collection and Testing Methods .................................................................................................................... 11
Household Survey .................................................................................................................................................................. 12
Data Analysis ......................................................................................................................................................................... 12
Scope of Data Collected During Pilot Study ........................................................................................................................... 12
PILOT STUDY RESULTS ............................................................................................................................................................... 14
Data Collected Using the LPE Form ....................................................................................................................................... 14
Water Samples ...................................................................................................................................................................... 19
Household Survey .................................................................................................................................................................. 23
DISCUSSION ............................................................................................................................................................................... 27
Determining LPE Exposure Areas .......................................................................................................................................... 27
Matching LPE and Non‐LPE Areas .......................................................................................................................................... 27
Describing and Justifying the Choice of Study Areas ............................................................................................................. 27
Water Samples ...................................................................................................................................................................... 27
Household Survey .................................................................................................................................................................. 28
Lessons Learned .................................................................................................................................................................... 29
EVALUATION OF PILOT STUDY AND PLANNING FOR NATIONAL STUDY .................................................................................... 31
Evaluation of Pilot Study ....................................................................................................................................................... 31
Conclusion ............................................................................................................................................................................. 31
Next Steps and Information for Prospective Utilities ............................................................................................................ 32
APPENDIX: PILOT STUDY LOW PRESSURE EVENTS AND DECISION‐MAKING ............................................................................. 33
LPE 1 ...................................................................................................................................................................................... 33
LPE 2 ...................................................................................................................................................................................... 34
LPE 3 ...................................................................................................................................................................................... 35
LPE 4 ...................................................................................................................................................................................... 37
LPE 5 ...................................................................................................................................................................................... 39
LPE 6 ...................................................................................................................................................................................... 40
REFERENCES ............................................................................................................................................................................... 42
List of Figures
Figure 1. Pilot study procedures and utility and study team roles ............................................................................................ 10
Figure 2. The distribution of turbidity, HPC, TOC, aerobic endospores, and total chlorine and the correlations between the
parameters ........................................................................................................................................................................ 21
Figure 3. The distribution of total hardness, alkalinity, ph, temperature, and total chlorine and the correlations between the
parameters ........................................................................................................................................................................ 22
Figure 4. Number and day of survey response, by mode of response ...................................................................................... 24
List of Tables
Table 1. Attribute classification for selection of LPE exposed and unexposed areas ................................................................ 11
Table 2. Summary of pilot data collection and how data were used in the pilot evaluation .................................................... 13
Table 3. Event, response, and repair type for six pilot LPEs ...................................................................................................... 16
Table 4. Event and repair description for six pilot LPEs ............................................................................................................. 16
Table 5. Characteristics of infrastructure, water source, and soil type for six pilot LPEs .......................................................... 17
Table 6. Field description of how households were affected by the six pilot LPEs ................................................................... 17
Table 7. Study LPE‐area pressure readings for six pilot LPEs ..................................................................................................... 18
Table 8. Distribution of field and grab sample water quality parameters following six LPES.................................................... 20
Table 9. Distribution of water quality parameters by exposure status from six LPEs ............................................................... 20
Table 10. Detection rates using ultrafiltration ........................................................................................................................... 22
Table 11. Summary of pilot study data collection for six LPEs................................................................................................... 24
Table 12. Reported household water use .................................................................................................................................. 25
Table 13. Reported household characteristics .......................................................................................................................... 25
Table 14. Reported water change during 3‐week period following the LPE.............................................................................. 26
Table A 1. Exposed area selection for LPE 1 .............................................................................................................................. 33
Table A 2. Unexposed area selection for LPE 1 .......................................................................................................................... 34
Table A 3. Exposed area selection for LPE 2 .............................................................................................................................. 35
Table A 4. Unexposed area selection for LPE 2 .......................................................................................................................... 35
Table A 5. Exposed area selection for LPE 3 .............................................................................................................................. 36
Table A 6. Unexposed area selection for LPE 3 .......................................................................................................................... 36
Table A 7. Exposed area selection for LPE 4 .............................................................................................................................. 38
Table A 8. Unexposed area selection for LPE 4 .......................................................................................................................... 38
Table A 9. Exposed area selection for LPE 5 .............................................................................................................................. 39
Table A 10. Unexposed area selection for LPE 5 ........................................................................................................................ 40
Table A 11. Exposed area selection for LPE 6 ............................................................................................................................ 41
Table A 12. Unexposed area selection for LPE 6 ........................................................................................................................ 41
EXECUTIVE SUMMARY
A team from the Centers for Disease Control and Prevention (CDC) Waterborne Disease Prevention Branch
completed a successful pilot study for an epidemiologic study to assess whether individuals exposed to low
pressure events (LPE) in a water distribution system are at an increased risk for acute gastrointestinal illness
(AGI) or acute respiratory illnesses. The study design and protocols were piloted at one utility site and evaluated
by CDC. Information from the pilot evaluation is being used to streamline and improve the study processes,
protocol, and methods prior to the full study launch, expected in 2014. Data will be collected from 65 LPEs
across 4‐5 utility sites. This report describes the design, conduct, and results of a pilot study for the nationwide
study.
Introduction
Each year, approximately 240,000 water main breaks occur in the U.S., presenting opportunities for pathogen
intrusion into water distribution systems (Kirmeyer et al., 1994). Similar estimates are not available regarding
planned maintenance events and repairs. An epidemiologic study in Norway systematically examined LPEs,
including main breaks and planned maintenance events, and gastrointestinal illness using a prospective cohort
study design, and found that LPEs were associated with a 58% increased risk of AGI in the week following the
LPE (Nygard et al., 2007). A similar study in the U.S. is needed to gather relevant health data to determine if LPEs
are associated with illness. No existing U.S. data sources can be used to answer this research question.
Systematic data collection across many LPEs, with a study design tailored for the purpose of answering the
research question, is needed to identify the health impacts of LPEs.
To address this need, the CDC plans to conduct a multisite epidemiologic study to assess whether individuals
exposed to LPEs in water distribution systems are at an increased risk for AGI or acute respiratory illnesses. The
study goal is to determine whether individuals exposed to LPEs in the water distribution system are at an
increased risk for gastrointestinal or respiratory illness. The study aims are to 1) assess the association between
exposure to the LPE and self‐reported illness during a short observation period following the LPE, 2) compare
microbial indicator test results in water samples from LPE and non‐LPE (control) areas, and 3) describe the LPE
characteristics.
Study Design and Methods
A prospective cohort study design was implemented. Investigators identified LPE exposure areas and unaffected
control areas (non‐LPE areas). The control areas were matched to LPE areas on pipe material and size, drinking
water source, predominant housing type, and census resident age groups. Households from the LPE and non‐LPE
areas were randomly selected and mailed a survey to elicit information on water use, recent water service,
other activities and exposures, and illness symptoms during a short observation period following the LPE. To
avoid bias, the survey participants were blinded to the study goal and research question. In the multi‐site study
that will follow the pilot study reported here, survey responses from LPE and non‐LPE areas will be compared to
determine if individuals in the LPE‐areas are more likely to report illness symptoms. The analysis will control for
other factors, such as respondent age or recent travel, which could potentially influence the risk of illness.
1
An LPE was defined as a water service disruption event causing a presumed loss of water pressure in the
distribution system, including unplanned outages or planned maintenance events. For each LPE, water utility
personnel completed a standard LPE form, which collected information on the event and repair process, the
infrastructure, water source, and soil characteristics, and field descriptions of how the households in the area
were affected by the LPE, including field pressure measurements during and after the LPE. The study team
selected areas affected by the LPE and matched them with unaffected non‐LPE control areas. The utility’s
internal infrastructure maps and records were used to determine the pipe sizes, materials, and water sources in
the area and to create a ‘normal flow’ map (i.e., before the LPE). The event description information and the LPE
form were used to create an ‘event hydraulic’ map, outlining the boundaries of the LPE and non‐LPE areas and
any changes in flow direction or magnitude that occurred following the LPE. Utility personnel were also asked to
choose exposed and unexposed area engineering attributes from a list to describe and justify the choice of the
study areas. The utility’s water system maps were used to select areas with matching infrastructure
characteristics. U.S. 2010 Census data were used to select areas within the same census tract/block group, or
census tracts with matching resident age groups when LPE areas did not fit within the census area boundaries.
Utility personnel collected three ultrafiltration (UF) water samples and three grab samples from each area
(within 24 hours of the event for exposed area samples and within 48 hours for control area samples).
Ultrafiltration samples were analyzed at CDC and grab samples were analyzed by the utility laboratory. CDC
tested for the following analytes: total coliforms/E. coli, enterococci, C. perfringens, somatic coliphages, total
aerobic endospores, and human‐specific Bacteroides spp. The utility processed grab samples for total
coliforms/E. coli, enterococci, heterotrophic plate count (HPC), total hardness/alkalinity, and total organic
carbon (TOC). The utility also collected the following field water quality data: temperature, pH, and total
chlorine concentration. The utility business support team securely transferred contact information (i.e., names
and addresses) for residential customers living within the study areas to CDC, and the CDC team administered
the survey. A multiple contact mailing strategy was implemented (Dillman, 2007). Selected households received
up to five survey prompts from CDC by postal mail: an advance letter, initial survey packet, thank you/reminder
note, second survey packet, and final appeal letter. Respondents had the choice to respond by mail using a
provided postage‐paid return envelope or on‐line using a secure survey website.
The CDC team trained utility personnel prior to the pilot launch. The first part of the training involved practicing
the selection of LPE and non‐LPE areas over the phone and by email, using work‐ups of past breaks and repairs.
The CDC team worked with utility business support and IT personnel to develop a utility‐specific method of
creating customer contact lists from study area maps. The next phase of training consisted of the team
responding to mock events in the field. This gave the study team an opportunity to work through the protocol in
real‐time, and to work through challenges with event notification, data collection, and communication in a field
setting. Before the pilot launch, the CDC team traveled to the utility to conduct a project kick‐off meeting and
on‐site training. Additionally, the CDC team and utility communications staff worked together to a develop press
release, fact sheets, and other community outreach materials. The community outreach efforts were designed
to make the public aware of the study’s importance and legitimacy; the study goal was presented as
understanding the links between water use habits and health.
Pilot Results
Field and laboratory data were collected from six LPEs over a five‐month period. The LPE form was completed
for the six LPEs. Three of the events were main breaks, and three were planned repairs or maintenance events;
2
none of the LPEs resulted in drinking water advisories, such as boil water notices. Two of the main breaks were
likely caused by deterioration and aging of the infrastructure, and one of the main breaks was due to an external
contractor accidentally sawing through the main. The first planned event was a valve replacement to isolate a
residential area from a planned utility worksite for relocating a main, the second planned event involved putting
an abandoned line out of service, and the third planned event involved relocating a main within a two‐day
period. Two of the planned events involved maintenance activities over multiple sequential days; for these
events, each day’s event/repair was recorded on a separate LPE form. It was not possible to measure pressure
readings during the events without violating the observational study design (i.e., interfering with or delaying the
repairs to measure the pressure would have caused study‐related changes to service), but it was possible to
obtain normal pressure readings before or after each of the LPEs.
Twenty‐four UF samples and 36 grab samples were collected and analyzed. Aerobic endospores were frequently
detected in UF samples and total coliforms were detected in one UF sample. E. coli, enterococci, C. perfringens,
somatic coliphages, and human‐specific Bacteroides spp. were not detected in any UF sample. Total coliforms, E.
coli and enterococci were not detected in grab samples analyzed by the utility, but HPC, TOC, total hardness,
alkalinity, and specific conductance were quantified for each grab sample.
A total of 646 households were contacted, and the overall survey response rate was 37%. The response rates for
each event ranged from 32‐43% and did not appear to vary by event type or size. Additionally, the response
rates were similar in LPE (38%) and non‐LPE areas (36%). Almost three‐quarters of the households (74%)
reported using home tap water for drinking in the last 30 days, and nearly all reported using home tap water for
potable purposes (99.6%).
One‐third of households in LPE areas (33%) reported noticing low pressure during the three weeks following the
LPE, and 28% reported noticing a complete loss of water service. Households in non‐LPE areas reported noticing
low pressure during the three‐week period about half as often (16%) as households in LPE areas, and <1% of
households in non‐LPE areas reported noticing a complete loss of service. Households in both LPE (21%) and
non‐LPE areas (14%) reported noticing a change in tap water during the three‐week period of interest. About 5%
of households reported that they were told to boil their water before drinking it, with LPE areas reporting this
more often (8%) than non‐LPE areas (2%).
Discussion
For each event, it was feasible to use knowledge of the water system and hydraulic principles to select the LPE
and non‐LPE areas within a few days of the LPE. The LPE form was an effective method for systematically
collecting descriptive information about the LPE characteristics. The LPEs took place under varied circumstances,
so the form was a useful tool for distilling the many event details to capture common information that could be
aggregated across all LPEs. The pressure data on the form were frequently missing. The pressure readings add
value to the field data, but the scope of how the pressure measurements can be used is limited because the
data are often missing, and the field measurements cannot capture the variability within the system. If a utility
has automated pressure sensors already in use, it could be helpful to gather this information in conjunction with
the study data to provide additional context about the LPE. The engineering attribute assignments were helpful
for making comparisons across the different LPEs during the pilot data analysis phase.
3
Water sample collection in LPE and non‐LPE areas was successfully performed within designated response times.
However, weather and logistical constraints precluded testing UF samples for two of the six LPEs studied.
Physical and chemical water quality data were readily obtained for water samples, but only two microbial
parameters (aerobic endospores and HPC) were consistently detected and quantified in UF or grab samples. In
general, water quality data were similar between LPE and non‐LPE areas, with the exception of turbidity, for
which LPE‐area water samples averaged 2.1 NTU and non‐LPE areas averaged 0.7 NTU. These pilot data will be
used to develop a modified water quality analytical plan for the full study to better optimize effective water
quality data collection versus the time and expense required for sample analysis.
The customer service lists were a reliable source of contact information, with deliverable addresses available for
98% of contacted households. The strategy of cold contacting random samples of water utility customers by mail
was resource efficient; however, the trade‐off might have been a lower response rate. Power calculations were
developed with the assumption that a minimum response rate of 40% was needed to answer the primary
research question of whether the events are associated with AGI, and the pilot achieved a 37% overall response
rate. To improve response rates, the study team will make minor modifications to the survey procedures and
will increase efforts to promote the study in the participating communities. The additional efforts to promote
the study will be designed to help communicate the study’s legitimacy and importance; the study goal will be
presented as understanding the links between water use habits and health. The survey itself took customers, on
average, 13 minutes to complete, suggesting that the time burden for participation was low.
Most households in LPE areas did not report observing low pressure, complete service loss, or a change in tap
water during the observation period following the LPE, indicating that customers were usually unaware of their
exposure status. Non‐differential survey response in LPE and non‐LPE areas and across different events likely
indicated that there was limited selection bias. Additionally, if response rates had been substantially higher in
exposed LPE areas than unexposed non‐LPE areas, it would have raised concern that persons in LPE areas were
more concerned about their water quality, which would increase the possibility of bias in the responses about
water use and illness. The study design and procedures allowed the study team to collect the data needed to
meet the study goal and aims. There were scheduling and data collection challenges during the winter season or
during other busy periods at CDC or the utility. To address this, the study team will plan for scheduling conflicts
and consider a staggered scale‐up to multiple utility sites. During the utility training, the importance of early
event notification and communication between the utility and CDC team prior to field data collection will be
emphasized for facilitating data collection.
Plans for National Multi‐Site Study
The national multi‐site study is expected to be launched in 2014. To be eligible to participate, the utility must
use a secondary disinfectant, have a history of at least one LPE per month (main breaks or planned repairs), on
average, and be able to partner with CDC for a 12‐18 month period to complete data collection for
approximately 13 low pressure events.
Each utility’s participation will help researchers gain an understanding of the health impacts of routine and
unplanned distribution system low pressure events, and this understanding can be used to help focus resources
toward effectively maintaining the safety and durability of U.S. water systems. Additionally, participation can
help utilities gain a better understanding of their customers’ water use habits and experiences; CDC can provide
4
each utility with de‐identified summary survey data about household water use. The utility and customer results
will be combined with others, and the participating utilities will not be identified in any publications.
5
INTRODUCTION
Rationale for National Study of Health Effects Associated with Distribution System
Low Pressure Events (LPEs)
Water management and treatment advancements to make safe water widely available across the U.S. have
contributed to reduced illnesses and deaths from infectious diseases, considered one of the top 10 public health
achievements of the 20th century [(CDC, 1999a), (CDC, 1999b) (Cutler and Miller, 2005)]. However, aging
infrastructure undermines the ability of water distribution systems to deliver safe, pressurized drinking water to
homes, hospitals, schools, and businesses. Distribution system vulnerabilities, such as water main breaks, cross‐
connections, back‐flow, and pressure fluctuations can result in potential intrusion of microbes and other
contaminants into the distribution system (Borchardt et al., 2004; LeChevallier et al., 2003; Swerdlow et al.,
1992; Lambertini et al., 2011). National waterborne disease outbreak data show that contamination of water in
distribution systems is a risk factor for disease, accounting for approximately 10% of all deficiencies identified in
outbreaks occurring in public water systems (Craun et al., 2010).
Each year, approximately 240,000 water main breaks occur in the U.S., presenting opportunities for pathogen
intrusion into water distribution systems (Kirmeyer et al., 1994). There are an unknown number of planned
maintenance events and repairs. A study conducted in Norway systematically examined LPEs, including main
breaks and planned maintenance events, and gastrointestinal illness using a prospective cohort study design,
and found that LPEs were associated with a 58% increased risk of acute gastrointestinal illness (AGI) in the week
following the LPE (Nygard et al., 2007). An experimental study in non‐disinfected public groundwater systems in
Wisconsin assessed the impact of common distribution system repairs and maintenance activities on virus
concentrations in drinking water, and identified higher virus levels in tap water following some distribution
system events, including pipe installation, valve exercising, and cutting open mains (Lambertini et al., 2011). In
other intervention trials, the amount of gastrointestinal illness attributed to drinking municipal tap water has
ranged from less than 11% (Colford et al., 2005) in the U.S. to 34% (Payment et al., 1991) in Canada. This
variation might be due in part to differences in the quality of the underlying source water (Hellard et al., 2001),
treatment processes, participant blinding issues (Payment et al., 1991; Payment et al., 1997), or other factors
(Colford et al., 2006). A recent meta‐analysis concluded that tap water consumption can be associated with
endemic gastrointestinal illness when there are distribution system problems, including water outages, lack of
routine continuous service, or a lack of disinfectant residuals (Ercumen et al., 2014). Exposure and risk studies
are valuable for identifying potential mechanisms for distribution system contamination and for identifying
vulnerabilities in the distribution system, but these studies are often not directly linked to human health
outcomes, so they cannot be used to assess the LPE impact on community health. Systematic data collection
across many LPEs, with a study design tailored for the purpose of answering this research question, is needed to
identify the health impacts of LPEs.
The existing evidence base suggests that distribution system vulnerabilities are an important public health
problem, but there are no existing U.S. data sources that can be used to systematically assess whether people
become ill more often following LPEs in water distribution systems. Outbreak data cannot be used to measure
specifically the association between LPE occurrence and illness as LPEs usually impact small portions of the
distribution system, resulting in a relatively small number of exposed persons per LPE. An increase in the
occurrence of illness over baseline incidence in exposed persons would not typically come to the attention of
6
public health agencies. As an example, in a recent communitywide cryptosporidiosis outbreak linked to
contaminated drinking water reservoirs, an epidemiologic survey conducted once the outbreak was detected
found that 28% of the community was ill; however, the outbreak was only detected after 5 lab‐confirmed cases
came to the attention of the health department (DeSilva, 2014). Distribution system events are even more
challenging to detect with outbreak surveillance because contamination is often transient and affects a smaller
proportion of the population. These health events are unlikely to generate a signal that public health agencies
could detect. Additionally, states have different resources to investigate and report outbreaks, so outbreak data
are not collected or reported systematically on a national level.
To address this need, the Waterborne Disease Prevention Branch at the Centers for Disease Control and
Prevention (CDC) plans to conduct the first epidemiologic study in the U.S. to assess whether individuals
exposed to LPEs in the water distribution system are at an increased risk for AGI or acute respiratory illnesses.
The project’s success hinges on forming strong public health and water utility sector partnerships.
Goal and Aims for the National Study
The goal for the national study is to determine whether individuals exposed to LPEs in water distribution
systems are at an increased risk for gastrointestinal or respiratory illness.
The study aims are to 1) assess the association between exposure to the LPE and self‐reported illness during a
short observation period following the LPE, 2) compare microbial indicator test results in water samples from
LPE and non‐LPE (control) areas, and 3) describe the LPE characteristics.
Epidemiological Study Design
A prospective cohort study design, a type of observational study, was implemented. In an observational study,
the researchers observe the effect of an exposure on an outcome in a target population but do not influence the
exposure. This contrasts with an experimental design, such as a clinical trial, where the researchers assign
groups of people to an exposure, such as a drug or treatment, and observe the outcome. All research involving
human subjects requires an ethical approval process; the approvals for an experimental design are even more
stringent. Since this study has ethical approval to conduct an observational study, it is especially important that
water utilities participating in the study do not change their water service or repair procedures in any way as a
part of the study.
In a prospective cohort study design, the researchers select groups of people who have different exposure‐
status at baseline to observe over time. When an LPE (exposure) occurred, investigators identified areas
affected by the LPE (LPE areas), and similar, but unaffected control areas (non‐LPE areas). Because exposures are
not randomly assigned, confounders–that is, other factors that differ by exposure status and are associated with
illness–could make it appear that outcomes are different between exposed and non‐exposed areas, when the
exposure of interest does not actually cause the difference in outcome. To control for confounders that could
influence the apparent relationship between the LPE and the illness outcome, investigators matched the areas
on census resident age groups, pipe material and size, drinking water source, and main housing type, and
collected data on other potential confounders using the household survey, such as recent travel or respondent
age, for statistical adjustment during data analysis. The study team mailed surveys to random samples of
households from the LPE and non‐LPE areas to elicit information on water use, recent water service, other
activities and exposures, and illness symptoms during a short observation period following the LPE. To avoid
bias, the survey participants were blinded to the specific research question about LPEs. The study goal
7
presented to the public through community outreach, the study website, and the household survey materials
was to understand links between water use habits and health. In the multi‐site study that will follow the pilot
study reported here, survey responses from LPE and non‐LPE areas will be compared to determine if individuals
in the LPE‐areas are more likely to report AGI symptoms, controlling for other factors, such as respondent age or
recent travel, which could potentially influence the risk of AGI.
For purposes of the study, an LPE was defined as a water service disruption event or incident causing a
presumed loss of water pressure in the distribution system, including unplanned outages or planned
maintenance events. For each LPE, water utility personnel completed a standard LPE form describing the event
and repair process. The study team selected areas affected by the LPE and similar but unaffected non‐LPE
control areas. Utility personnel collected ultrafiltration water samples and grab samples from both areas;
ultrafiltration samples were analyzed at CDC and grab samples were analyzed by the utility laboratory. The
utility business support team securely transferred contact information (i.e., names and addresses) for residential
customers living within the study areas to CDC, and the CDC team administered the survey.
8
PLANNING AND PREPARATION FOR PILOT STUDY
Pilot Study Goals and Objectives
The prospective cohort study design and study protocols were piloted at one utility site. The pilot study
objectives were to determine if study procedures allow the team to answer the research questions and to
familiarize team members with study implementation and procedures. The pilot activities were monitored
during data collection and evaluated post‐pilot through indicators that broadly measured 6 outcomes: 1) CDC
team and utility select appropriate LPE and non‐LPE areas, 2) the standard LPE form describing the event is filled
out completely at the time of repair, 3) the water sample collection team follows standard operating
procedures, 4) the laboratory team follows testing, standard operating procedures, and maintains chain of
custody logs, 5) the CDC team follows survey mailing procedures and tracks household participation, and 6) the
survey procedures encourage participation and equitable study participation in LPE area and non‐LPE area
groups. Information from the pilot evaluation is being used to streamline and improve the study processes,
protocol, survey, and laboratory methods prior to the full study launch.
Preparations for Pilot Launch
The CDC team trained utility personnel prior to the pilot launch. The first part of the training involved practicing
the selection of LPE and non‐LPE areas over the phone and by email, using work‐ups of past breaks and repairs.
The goal was to increase the utility team’s familiarity with the study protocol, the CDC team’s familiarity with
the utility’s water system, and for both teams to practice communicating and working together.
Also during this phase of training, the CDC team worked with utility business support and IT personnel to
develop a utility‐specific method of creating customer contact lists from study area maps. Once the method was
developed, study personnel practiced securely transferring customer data to CDC.
The next phase of training consisted of the team responding to mock events in the field. When an eligible event
occurred, field teams notified the utility project manager, and the study team made a site visit, filled out the LPE
form, and practiced selecting LPE and non‐LPE areas with the CDC team. This gave the study team an
opportunity to work through the protocol in real‐time, and to work through challenges with event notification,
data collection, and communication in a field setting.
Before the pilot launch, the CDC team traveled to the utility to conduct a project kick‐off meeting and on‐site
training. CDC staff visited the utility site, toured facilities, and delivered the ultrafiltration water collection
supplies. During the kick‐off meeting, CDC staff presented the study motivation, goals, and methods to the
entire utility community. Additionally, CDC staff trained the field operators on LPE form data collection and
trained utility staff on ultrafiltration water sampling.
The CDC team and utility communications staff worked together to develop a press release, fact sheets, and
other community outreach materials. The goal of these community outreach efforts was to make the public
aware of the study’s importance and legitimacy to support survey completion. CDC contact information was
included on the outreach materials at the utility’s request, and the CDC call center was prepared to field study
inquiries.
9
Establishment of Pilot Study Procedures and Roles
Figure 1 is a flowchart that outlines the pilot study procedures and utility and study team roles. The field
operators were responsible for notifying the utility project manager when an LPE occurred. Operators were
asked to respond to all LPEs using their normal procedures. The utility project manager was responsible for
notifying the CDC team of the LPE; the CDC team was responsible for determining if the LPE was eligible for
study inclusion and for giving the go‐ahead to the utility to proceed with data collection. If the event was
eligible, the utility project manager conducted a site visit to observe the event and to fill out the standard LPE
form, gathering information from the operators as needed. Following the site visit, the utility project manager
provided the CDC team with the initial event description, LPE form, and a general map of the event area. The
utility and CDC team worked together to determine the extent of the LPE and non‐LPE areas, and to select
matching study areas. Following the event, the utility team collected water samples from the study areas and
transferred samples to the appropriate laboratories, following sample collection and handling protocols. The
laboratories followed their standard procedures for processing samples and managing data. The utility project
manager provided the utility business support team member with a GIS map of the final study areas; the map
was linked to a customer service database and lists of contact information were generated. The business
support team member securely transferred the contact information to CDC using an encrypted FTP site. The CDC
team retrieved the customer service list, selected the survey sample, and began household survey
administration. The utility project manager also securely transferred utility grab sample results to CDC using the
encrypted FTP site.
•
•
•
UTILITY FIELD CREWS
Notify utility project manager of LPE
Respond to LPE
Fill out pages 1‐2 of LPE form
DURING LPE
•
•
UTILITY and CDC STUDY TEAMS
Determine if LPE is eligible for study
Select LPE and non‐LPE study areas
AFTER WATER SERVICE RESTORED (LPE END)
•
•
•
•
UTILITY STUDY TEAM
Collect 100‐L ultrafiltration samples
Collect grab samples
Fill out pages 3‐4 of LPE form
Transfer samples to appropriate labs
•
•
UTILITY BUSINESS SUPPORT
Link GIS maps of study areas to customer database
Transfer customer contact information for
selected study areas to CDC using sFTP site
•
•
UTILITY LAB
Process grab samples
Manage lab data
•
•
CDC LAB
Process ultrafiltration
samples
Manage CDC lab data
•
•
•
CDC EPI TEAM
Take survey sample
Begin survey mailings
Manage survey data
Figure 1. Pilot study procedures and utility and study team roles
10
Selection of LPE and Non‐LPE Areas
The most challenging aspect of the protocol was the selection of the matching LPE and non‐LPE areas. During
each event, the study team selected the LPE and non‐LPE areas using knowledge of the water system and
hydraulic principles; a standard list of engineering attributes was developed to guide decision‐making (Table 1).
Table 1. Attribute classification for selection of LPE exposed and unexposed areas
Exposed area attributes
1. Areas with known lower steady state pressures
2. Smaller diameter mains nearby or in direct hydraulic connection to LPE
3. Higher elevation than main break location (assuming there is no nearby storage to compensate for the
elevation)
4. Near lower flow areas such as dead ends, pressure zone boundaries
5. Away from pressure release valves, pumps, or storage facilities that float on the system, which would
allow water to be released into the system to compensate for lower pressure caused by LPE
6. Other, please describe (for example, flow direction in surrounding areas changed as a result of the
repair)
Unexposed area attributes
1. Nearby but different pressure zone
2. No recent main breaks or LPEs in the vicinity
3. If in same pressure zone, served by pump or storage facility that floats on system
4. If in same pressure zone, area served by larger diameter mains with routinely good steady state
pressures
5. Areas in the middle of the grid (away from low flow sections, dead ends, pressure zone boundaries)
6. Other, please describe
Water Sample Collection and Testing Methods
For each LPE, utility study personnel collected distribution system water samples from residential hose bibs in
both the LPE area and the non‐LPE area. Water sample sites (selected by the water utilities) were not linked to
household survey addresses. These samples were intended to be representative of the water in the distribution
system in areas affected by the LPE and in areas serving the unaffected addresses included in the study. The
exact location of all samples were determined by the water utility’s field team and had to meet specific criteria.
Three, ~100‐L drinking water samples were collected from the exposed area on the same day as the LPE: one
upstream of LPE within five service connections, one downstream of LPE within five service connections, and
one at the water utility personnel’s discretion within the exposed area. In addition, three ~100‐L drinking water
samples were collected from the unexposed area within 48 hours of the LPE.
Prior to sample collection, the following water quality parameters were measured and recorded at each site:
water temperature by Standard Method (SM) 2550B (Field) (for all Standard Methods, please refer to APHA,
2005), total chlorine according to SM 4500CL‐G (Field), pH by SM 4500H‐B (Field), turbidity by SM 2130B (Field)
and specific conductance by SM 2510B (Field). Grab samples for water quality testing were collected in sterile
bottles, according to analyte method requirements, and large‐volume (~100 L) water samples were collected
using dead‐end ultrafiltration (Smith and Hill, 2009). Samples were shipped in coolers to the CDC Environmental
Microbiology Laboratory in Atlanta, as well as the utility laboratory, for testing within 30 hours of collection. The
utility processed grab samples for the following analytes: total coliforms and E. coli (SM 9223B), enterococci
11
[American Society for Testing and Materials (ASTM) # D6503 99], heterotrophic plate count (HPC) (SM 9215D),
total hardness (SM2340B), alkalinity as CaCO3 (SM 2320B), calcium (EPA 200.7‐DW), magnesium (EPA 200.7‐
DW) and total organic carbon (TOC) (SM 5310C). The CDC Environmental Microbiology Laboratory processed the
ultrafilter samples by backflushing with a solution consisting of 0.01% Tween™80, 0.01% sodium polyphosphate
(NaPP), and 0.001% Antifoam Y‐30 (Smith and Hill, 2009). Each concentrated water sample was analyzed using
the following enumerative methods: total coliforms and E. coli, by SM 9223B ‐ Colilert Quanti‐Tray 2000 (IDEXX
Laboratories) enterococci by Enterolert Quanti‐Tray 2000 (IDEXX Laboratories), C. perfringens by membrane
filtration and culture onto mCP agar (USGS, 2007), and aerobic endospores according to SM 9218B. For somatic
coliphages and human‐specific Bacteroides spp. analysis, the concentrated water samples were further
concentrated by polyethylene glycol (PEG) precipitation (Hill et al., 2010), and the final pellet was then divided
for analysis. One‐half of the pellet was analyzed for somatic coliphage by single‐agar layer plaque assay (SM
1602), and 750 µL was extracted for human‐specific Bacteroides spp. testing by real‐time polymerase chain
reaction (PCR) (Haugland et al., 2010).
Household Survey
A multiple contact mailing strategy was implemented (Dillman, 2007). Selected households received 5 survey
prompts from CDC by postal mail: an advance letter, initial survey packet, thank you/reminder note, second
survey packet, and final appeal letter. Respondents had the choice to respond by mail using a provided postage‐
paid return envelope or on‐line using a secure survey website, accessed using an ID and password provided in
the survey packet.
Data Analysis
All analyses were performed in SAS version 9.3 (SAS Institute Inc., Cary NC) using a p‐value <0.05 to indicate
statistical significance. Levels of laboratory parameters were summarized as untransformed (crude) means and
standard deviations; for values below the limit of detection (LOD), ½ the LOD value was substituted (Hornung
and Reed, 1990). Correlations between parameters were evaluated using Spearman correlations. To test for
statistically significant differences in laboratory parameters by LPE status, appropriate data transformations
were identified to achieve approximately normal distributions (PROC TRANSREG). The transformed values were
modeled using a random effects model (PROC MIXED) to account for the matched design and non‐independence
of observations from the same LPE. Survey data were summarized using descriptive percentages of both
household‐level and individual‐level data, but no statistical comparisons between LPE and non‐LPE area
responses were performed because there was insufficient statistical power due to the small scope of the pilot
study.
Scope of Data Collected During Pilot Study
Since the pilot study was designed to test the design and procedures for the national study, rather than to
answer the research questions for the national study, the scope was much smaller. Pilot study data collection
was not intended to provide an evidence base for making inferences about the research questions. The pilot
study data are being used to inform modifications to study processes and methods prior to the full launch of the
national study. Table 2 summarizes the pilot data collected and explains how the data are being used to improve
study procedures before the full study launch.
12
Table 2. Summary of pilot data collection and how data were used in the pilot evaluation
Database
Data description
Purpose for evaluation
LPE form
Access database
Pages 1‐2 of LPE form – event and
repair description
Form and data quality
Pages 3‐4 of LPE form – parameters
for water samples, chain of custody
Form and data quality; exploratory lab
analyses; chain of custody compliance
CDC event records
CDC network
drive/hard copies
Records of emails, maps,
demographic information, checklists
for all events
Burden of work for utility; format and quality
of field data; data management and
organization
Utility laboratory
Excel booklets
Grab sample results
Exploratory lab analyses; which parameters to
include in full study; give clarification on
when/where to take samples
CDC laboratory
Access database
Ultrafiltration sample results
Exploratory lab analyses; which parameters to
include in full study; give clarification on
when/where to take samples
Household survey
SQL database
Secure storage for paper and web
survey data
Data management; format and data quality;
content; response rate and timing
Household survey
Access interface
Secure storage for customer contact
information, mailing schedule, and
tracking mailing and participation
Mailing management; response rate and
timing
Climate data
SAS database
NOAA quality controlled local climate Plan for integrating data into study analyses
data
13
PILOT STUDY RESULTS
The study team collected field and laboratory data from the six LPEs from December 2013 to March 2014. Data
collection for the household survey administration was completed in May 2014. A total of 646 households were
surveyed, LPE forms were completed for each event, and 24 ultrafiltration samples and 36 grab samples were
collected and analyzed. Data were collected and stored securely in several databases, housed at CDC.
Detailed descriptions of each LPE and the decision making process for selecting the study areas are in the
Appendix. The utility’s internal infrastructure maps were used to determine the pipe sizes, materials, and water
sources in the area and create a ‘normal flow’ map (i.e., before the LPE). The event description information and
the LPE form were used to create an ‘event hydraulic’ map, outlining the boundaries of the LPE and non‐LPE
areas and any changes in flow direction or magnitude that occurred following the LPE. The utility’s engineering
department modeled the break or repair scenarios for three events; the hydraulic modeling results supported
the study team’s choices of the areas. As quality assurance for the pilot study, it was helpful to use the hydraulic
models to check the choice of the study areas. However, the modeling results were unavailable until a few days
after the event, so it would not have been possible to use the models in real‐time to select the areas. It was also
not possible to obtain the field pressure measurements during the events. During the emergency events, the
operators were focused on responding to the event, and interfering with their work to obtain the pressure
readings would have caused study‐related delays in restoring water service. During the planned events, the
study team was able to demonstrate low pressure by verifying no water service at home hose bibs but was not
able to obtain pressure measurements. The study team was able to obtain ‘normal’ pressure readings once the
water service was restored.
The utility’s water system maps were used to select areas with similar pipe size, material, and water source, and
the repair site visit was used to locate areas with a complete or partial main housing type match. For 4/6 events,
it was not possible to locate LPE and non‐LPE areas within the same census block group or tract; the areas
potentially exposed to the LPE were usually larger than the census area boundaries. As an alternative, the areas
were matched on the census age groups (i.e., percent of people in the census tract <18 years and percent of
people in the census tract 65 years and older).
Although a few items were missing from the LPE forms, such as the description of the pipe interior when the
repair prevented close observation, LPE forms were nearly complete for all events. The customer service lists
were securely transferred to CDC for all LPEs. A late customer list delayed survey administration for one event;
this event took place over the winter holiday season, which made it difficult to coordinate personnel leave
schedules. The field water parameters, chain of custody information, and utility grab sample results were
complete for all water samples. For two events, the ultrafiltration shipments did not arrive at CDC within the 30
hour processing window. One shipment was lost during a winter storm that shut down Atlanta, and one was not
collected at CDC’s request because of a scheduling conflict with another CDC resource‐intensive project.
Data Collected Using the LPE Form
During the LPEs, the operators and utility project manager completed the LPE form to describe the event and
the repair procedures. The LPE form provides a standardized mode to collect descriptive information about the
LPE characteristics. Three of the events were main breaks, and three were planned repairs or maintenance
events (Table 3). Two of the main breaks were likely caused by deterioration and aging of the infrastructure, and
14
one of the main breaks was due to an external contractor accidentally sawing through the main. The first
planned event was a valve replacement to isolate a residential area from a utility worksite for relocating a main,
the second planned event involved putting an abandoned line out of service, and the third planned event
involved relocating a main twice in a two‐day period. Two of the planned events involved maintenance activities
over multiple sequential days; for these events, each day’s events were recorded on a separate LPE form.
For one main break, the utility project team assessed that there was high potential for contamination, for a
second main break the utility project team assessed that there was moderate potential for contamination, and
for a third main break and all three planned events, the utility project team assessed that were was low
potential for contamination (Table 4). In two of the main breaks, the pipe was submerged in trench water before
or during the repair; in one main break, the pipe was submerged in trench water while the operators were
tightening the bolts but not during pipe installation. The replacement parts were swabbed with chlorine in 3/6
events, the main was flushed after the repair in 5/6 events, and the main was chlorinated after the repair in 0/6
events. For all of the events, no precipitation was reported during the repair, and no sewage lines or reclaimed
lines were adjacent or in close proximity to the main being repaired (data not shown).
The (affected) pipe sizes varied from 2”‐ 8” diameter, and the pipe ages from < 20 years to approximately 50
years (Table 5). The pipe depth ranged from 3’‐ 4.5’, and the soil type was usually sand. The pipe material was
PVC in four events, ductile iron in one event, and galvanized material in one event. The water source was a
surface and groundwater blend for five events, and 100% groundwater for one event.
Each event had a different extent of exposure, including the duration of low pressure and the number of
households affected. For the main breaks, the utility estimated the number of households affected by the break
before it was isolated and in the repair area (Table 6). For the planned events, the utility estimated the number
of households in areas that were completely or partially valved off for the procedure. The main housing type
was often single family homes; condos or apartments were also present in two events. The number of
households exposed to low pressure ranged from 16 to 166, and the duration of low pressure ranged from 45
minutes to seven hours. There were no reports of a boil water advisory or notice administration. The team was
unable to take pressure readings during the events (Table 7). In the three planned events, and in one of the
main breaks that caused minimal damage, the study team was able to demonstrate low pressure once the area
was valved‐off by verifying no water service at home hose bibs or assuming zero pressure when the area was
completely valved off. In two of the breaks, where water flooded the street, the study team was unable to
estimate the pressure at all. The study team was able to get normal pressure readings for all six events.
15
Table 3. Event, response, and repair type for six pilot LPEs
LPE
Main break
Valve replacement
Main break
Main break
Abandon line
Main relocation
Cause
Deterioration, blowout in old, weak pipe; water rushed into
street, damaged home and tree
Main needed to be moved in preparation for installation of new
storm drain; valve broke during repair causing larger area than
expected to be valved off
Older pipe leaking at bell and spigot joint; contractor installing
sidewalk reported leak
Contractor burying phone lines sawed through 6" main, caused
blowout
Cross‐connection between out‐of‐service line & in‐service line
caused leaking in street; repair to abandon line and stop leak
Main needed to be moved in preparation for installation of
traffic signal pole
Response
Emergency
Planned
maintenance
Emergency
Emergency
Planned repair
Planned
maintenance
Repair type
Clamp repair, but pipe broke again after
clamped
Valve installation; valve off area (multi‐
day)
Cut and replace section of pipe; valves
shut down to isolate break
Cut and replace section of pipe
Abandon old main by cutting it and
adding valve; valve off area
Cut and replace and offset section of
pipe; valve off area (multi‐day)
Table 4. Event and repair description for six pilot LPEs
Main break
Operator assessment
of potential for
contamination
High
Valve replacement
Main break
Low
Low
Main break
Moderate
Abandon line
Low
Main relocation
Low
LPE
Pipe in trench water
Parts swabbed
with Chlorine
Main flushed Main chlorinated
after repair
after repair
Yes ‐ water from main before and during
repair
No
Yes ‐ water in pit when tightening bolts, not
during installation
Yes ‐ water from main before repair crew
arrived
Yes for out‐of‐service line, but homes
connected to in‐service line
No
No
Yes
No
Yes
No
Yes
Yes
No
No
Yes
Yes
No
No
No
No
Yes
Yes
No
16
Table 5. Characteristics of infrastructure, water source, and soil type for six pilot LPEs
LPE
Pipe
dia.
Est. Pipe age
(years)
Pipe
depth
Material
Pipe
interior a
Operator’s description
of pipe interior
Water
source
Soil
Main break
8”
33
3'
PVC (white)
‐‐
Did not see
Blended
Sand, clay
Valve replacement 6”
Don't know
3'
Ductile Iron
2
No tubercles, fairly
clean
Blended
Sandy, shells (beach)
Main break
6”
48
3'
PVC
1
Brown, slimy biofilm
Blended
Sand
Main break
6”
48
4'
PVC
1
Light coating of biofilm,
good condition
Blended
Sand
Abandon line
2”
45
3', 6"
Galvanized
3
Very corroded on
outside and rusty and
pitted on inside
Blended
Sand
Main relocation
8”
<20
4', 6"
PVC
1
100%
Sand
Good condition, no
tubercles. Small amount Groundwater
of smooth biofilm
a
Interior condition of pipe rated on a scale of 1 to 5, with 1 = smooth and 5 = highly tuberculated
Table 6. Field description of how households were affected by the six pilot LPEs
LPE
Exposed area
Area affected by break &
valved off area
Valve replacement Valved off area
Main break
Main housing
Single family homes;
condos
Single family homes;
condos/apartments
Single family homes
No. homes exposed to Low pressure
low pressure
duration (hr)
154
7
No. homes
lost service
31
Loss of
service (hr)
0.5
85
85
3
3
Area affected by break &
23
1.8
23
1.8
valved off area
Area affected by break &
3
5.5
Main break
Single family homes
166
3.5 a
valved off area
Abandon line
Valved off area
Single family homes
16
0.75
16
0.75
Main relocation
Valved off area
Single family homes
130
4
130
4
a
One‐hundred‐sixty‐six homes experienced low pressure for 3.5 hours; three of these experienced an additional 5.5 hours of complete service loss.
Main break
17
Table 7. Study LPE‐area pressure readings for six pilot LPEs
Pressure at
Means of verification
break/work site (psi)
Unable to measure because of
Main break
Area affected by break & valved off area
‐‐
damage from main break
Valve replacement Valved off area
0
Assumed 0 because valved off
Opened home hose bibs, verified no
Main break
Area affected by break & valved off area
0
water
Main break
Area affected by break & valved off area
‐‐
Unable to measure
Opened home hose bibs, verified no
Abandon line
Valved off area
0
water
Main relocation
Valved off area
0
Assumed 0 because valved off
a
Pressure readings taken before the area was valved off for the planned repair.
Type
Exposed area
Pressure after
break/work (psi)
76; 83
60; 64
69; 69; 67
53; 53;64
53a; 65a; 65a
83; 82; 80
18
Water Samples
Field water quality data and grab sample testing data were obtained for all six LPEs. Water quality data from the
UF sampling was also obtained for four of the six LPEs. Overall mean levels of the various parameters for the
field and grab sample water quality parameters are summarized in Table 8.
Crude mean values by exposure status for the six events and p‐values from random effects models are shown in
Table 9. Most of the water quality parameters had similar crude mean values between LPE and non‐LPE areas;
however, the p‐values indicate a statistically significant (p<0.05) difference by exposure status for some
parameters. This is mainly because the random effects models isolate the within‐event differences between the
water in exposed and unexposed areas while taking into account the variability in water characteristics between
events. Total hardness, specific conductance, turbidity, and TOC varied significantly (p<0.05) between LPE and
non‐LPE areas; the only parameter for which the crude means appeared meaningfully different was turbidity
(2.1 NTU in LPE areas vs. 0.7 in non‐LPE areas, p = 0.001). For total hardness, specific conductance, and TOC, the
crude means were similar and the standard deviations overlapped substantially, indicating that the variability
associated with the LPE was smaller than variability between events in the values of these parameters. Mean
HPC values were higher in LPE areas than non‐LPE areas, though not significantly. Because the event involving a
groundwater source had higher TOC levels compared to the five events involving blended water and appeared
to be an outlier for this parameter (not shown), the comparisons were repeated after excluding the
groundwater‐associated event. TOC levels remained significantly different between exposed and unexposed
areas, mean levels were lower in both exposed and unexposed areas (1.4 and 1.3 mg/L), and standard
deviations were much smaller (0.1 and 0.1).
Statistical correlations between parameters using all data (i.e., with LPE areas and non‐LPE areas combined)
were also tested (Figures 2‐3). Figure 2 shows the distribution and the correlations among turbidity, HPC, TOC,
aerobic endospores, and total chlorine. Turbidity and HPC were positively correlated (r = 0.33, p‐value = 0.04)
and HPC and total chlorine were inversely correlated (r = ‐0.32, p = 0.06). TOC and HPC were also found to be
inversely correlated (r = ‐0.41, p‐value = 0.01). However, after excluding the event involving a groundwater
source, this correlation was reduced in magnitude and was no longer statistically significant (r = ‐0.30, p = 0.11).
Figure 3 shows the distribution of total hardness, alkalinity, pH, temperature, and total chlorine and the
correlations between the parameters. Many of the field parameters were statistically correlated. Total hardness
was correlated with alkalinity, pH, temperature, total chlorine, TOC, and specific conductance. Additionally,
alkalinity was correlated with temperature, TOC, and specific conductance, and pH was correlated with specific
conductance.
The detection rates of the UF samples are listed in Table 10. There was 1 total coliform positive from an LPE‐area
sample (n = 24). No samples were positive for E. coli, enterococci, C. perfringens, somatic coliphages, or human‐
specific Bacteroides. Aerobic endospore testing yielded the most detections for UF samples. Seventy‐five
percent of the samples were positive (18/24) for aerobic endospores. There was not an apparent difference in
aerobic endospore detections or concentrations between LPE area and non‐LPE area samples.
19
Table 8. Distribution of field and grab sample water quality parameters following six LPEs (n = 36)
Parameter
Mean
Calcium (mg/L)
Magnesium (mg/L)
Turbidity (NTU)
Total Hardness (mg/L)
Alkalinity (as CaCO3 mg/L)
pH
Temperature (C)
Total Chlorine (mg/L)
TOC (mg/L)
HPC (CFU/mL)
Specific conductance (µmhos/cm)
77.0
6.4
1.4
218.5
167.2
7.6
21.9
3.4
1.7
128.9
555.3
SD Median
3.4
0.8
1.3
11.0
18.1
0.1
1.6
0.5
0.7
150.5
47.8
77.1
6.7
1.1
221.0
160.0
7.7
21.9
3.6
1.5
56.5
571.0
Min
Max
25th
Pctl
75th
Pctl
70.0
4.7
0.2
194.0
140.0
7.4
18.3
2.0
1.2
3.0
448.0
82.7
7.2
7.0
236.0
200.0
8.0
25.7
4.2
3.4
640.0
611.0
76.2
5.9
0.7
217.0
150.0
7.6
21.0
3.2
1.4
12.5
534.0
79.2
7.1
1.4
224.5
180.0
7.8
23.3
3.7
1.5
200.0
590.0
Table 9. Distribution of water quality parameters by exposure status from six LPEs
Parameter
LPE areas
N Mean
SD Median
Calcium (mg/L)
18
76.5
3.0
76.6
Magnesium (mg/L)
18
6.3
0.8
6.7
a
Total Hardness (mg/L)
18 217.1 10.2
220.5
18 167.2 18.7
160.0
Alkalinity (as CaCO3 mg/L)
pH
18
7.6
0.2
7.6
18
21.9
1.8
21.9
Temperature (C)
Total Chlorine (mg/L)
18
3.5
0.5
3.6
Specific conductance (µmhos/ cm) a 18 545.3 49.4
568.0
Aerobic endospores (CFU/100 mL)
12
0.3
0.5
0.0
Turbidity (NTU) a
18
2.1
1.6
1.4
TOC (mg/L) a
18
1.8
0.8
1.5
HPC (CFU/mL)
18 163.6 182.0
120.0
a
Statistically significant difference between LPE and non‐LPE area
N
18
18
18
18
18
18
18
18
12
18
18
18
Non LPE‐areas
P‐value
Mean
SD Median
77.5 3.7
77.3
6.4 0.9
6.6
219.9 12.0 221.0
0.02
167.2 18.1 160.0
1.00
7.7 0.1
7.7
0.07
21.9 1.4
21.8
0.99
3.4 0.4
3.4
0.34
565.2 45.4 583.0 <0.001
0.2 0.4
0.0
0.65
0.7 0.4
0.8 0.001
1.7 0.8
1.4
0.01
94.2 104.6
46.0
0.85
20
HPC
(CFU/mL)
TOC
(mg/L)
Aerobic endospores
(CFU/100mL)
Total Chlorine
(mg/L)
Total Chlorine
(mg/L)
Aerobic endospores
(CFU/100mL)
TOC
(mg/L)
HPC
(CFU/mL)
Turbidity
(NTU)
Turbidity
(NTU)
Figure 2. The distribution of turbidity, HPC, TOC, aerobic endospores, and total chlorine and the correlations
between the parameters. Data for all water samples, including those collected from LPE areas and non‐LPE
areas, are included. Scatter plots indicate correlations between parameter indicated on row and parameter
indicated in column. The histograms show the overall distribution of the indicated parameter across all
samples.
21
Total Hardness
(mg/L)
Alkalinity
(CaCO3 mg/L)
pH
Water Temp.
(C)
Total Chlorine
(mg/L)
Total Chlorine Water Temp.
(mg/L)
(C)
pH
Total Hardness
Alkalinity
(mg/L)
(CaCO3 mg/L)
Figure 3. The distribution of total hardness, alkalinity, pH, temperature, and total chlorine and the
correlations between the parameters. Data for all water samples, including those collected from LPE areas
and non‐LPE areas, are included. Scatter plots indicate correlations between parameter indicated on row and
parameter indicated in column. The histograms show the overall distribution of the indicated parameter
across all samples.
Table 10. Detection rates using ultrafiltration (n = 24)
UF‐associated tests
Number of detections
Total coliforms
1
E. coli
0
Enterococci
0
C. perfringens
0
Somatic coliphages
0
Total aerobic endospores
18
Bacteroides spp.
0
%
4.2
0
0
0
0
75.0
0
22
Household Survey
A total of 646 households were contacted, and the overall survey response rate was 37% (Table 11). Two
percent of selected households could not be contacted by mail because the addresses were undeliverable. The
response rates for each event ranged from 32‐43% and did not appear to vary by event type or size.
Additionally, the response rates were similar in LPE (38%) and non‐LPE areas (36%).
Throughout the pilot, the CDC team tracked the day of completed survey receipt and the mode of response.
There was an apparent boost in survey response following each of the mail prompts from CDC (Figure 4). The
majority of respondents chose to respond by mail (70%); 30% completed the survey using the secure website.
The website surveys took an average of 13.3 minutes to complete, a median of 11 minutes, a minimum of 5
minutes, and a maximum of 56 minutes.
Almost three‐quarters of the households (74%) reported using home tap water for drinking in the last 30 days,
and nearly all reported using home tap water for potable purposes (99.6%), defined as: rinsing produce,
brushing teeth, drinking, making ice, mixing cold drinks, or mixing infant formula (Table 12). The majority of
households reported using tap water for cooking or making hot drinks (96%), and 18% reported using tap water
for medical uses, defined as: nebulizer, neti pot, or contact lens care. Reported home tap water uses were
similar between LPE and non‐LPE areas.
Most households (71%) reported tap water as a type of water household members drink most often at home.
Thirty‐six percent (36%) reported bottled water as a type of water household members drink most often at
home, and 29% reported only commercially bottled water as the type of water household members drink most
often at home. This differed some in LPE and non‐LPE areas, with households in LPE areas reporting tap water
use for drinking more often and exclusive bottled water use less often.
A minority of households reported having a private well (7%), which was similar in LPE and non‐LPE areas. Over a
third of households (35%) reported using a water softener, and water softener use was reported more often in
non‐LPE areas than in LPE areas. The majority of households reported using some type of point‐of‐use water
filter (68%).
The housing type reported most often was a house (89%), followed by an apartment or condominium (7%)
(Table 13). The reported housing type differed in LPE and non‐LPE areas, with almost all households in the non‐
LPE areas reporting living in a house (97%), compared to about 80% of households in LPE areas. The remaining
households in the LPE areas reported living in an apartment, condominium, duplex, villa, or cluster home.
Housing type was a matching characteristic, so it is not clear if the match was not good or if respondents did not
fully represent the selected areas. Sixty‐four percent (64%) of households reported pets living in the home or
yard, with more households in LPE areas reporting having pets (72%) compared to non‐LPE areas (56%). Very
few households reported livestock or animal enclosures within 50 yards of the household.
One‐third of households in LPE areas (33%) reported noticing low pressure during the three weeks following the
LPE, and 28% reported noticing a complete loss of service (Table 14). Households in non‐LPE areas reported
noticing low pressure during the 3‐week period about half as often (16%), and <1% of households in non‐LPE
areas reported noticing a complete loss of service. Households in both LPE (21%) and non‐LPE areas (14%)
reported noticing a change in tap water during the 3‐week period of interest. About 5% of households reported
23
that they were told to boil their water before drinking it, with LPE areas reporting this more often (8%) than
non‐LPE areas (2%). The utility did not conduct any boil water notices, orders, or advisories.
Table 11. Summary of pilot study data collection for six LPEs
Event
Size
Main break
Valve replacement
Main break
Main break
Abandon line
Main relocation
Large
Medium
Small
Large
Small
Large
Households Response, n (%)
Surveyed, n
160
65 (43)
82
26 (33)
56
19 (35)
160
59 (38)
28
12 (43)
160
52 (33)
40
Survey 1
leaves CDC
Number of respondents
35
30
Thank you/
reminder
25
Final appeal
Survey 2
20
15
10
5
0
1
3
5
7
9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45
Day
Mail
Web
Figure 4. Number and day of survey response, by mode of response (n = 231). Two additional mail surveys
were received outside of the 6‐week window and are not included in this figure or in survey analyses.
24
Table 12. Reported household water use (n = 231 households)
Overall
LPE area
Non‐LPE area
Characteristic
(n = 231)
(n = 117)
(n = 114)
%
%
%
Home tap water use in last 30 days
Drinking
74.0
76.1
71.9
Potable uses (Rinsing produce, brushing teeth, drinking,
making ice, mixing cold drinks, mixing infant formula)
99.6
99.2
100.0
Hot water consumption (cooking, making hot drinks)
96.1
94.9
97.4
Medical uses (nebulizer, neti pot, contact lens)
17.8
18.8
16.7
At home, water type household members drink most often
Any tap water
70.6
76.1
64.9
Any bottled water
36.4
31.6
41.2
Only bottled water
28.6
23.1
34.2
Home has private well
6.9
7.7
6.1
Home has water softener
34.6
27.4
42.1
Home uses water filter
68.4
71.8
64.9
Table 13. Reported household characteristics (n = 231 households)
Characteristic
Housing type
House
Apartment or condominium
Mobile home
Other (duplex, villa, or cluster home)
Pets living in home or yard
Livestock or animal enclosures within 50 yards of household
Livestock
Cats/dogs
Overall
(n = 231)
%
88.7
6.9
0.0
3.5
64.1
1.3
0.9
LPE area
(n = 117)
%
80.3
12.0
0.0
6.8
71.8
0.0
1.7
Non‐LPE area
(n = 114)
%
97.4
1.8
0.0
0.0
56.1
2.6
0.0
25
Table 14. Reported water change during 3‐week period following the LPE (n = 231
households)
Overall
(n = 231)
%
24.7
6.9
LPE area
(n = 117)
%
33.3
4.3
Non‐LPE area
Characteristic
(n = 114)
%
Noticed low pressure
15.8
Don't know
9.7
Noticed complete loss of service
14.7
28.2
0.9
Don't know
1.3
0.9
1.8
Noticed change in tap water a
17.3
20.5
14.0
Change in odor
7.8
8.6
7.0
Change in taste
5.6
7.7
3.5
Change in color
7.8
10.3
5.3
Told to boil water before drinking it
4.8
7.7
1.8
Don't know
1.3
1.7
0.9
a
One household from non‐LPE area wrote in other category of calcium deposits
26
DISCUSSION
Determining LPE Exposure Areas
It was feasible to use knowledge of the water system and hydraulic principles to select the LPE and non‐LPE
areas within a few days. It was essential that the areas were selected quickly and correctly, so that utility
business support had enough time to construct the customer service list and transfer it to CDC on time for the
first survey mailing. Incorrectly classifying the LPE and non‐LPE areas would violate a fundamental assumption of
the epidemiologic cohort study design, and delayed survey administration could result in recall bias or
decreased participation. The standardized list of engineering attributes worked well for describing and justifying
the selection of the study areas. Alternative methods for selecting study areas would have been to select areas
with a pre‐determined pressure measurement threshold, such as < 20 psi, or to use hydraulic modeling to select
areas with pressure loss. Field pressure measurements were not used to determine LPE eligibility because of
challenges in consistently obtaining the pressure readings, especially during emergency repairs. Additionally, not
all utilities have hydraulic modeling capacity and given the study time constraints, it was preferable to use the
field assessment from the operators to define the study areas, rather than a hydraulic model.
Matching LPE and Non‐LPE Areas
For each event, it was feasible and straightforward to match the study areas on the infrastructure characteristics
of pipe size and material, main housing type, and water source. The census area match was not always possible
because some events were larger than anticipated and did not fit neatly within the census boundaries; however,
the CDC team was able to use census data to find comparable populations in different census tracts. In the
national study, the infrastructure matching characteristics will be prioritized over the census area match
because differences in respondent demographics can be adjusted for in the analysis using the self‐reported
information from the survey.
Describing and Justifying the Choice of Study Areas
The LPE form was a useful tool for distilling the many event details to capture common information that could
be aggregated across all LPEs.
The engineering attributes were designed as guidelines to help the utility teams select adjacent, less obvious
LPE‐areas (exposed areas) that should be avoided for the control non‐LPE (unexposed) area selection. The
standard attributes list was helpful for distilling the many event details down to the factors that directly related
to whether an area was or was not affected by the LPE. In addition to helping with the selection of the study
areas, since each LPE took place under varying circumstances, it was helpful to use the attribute assignments to
facilitate comparisons across the different LPEs during the pilot data analysis phase.
Water Samples
Several statistically significant differences by LPE status were observed (see Table 9), although the crude mean
values sometimes seemed similar and standard deviations were large. Variance components analysis revealed
that all water sample parameter values were highly correlated within each event, whereas the differences
between the events were comparatively large (data not shown). It is statistically possible, especially with non‐
independent samples, for the ranges of values in comparison groups to overlap substantially but still be
statistically significantly different (Van Belle and Steven, 1998). When matching is present, as in this study
27
design, it is possible to have an even greater overlap between the distributions of values and still maintain
statistical significance because values are compared within the matched sets rather than across all values (in this
case, within LPEs). The crude means were presented in this report to summarize the general levels of these
parameters in the water samples, but they do not adequately take into account the distributions of the
parameters or the study design.
The pilot data were also used to inform changes to the analytical suite of indicators and water parameters. HPC
concentrations were higher in LPE‐area water samples, but the variability for this parameter was substantial. It is
possible that adding testing for adenosine triphosphate (ATP), a molecule found in living cells, to the analytical
suite could provide similar differential data because the test detects live microorganisms, but with less
variability because it is a non‐culture parameter.
Many statistical correlations among the field parameters were observed, although the small size of the dataset
limits the conclusions that can be drawn and the reasons for many of these correlations are not clear. In the
multi‐site study, additional analyses of the water quality and microbial indicators data will be explored, including
comparing samples collected upstream and downstream of the repair site and correlating water quality
parameters with household survey data.
Household Survey
In the pilot, the survey response rates were similar among LPE and non‐LPE areas and across events, but the
overall response rate of 37% was lower than expected. Non‐differential response in LPE and non‐LPE areas and
across different events suggests the survey procedures encouraged equitable participation in the LPE and
control areas, which likely indicated that there was limited selection bias. Additionally, if response rates had
been substantially higher in exposed LPE areas than unexposed non‐LPE areas, it would have raised concern that
persons in LPE areas were more concerned about their water quality, which would increase the possibility of
information bias in the responses about water use and illness. The multiple mailing contact strategy encouraged
participation, evidenced by boosts in survey response following the mail prompts.
The customer service lists were a reliable source of contact information, with deliverable addresses available for
98% of contacted households. The pilot survey response rate of 37% was comparable to recent CDC surveys
about waterborne illnesses. A CDC survey that was administered after a community‐wide drinking water
outbreak used water utility billing information to contact households for a postal survey and attained a 33%
response rate (Ailes et al., 2013). A CDC and National Park Service survey about illness, including AGI, and injury
among backcountry travelers attained a 39% response rate among respondents that consented to participate
prior to the survey administration; respondents specified a preferred method of either e‐mail or postal contact,
and researchers implemented a multiple contact strategy (Rajasingham et al., 2013). Additionally, the pilot
survey response rate was similar to a series of population studies on gastrointestinal illness and drinking water
consumption patterns in Canada, in which researchers used a similar strategy of selecting random samples of
households from contact lists, first contacting households with an advance letter, and then utilizing a multiple
contact strategy, although these studies used a telephone mode of survey administration (44%) (Jones et al.,
2007; Thomas et al., 2006). The strategy of cold contacting random samples of water utility customers by mail
was resource efficient compared to conducting in‐person interviews; however, the trade‐off might have been a
lower response rate. Higher response rates are more likely to be achieved in in‐person interviews (59%) (Arnold
et al., 2013). Power calculations were developed with the assumption that a minimum response rate of 40% was
needed to answer the primary research question of whether the events were associated with AGI. The goal for
28
the pilot was to reach a 60% response rate to ensure generalizability of findings and to provide enough
statistical power to conduct stratified analyses by utility disinfectant type. System data demonstrated that the
web‐based survey took customers, on average, 13 minutes to complete, suggesting that the time burden for
participation was low.
The majority of survey respondents reported drinking and using tap water for potable purposes, and this was
similar in the LPE and non‐LPE areas. Most households in LPE areas did not report observing low pressure,
complete service loss, or a change in tap water during the three weeks following the LPE, indicating that
customers were usually unaware of their exposure status. Respondents reported noticing low pressure in LPE‐
areas about twice as often as in non‐LPE areas, although less than 1% of respondents in the non‐LPE areas
reported noticing a complete water service loss. A minority of households mistakenly thought they were under a
boil water notice, advisory, or order. This was uncommon, but occurred more in LPE‐areas, so it is possible that
customers misinterpreted a work notice communication, such as a door hanger, from the utility.
Lessons Learned
Through the process of piloting the procedures for selecting the study LPE and non‐LPE areas, the utility and CDC
team were able to communicate and work together effectively; this will be especially important during the
multi‐site study to ensure consistency across the different utility sites. The pilot also highlighted several
communication challenges. The utility operators were not always able to notify the utility project manager of
eligible events early enough into the repair for the study team to observe the full event/repair during the site
visit. Similarly, the utility sometimes gave delayed notification to the CDC team. This was especially difficult to
prevent in emergency response situations. The importance of timely event notification will be emphasized
during utility trainings. The area selection process was much easier when the utility team was able to make a
site visit and observe the entire event/repair and when the CDC team was able to assist with the study area
selection while the utility team was still in the field.
The engineering attributes list and LPE form will be revised to streamline data collection and reduce the burden
on the utility staff. The pressure data on the LPE form were frequently missing. It was not possible to measure
pressure readings during the events without violating the observational study design (i.e., interfering with or
delaying the repairs to measure the pressure would have caused study‐related changes to service). The pressure
readings add value to the field data, but the scope of how the pressure measurements can be used in this study
is limited because of how the data are collected. The three field readings measured during the event can provide
snapshot measurements but cannot capture the variability within the system. This, coupled with frequently
missing data, limits the ability to use the pressure readings to measure LPE‐exposure or to relate the data to the
human health outcomes. It might be helpful to use the pressure measurements in real‐time to help identify the
extent of the LPE‐areas or to consider the duration or extent of low pressure in the analyses, but the study
design does not lend itself to that, and that information is not necessarily needed to answer the study question.
For study purposes, other information, such as verifying a loss of water service at home hose bibs or checking
for customer complaints of lost service, can be used to corroborate the selection of the LPE‐areas using the
engineering attributes. If a utility has automated pressure sensors already in use, it could be helpful to gather
this information in conjunction with the study data to provide additional context about the LPE.
Most of the data collection for the pilot took place during the winter, which allowed the study procedures to be
tested under worst‐case conditions; the utility was in its busy season, responding to multiple main breaks and
needs for repairs, there were two extreme weather events that shut down the Atlanta CDC office, and survey
29
administration fell within the winter holiday season, when residents were potentially not home to receive the
survey. These seasonal issues will be considered when planning for the scale‐up to multiple utility sites. Possible
solutions include staggered utility enrollment and a study break during the winter holiday season.
Additionally, because the CDC Environmental Microbiology Laboratory must sometimes respond to public health
emergencies that temporarily limit the lab’s capacity for research projects, it will not be feasible to collect
ultrafiltration samples for each of the 65 LPEs in the multi‐site study. Ultrafiltration samples will be collected
from about 2/3 of all events to allow for efficient progress on the epidemiologic study at times when the CDC lab
is unable to receive samples. Utilities will work with CDC to verify event eligibility before proceeding with a study
response; at that time, CDC will let them know whether to collect the ultrafiltration samples.
Survey response rates were lower than anticipated. To improve response rates, the study team will make minor
modifications to the survey procedures and will increase efforts to promote the study in the participating
communities. Nearly half of the page views for the study website (46%) occurred in the 30 days following the
press release, before any survey materials were mailed out, suggesting that additional ongoing publicity in the
study communities has potential to motivate participants. For the multi‐site study, the study team will
implement additional community outreach throughout the study period to improve community acceptance of
the study and to boost response rates.
The majority of respondents chose to return the survey by postal mail, using the provided return envelope. The
data quality of the web surveys was higher than the paper surveys because data verification rules and question
skip patterns were built into the survey interface. Since the web survey instructions and access information were
printed on the survey materials, rather than sent electronically to customers, it might have been inconvenient
for respondents to access a computer, type the link to the website, and log‐in to take the survey. To encourage
web survey participation and possibly increase overall survey response rates, the study team will send the
survey link electronically to customers that have email addresses on file (approximately 10% of pilot utility
customers).
30
EVALUATION OF PILOT STUDY AND PLANNING FOR NATIONAL STUDY
Evaluation of Pilot Study
Six outcomes were used to evaluate the success of the pilot study.
1) CDC team and utility select appropriate LPE and non‐LPE areas
The operator and study team’s selection of the matched LPE and non‐LPE areas was completed accurately and
efficiently using knowledge of the water system and hydraulic principles. Hydraulic models were developed for
three events, and results corroborated the selection of the study areas based on the field assessment and
knowledge of the water system.
2) The standard LPE form describing the event is filled out completely at the time of repair
The LPE form was nearly complete for all six events, with the exception of a few items that the team was unable
to measure or observe, such as pressure readings and interior conditions of pipes.
3) The water sample collection team follows standard operating procedures
Grab water sample collection was complete for all events. UF sample collection was incomplete for two events,
due to a winter storm and CDC laboratory staffing capacity. All standard operating procedures were followed
for water sample collection and shipment.
4) The laboratory team follows testing, standard operating procedures, and maintains chain of custody logs
The water sample chain of custody information and field water parameters were complete for all events. Utility
and CDC laboratories followed laboratory procedures and maintained appropriate records.
5) The CDC team follows survey mailing procedures and tracks household participation
A secure web survey and database were developed specifically for this project to manage utility customer data,
collect survey data, and track participation. CDC contacted over 600 households following a total of six events.
The study team maintained the privacy of the utility customer data, and tracked participation.
6) The survey procedures encourage participation and equitable study participation in LPE area and non‐LPE
area groups.
Survey response rates were in line with studies that used similar survey administration methods. Response rates
were similar in LPE (exposed) and non‐LPE (unexposed) areas, suggesting there was limited bias. The survey item
response rate was high.
Conclusions
The pilot was effective for testing the study design and procedures but the findings are not necessarily
generalizable to all utilities that might participate in the multi‐site study. The study was piloted at one utility site,
and utility practices and capacities to accommodate research projects vary. The pilot utility is located in the
southeast region of the U.S. and serves approximately 427,000 people (EPA, 2014). Utilities of different sizes and
in different regions might encounter additional or different challenges in applying the study protocols, so the
31
utility roles and study protocols will need to be adapted to each site. An additional difference is that the pilot
field data collection took place during December to March, and the multi‐site study will be implemented year‐
round. The frequency of main breaks and repairs is expected to vary seasonally.
Additionally, the pilot was much smaller in scope than the multi‐site study. The multi‐site study will collect data
from 65 events, whereas data were collected from six events for the pilot. This limits the inferences that can be
made about the data. For example, the pilot utility was a chloraminated system, and the study areas were most
often served by blended source water (five events) and groundwater (one event). Thus, the pilot laboratory
findings might not be generalizable to systems or events with different water characteristics.
Overall, the pilot study demonstrated that the study design and procedures will allow the study team to collect
the data needed to meet the study goal and aims. Field, laboratory, and epidemiologic components of the study
were well‐coordinated and data collection proceeded smoothly. Following minor modifications to the study
materials and protocols, the study team will be ready to implement the full study to determine whether LPEs are
associated with illness.
Next Steps and Information for Prospective Utilities
The multi‐site study is expected to be launched in 2014. To be eligible to participate, the utility must use a
secondary disinfectant, have a history of at least one LPE per month (main breaks or planned repairs), on
average, and be able to partner with CDC for a 12‐18 month period to complete data collection for
approximately 13 low pressure events.
Additionally, the utility should be able to use knowledge of the water system and hydraulics to identify LPE and
non‐LPE areas following the event and communicate the area selection to CDC using basic hydraulic maps (i.e.,
pipe diameters, normal flow direction, changes to flow during the event, areas valved‐off during the repair). The
maps can be hand‐drawn or computer‐generated and should help explain the reasoning behind the choice of
the study areas. The utility must be able to generate contact information lists for residential customers within
the study areas and securely transfer the lists to CDC. The CDC team will help set up the method for securely
transferring the customer contact information to CDC.
The project’s success hinges on forming strong public health and water utility sector partnerships. Participation
from utilities across the country will help researchers gain an understanding of the health impacts of routine and
unplanned distribution system low pressure events, and this understanding can be used to help focus resources
toward effectively maintaining the safety and durability of U.S. water systems. Additionally, participation can
help utilities gain a better understanding of their customers’ water use and experiences; CDC can provide each
utility with de‐identified summary survey data about household water use. The utility and customer results will
be combined with others, and the participating utilities will not be identified in any publications.
For more information, please contact the project team:
Julia Gargano (CDC Epidemiologist)
[email protected]
32
APPENDIX: PILOT STUDY LOW PRESSURE EVENTS AND DECISION‐MAKING
LPE 1
Low pressure was caused by 1) a blowout in an 8” PVC main, and 2) the repair process, which involved partially
or completely valving off an area to isolate the break and replace 10’ of pipe. Operators attempted to conduct a
clamp repair; the pipe broke again after it was clamped, which complicated the repair. The cause of the main
break was likely deterioration and aging. The break caused water to flood the street; water undermined the
foundation of a home and a tree. The utility study team assessed that there was high potential for
contamination. Muddy trench water, primarily water from the main, entered the pipe before and during the
repair. An estimated 154 homes were exposed to low pressure either during the break or when the operators
valved off the surrounding areas to isolate the break site. The total time of low pressure was estimated to be
seven hours. Of these 154 homes, an estimated 31 lost complete water service when the operator completely
valved off a smaller area to isolate the repair site and cut and replace the section of pipe; these homes were
without service for approximately 30 minutes. The area was flushed following the repair.
The selection of the exposed and unexposed study areas was challenging because the break took place in a
looped portion of the system. Since the loop was fed by large mains (i.e., 48”, 12”, 12”) and the break and repair
areas were nearly in a dead end off of the loop, it was reasonable to assume that the normal flow direction in
the loop would not change as a result of the break, and that there would be limited impact to areas outside of
the valved‐off area once the break area was isolated. The following exposed area attributes were used to
describe the choice of the exposed area.
Table A 1. Exposed area selection for LPE 1
Selected exposed area attributes
Reasoning
2. Smaller diameter mains nearby or in
direct hydraulic connection to LPE.
True, there were smaller diameter mains nearby or in direct
hydraulic connection to the main that had the break.
4. Near lower flow areas such as dead
ends, pressure zone boundaries.
True, the area near the break was nearly in a dead end.
5. Away from PRVs, pumps, or storage
facilities that float on the system, which
would allow water to be released into the
system to compensate for lower pressure
caused by LPE
True
6. Other, please describe (for example,
flow direction in surrounding areas changed
as a result of the repair).
The selected exposed area was partially or completely valved
off to complete the repair.
The selection of the unexposed area was challenging because the area was selected within the same census
tract, for demographic matching purposes, and with the same pipe size and material, water source, and housing
type as the exposed area. These matching criteria led the study team to select an area in close geographic
proximity to the exposed area, which meant that the unexposed area was also part of the looped system.
Although it was expected that there would be limited impact to areas outside of the repair area once the break
33
area was valved off, it was important to be sure that the selected unexposed area was not affected before the
break was isolated. The unexposed area attributes were used to describe the choice of the unexposed area.
Table A 2. Unexposed area selection for LPE 1
Selected unexposed area attributes
Reasoning
2. No recent main breaks or LPEs in the
vicinity.
True, also no customer complaints.
4. If in same pressure zone, area served by True, fed from the N or S, depending on demand, by
12” mains in the loop.
larger diameter mains with routinely good
steady state pressures.
6. Other, please describe
Exposed area nearly in a dead end, so impact to
surrounding areas could be limited.
Since the validity of the choices of study areas depended on several assumptions, hydraulic models, developed
by the utility’s engineering department, for three scenarios (i.e., normal ‐ pre‐break, during break ‐ before area
valved off, and during repair‐ after area valved off) were used to evaluate the choices made based on the field
assessment and engineering principals. Overall, the models supported choices of exposed and unexposed areas,
of normal flow direction, and the assumption that flow direction in the loop did not change during the event.
The model showed small pressure loss in the exposed area (~ 5 psi) and at the boundary of the unexposed area
(~1 psi) during the break, before the area was valved off. The model showed limited impact to areas outside of
valved‐off area once the break area was valved off.
LPE 2
Low pressure was caused by an area being valved‐off and restored multiple times over a two‐day period as part
of a larger planned maintenance event. The operators installed several new valves in the affected area, so they
could isolate the area for a procedure to relocate a main to accommodate new storm drain installation. The new
valves allowed the operator to complete the main relocation within an isolated area that did not contain any
service connections to homes. On the day before the event, the operator flushed the area and exercised the
valves to prepare for the procedure; the area was valved off for several hours. On the day of the event, the
operator tried to isolate the area, but a valve broke and needed replacement. A larger area was valved off, the
operator replaced the broken valve, and restored service to some of area. The operator installed a new valve to
isolate the street from the planned work site for the main relocation. Water flowed out of the main during
installation. The main was flushed and service was restored. The main relocation was completed the next day,
with no service connections affected. The utility study team assessed that there was low potential for
contamination. The main was flushed extensively on the days before and after the repair. There was a small
amount of flow (i.e., positive pressure) during the installation of the new valve. Pumps removed water from the
pit, so that pipe ends were never submerged in water. Pipes and valves were swabbed with chlorine before
installation of the new valve. An estimated 85 homes were exposed to low pressure when operators valved off
the area over the two‐day period. The approximate time of water loss was three hours on the day of the event.
The selection of the exposed area was relatively straightforward for this event, since it only included areas that
had been valved off by the utility operator. The following exposed area attribute was selected:
34
Table A 3. Exposed area selection for LPE 2
Selected exposed area attributes
Reasoning
6. Other, please describe (for example,
flow direction in surrounding areas changed
as a result of the repair).
The exposed area was valved off and restored multiple times
over two‐day period.
The exposed area did not meet study eligibility criteria of 80% occupied housing within the census tract; the area
was located within a strip of beach area in a tract with 52% occupied housing. However, utility water usage data
suggested ~70‐90% of homes had normal residential usage in the last 60 days, so the event was included. The
unexposed area was similar to the exposed area; it was also located on an adjacent beach strip within a census
tract that had 69% occupied housing. Both of these strips were fed predominantly from the N to S by 16” mains.
Additionally, the exposed area was fed by a subaqueous line from the unexposed area. These exposed and
unexposed areas were not in the same census tracts, but the census age groups were similar (i.e., percent of
census tract less than 18 years old and percent of census tract 65 years and older). The areas had the same
water source, and similar pipe sizes and materials; the unexposed area had slightly larger pipes. The unexposed
area attributes were used to describe the choice of the unexposed area.
Table A 4. Unexposed area selection for LPE 2
Selected unexposed area attributes
Reasoning
4. If in same pressure zone, area served by
True, served by 16” main, flow from N to S.
larger diameter mains with routinely good steady
state pressures.
LPE 3
Low pressure was caused by the repair process to stop a leak at the bell and spigot joint of an older (i.e., ~50
years old) 6” PVC main. A contractor installing sidewalks discovered a leak when digging up a driveway and
reported it to the water utility; during the repair, residents informed operators that the ground in the area had
been soggy for years, suggesting that the leak might have been present for several years. The cause of the leak
was likely due to the gaskets on the joint deteriorating. Operators valved off the area and cut and replaced a 3’
section of pipe. The utility study team assessed that there was low potential for contamination. The water in the
repair trench was below the pipe level during the pipe installation. The new pipe dipped into the pit water while
the bolts were being tightened, but the utility personnel did not feel this would create risk for contamination
since the bolts were already sealed. An estimated 23 homes experienced low pressure due to a loss of service
when the operators valved off the area to complete the repair. The homes were without service for a little less
than two hours. The area was flushed following the repair.
The selection of the exposed study area was relatively straightforward. The neighborhood selected as the
exposed area was fed by a single 12” main; smaller mains off of this line fed homes in a circular flow pattern.
One of these smaller mains was the main with the leak. Because of these hydraulic circumstances, it was
assumed that all of the homes in the neighborhood were affected by the event and no additional areas outside
of the neighborhood would be impacted. The following exposed area attributes were used to describe the
choice of the exposed area.
35
Table A 5. Exposed area selection for LPE 3
Selected exposed area attributes
Reasoning
1. Areas with known lower steady state pressures.
True
2. Smaller diameter mains nearby or in direct hydraulic
connection to LPE.
True, smaller diameter mains in neighborhood
with circular flow pattern (exposed area)
3. Higher elevation than main break location (assuming
there is no nearby storage to compensate for the
elevation).
True, there is a downhill slant from N to S in this
area.
5. Away from PRVs, pumps, or storage facilities that
float on the system, which would allow water to be
released into the system to compensate for lower
pressure caused by LPE
True
6. Other, please describe (for example, flow direction in The exposed area was completely valved off to
surrounding areas changed as a result of the repair).
complete repair; utility team member verified
homes had no water service at hose bibs
The unexposed area selection was also straightforward. The unexposed area was upstream of the exposed area,
fed off of the same 12” main that fed the exposed area, and was also a smaller neighborhood fed by small mains
with a circular flow pattern. Both areas had similar housing type, pipe sizes and materials, and the same water
source. The areas were in different census tracts, but the census age groups were similar (i.e., percent of census
tract less than 18 years old and percent of census tract 65 years and older). Both census areas had over 80%
occupied housing. The unexposed area attributes were used to describe the choice of the unexposed area.
Table A 6. Unexposed area selection for LPE 3
Selected unexposed area attributes
Reasoning
1. Nearby but different pressure zone.
The unexposed area is uphill from the exposed area
and is closer to the pump station, so there might be a
difference in pressures.
3. If in same pressure zone, served by pump or
storage facility that floats on system.
True, served by pump station
4. If in same pressure zone, area served by larger
diameter mains with routinely good steady state
pressures.
True, served by 12” main
6. Other, please describe
The unexposed area is upstream of the exposed area;
no reversal of flow or additional affected areas outside
of exposed neighborhood, so confident area is
upstream.
36
LPE 4
Low pressure was caused by 1) a pump station that directly fed the study areas going into “fill mode” causing a
slight, ~ 7 psi, drop in pressure at the pump station, 2) a blowout in a 6” main, and 3) the repair process, which
involved completely valving off a small area to isolate the break and replace a section of pipe. The main break
occurred after a contractor who was burying phone lines accidentally sawed through a 6” main. During the
break, water from the north and south and from small mains in the exposed neighborhood traveled to the break
site and flooded the road. The utility study team assessed that there was moderate potential for contamination.
The cut pipe was submerged in drinking water from the pipe before the repair crews arrived and before the
break was valved off, for about 3.5 hours. A pump was used to remove water from the pit before the repair
started. An estimated 166 homes were exposed to low pressure for about 3.5 hours during the break, before the
area was valved off. Of these 166 homes, three lost complete water service when the operator valved off a
smaller area to isolate the repair site; these homes were without service for an additional 5.5 hours. During this
time, the remaining homes were back‐fed from surrounding areas, so most residents had water service. The
area was flushed following the repair.
The selection of the exposed and unexposed areas was complicated by the presence and activity of the nearby
pump station, since this pump station directly fed both the exposed and unexposed areas. The study team was
concerned that the unexposed area could have potentially experienced a loss of pressure or flow change when
the pump station experienced a dip in pressure, either before or during the break. Utility operators assessed
that the dips in pressure at the pump station would not have affected the unexposed area. Through this inquiry,
the study team learned that the pump stations at this utility do not automatically adjust for drops in pressure in
the system, and that it is important to follow up with the operator before using this unexposed area attribute. A
second concern was the uncertainty of how the pump station being in “fill mode” affected the surrounding
areas during the break. The utility operator consensus was that since the break was in the main directly fed by
the pump station, “fill mode” allowed the pump station to contribute water to the break, but could not
compensate for pressure loss in the exposed subdivision. The break was in the main feeding the subdivision, and
the small, dead end mains in the subdivision allowed water to rush from the surrounding area to the break site;
this entire subdivision was selected as the exposed area. The following exposed area attributes were used to
describe the choice of the exposed area.
37
Table A 7. Exposed area selection for LPE 4
Selected exposed area attributes
Reasoning
1. Areas with known lower steady state
pressures.
True, area served by this pump station is known low
pressure area
2. Smaller diameter mains nearby or in direct True, smaller mains in subdivision (exposed areas) served by
hydraulic connection to LPE.
main with break.
4. Near lower flow areas such as dead ends,
pressure zone boundaries.
True, smaller mains in subdivision (exposed areas), were
dead ends, which caused water to rush out of area from
break site.
6a. Other, please describe (for example, flow
direction in surrounding areas changed as a
result of the repair).
Pump station went into fill mode just before break, which
caused ~7 psi dip in pressure at pump station before the
break
6b. Other, please describe (for example, flow
direction in surrounding areas changed as a
result of the repair).
Customer complaint of very low water pressure about two
hours after the break, before repair was completed
The unexposed area had the same water source from the pump station and similar housing type and pipe sizes
and materials. The exposed and unexposed areas were in different census tracts, but the census age groups
were similar (i.e., percent of census tract less than 18 years old and percent of census tract 65 years and older).
Both areas had over 80% occupied housing. The following unexposed area attributes were used to describe the
choice of the unexposed area.
Table A 8. Unexposed area selection for LPE 4
Selected unexposed area attributes
Reasoning
2. No recent main breaks or LPEs in the
vicinity.
True, also no customer complaints in area
3. If in same pressure zone, served by pump
or storage facility that floats on system.
True, operator closed fill valve after break,
allowing pressure at pump station to increase (no
automatic response to pressure drop).
4. If in same pressure zone, area served by
larger diameter mains with routinely good
steady state pressures.
True, served by 24” main from pump station and
16” main from north.
The study team suggested that exposed area attributes 1 and 6a could have also been applied to the unexposed
area. The utility operators felt that the unexposed area would not be affected. A hydraulic model, developed by
the utility’s engineering department, was used to estimate pressures in the exposed and unexposed study areas
during the break scenario. The hydraulic model estimated very low pressures in the exposed area subdivision
during the break (4‐11 psi). The hydraulic modeling results also showed that the unexposed area might have
been affected by a ~3 psi drop in pressure, but that the overall pressure estimates remained high (61‐70 psi).
Because the differential exposure in the two areas was so large and the exposure of a 3 psi drop could be of
similar magnitude of diurnal pressure variation, the hydraulic model supported the utility operators’ choice of
exposed and unexposed areas.
38
LPE 5
Low pressure was caused by an operator valving off a small area to install a valve to put an abandoned line out
of service. The abandoned line was still connected to the line with home service connections and was leaking
into the street. The utility team assessed that there was very low potential for contamination because the valve
installation was on the abandoned main, which did not have service connections to homes. However, the
abandoned main was in the same pit as the in‐service line, and there was a cross‐connection somewhere
between the lines, which was the source of the leak. Following the repair, a customer complained of a black
substance in the drinking water; the utility team believed it was biofilm but could not flush the area because it
was a dead end and the blow off valve for flushing was corroded shut. An estimated 16 homes were exposed to
low pressure for about 45 minutes when operators valved off the area.
The selection of the exposed area was relatively straightforward, since it was limited to the area that was valved
off by the utility operator. The exposed area had a uni‐directional flow pattern and there were no alternative
water feeds to the area, so it was assumed that no additional areas outside of valved off area should have been
affected by the event. The following attributes applied, although #2, 4, 5, and 6a might not have been directly
related to why the area was selected as exposed to low pressure during the event.
Table A 9. Exposed area selection for LPE 5
Selected exposed area attributes
Reasoning
2. Smaller diameter mains nearby or in
direct hydraulic connection to LPE.
True, 4” line with service connections to homes connected to
older main
4. Near lower flow areas such as dead
ends, pressure zone boundaries.
True, dead end street with lower demand; low flow area
5. Away from PRVs, pumps, or storage
facilities that float on the system, which
would allow water to be released into the
system to compensate for lower pressure
caused by LPE.
True
6a. Other, please describe (for example,
flow direction in surrounding areas
changed as a result of the repair).
Blow off at end of street inoperable; unable to flush to end of
street. Not able to flush for several years
6b. Other, please describe (for example,
flow direction in surrounding areas
changed as a result of the repair).
Exposed area valved off during repair; homes lost water service.
It was straightforward to match the exposed and unexposed areas on the infrastructure characteristics. The
exposed area was on a peninsula with water‐front properties. The unexposed area was also made up of several
similar peninsulas. The flow patterns in both areas were similar; both were dead‐ends, fed by a 24” main that
reduced to 6” mains that served the areas uni‐directionally. Both areas had similar housing types, pipe sizes and
materials, and the same water source from a nearby pump station. The areas were not in the same census tract,
and the census age groups did not match exactly. A greater proportion of the exposed tract population was 65
years and older, although the areas had similar proportions of the population under 18 years old. The exposed
area had 76% occupied housing, and the unexposed area had 80% occupied housing.
39
The unexposed area was upstream of the exposed area and closer to the pump station. The 24” main from the
pump station serviced both areas but served the unexposed area first. The following unexposed area attributes
were used to describe the choice of the unexposed area.
Table A 10. Unexposed area selection for LPE 5
Unexposed area attributes
Reasoning
3. If in same pressure zone, served by
pump or storage facility that floats on
system.
True, close to pump station
4. If in same pressure zone, area served True, close to 24” main that services both areas
by larger diameter mains with routinely
good steady state pressures.
6. Other, please describe
Upstream from the exposed area and closer to pump
station; no alternative feed to exposed area, so
confident unexposed area is upstream.
In addition to the study matching characteristics, several other factors were observed that might have impacted
the quality of the exposed/unexposed area match for this event. The exposed area had not been flushed for
several years because the blow‐off valve was corroded, and there was a cross‐connection between the
abandoned main line and the in‐service line that had been present for several years. It was not possible to know
if the unexposed area had similar issues (i.e., the cross connection and inability to flush for several years).
LPE 6
Low pressure was caused by an area being valved‐off and restored twice over a two day period to relocate a
main to accommodate future installation of traffic light posts. For both event days, the utility study team
assessed that there was low potential for contamination. On the first day, a small amount of dirt entered the
pipe when it was cut, but the pipe was well flushed following the repair. On the second day, the utility study
team did not go to the repair site, but a follow up inquiry with the utility inspector indicated that the repair went
smoothly. An estimated 130 homes were exposed to low pressure when the operators valved off the area on
the first day; a subset of these homes were also exposed on the second day. The two procedures were
performed on the same main, and the approximate time of water loss was four hours on each day. The main
was flushed extensively to drain air from pipe on the first day, and the main was also flushed following the
procedure on the second day.
The selected exposed area was the area that the operators valved off to complete the first procedure to relocate
the main. The exposed area was fed uni‐directionally, and there were no alternate feeds to the area, so it was
assumed that no additional areas outside of the repair area should have been affected by the event. The
following exposed area attribute was selected:
40
Table A 11. Exposed area selection for LPE 6
Exposed area attributes
Reasoning
6. Other, please describe (for
example, flow direction in
surrounding areas changed as a
result of the repair).
Exposed area valved off during repair; homes lost water service.
There was no alternate feed to the area, so there was no
additional exposed area.
The unexposed area was nearby but upstream of the valves that were closed to isolate the exposed area. Both
exposed and unexposed areas were fed uni‐directionally from the same main; 8” and 6” mains fed the areas.
The flow patterns within the areas were slightly different. The exposed area was looped; the unexposed area
had a looped section but also had several dead ends. The pipe sizes and materials were similar in both areas,
and the water source was the same. The main housing types in the areas were similar (i.e., single family homes),
but not an exact match. Most of the exposed area and all of the unexposed area had concrete block homes. Part
of the exposed area had pre‐fabricated homes. Both areas were in the same census block group, so it was
reasonable to assume that the demographic characteristics of the populations in the exposed and unexposed
areas were similar. The census housing occupancy was 87%. The following unexposed area attributes were used
to describe the choice of the unexposed area.
Table A 12. Unexposed area selection for LPE 6
Unexposed area attributes
Reasoning
3. If in same pressure zone, served by
pump or storage facility that floats on
system.
True, served by water source before exposed area.
4. If in same pressure zone, area served by
larger diameter mains with routinely good
steady state pressures.
True, served by 48" to 12" to 8" mains from water
source.
6. Other, please describe
Unexposed area is upstream of the valves that were
shut off to isolate the exposed area. Both areas fed
uni‐directionally, so confident unexposed area was
upstream.
41
REFERENCES
Kirmeyer, G.; Richards W.; Smith C.D. An assessment of water distribution systems and associated research
needs. AWWARF, Denver, CO. 1994.
Nygard, K.; Wahl, E.; Krogh, T.; Tveit, O.A.; Bohleng, E.; Tverdal, A.; Aavitsland, P. Breaks and maintenance work
in the water distribution systems and gastrointestinal illness: a cohort study. International Journal of
Epidemiology. 2007 Aug;36(4):873‐880.
Dillman, D.A. Mail and Internet Surveys: The Tailored Design Method. 2nd ed. New Jersey: John Wiley & Sons,
Inc; 2007.
CDC. Ten great public health achievements – United States, 1990 – 1999 Morbidity and Mortality Weekly
Reports. 1999a;48: 241‐243.
CDC. Achievements in public health, 1990 – 1999: control of infectious diseases. Morbidity and Mortality Weekly
Reports 1999b; 48: 621‐629.
Cutler, D.; Miller, G. The role of public health improvements in health advances: the twentieth‐century United
States. Demography. 2005;42: 1‐22.
Borchardt, M.A.; Haas, N.L.; Hunt, R.J. Vulnerability of drinking‐water wells in La Crosse, Wisconsin, to enteric‐
virus contamination from surface water contributions. Applied and Environmental Microbiology. 2004
Oct;70(10):5937‐5946.
LeChevallier, M.W.; Gullick, R.W.; Karim, M.R.; Friedman, M.; Funk, J.E. The potential for health risks from
intrusion of contaminants into the distribution system from pressure transients. J Water Health. 2003
Mar;1(1):3‐14.
Swerdlow, D.L.; Woodruff, B.A.; Brady, R.C.; Griffin, P.M.; Tippen, S.; Donnell, H.D., Jr.; Geldreich, E.; Payne, B.J.;
Meyer, A., Jr.; Wells, J.G.; et al. A waterborne outbreak in Missouri of Escherichia coli O157:H7 associated with
bloody diarrhea and death. Ann Intern Med. 1992 Nov 15;117(10):812‐819.
Lambertini, E.; Spencer, S.K.; Kieke, B.A., Jr.; Loge, F.J.; Borchardt, M.A. Virus contamination from operation and
maintenance events in small drinking water distribution systems. Journal of Water and Health. 2011
Dec;9(4):799‐812.
Craun, G.F.; Brunkard, J.M.; Yoder, J.S.; Roberts, V.A.; Carpenter, J.; Wade, T.; Calderon, R.L.; Roberts, J.M.;
Beach, M.J.; Roy, S.L. Causes of outbreaks associated with drinking water in the United States from 1971 to
2006. Clinical Microbiology Reviews. 2010 Jul;23(3):507‐528.
Colford, J.M.; Wade, T.J.; Sandhu, S.K.; Wright, C.C.; Lee, S.; Shaw, S.; Fox, K.; Burns, S.; Benker, A.; Brookhart,
M.A.; van der Laan, M.; Levy, D.A. A randomized, controlled trial of in‐home drinking water intervention to
reduce gastrointestinal illness. American Journal of Epidemiology. 2005 Mar 1;161(5):472‐482.
Payment, P.; Richardson, L.; Siemiatycki, J.; Dewar, R.; Edwardes, M.; Franco, E. A randomized trial to evaluate
the risk of gastrointestinal disease due to consumption of drinking water meeting current microbiological
standards. Am J Public Health. 1991 Jun;81(6):703‐708.
42
Hellard, M.E.; Sinclair, M.I.; Forbes, A.B.; Fairley, C.K. A randomized, blinded, controlled trial investigating the
gastrointestinal health effects of drinking water quality. Environ Health Perspect. 2001 Aug;109(8):773‐778.
Payment, P.; Siemiatycki, J.; Richardson, L.; Renaud, G.; Franco, E.; Prevost, M. A prospective epidemiological
study of gastrointestinal health effects due to the consumption of drinking water. International Journal of
Environmental Health Research. 1997 Mar;7(1):5‐31.
Colford, J.M., Jr.; Roy, S.; Beach, M.J.; Hightower, A.; Shaw, S.E.; Wade, T.J. A review of household drinking water
intervention trials and an approach to the estimation of endemic waterborne gastroenteritis in the United
States. J Water Health. 2006;4 Suppl 2:71‐88.
Ercumen, A.; Gruber, J.S.; Colford, J.M., Jr. Water Distribution System Deficiencies and Gastrointestinal Illness: A
Systematic Review and Meta‐Analysis. Environmental Health Perspectives. 2014 Mar 21.
DeSilva, M. EIS Conference Presentation. 2014.
American Public Health Association (2005). Standard Methods for the Examination of Water and Wastewater
(21st ed.): American Public Health Association (APHA), American Water Works Association (AWWA) and Water
Environment Federation (WEF).
Smith, C.M.; Hill, V.R. Dead‐end hollow‐fiber ultrafiltration for recovery of diverse microbes from water. Applied
and Environmental Microbiology. 2009 Aug;75(16):5284‐5289.
United States Geological Survey, Ohio Water Science Center (2007). Analysis of Clostridium perfringens in
Environmental Water Samples. Retrieved from
http://oh.water.usgs.gov/micro_methods_clostridium_perfringens.htm
United States Environmental Protection Agency (2001). Method 1602: Male‐specific (F+) and Somatic Coliphage
in Water by Single Agar Layer (SAL) Procedure. (EPA 821‐R‐01‐029). Washington, D.C.: Retrieved from
http://www.epa.gov/microbes/documents/1602ap01.pdf.
Hill, V.R.; Mull, B.; Jothikumar, N.; Ferdinand, K.; Vinje, J. Detection of GI and GII Noroviruses in Ground Water
Using Ultrafiltration and TaqMan Real‐time RT‐PCR. Food Environ. Virol. 2010; 2:218‐224.
Haugland, R.A.; Varma, M.; Sivaganesan, M.; Kelty, C.; Peed, L.; Shanks, O.C. Evaluation of genetic markers from
the 16S rRNA gene V2 region for use in quantitative detection of selected Bacteroidales species and human fecal
waste by qPCR. Systematic and Applied Microbiology. 2010 Oct;33(6):348‐357.
Hornung, R.W.; Reed, L.D. Estimation of average concentration in the presence of nondetectable values. Appl
Occup Environ Hyg. 1990; 5: 46‐51.
Van Belle, G.M., Steven P. STRUTS: Statistical Rules of Thumb. Seattle, WA: University of Washington; 1998.
Ailes, E.; Budge, P.; Shankar, M.; Collier, S.; Brinton, W.; Cronquist, A.; Chen, M.; Thornton, A.; Beach, M.J.;
Brunkard, J.M. Economic and health impacts associated with a Salmonella Typhimurium drinking water
outbreak‐Alamosa, CO, 2008. PloS One. 2013;8(3):e57439.
Rajasingham A.; Ayers T.; Derado G.; Stockman L.J.; Larsen G.; Higgins C.; Beach M.J.; Roy S.L., CDC Report
(2013). Yellowstone National Park Backcountry Study: Study of Illness and Injury among Backcountry Users.
43
Jones, A.Q.; Majowicz, S.E.; Edge, V.L.; Thomas, M.K.; MacDougall, L.; Fyfe, M.; Atashband, S.; Kovacs, S.J.
Drinking water consumption patterns in British Columbia: an investigation of associations with demographic
factors and acute gastrointestinal illness. The Science of the Total Environment. 2007 Dec 15;388(1‐3):54‐65.
Thomas, M.K.; Majowicz, S.E.; MacDougall, L.; Sockett, P.N.; Kovacs, S.J.; Fyfe, M.; Edge, V.L.; Dore, K.; Flint, J.A.;
Henson, S.; Jones, A.Q. Population distribution and burden of acute gastrointestinal illness in British Columbia,
Canada. BMC Public Health. 2006;6:307.
Arnold, B.F.; Schiff, K.C.; Griffith, J.F.; Gruber, J.S.; Yau, V.; Wright, C.C.; Wade, T.J.; Burns, S.; Hayes, J.M.;
McGee, C.; Gold, M.; Cao, Y.P.; Weisberg, S.B.; Colford, J.M. Swimmer Illness Associated with Marine Water
Exposure and Water Quality Indicators Impact of Widely Used Assumptions. Epidemiology. 2013 Nov;24(6):845‐
853.
EPA. SDWIS Envirofacts
http://oaspub.epa.gov/enviro/sdw_report_v2.first_table?pws_id=FL6521405&state=FL&source=Purch_surface_
water&population=426877&sys_num=0; [modified February 10, 2014May 22, 2014]
44
| File Type | application/pdf |
| File Title | Microsoft Word - 4390_PilotEvaluationReport |
| Author | alang |
| File Modified | 2016-02-16 |
| File Created | 2014-09-25 |