Published online November 30, 2007
PEDIATRICS
Vol. 120
No. 6
December 2007, pp.
e1472-e1475
(doi:10.1542/peds.2007-0283)
Pediatric Emergency Department Visits for Diarrheal Illness Increased After Release of Undertreated Sewage
Ryan L. Redman, MDa,
Cheryl A. Nenn, MSb,
Daniel Eastwood, MSa and
Marc H. Gorelick, MD, MSCEa
a Emergency Department, Medical College of Wisconsin, Milwaukee, Wisconsin
b Friends of Milwaukee's Rivers, Milwaukee, Wisconsin
 |
ABSTRACT
|
|---|
CONTEXT. Contamination of local waterways may occur through release of partially treated sewage. The Environmental Protection Agency has recently reviewed regulatory standards for this practice. However, the health effects of these events have not been well studied.
OBJECTIVE. Our goal was to identify any increase in visits to a pediatric emergency department for diarrheal illness after sewage bypass into Lake Michigan.
METHODS. The study was conducted as a retrospective, observational time-series analysis in a tertiary care children's hospital emergency department with an annual volume of 
45000 visits. We collected data for 2002–2004 pertaining to the daily number of emergency department visits for children (aged <19 years) for diarrheal illness (using specified International Classification of Diseases, Ninth Edition codes as a reference). Daily diarrheal illness visits were the dependent variable in a time-series model. The primary independent variable was the occurrence of a sewage-bypass event in the 3 to 7 preceding days. Potential confounders included the season and daily rainfall. Separate models were created for visits from people living in zip codes that used Lake Michigan drinking water and those who used other water sources.
RESULTS. Over the 3-year study period, there was a mean of 5.0 ± 3.8 (SD) daily visits for diarrheal illness from people who lived in zip codes that used Lake Michigan drinking water and 1.2 ± 1.4 (SD) from outside that area. There were 6 sewage-bypass events identified. After adjusting for the season and rainfall, there was a significant increase of 2.5 to 2.7 visits only from people who lived in zip codes that used Lake Michigan drinking water after the 2 largest of the 6 bypass events.
CONCLUSIONS. Emergency department visits for diarrheal illness increased significantly after 2 events of release of partially treated sewage into area waterways. These data suggest a potentially harmful effect of such practices.
Key Words: gastrointestinal infections • environmental health
Abbreviations: ED—emergency department • ARIMA—autoregressive integrated moving average • ICD-9— International Classification of Diseases, Ninth Edition • CHW—Children's Hospital of Wisconsin • CI—confidence interval
Contamination of local waterways by untreated or partially treated sewage may affect public health through dissemination of waterborne pathogens.1 In the practice known as secondary bypass, or blending, sewage proceeds to primary treatment, where solids, hydrophobic compounds, and sediment are removed. From there, up to 20% of the sewage stream bypasses the usual secondary treatment with biological agents (where most pathogens are removed) and is directly diverted to the final step in the process. In this final step, the diverted sewage is blended with sewage from secondary treatment, and then disinfecting agents such as chlorine are added, and the partially treated or "blended" sewage is discharged into local watershed areas.2 During these events, the effluent will commonly meet permit limits and water-quality criteria: however, previous studies have raised the question of whether currently used water-quality parameters are adequate for identifying the presence of human gastrointestinal pathogens.3,4
Although not generally currently allowed under existing federal regulation, bypass is permitted in areas served by combined sewers (those that carry both waste and rainwater) under certain circumstances by a number of state regulatory agencies.5 The number of such blending permits is unknown, but there are an estimated 746 communities in 32 states served by combined sewers. The Environmental Protection Agency recently proposed revisions to its peak wet-weather policy that would establish national standards to allow this practice.6
The effect of secondary sewage bypass on human health is largely unknown.2 Many investigations have addressed the role of specific waterborne pathogens in outbreaks of diarrheal illness.7–10 However, no studies have looked specifically at the health effects of secondary bypass. The purpose of this study was to determine if there is an association between secondary sewage bypass and emergency department (ED) visits for gastroenteritis among children in the Milwaukee, Wisconsin, metropolitan area.
|
METHODS
|
|---|
The study was conducted as a retrospective observational time-series analysis. We obtained data for a 3-year period, from January 1, 2002, to December 31, 2004, from the computerized billing system at the Children's Hospital of Wisconsin (CHW), a tertiary care hospital located in suburban Milwaukee with 45000 annual ED visits. It is the only children's hospital in southeastern Wisconsin and accounts for >40% of all ED visits for children <18 years old in Milwaukee County. The primary outcome variable was the daily number of visits for diarrheal illness to the CHW, which was based on any of the following International Classification of Diseases, Ninth Revision (ICD-9) codes being recorded as the ED discharge diagnosis: specified gastrointestinal infections (ICD-9 codes 001–009.9), unspecified gastroenteritis (ICD-9 code 558.9), or diarrhea (ICD-9 code 787.91).
The main independent variable was the occurrence of a secondary sewage-bypass event as reported by the Milwaukee Metropolitan Sewerage District. We recorded the event date and volume of sewage diverted, as well as counts of fecal coliforms in the plant effluent. When the data were available, we also recorded information on counts of Giardia cysts in the effluent as reported by the Milwaukee Health Department. Because sewage bypass usually occurs during periods of wet weather, daily rainfall totals at General Mitchell International Airport in Milwaukee were obtained from the National Oceanographic and Atmospheric Association.
Because the data set consists of values of the variables that were recorded at regular intervals over a long period of time, the observations in such a series are correlated. A technique for the analysis of such time-series data is the autoregressive integrated moving average (ARIMA) model.11 By considering serial autocorrelation, we can account for the effect of underlying perturbations in the data (eg, local community outbreaks of disease). We used a type of ARIMA model called an intervention model to estimate the effect of the events in question. In this model, the input series is an indicator variable that denotes the event. In this case, we flagged dates 3 to 7 days after a sewage-bypass event, a lag that was chosen a priori on the basis of clinical experience as well as epidemiologic data from previous studies.12 Thus, an event variable is a step function, and it estimates the impact of the occurrence of an event that affects the response series. Because the volume of sewage and type of bypass differed for each event, a unique indicator variable was created for each of the events. The primary outcome variable of interest was the number of visits for diarrheal illness on a given date. We used a square-root transform to normalize the visit data. To adjust for potential confounding, we included variables for season (winter versus nonwinter) and mean rainfall in the 3 to 5 days before each day in the model. Separate models were generated for visits from zip codes that used Lake Michigan drinking water sources and for those in which drinking water came from other sources (primarily wells). A small number of visits from zip codes that used mixed water service were excluded from the analysis.
At each lag in time, the autocorrelation and partial autocorrelations were plotted and visually inspected, and model parameters for relevant lags were created. ARIMA model coefficients and 95% confidence intervals (CIs) were calculated for each of the factors under study. The adequacy of the final model was assessed by checking the residual plots for white noise, that is, residuals were distributed randomly and normally over time (Portmanteau Q test).13
|
RESULTS
|
|---|
During the study period, there were 8 secondary sewage-bypass events (Table 1). Two of these events involved the release of small amounts of sewage and were immediately preceded by a larger event within 2 days. In both of these instances, we combined the 2 adjacent releases and considered them to be a single event. Levels of fecal coliforms in the sewage-plant effluent were elevated but within permissible limits after all events. Increased levels of Giardia cysts were also found in the effluent.
During the study period, there was a mean of 5.0 daily visits for gastroenteritis from the people who lived in zip codes that used Lake Michigan drinking water and 1.2 from those who lived in zip codes outside the Lake Michigan water service area. This ratio (80% of visits from zip codes that used Lake Michigan drinking water) is reflective of the relative frequency of all visits to the CHW ED (82.6% from these zip codes) during the study period. Results from the ARIMA models are shown in Table 2. Significant increases in visits were seen by people who lived in zip codes that used Lake Michigan drinking water after the 2 largest events, after adjusting for season and rainfall. The average daily increase for these 2 events after adjusting for the mean and transforming back to the original scale of the data, was 2.5 visits (95% CI: 0.4 to 4.9) and 2.7 visits (95% CI: 0.0 to 5.9), respectively, which was a relative increase of 50% (26 additional cases). No significant increase was seen after any of the events from the people who lived in zip codes that used drinking water from non–Lake Michigan sources.
|
DISCUSSION
|
|---|
For 2 of the 6 secondary bypass events, there was a statistically significant increase in the number of visits to the CHW pediatric ED for diarrheal illness after controlling for both season and rainfall, which were seen as 2 potential confounders. Although the limited data we have cannot establish a causal relationship, this possibility is suggested by the fact that the association was observed only from people who lived in zip codes that used Lake Michigan drinking water sources (specificity); that the 2 events with a significant association with increased visits were the 2 largest-volume events (dose-response); and that there were elevated levels of fecal coliforms and Giardia cysts detected in the plant effluent (biological plausibility). It should be emphasized that the levels, although elevated, were within acceptable mandated limits. However, previous work has shown that even water that is free of detectable fecal coliform bacteria may cause human disease,3 mainly because there is no routine testing for viral pathogens.
The absolute magnitude of increased risk in this population is quite small, with just 26 cases occurring over a 3-year period. However, the 50% relative increase suggests that the type of sewage release we studied may have a more significant impact in communities where the practice is more common. For example, Knoxville, Tennessee, had 86 reported releases of blended sewage in the first 10 months of 2003.14 Moreover, it is worth noting that we are only able to examine the burden of disease associated with large, reported, discrete events of release of partially treated sewage such as those under study; our results should not be interpreted to suggest that the overall attributable risk of contaminated water is small. Recent work by Messner et al15 from the Environmental Protection Agency suggested an annual incidence of >16 million cases of acute gastroenteritis that is attributable to drinking water in the United States.
There are important limitations to this study. There were no patient-level data obtained for the visits. It is, therefore, impossible to determine the specific sources of drinking water or other possible environmental exposures for these children or to examine for differential effects on the basis of patient characteristics such as age. The specific causes of illness are not known, because microbiologic testing is not routinely performed for children with uncomplicated gastroenteritis. Diagnoses are based on clinical impression, and there is the potential for misclassification. Nondifferential misclassification would tend yield a biased result toward the null. If clinicians were more vigilant in diagnosing cases after a bypass event that could cause our results to be biased in a positive direction; however, these events were not publicized at the time they occurred. Other sewage release (eg, combined sewer overflows) may also occur around the time of secondary sewage bypass, which could also influence the incidence of diarrheal illness. Although we were able to adjust for seasonal effects and rainfall, there are likely to be other unmeasured confounders that we did not include in our model. Finally, our data represent observations from only 1 hospital ED. It is possible that the inclusion of data from other hospitals in the area would have influenced the results in a different manner.
|
CONCLUSIONS
|
|---|
We have observed an association between the release of partially treated sewage into a drinking-water source in a metropolitan area and subsequent child ED visits for diarrheal illness. Although these results are preliminary and cannot themselves demonstrate a causal relationship, they suggest a need for additional study in this area to inform ongoing policy debate.
|
FOOTNOTES
|
|---|
Accepted May 10, 2007.
Address correspondence to Ryan L. Redman, MD, Department of Pediatric Emergency Medicine, East Tennessee Children's Hospital, 2018 Clinch Avenue, Knoxville, TN 37901. E-mail: ryan_redman{at}hotmail.com
The authors have indicated they have no financial relationships relevant to this article to disclose.
|
REFERENCES
|
|---|
- Blackburn BG, Craun GF, Yoder JS, et al. Surveillance for waterborne-disease outbreaks associated with drinking water: United States, 2001–2002.
MMWR Surveill Summ. 2004;53
:23
–45[Medline]
- US Environmental Protection Agency, Office of Water. Report to Congress: impacts and control of CSOs and SSOs. Available at: http://cfpub.epa.gov/npdes/cso/cpolicy_report2004.cfm. Accessed May 26, 2006
- 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;81
:703
–708[Abstract/Free Full Text]
- National Academy of Sciences.
From Monsoons to Microbes: Understanding the Ocean's Role in Human Health. Washington, DC: National Academies Press; 2000
- US Environmental Protection Agency, Office of Enforcement and Compliance Assurance. Enforcement alert. Available at: www.epa.gov/Compliance/resources/newsletters/civil/enfalert/bypass.pdf. Accessed May 26, 2006
- 68
Federal Register63042
–63052 (2003)
- Castor ML, Beach MJ. Reducing illness transmission from disinfected recreational water venues: swimming, diarrhea and the emergence of a new public health concern.
Pediatr Infect Dis J. 2004;23
:866
–870[CrossRef][ISI][Medline]
- MacKenzie WR, Schell WL, Blair KA, et al. Massive outbreak of waterborne cryptosporidium infection in Milwaukee, Wisconsin: recurrence of illness and risk of secondary transmission [published correction appears in Ann Intern Med. 1996;158].
Clin Infect Dis. 1995;21
:57
–62[ISI][Medline]
- Goldstein ST, Juranek DD, Ravenholt O, et al. Cryptosporidiosis: an outbreak associated with drinking water despite state-of-the-art water treatment [published correction appears in Ann Intern Med. 1996;125:158].
Ann Intern Med. 1996;124
:459
–468[Abstract/Free Full Text]
- Kuusi M, Klemets P, Miettinen I, et al. An outbreak of gastroenteritis from a non-chlorinated community water supply.
J Epidemiol Community Health. 2004;58
:273
–277[Abstract/Free Full Text]
- Box G, Jenkins G.
Time Series Analysis: Forecasting and Control. San Francisco, CA: Holden-Day; 1976
- Schwartz J, Levin R, Hodge K. Drinking water turbidity and pediatric hospital use for gastrointestinal illness in Philadelphia.
Epidemiology. 1997;8
:615
–620[CrossRef][ISI][Medline]
- Ljung G, Box G. On a measure of lack of fit in time series models.
Biometrika. 1978;65
:297
–303[Abstract/Free Full Text]
- Dorfman M.
Swimming in Sewage. New York, NY: Natural Resources Defense Council; 2004. Available at: www.nrdc.org/water/pollution/sewage/sewage.pdf. Accessed April 17, 2007
- Messner M, Shaw S, Regli S, Rotert K, Blank V, Soller J. An approach for developing a national estimate of waterborne disease due to drinking water and a national estimate model application.
J Water Health. 2006;4(suppl 2)
:201
–240
PEDIATRICS (ISSN 1098-4275). ©2007 by the American Academy of Pediatrics
TOP
ABSTRACT
METHODS
