Grantee Research Project Results
Final Report: A Prospective Epidemiological Study of Gastrointestinal Health Effects Associated with Consumption of Conventionally Treated Groundwater
EPA Grant Number: R830376Title: A Prospective Epidemiological Study of Gastrointestinal Health Effects Associated with Consumption of Conventionally Treated Groundwater
Investigators: Moe, Christine L. , Moll, Deborah , Nilsson, Kenneth , Hooper, Stuart
Institution: Emory University , Centers for Disease Control and Prevention , University of South Florida
EPA Project Officer: Page, Angela
Project Period: October 1, 2002 through September 30, 2005 (Extended to September 30, 2008)
Project Amount: $1,820,900
RFA: Microbial Risk in Drinking Water (2001) RFA Text | Recipients Lists
Research Category: Water , Drinking Water , Human Health
Objective:
The overall goal of this study was to estimate the risks of endemic gastrointestinal illness (GI) associated with the consumption of conventionally treated groundwater (GW) in the U.S. and to determine the relative contributions of source water quality, treatment efficacy and distribution system vulnerability to endemic waterborne disease.
From October 2002- July 2005, we attempted to launch an epidemiological study to address this goal by recruiting and randomizing 1,000 households (HH) in an area served by a large GW distribution system into one of three study groups: 1) HH who drink tap water, 2) HH provided with bottled treated water collected at the water treatment plant effluent, or 3) HH provided treated water collected at the water treatment plant effluent and then bottled after ozonation and reverse osmosis. Families were asked to record information on their health symptoms in health diaries, and this information was to be collected via bi-weekly telephone calls for 12 months. In addition, we proposed extensive sampling of the source water, the treated water, and water throughout the distribution system to be tested for a panel of microbial indicator organisms (total coliforms, heterotrophic plate count bacteria, E. coli, C. perfringens, P. aeruginosa, H. hydrophila) and residual chlorine.
Because of the difficulty encountered in recruiting the target number of families in the study area who drink tap water (without any point-of-use treatment device), we modified our study design in December 2004 to recruit HH and provide them with either 1) bottled tap water collected from the distribution system, 2) bottled treated water collected at the treatment plant effluent, or 3) treated water collected at the treatment plant effluent and bottled after ozonation and RO. Because of the increased cost of this modified design and the need to restart the recruitment process, the EPA asked us to discontinue the epidemiology study and pursue alternative designs to achieve the study goal.
The study design was modified again in August 2005 and focused on identifying specific risk factors that impact distribution system water quality and characterizing the impact of specific short-term events.
Specific Aims:
1) Monitor the distribution system for low pressure events using high-resolution pressure loggers.
2) Examine changes in water quality from the treatment plant to the distribution system
3) Continuously monitor physical and chemical water quality in selected vulnerable points of the distribution system and measure water quality immediately after identified low pressure events or deviations in turbidity and chlorine residual by collecting and analyzing large volume water samples for physical and microbial indicators of water quality.
4) Monitor water quality immediately after short-term events such as main breaks or repairs, scheduled flushing events, and fire flow tests
Secondary Objectives:
-Compare results from grab samples collected to satisfy regulatory requirements to those collected by continuous flow methods from the same location at the same time (such as for total coliforms).
-Assess various methods of measuring microbiological water quality after a pressure loss event.
-Compare the water quality results obtained from the event samples to the vulnerability assessment model previously completed for this distribution system, so that the model can be evaluated and specific improvements to the vulnerability assessment methodology can be proposed.
Summary/Accomplishments (Outputs/Outcomes):
A. Epidemiologic Study
From April – October 2004, the study focused on recruitment of households for a randomized water intervention study. The original study design included three study groups: 1) a tap water group, 2) a bottled water group 1 (bottled tap water + reverse osmosis and ozone), and 3) a second bottled water group (bottled tap water). The study site was water utility with about 80,000 connections. Recruitment efforts included sending14,000 recruitment letters, making 45,000 telephone calls as well as a media campaign via newspaper articles, radio and TV announcements, distributing posters and brochures, and making presentations to local environmental groups, churches, homeowners associations, etc. Study staff contacted 34,591 households by telephone. However, only 0.8 % of contacts were interested and eligible to participate in the study. The main reason why households were ineligible to participate in the study was because of their drinking water habits. One of the study eligibility requirements was that household members must currently drink primarily tap water AND not use water softeners or point-of-use treatment devices because these may alter the microbiological quality of the water. Of the 14,200 households that we collected data on specific drinking water habits, 59% used a point-of-use filtration device, 31% consumed bottled water, 8% had a home water softener and 0.1% reported drinking boiled water. Only 2.2% of these 14, 200 households reported drinking tap water, and some of these households were not eligible to participate in the study because of health reasons.
Drinking water habits have been changing rapidly in the past decade. In some parts of the US, a low percentage of population may drink unfiltered tap water. In this study area, the sulfur taste of the groundwater seemed to be acceptable for long-time residents but not for new residents who had moved to the area from other places. In addition, we learned that aggressive telemarketing of bottled water and home water treatment devices in study area made people wary of phone calls asking about their drinking water habits – even though we identified ourselves as calling from Emory University. In February 2005, the state Attorney General had issued a warning about fraudulent offers for “free water testing” in order to sell bottled water or water treatment devices – which indicates the magnitude of the problem in the study area. Finally, we learned that there was very limited interest in participating in a research study with no incentives.
These findings have important implications for planning drinking water epidemiology studies in the United States. Perhaps the greatest challenge to conducting these types of studies will be recruiting conscientious, committed study subjects who understand the goals of the study and are willing to comply with the study protocol and ingest the type of water assigned to their study group (finished water from the treatment plant, distribution system tap water or water with additional treatment) and accurately record their health symptoms over a period of months. Studies of public drinking water supplies will become increasingly difficult to conduct because of the rapidly growing portion of the population that regularly consumes bottled water or already uses a household water treatment device. Future studies population-based studies may need to have a larger than expected population pool to recruit from, may need to offer incentives for participation and may need to use tactics like celebrity endorsements to generate enough community support to enable recruitment of the necessary sample size.
B. Distribution System Monitoring Study
Specific Aim 1. Monitor the distribution system for low pressure events using high-resolution pressure loggers.
Pressure monitoring was performed using Teloger HRD-31 high resolution loggers. These loggers were normally set to a sampling interval of 1/sec, and a recording interval of 1/2 min. A total of 2-6 loggers were placed throughout the distribution system and at the water treatment plant from February 2006 through September 2007. During this period, a total of 32 low pressure events were detected by one or more loggers. Almost half of these events (15) affected all the loggers in the system at the time. The other 17 events were only detected by one logger. The minimum recorded pressure during these events ranged from 0 – 33 psi. For eight events, the recorded pressure in the distribution system was zero. For 16 events, the recorded pressure was 11-20 psi and for 5 events, the recorded pressure was <10 psi. In six cases, the pressure drop appeared to be associated with a flushing event. For the other pressure loss events, the cause was either not identified or was associated with a power loss or construction. The pressure loss events occurred mainly in the spring and summer with15 (48%) of the events in April-May and 9 (29%) of the events in June-August.
Specific Aim 2. Examine changes in water quality from the treatment plant to the distribution system
Methods:
A programmable, automated monitoring and sampling (AMS) device was designed and constructed in order to collect large volume water samples in the distribution system immediately after possible intrusion events or sudden changes in water quality. This device performed continuous monitoring of total chlorine residual, turbidity, pH, conductivity, oxidation reduction potential and temperature. The device also had four 20-L carboys in series that were filled during a sampling event. Total sample volume was approximately 90 liters. The device had a cell phone modem to allow two-way communication at any time and monitor real-time data as well as initiate sample collection when necessary. The AMS was semi-portable and was installed at water treatment plant and five points in the water distribution system from May 2006 through September 2007.
Source and treated water quality were collected at the Water Treatment Plant between May 2006 and September 2007. Both grab (2-L) and large volume (90 L) samples were collected at each sampling event. A total of 29 source water samples and a total of 48 treated water samples were collected (both grab and large volume). Although the objective was to collect paired source and treated water samples, if collecting samples at both locations was not possible due to lab capacity limitations, then paired grab and large volume samples were collected at the treated water sampling point only. Treated water samples were typically collected every one to two weeks, with an average sampling rate of 10 days. Source water samples were collected on average every 17 days. This continuous sampling program was designed to capture any seasonal variability in water quality and to provide data to be compared to distribution system water quality.
Background distribution system (DS) water quality samples were defined as any sample (grab or large volume), collected at an AMS location or at a fire hydrant prior to a flushing event, that is not associated with any known distribution system event. Under this definition, 33 pairs of large volume and grab DS background samples were collected from September 2006 through September 2007. Of these, 12 were collected at fire hydrants on the day prior to a flushing event. The remaining 24 samples were collected at AMS sites, as part of routine background sample collection. These background DS samples allowed us to examine water quality in the distribution system during normal conditions and compare that to DS water quality during distribution system events – such as a flushing event or pressure loss event.
All water samples were immediately taken to the laboratory to for same-day analysis of E. coli, total coliforms, P. aeruginosa, C. perfringens, A. hydrophila, heterotrophic plate count (HPC), somatic coliphage, and MS2 coliphage. Large volume samples were concentrated by ultrafiltration.
Microbiological Water Quality
The most striking finding is that the overall microbiological quality of the source water was better than that of the treated water. Ten percent of the large volume source water samples were positive for total coliforms but all three of these samples had less than 0.2 cfu/100 ml. However, 29% of the large volume samples were positive for total coliforms, and five (36%) of these samples were greater than 0.7 cfu/100 ml. The highest concentration of total coliforms detected in the treated water was 3.2 cfu/100 ml. E. coli were only detected in one sample of source water and not in any samples of treated water. Heterotrophic plate count (HPC) bacteria were the most frequently detected indicator organism in large volume samples of both source and treated water. HPC concentrations were consistently higher in the samples from the distribution system compared to samples of treated water collected at the treatment plant. A. hydrophila and P. aeruginosa were detected frequently in both source water (45% and 32% of large volume were samples positive, respectively) and treated water (55% and 31% of large volume samples were positive, respectively). P. aeruginosa detections in the source water were highly seasonal with all the positive samples occurring between September through February. However, P. aeruginosa was detected in treated water throughout the year. Concentrations of A. hydrophila and P. aeruginosa were higher in the distribution system than in the source water or treated water samples collected at the treatment plant – indicating microbial growth in the distribution system. Somatic coliphage were detected once in source water and once in treated water, and MS2 coliphage were detected once in treated water. C. perfringens, which was measured as an indicator of intrusion, was not detected at any time in either grab or large volume samples of the source and treated water.
Specific Aim 3. Continuously monitor physical and chemical water quality in selected vulnerable points of the distribution system and measure water quality immediately after identified low pressure events or deviations in turbidity and chlorine residual by collecting and analyzing large volume water samples for physical and microbial indicators of water quality
Physical and Chemical Water Quality
The AMS device collected data on pressure, total chlorine residual, turbidity, redox, conductivity and pH at five locations in the water distribution system. The time at each site varied from 3 months (site 1) to 4 weeks (sites 2-4) to 2 weeks (site 5). Little temporal variation was observed in redox, conductivity, temperature and pH. However, there was temporal variation in pressure as well as differences between the five sites. Some sites had temporal variations in turbidity ranging from <1-5 NTU with a few spikes above 5 NTU. Total chlorine residual also varied between sites and over time – although some of this variation was due to problems with the AMS chlorine sensor.
The AMS was programmed to detect distribution system events as defined by an increase in turbidity (>10 NTU), a decrease in pressure (<20 psi), or a decrease in total chlorine residual. Samples were successfully collected after six distinct distribution system events. Three of these events were pressure drops, two were due to a sudden drop in total chlorine residual, and one was in response to a spike in turbidity. Careful examination of the sensor data during these events indicates that while one parameter, such as pressure, may be the first indication of an event, often other parameters were also affected such as turbidity, chlorine residual or conductivity.
Specific Aim 4. Monitor water quality immediately after short-term events such as main breaks or repairs, scheduled flushing events, and fire flow tests
Impact of Flushing Operations on Water Quality
The AMS device was also used to study the impact of flushing operations on water quality. Twelve distribution system flushing events at 11 different sites were examined during this study. Each of these sites was flushed on a regular basis by the water utility, but some of the sites were flushed frequently whereas others were infrequently flushed. Each flushing event consisted of three phases: pre-flushing, flushing and post-flushing. Turbidity and pressure were continuously monitored throughout each event. Additionally, water samples were collected during each phase of each event to characterize the water quality before, during and after each flushing. The 100 liter water samples were collected before, during and after each flushing event, concentrated by ultrafiltration, and analyzed for total coliforms, E. coli, P. aeruginosa, A. hydrophila, C. perfringens, and coliphage. The 2L grab samples were also collected during each phase and analyzed for the same indicator organisms to compare to the large-volume concentrated samples.
Total event durations ranged from 23 hours to 51 hours and flushing event durations ranged from 10.5 minutes to 29.5 minutes. The flushing flow rate ranged from less than 1900 liters per minute (Lpm) to greater than 7700 Lpm. The average recorded pressure ranged from 4 to 34 psi, and turbidity ranged from 0.21 to 7.05 NTU. On average the system turbidity increased after flushing. Across all events, the average turbidity before the flushing event was 4.1 NTU and the turbidity after flushing was 5.0 NTU. In addition, the average turbidity during flushing was higher than pre-flushing levels and slightly lower than the average post-flushing turbidity level (4.9 NTU).
Microbiological water quality declined during the flushing events but improved post-flushing. Of the 24 pre-flush samples collected (12 grab samples and 12 large volume samples), 4 (33%) of the grab samples and 4 (33%) of the large volume samples were positive for P. aeruginosa, and 3 (25%) of the grab samples and 4 (33%) of the large volume samples were positive for were positive for A. hydrophila. A total of 36 grab samples and 12 large volume samples were collected during flushing. Five (14%) of the grab samples were positive for P. aeruginosa and 10 (28%) were positive for A. hydrophila. However, 4(33%) of the 12 large volume flushing samples were positive for total coliforms, and 2 (17%) were positive for P. aeruginosa and 6 (50%) were positive for A. hydrophila. A total of 12 grab samples and 12 large volume samples were collected during the post-flushing phase. Only one post-flushing grab sample was positive (P. aeruginosa) and 4 post-flushing large volume samples were positive (1 for total coliforms and 3 for P. aeruginosa). No E. coli or coliphage were detected in any of the flushing event samples.
Impact of Sample Size on Water Quality Assessment
Throughout this study, almost all samples collected at all sites were collected in duplicate: an 80-100 L “large volume” sample and a 2-L sample. The purpose of the duplicate sample collection of different volumes was to establish how well the water quality measured in the grab samples (2-L) matched that measured in the large volume samples (80-100 L). The grab samples approximated the volume of typical grab samples collected for routine water quality monitoring – the 2-L volume was the minimum needed for adequate analysis of all the analytes. The large volume samples were processed by ultrafiltration, ultimately concentrating them to a volume of about 2 L.
The collection of large volume samples followed by concentration using ultrafiltration proved to be a sensitive and feasible method of examining microbiological water quality. The large volume samples consistently had higher rates of positive detections compared to the 2-liter grab samples for all the indicator organisms tested. The microbial concentrations measured in the large volume samples were also consistently higher than those measured in the grab samples.
Acknowledgements: We are grateful to Dr. Lillian Stark and the staff of the Florida Department of Health Laboratory for the microbiological analyses of the water samples collected during this study. These microbiological analyses were made possible by the generous support of the Water Research Foundation (Project number 03049). We thank Dr. Luke Mulford, Arnold Becken, Jim Jeffers, J. McCary, Mark Lehigh and Norman Vik of the Hillsborough County Water Resource Services; Dr. Jim Uber of the University of Cincinnati; Dr. Ricardo Izurieta of the University of South Florida; Renea Doughton, Allison King, Laura Kovalchick and Hannah Cluck of Emory University; and Misha Hasan, Alice Fulmer, Djanette Khiari, Chris Rayburn, and Peggy Falor of the Water Research Foundation for their collaboration and support. Finally, we are very grateful to Angela Page of the USEPA for all her valuable guidance and support during this study.
Journal Articles:
No journal articles submitted with this report: View all 12 publications for this projectSupplemental Keywords:
water, drinking water, distribution system, exposure, risk, health effects, human health, pathogens, epidemiology, modeling., RFA, Health, Scientific Discipline, Water, Environmental Chemistry, Health Risk Assessment, Epidemiology, Risk Assessments, Biochemistry, Drinking Water, groundwater disinfection, health effects, microbial contamination, bacteria, human health effects, waterborne disease, other - risk assessment, exposure, microbial effects, treatment, human exposure, microbial risk, water disinfection, groundwater contamination, water quality, dietary ingestion exposures, drinking water contaminants, drinking water treatment, human health, gastrointestinal health, groundwater, gastrointestinal health effects, exposure assessmentRelevant Websites:
Progress and Final Reports:
Original AbstractThe perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.
Project Research Results
- 2007
- 2006 Progress Report
- 2005 Progress Report
- 2004 Progress Report
- 2003 Progress Report
- Original Abstract
1 journal articles for this project