Grantee Research Project Results
2012 Progress Report: Measures of Distribution System Water Quality and Their Relation to Health Outcomes in Atlanta
EPA Grant Number: R834250Title: Measures of Distribution System Water Quality and Their Relation to Health Outcomes in Atlanta
Investigators: Moe, Christine L. , Tolbert, Paige , Klein, Mitchel , Uber, Jim , Hooper, Stuart , Sarnat, Stefanie Ebelt , Levy, Karen
Current Investigators: Moe, Christine L. , Sarnat, Stefanie Ebelt , Kirby, Amy , Levy, Karen , Klein, Mitchel , Tolbert, Paige
Institution: Rollins School of Public Health, Emory University , University of Cincinnati
Current Institution: Rollins School of Public Health, Emory University
EPA Project Officer: Page, Angela
Project Period: July 1, 2009 through June 30, 2012 (Extended to June 30, 2014)
Project Period Covered by this Report: July 1, 2011 through June 30,2012
Project Amount: $599,756
RFA: Innovative and Integrative Approaches for Advancing Public Health Protection Through Water Infrastructure Sustainability (2008) RFA Text | Recipients Lists
Research Category: Pollution Prevention/Sustainable Development , Sustainable and Healthy Communities , Water
Objective:
The investigators proposed research to link health outcomes, quantified through emergency department (ED) visits for gastrointestinal (GI) illness, to distribution system water quality and infrastructure characteristics. In addition, the investigators proposed an integrative approach for characterizing health risks associated with microbial contamination of drinking water distribution systems through the development of vulnerability assessments and the implementation of an innovative automated continuous water quality monitor for use in distribution systems.
This study has four aims:
Aim A (Spatially refined analysis): Refine the previously conducted analyses assessing the association between rates of ED visits for GI illness and estimated residence times of drinking water serving the study area of two utilities, using geocoded patient address data to identify the closest node in the distribution system.
Aim B (1993-2004, 2005-2009, 1993-2009 analysis): Extend the previously-conducted analyses assessing the association between rates of ED visits for GI illness and estimated residence times of drinking water serving the service area of two utilities by considering data from these utilities covering 2005 through 2009.
Aim C (Vulnerability assessment): Assess the association between rates of ED visits for GI illness and exposure metrics derived from the results of the vulnerability assessment. Compare these associations to those observed when only chlorine residual or water age metrics are considered as the exposure. One distribution system will be studied.
Aim D (AMS evaluation of vulnerability assessment results): Use an automated monitoring system (AMS) to examine the results of the vulnerability assessment by comparing continuously monitored water quality in areas of the distribution system predicted to be “high risk” for microbial contamination to areas predicted to be “low risk.”
In the second year of the study, we developed goals for each of the four aims to streamline the progress of the project. In the following sections, we will discuss the progress toward these goals, and we will also outline the goals for the fourth year of the project.
In the third year of this grant, our major goals, as outlined in our Year 2 Annual Report, were to:
- Complete analysis for Aim A: to refine the previously conducted analyses assessing the association between rates of ED visits for GI and estimated residence times of drinking water using geocoded patient address data to identify the closest node in the distribution system.
- Develop water residence time estimates for two drinking water utilities in metro Atlanta through 2009 under Aim B.
- Perform analysis for Aim B: to extend the previously conducted analyses assessing the association between rates of ED visits for GI illness and estimated residence times of drinking water serving the service area of the two utilities by considering data from these utilities covering 2005 through 2009.
- Complete a vulnerability assessment of the distribution system to pathogen contamination for one selected utility (Aim C).
- Initiate continuous monitoring of distribution system water quality at the same utility for which the vulnerability assessment was conducted (Aim D).
- Write a manuscript that describes the results for Aims A and B and submit for publication.
Progress Summary:
Aim A: We have completed analyses for Aim A. Much of Year 3 was devoted to checking and merging of datasets (i.e., data on estimates of water age at nodes in the hydraulic model with geocoded data on emergency department visits) in GIS. This took longer than expected because we had to reconcile some unexpected differences between the datasets. After this was resolved, we spent time developing the epidemiologic models to assess the association between water age and emergency department visits for GI illness. We expect to write up these results for publication in the coming year.
Aim B: We have obtained and processed the relevant health outcome data to carry out the analyses for Aim B, and we have obtained the water production data from both of our collaborating metro Atlanta water utilities, but revised water residence time analyses have not yet been completed by our subcontractor, Dr. James Uber at the University of Cincinnati. Despite repeated requests in the past year, Dr. Uber has not provided these water residence time estimates, and we have been unable to make further progress on this aim. Because the funding for this study will be depleted in December 2012, we are no longer able to complete this study aim.
Aim C: The assessment of microbial health risks due to low-pressure intrusion events has received increased attention in recent years, due to concerns that drinking water may contribute a significant fraction of gastrointestinal disease burden within a community. While recent efforts to estimate the health risks from low-pressure intrusion events have shown impressive progress, they would benefit from a general risk assessment framework for water distribution systems. We selected the City of Atlanta water utility for performing the risk assessment because its water distribution system is representative of many older cities in the eastern United States.
Our subcontractor, Dr. James Uber at the University of Cincinnati, was responsible for developing a risk assessment framework, for the City of Atlanta water system based on distribution system hydraulic and water quality models. Last year, he informed us that the framework was “90% complete” and sent us the following description:
The framework includes the following features:
- Simulation of pathogen transport using general, multi-species fate and transport models.
- General specification of the key characteristics of the vulnerability assessment model, using a specialized language for Monte-Carlo simulation.
- Database support for flexible and efficient storage of Monte Carlo ensemble results, for post-processing of vulnerability assessment model results.
- Efficient, yet accurate, simulation of large numbers of intrusion scenarios, using the theory of linear superposition.
- Support for accurate and efficient simulation of short duration intrusion events (as required for low pressure events, ranging in duration from seconds to minutes), and accurate capture of instantaneous concentrations as well as statistical descriptions of concentration variations.
- Development of an open-source programming project to ensure that the framework continues to meet changing research needs.
In May 2011, Dr. Uber presented an update on progress on this framework at the American Society of Civil Engineers World Water Congress in Palm Springs, CA.
Dr. Uber’s scope of work included completing the framework and applying it to the City of Atlanta distribution system, as the first comprehensive analysis with this framework. Our goal was to then use this risk assessment in our epidemiologic analyses to compare the rates of ED visits in geographic areas of greater and lesser risk as identified by the assessment and compare the results of these analyses to our previous analyses based on water residence time.
Despite repeated requests during this past year, we have not received this risk assessment from Dr. Uber and have not been able to move forward on this specific aim. Because the funding for this study will be depleted in December 2012, we are no longer able to complete this study aim.
Aim D: We deployed the Automated Monitoring System (AMS) at the City of Atlanta water treatment plant and completed baseline monitoring and sampling.
- We installed the AMS at the COA treatment plant on November 29, 2011.
- We monitored water quality parameters—pH, total chlorine, conductivity, redox, turbidity, and temperature—of finished water before it enters the distribution system.
- We collected five large 90 L samples and one grab sample and analyzed for a panel of microorganisms, including Escherichia coli, total coliform, Aeromonas hydrophila, heterotrophic plate count (HPC) bacteria, Pseudomonas aeruginosa, Clostridium perfringens, and male-specific coliphage.
- We identified a location, a fire station, in the distribution system for AMS deployment. The site selection was based on three criteria: main break clusters, water residence time at node level, and frequency of low pressure events.
-
We collaborated with the fire department on the logistics of the installation as well as operation, maintenance, and sampling. We installed the AMS at the fire station on September 14, 2012. This site represents a geographic location with long water residence time. Data collection at this site is ongoing.
Results to Date
We have completed the analysis for Aim A (spatially refined analysis), and are currently in the process of writing a manuscript to describe the results. In GIS, we joined geocoded data on emergency department (ED) visits from 14 hospitals with node-level data on water residence time (WRT) from nodes lying within the utility service area and with >0 water demand. We then imported the joined dataset to SAS, where we carried out logistic regression to evaluate the impact of WRT on ED GI admissions, controlling for potential confounders, including year, day of week, hospital, distance of residence to hospital, patient age and Medicaid status, and census block demographics.
The refined exposure assessment, assigning each patient the WRT from the closest node in the distribution system to their geocoded residential address, resulted in similar results from those published by Tinker et al. (2009) for Utility 1, but the effect of long WRT on ED visits for GI was not observed for Utility 2. To investigate this new result, we ran two additional analyses that provided further insight. We observed that the average WRTs for Utility 1 (47.2 hrs) were much longer than for Utility 2 (18.9 hrs). First, we applied the 90th percentile WRT cutpoint for Utility 1 to the Utility 2 data as the criterion for “long” WRT, and we observed for Utility 2 a similar effect estimate for the impact of WRT on GI illness as observed previously by Tinker et al. (2009). However, the confidence intervals were extremely large because very few nodes in the Utility 2 DS had such long WRT (Table 1).
Second, we further examined the absolute effect of WRT in terms of number of hours in the DS (Table 2). This more detailed analysis provides further support for the importance of considering the impact of absolute number of hours for WRT rather than just examining the relative length in the DS for a particular utility. We could not have gained this important insight without the updated, more spatially explicit, exposure assessment and analysis.
Table 1. Associations between water residence time and GI illness, comparing the effects for short (<10th percentile) and long (>90th percentile) WRT vs. intermediate values (10-90th percentile).
Tinker et.al. (2009)* | Aim A refined expososure assessment | ||||
---|---|---|---|---|---|
Water Residence TIme | Utility 1 | Utility 2 | Utility 1 | Utility 2 | Utility 2 (using Utility 1 WRT cutpoints) |
Short | 0.95 (0.88-1.02) | 0.992 (0.95-1.06) | 1.00 (0.95-1.06) | 0.998 (0.96-1.04) | 0.996 0.97-1.02) |
Intermediate | 1.0 | 1.0 | 1.0 | 1.0 | 1.0 |
Long | 1.045 (1.00-1.09) | 1.049 (1.02-1.078) | 1.07 (1.02-1.13) | 0.98 (0.94-1.02) | 1.25 (0.96-1.64) |
* Tinker et.al. (2009) used ZIP code average WRT as the exposure measure. Aim A used node-level WRT as the exposure measure. |
Table 2. Associations between water residence time and GI illness, comparing the effects in WRT in 12 hour increments vs. WRT of less than 12 hrs.
Water Resistance Time | Utility 1 | Utility 2 |
---|---|---|
<12 hrs | 1.00 | 1.00 |
12-24 hrs | 0.997 (0.93-1.08) | 1.02 (0.99-1.05) |
24-36 hrs. | 1.01 (0.93-1.09) | 0.997 (0.95-1.04) |
36-48 hrs. | 0.94 (0.87-1.01) | 1.01 (0.96-1.08) |
48-60 hrs. | 1.03 (0.95-1.11) | 0.92 (0.82-1.03) |
60-72 hrs. | 0.92 (0.84-1.01) | 1.001 (0.83-121) |
72-84 hrs. | 0.96 (0.86-1.08) | 1.06 (0.86-1.30) |
84-96 hrs. | 0.96 (0.83-1.10) | 1.08 (0.76-1.53) |
>96 hrs. | 1.05 (0.97-1.14) | 1.19 (0.97-1.46) |
Total WRT from the water treatment plant to the household includes: (1) the WRT from the treatment plant to the distribution system (DS) node, and (2) the WRT from the DS node to the household. While our analyses above considered the first part of the exposure pathway, the geocoded database also allowed us to examine the second part of the exposure pathway using the Euclidean distance from the DS node to the household address of the ED patient, as a proxy for WRT from the DS node to the household. For Utility 1, we found an increased risk of GI illness for patients living in houses >500 ft. from the node compared to patients living in houses <100 ft from the node. While we did not see the same effect for Utility 2, the results are suggestive of node-household WRT as being an important part of the exposure pathway for drinking water-related GI illness. This analysis would not have been possible without the geocoded dataset produced for Aim A.
Table 3. Associations between node-household distance and GI illness.
Utility 1 | Utility 2 | |
---|---|---|
100-500 ft vs. <100 ft. | 1.021 (0.98-1.06) | 0.98 ()0.95-1.01 |
>500 ft vs. <100 ft. | 1.091 (1.03-1.15) | 0.96 (0.92-1.00) |
Future Activities:
- We plan to complete the manuscript describing the analyses for Aim A in early 2013 and submit this work for publication.
- For Aim D, we plan to move the AMS device to one additional location in the City of Atlanta water distribution system. We have been analyzing historic data on main breaks in the system and identifying geographic clusters of these events. We plan to move the AMS device in January 2013 to an area that represents greater vulnerability to main breaks and will collect physical and chemical water quality data for a period of 2 months. All of the AMS data are currently being analyzed by a graduate student in our M.P.H. program as the basis for her thesis. We expect these analyses to be completed in the spring or summer of 2013. We plan to write a manuscript describing these analyses and submit this for publication by the end of Year 4.
- At this time, we do not anticipate being able to conduct the analyses for Aims B and C because we have not received the necessary information from our subcontractor, Dr. Uber. Although we have identified other collaborators who could do the risk assessment for the City of Atlanta water distribution system (Aim C), we do not have funding to recruit another person to conduct these analyses. We plan to seek other sources of funding to complete this work.
Journal Articles:
No journal articles submitted with this report: View all 14 publications for this projectSupplemental Keywords:
Water, drinking water, exposure, risk, health effects, human health, vulnerability, population, indicators, public policy, epidemiology, engineering, modeling, monitoring, Georgia, GAProgress 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
- Final Report
- 2013 Progress Report
- 2011 Progress Report
- 2010 Progress Report
- 2009 Progress Report
- Original Abstract
1 journal articles for this project