Final Report: Drinking Water Quality and Emergency Visits for Gastroenteritis in Atlanta

EPA Grant Number: R831629
Title: Drinking Water Quality and Emergency Visits for Gastroenteritis in Atlanta
Investigators: Tolbert, Paige , Amirtharajah, A. , Flanders, Dana , Hooper, Stuart , Klein, Mitchel , Moe, Christine L. , Singer, Philip C. , Tinker, Sarah , Uber, Jim
Institution: Emory University , Georgia Institute of Technology , University of Cincinnati , University of North Carolina at Chapel Hill
EPA Project Officer: Page, Angela
Project Period: September 1, 2004 through August 31, 2007 (Extended to August 31, 2008)
Project Amount: $1,223,366
RFA: Microbial Risk in Drinking Water (2003) RFA Text |  Recipients Lists
Research Category: Drinking Water , Water , Health Effects

Objective:

The objective of this research was to assess the relationship between drinking water quality and the incidence of gastrointestinal (GI) illness and in particular to focus on the role of the distribution system in the degradation of water quality.

The research took advantage of an extensive database containing information on emergency department (ED) visits in the 20-county Atlanta-metro area from 1993 through 2004. Water quality data were assembled from a variety of sources within the five-county Atlanta-metro area. The relationship between drinking water quality and GI illness was examined by temporal and spatial analyses of counts of ED visits for GI complaints with levels of estimated water quality, both in the general population and in susceptible subpopulations. The modifying effect of the drinking water distribution system on the quality of the water measured as it leaves the treatment plant was examined. The impact of changes in treatment methods on water quality and levels of endemic GI illness in Atlanta was considered. The proportion of risk for GI illness attributable to microbial contamination at the treatment plant and within the distribution system was estimated.

Summary/Accomplishments (Outputs/Outcomes):

Overall objectives and aims

We addressed the objectives of our project through three primary analyses, the purposes of which were to:

  • Assess the association between rates of ED visits for GI illness and estimated residence times of drinking water serving the study area;
  • Assess heterogeneity in the rates of ED visits for GI illness among the service areas of different drinking water treatment plants; and
  • Assess the temporal association between daily counts of ED visits for GI illness and drinking water turbidity.

Within the context of these primary analyses, we were also able to address our other research aims, which were to:

  • Assess whether changes in drinking water treatment practices over time were reflected in changes in drinking water quality or the incidence of GI illness;
  • Assess the association between drinking water quality and GI illness among susceptible subgroups; and
  • Estimate the population attributable fraction of ED visits for GI illness due to microbial contamination of drinking water.

Aim 1

In the analysis in which we addressed the first of the above-listed research aims, we examined whether the average water residence time for a Zip Code was related to the proportion of ED visits for GI illness among residents of that Zip Code. ED data were collected from all hospitals located in the five-county Atlanta-metro area Crom 1993 to 2004. Two of the largest drinking water utilities in the city, together serving 1.7 million people, were considered. People served by these utilities had almost 3 million total ED visits, 164,937 of them for GI illness. For one utility, two hydraulic models were used to estimate water residence time; one covered 1996 to 1998 and another covered 1999 to 2003. A single hydraulic model was used to estimate water residence times from 1993 to 2004 for the other utility. The relationship between water residence time and GI illness risk was assessed using unconditional logistic regression, controlling for potential confounding factors, including age, year, hospital, distance from Zip Code to hospital, and markers of SES. Analyses were stratified by the hydraulic model used to derive the residence time estimates. The odds ratios comparing GI illness risk among residents of Zip Codes with short average water residence time (10th percentile or less) to risk among those in Zip Codes with intermediate average water travel time (11th to 89th percentile) suggested little or no association. However, odds ratios for residents of Zip Codes in the top decile of average water residence times compared to residents of Zip Codes with intermediate average residence times suggested modestly increased risk for GI illness (odds ratio (OR) for Utility 1 = 1.07, 95% confidence interval (CI) = 1.03, 1.10; OR for Utility 2 = 1.05, 95% CI = 1.02, 1.08). Effect measure modification by year was suggested for both utilities and by season and age group for Utility 2. The results suggest that drinking water contamination in the distribution system may contribute to the burden of endemic GI illness.

Table 1: Descriptive statistics of estimated drinking water residence time by year, in hours, summarized over Zip Codes, Atlantaa

Short water residence time exposure categoryb
Mean (Min – Max)
Intermediate water residence time categoryb
Mean (Min – Max)
Long residence water time categoryb
Mean (Min – Max)
Utility 1,
Hydraulic Model 1c
 6.83 (4.54 – 8.28) 21.96 (8.34 – 36.37) 47.40 (37.61 – 68.46)
Utility 1,
Hydraulic Model 2d
10.12 (9.14 – 11.18) 33.43 (11.19 – 51.78) 74.41 (52.20 – 88.36)
Utility 2e 5.85 (4.74 – 8.57) 18.45 (8.62 – 40.29) 60.39 (40.31 – 144.06)

a Two utilities considered

b Short water residence time defined as ≤ 10th percentile of all water residence time estimates, stratified by hydraulic

model; intermediate = 11th to 89th percentile; long = ≥ 90th percentile

c 1996 – 1998

d 1999 – 2003

e 1993 – 2004

Figure 1:  Risk ratio estimates for gastrointestinal illness emergency department visits among people living in Zip Codes with short* drinking water residence times and long* drinking water residence times compared to intermediate* drinking water residence times, metro Atlanta, stratified by utility

* Short water residence time defined as ≤ 10th percentile of all water residence time estimates, stratified by

hydraulic model; intermediate = 11th to 89th percentile (referent category); long = ≥ 90th percentile

Two utilities considered

Figure 2: Risk ratio estimates for gastrointestinal illness emergency department visits among people living in Zip Codes with long* drinking water residence times compared to intermediate* drinking water residence times, metro Atlanta, stratified by year

* Intermediate water residence time defined as 11th to 89th percentile of all water residence time estimates,

stratified by hydraulic model (referent category); long = ≥ 90th percentile

Two utilities considered

Figure 3: Risk ratio estimates for gastrointestinal illness emergency department visits

among people living in Zip Codes with short* drinking water residence times and long*

drinking water residence times compared to intermediate* drinking water residence times,

metro Atlanta, stratified by season, Utility 2 only

* Short water residence time defined as ≤ 10th percentile of all water residence time estimates, stratified by

hydraulic model; intermediate = 11th to 89th percentile (referent category); long = > 90th percentile

Results for one utility presented here.

Figure 4: Risk ratio estimates for gastrointestinal illness emergency department visits among people living in Zip Codes with short* drinking water residence times and long* drinking water residence times compared to intermediate* drinking water residence times, metro Atlanta, stratified by age group, Utility 2 only

* Short water residence time defined as ≤ 10th percentile of all water residence time estimates, stratified by

hydraulic model; intermediate = 11th to 89th percentile (referent category); long = > 90th percentile

Results for one utility presented here.

Aim 2

To address the second study aim, we compared rates of ED visits for GI illness among people served by different drinking water treatment plants. Using Poisson regression, we analyzed data on almost 4 million ED visits from 27 hospitals. We also examined treatment plant attributes and changes in treatment methods over time that could plausibly contribute to differences in finished water quality and risk of GI illness. We observed modest differences in the rates of ED visits for GI illness among those served by different treatment plants. People living in the service area of one plant had a small but statistically significant elevation in the rate of ED visits for GI illness relative to people living in the service areas of the other plants (Rate Ratio (RR) = 1.05, 95% CI = 1.03, 1.08). We identified several operational conditions that may have contributed to poorer water quality at this plant during certain parts of the study period. Over time, several improvements were made to the plant and its operation, and these changes may be reflected in the reduced relative rates we observed for this plant during the later parts of our study period. These results suggest that heterogeneity of rates for GI illness among people served by different treatment plants in Atlanta may be partially explained by differences in treatment plant operation and resulting water quality.

Table 2.  Adjusted rate ratio estimates for emergency department visits for gastrointestinal illness among people served by any one drinking water treatment plant compared to all other treatment plants, and attributes of the drinking water treatment plants in the five-county metro-Atlanta area, 1993-2004.

 
Treatment Plant
 
Total Non-Injury ED Visits
 
ED Visits for GI Illness (% of Total)
 
Rate Ratio*,†,‡
 
Production Capacity (MGD§)
 
Source Water Type
 
Treatment Method||
 
Disinfectant
 
Coagulant
 
Watershed Susceptibility#
 
Mean Daily Maximum Raw Water Turbidity** (NTU††)
 
Mean Daily Average Filtered Water Turbidity‡‡ (NTU)
A 193,186 18,192 (9.4) 0.98 (0.96, 1.00§§) 90 River Conventional Chlorine Alum Medium 6.5 0.03
B 755,806 68,040 (9.0) 0.99 (0.98, 1.01) 128 River Conventional Chlorine Alum Medium 13.5 0.06
C|| || 45,745 3,301 (7.2) 1.03 (0.99, 1.07) 65 River Conventional Chlorine Alum High 55.0 0.07
D## 27,306 2,172 (8.0) 1.00 (0.94, 1.10) Plant 1 = 20 Plant 2 = 10 Creek-fed reservoir; LAS *** Conventional Chlorine, UV Alum Medium 4.2 0.08
E|| || 505,951 36,561 (7.2) 0.98 (0.96, 1.00§§) 137 River Conventional Chlorine Alum High 11.9 0.10
F 580,703 53,266 (9.2) 1.05 (1.03, 1.08) 150 Lake Direct filtration Chlorine, ozone Polymer, ferric chloride††† Low - medium 1.5 0.17
G§§ 418,674 33,969 (8.1) 1.00 (0.99, 1.02) 86 River Conventional Chlorine Alum High 22.0 0.06
H§§ 187,451 15,461 (8.2) 1.02 (0.99, 1.04) 72 Lake Conventional Chlorine Alum Medium 10.3 0.08

* Adjusted for age group, year, season, hospital, distance from Zip Code to treatment plant, distance from Zip Code to hospital, Zip Code median income, Zip Code percent minority, Medicaid payment status.

† Comparing that plant to all other plants.

‡ 95% confidence interval given in parentheses.

§ Million gallons per day.

|| Conventional treatment: Coagulation, flocculation, sedimentation, filtration, disinfection. 

# Standardized assessments conducted by the Atlanta Regional Commission and Georgia Mountains Regional Development Center; based on number of point sources of pollution and measures of non-point source pollution. ** Mean daily maximum raw water turbidity values from hourly measurements taken from 1/1/2002 to 12/31/2004, except for Plant D, for which measurements were only available beginning 3/1/2002.

†† Nephelometric Turbidity Units.

‡‡ Mean daily average filtered water turbidity values from hourly measurements taken from 7/1/1993 to 12/31/2004, except for Plant D, which did not begin operation until late 10/1999, Plant C, for which measurements were available from 1/1/1993 to 12/31/2004, and Plant F, for which measurements were available from 7/1/1993 to 12/31/1999 and 1/1/2002 to 12/31/2004.

§§ Although the 95% confidence intervals include 1.00, the rate ratio estimates for Plants A and E are statistically significant at the traditional alpha value of 0.05. The p-value for these estimates are less than 0.05; the upper 95% confidence intervals equal 1.00 due to rounding.

|| || Plants C and E are operated by the same utility; Plants G and H are operated by the same utility.

## Two plants considered together because their water was mixed prior to distribution.

*** Land application system.

††† Plant used polymer only until 1998, when ferric chloride was added as additional coagulant.

Figure 5: Rate ratioa,b estimates for emergency department visits for gastrointestinal illness among people served by Plant F compared to all other treatment plants, 1993 – 2004

Improved backwash treatment

 
Discovered problem with filters and initiated replacementc
 
Plant expansion while plant was still running
 
Ozone used as a disinfectant in addition to chlorine
 
Only chlorine used as a disinfectant
 

Aim 3

In our analysis to address the third aim listed above, we used Poisson time-series methods to examine the relationship between turbidity levels of raw and filtered surface water measured at the treatment plants operated by the participating utilities, and over 240,000 ED visits for GI illness among the population served by these plants. For filtered water turbidity, the results were consistent with no association with ED visits for GI illness. We observed a modest association between raw water turbidity and ED visits for GI illness. This association was not observed for all treatment plants in plant-specific analyses. The observed increased risk of GI illness with increased raw water turbidity was strongest among children age 5 years and younger. Our results suggest that source water quality may contribute modestly to endemic GI illness in the study area. The association between turbidity and ED visits for GI illness was only observed when raw water turbidity was considered; filtered water turbidity may not serve as a reliable indicator of modest pathogen risk at all treatment plants.

Table 3. Rate ratio estimates and 95% confidence intervals from a priori models for the association of daily drinking water turbidity measures with emergency department visits for gastrointestinal illness in Atlanta, 1993 - 2004

Average Filtered
Water Turbidity*
Maximum Filtered
Water Turbidity*
Minimum Raw
Water Turbidity
Maximum Raw
Water Turbidity
 
Summary
0.98 (0.96, 1.01) 0.99 (0.98, 1.01) 1.06 (1.04, 1.08) 1.02 (1.01, 1.03)
 
Stratified by Drinking Water Treatment Plant
Plant A 0.64 (0.32, 1.26) 0.79 (0.56, 1.11) 1.08 (0.90, 1.30) 1.10 (0.99, 1.23)
Plant B 1.05 (0.92, 1.20) 1.01 (0.95, 1.07) 1.07 (1.03, 1.11) 1.02 (1.00, 1.04)
Plant C 0.96 (0.81, 1.15) 1.23 (0.69, 2.22) 1.05 (0.97, 1.14) 1.01 (0.98, 1.03)
Plant D 0.36 (0.12, 1.14) 0.47 (0.22, 1.01) 0.63 (0.26, 1.53) 0.81 (0.41, 1.62)
Plant E 0.98 (0.95, 1.00) 0.98 (0.97, 1.00) 1.08 (1.04, 1.12) 1.06 (1.03, 1.09)
Plant F 1.00 (0.94, 1.07) 1.01 (0.97, 1.04) 2.07 (0.61, 7.01) 1.29 (0.48, 3.48)
Plant G 1.70 (1.28, 2.26) 1.26 (1.07, 1.47) 1.06 (1.02, 1.09) 1.02 (1.00, 1.04)
Plant H 0.68 (0.45, 1.03) 0.85 (0.68, 1.06) 0.93 (0.83, 1.03) 0.94 (0.86, 1.03)
 
Stratified by Age Group
0 – 5 years 0.94 (0.90, 1.00) 0.98 (0.95, 1.01) 1.11 (1.08, 1.15) 1.04 (1.03, 1.06)
6 – 18 years 1.00 (0.93, 1.08) 1.00 (0.95, 1.04) 1.01 (0.94, 1.08) 0.99 (0.96, 1.02)
19 – 64 years 0.99 (0.96, 1.02) 0.99 (0.97, 1.01) 1.03 (1.00, 1.06) 1.01 (0.99, 1.02)
65+ years 1.00 (0.92, 1.09) 0.99 (0.94, 1.04) 0.99 (0.92, 1.06) 1.02 (0.99, 1.06)

* Unit of change for rate ratio estimates for filtered water turbidity measures was 0.1 NTU over 21 days encompassing day of emergency department visit and preceding 20 days

Unit of change for rate ratio estimates for raw water turbidity measures was 10 NTU over 21 days encompassing day of emergency department visit and preceding 20 days

Susceptible subgroups were considered in the context of each of the analyses described above. We observed stronger associations between long water residence time and GI illness among children age 18 years and younger, although this result was only observed for one of the two utilities considered (Figure 4). The strongest associations between treatment plant of service and ED visits for GI illness were observed for children (0 to 18 years) and the elderly (65+ years) (Figure 6). We observed little association of GI illness with filtered water turbidity measures when each age group was considered separately. For raw water turbidity, the association for children age five years and younger was somewhat stronger than for the other age groups (Table 3).

Figure 6:  Rate ratio estimates*, † for emergency department visits for gastrointestinal illness among people served by any one drinking water treatment plant compared to all other treatment plants in the five-county metro-Atlanta area, 1993-2004, stratified by age group.

 

* Adjusted for year, season, hospital, distance from Zip Code to treatment plant, distance from Zip Code to hospital, Zip Code median income, Zip Code percent minority, Medicaid payment status.  

95% confidence intervals given by vertical lines. 

Plant D began operation in November 1999.

Using the results of our analyses, we can make rough approximations for the population attributable fraction of ED visits for GI illness due to microbial contamination of drinking water.

  • 0.5 to 1.3 percent of ED visits for GI illness among the study population were attributable to exposure to water with long residence time, assuming a causal association.
    • 1.1 percent of ED visits for GI illness among the study population was attributable to exposure to drinking water treated and distributed by Plant F, assuming a causal relationship.
  • Approximately 2 percent of ED visits for GI illness in the study area were attributable to exposure to 10 NTU increases in raw water turbidity, assuming a causal relationship.

As always, caution must be used when interpreting the results of these analyses, as alternative explanations exist. The first two analyses described were particularly susceptible to confounding by spatial covariates, and while variables were included in the regression models in an attempt to accommodate this source of bias, residual confounding likely remained. While the third analysis, for which a time-series was conducted, was not as susceptible to spatial confounding as the other studies, this analysis was vulnerable to temporal confounding. Indicators for day-of-week and temperature and cubic splines to account for long-term time trends were included in the analytical model to address these concerns, but residual confounding may have remained.

Roles for both the raw water source and the distribution system as sites of drinking water contamination are suggested by the results of the analyses. Filtered water turbidity, a primary water quality measure used by utilities, did not appear to predict risk. Overall, the results of this study suggest that a low level of GI illness in Atlanta may be attributable to drinking water exposure, particularly among young children and the elderly. We believe the results of this study support the need for future research regarding:

  • The association between water residence time and GI illness in additional utility settings.
  • The use of hydraulic models to inform utilities about the areas of the distribution system with the greatest susceptibility to pathogens and whether monitoring at these locations may be more effective than random monitoring throughout the distribution system.
  • Other methods of assessing pathogen contamination in filtered water in addition to turbidity monitoring. While turbidity may serve as a reliable indicator of large-scale pathogen contamination, it may not be completely effective for monitoring low levels of contamination that are still capable of causing human disease.


Journal Articles on this Report : 2 Displayed | Download in RIS Format

Other project views: All 13 publications 2 publications in selected types All 2 journal articles
Type Citation Project Document Sources
Journal Article Tinker SC, Moe CL, Klein M, Flanders WD, Uber J, Amirtharajah A, Singer P, Tolbert PE. Drinking water residence time in distribution networks and emergency department visits for gastrointestinal illness in Metro Atlanta, Georgia. Journal of Water and Health 2009;7(2):332-343. R831629 (Final)
  • Abstract from PubMed
  • Abstract: Journal of Water and Health Abstract
    Exit
  • Journal Article Tinker SC, Moe CL, Klein M, Flanders WD, Uber J, Amirtharajah A, Singer P, Tolbert PE. Drinking water turbidity and emergency department visits for gastrointestinal illness in Atlanta, 1993-2004. Journal of Exposure Science & Environmental Epidemiology 2010;20(1):19-28. R831629 (2007)
    R831629 (Final)
    R829213 (Final)
    R830376 (2006)
  • Abstract from PubMed
  • Full-text: Nature Publishing Group-Full Text HTML
    Exit
  • Abstract: Nature Publishing Group-Abstract
    Exit
  • Other: Nature Publishing Group-PDF
    Exit
  • Supplemental Keywords:

    water, drinking water, exposure, risk, health effects, human health, population, sensitive populations, children, elderly, pathogens, dose-response, public policy, socioeconomic, epidemiology, modeling, Georgia, GA,, RFA, Health, Scientific Discipline, INTERNATIONAL COOPERATION, Water, POLLUTANTS/TOXICS, Environmental Chemistry, Health Risk Assessment, Epidemiology, Risk Assessments, Drinking Water, Microorganisms, groundwater disinfection, health effects, microbial contamination, bacteria, human health effects, waterborne disease, other - risk assessment, Atlanta, treatment, human exposure, microbial effects, water quality, microbial risk, water disinfection, groundwater contamination, dietary ingestion exposures, drinking water contaminants, drinking water treatment, human health, gastrointestinal health, groundwater, gastrointestinal health effects

    Progress and Final Reports:

    Original Abstract
  • 2005 Progress Report
  • 2006 Progress Report
  • 2007 Progress Report