w a t e r r e s e a r c h 5 9 ( 2 0 1 4 ) 2 3 e3 6

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.elsevier.com/locate/watres

Effect of submarine groundwater discharge on bacterial indicators and swimmer health at Avalon Beach, CA, USA Vincent M. Yau a, Kenneth C. Schiff b, Benjamin F. Arnold c, John F. Griffith b, Joshua S. Gruber c, Catherine C. Wright c, Timothy J. Wade d, Susan Burns e, Jacqueline M. Hayes e, Charles McGee f, Mark Gold g, Yiping Cao b, Alexandria B. Boehm h, Stephen B. Weisberg b, John M. Colford Jr.c,* a

School of Public Health, University of California, Berkeley and Kaiser Permanente Division of Research, United States b Southern California Coastal Water Research Project, United States c School of Public Health, University of California, Berkeley, United States d United States Environmental Protection Agency, National Environmental Health Effects Research Laboratory, Chapel Hill, United States e University of California, Berkeley, Survey Research Center, United States f Orange County Sanitation District, United States g Institute of Environment and Sustainability, University of California, Los Angeles, United States h Environmental and Water Studies, Dept. Civil & Environmental Engineering, Stanford University, Stanford, CA 94305, United States

article info

abstract

Article history:

Use of fecal indicator bacteria (FIB) for monitoring beach water quality is based on their co-

Received 6 November 2013

occurrence with human pathogens, a relationship that can be dramatically altered by fate

Received in revised form

and transport processes after leaving the human intestine. We conducted a prospective

21 February 2014

cohort study at Avalon Beach, California (USA), where the indicator relationship is

Accepted 18 March 2014

potentially affected by the discharge of sewage-contaminated groundwater and by solar

Available online 29 March 2014

radiation levels at this shallow, relatively quiescent beach. The goals of this study were to determine: 1) if swimmers exposed to marine water were at higher risk of illness than non-

Keywords:

swimmers; 2) if FIB measured in marine water were associated with swimmer illness, and;

Indicator organisms

3) if the associations between FIB and swimmer health were modified by either submarine

Water quality

groundwater discharge or solar radiation levels. There were 7317 individuals recruited

Marine water

during the summers of 2007e08, 6165 (84%) of whom completed follow-up within two

qPCR

weeks of the beach visit. A total of 703 water quality samples were collected across mul-

Gastrointestinal illness

tiple sites and time periods during recruitment days and analyzed for FIB using both culture-based and molecular methods. Adjusted odds ratios (AOR) indicated that swimmers who swallowed water were more likely to experience Gastrointestinal Illness (GI Illness) within three days of their beach visit than non-swimmers, and that this risk was

* Corresponding author. School of Public Health, University of California, Berkeley, 113A Haviland Hall, MC #7358, Berkeley, CA 947207358, United States. E-mail address: [email protected] (J.M. Colford). http://dx.doi.org/10.1016/j.watres.2014.03.050 0043-1354/ª 2014 Elsevier Ltd. All rights reserved.

24

w a t e r r e s e a r c h 5 9 ( 2 0 1 4 ) 2 3 e3 6

significantly elevated when either submarine groundwater discharge was high (AOR [95% CI]:2.18 [1.22e3.89]) or solar radiation was low (2.45 [1.25e4.79]). The risk of GI Illness was not significantly elevated for swimmers who swallowed water when groundwater discharge was low or solar radiation was high. Associations between GI Illness incidence and FIB levels (Enterococcus EPA Method 1600) among swimmers who swallowed water were not significant when we did not account for groundwater discharge, but were strongly associated when groundwater discharge was high (1.85 [1.06, 3.23]) compared to when it was low (0.77 [0.42, 1.42]; test of interaction: P ¼ 0.03). These results demonstrate the need to account for local environmental conditions when monitoring for, and making decisions about, public health at recreational beaches. The views expressed in this article are those of the authors and do not necessarily reflect the views or policies of the U.S. Environmental Protection Agency. ª 2014 Elsevier Ltd. All rights reserved.

1.

Introduction

Public health risk from exposure to sewage-contaminated recreational waters has been recognized in the United States since the 1920’s (Simons et al., 1922; Stevenson, 1953). The wide variety of pathogens that may be present in recreational waters makes testing for individual pathogens impractical. For example, more than 100 human enteric viruses may be transmitted by human feces and testing for each of these on a regular basis would be logistically infeasible (Puig et al., 1994). Instead, public health monitoring networks measure a limited set of fecal indicator bacteria (FIB) that correlate with the presence of fecal contamination and human health risk. These FIB include organisms such as Enterococcus, fecal coliforms, and total coliforms. Based on a series of epidemiologic studies, the U.S. EPA issued recommended national regulatory criteria for acceptable levels of FIB in recreational waters to protect public health (U.S. EPA, 2012). One challenge in applying national criteria to individual beaches is that the relationship between FIB and pathogens that co-occur in sewage can be altered by differential fate and transport after discharge. The result may be a dissimilar FIBhealth relationship than that used to inform national criteria. In some instances, the FIB source may discharge directly to the surf zone, indirectly to a fresh surface water source (i.e., creek or river) that then discharges to the surf zone (Haile et al., 1999), and/or through the sand via submarine groundwater discharge (Boehm et al., 2003). Many beaches have multiple contamination sources, with the dominant source varying depending on environmental conditions. For example, FIB contributions from land-based runoff may dominate when flow is high, but feces from birds at the beach may dominate at other times, with flow rate affecting the human health relationships with FIB (Converse et al., 2012; Colford Schiff et al., 2012; Schoen and Ashbolt, 2010; Haile et al., 1999; Calderon and Mood, 1991). Similarly, the dominant FIB source at a beach impacted by sewagecontaminated groundwater may vary with submarine groundwater discharge rates (Boehm et al., 2009; Russell et al., 2013). Health relationships have been well-documented for direct discharges from publicly owned treatment works to

beaches and inputs to fresh surface water sources that discharge to a beach (see Wade et al., 2003 for a review). However, no epidemiology study of FIB discharged to a beach site via contaminated groundwater has been conducted. Avalon Beach, located on Catalina Island 42 km west of Los Angeles, is one of the most polluted marine beaches in California (Heal the Bay 2012, California Department of Public Health 2011). The quiescent beach is approximately 500 m long and protected from waves by jetties. The City of Avalon has a population of approximately 3500 year-round residents, but swells with roughly one million visitors in the summer tourist season. Avalon Beach suffers from an aging infrastructure that utilizes salt water in its sewerage system, which contributes to corrosion (Boehm et al., 2003). The City of Avalon and the State of California have invested in inspecting, repairing, and replacing much of the sewerage infrastructure, but beach water quality has remained poor (Heal the Bay, 2012). Previous work has investigated the sources, fate, and transport of microbial pollution at this beach. For example, Boehm et al. (2003) detected human-specific bacteria and enterovirus in shoreline samples and beach groundwater samples at Avalon Beach. Using a mass balance model of FIB, Boehm et al. (2009) found that inputs from submarine groundwater discharge to the surf zone, combined with losses due to solar inactivation and mixing, explained the hourly time series of Escherichia coli and enterococci measured adjacent to the shoreline. In this model, the surf zone was considered a well-mixed prism that changed volume with the tides and alongshore transport was neglected. Thus, residence time of beach water within the prism was related to the tidal condition and was estimated to have a median of 7 h. These results suggest that submarine groundwater discharge contaminated with sewage exfiltrates at the shoreline at Avalon Beach and represents a potentially important source of infectious material to swimmers. Another potential influence on the levels of infectious material found in marine waters that may be less beachspecific is solar radiation. UVB and UVA radiation in particular have been associated with the inactivation of aquatic organisms including bacteria and waterborne human pathogens (Ha¨der et al., 2011; Noble et al., 2004). Solar UV may limit motility, orientation, activity, and damage the DNA of such

w a t e r r e s e a r c h 5 9 ( 2 0 1 4 ) 2 3 e3 6

organisms (Ha¨der et al., 2011, Kaiser and Herndl, 1997; Joux et al., 1999, Vincent and Roy, 1993). Reduced infectivity and increased decay rates of viruses has also been documented when solar radiation levels are elevated (Suttle and Chen, 1992; Weinbauer et al., 1997). The goal of the present study was to investigate swimmer health risk and its relationship to FIB at the beach, and document how changing submarine groundwater discharge (SGD) and solar radiation may act as modifiers of the relationship. The specific objectives were to determine: 1) if swimmers exposed to marine water were at higher risk of illness than non-swimmers; 2) if FIB concentrations measured in marine water were associated with swimmer illness, and; 3) if the associations between FIB concentrations and swimmer health were modified by either SGD or solar radiation.

2.

Materials and methods

We conducted a prospective cohort study at Avalon Beach using a similar design to prior studies (Wade et al., 2006, 2008, 2010; Colford Wade et al., 2007; Colford Schiff et al., 2012; Arnold et al., 2014). Study participants were recruited into the study on the day of their beach visit. On the same day, water samples were collected three times at four different locations for FIB concentrations to capture swimmer exposure. Physical environmental conditions such as solar radiation levels, precipitation levels, and water temperature were captured at the beach using sensors and instruments to automatically record data. Information on new illnesses among participants was obtained by phone interview 10e14 days after recruitment on the beach.

2.1.

Recruitment

Participants were recruited on 61 days during weekends and holidays over the summer months in 2007 and 2008. Study eligibility criteria included: 1) no prior participation in the study; 2) a family member older than 18 at the beach, and; 3) a home address in the United States, Canada, or Mexico. Other recorded data included the water sampling site closest to the participant, when the participant entered the water (if ever), other beach activities, prior illnesses or health conditions, and intensity of water contact. Ten to fourteen days after initial enrollment, participants were telephoned and the study collected additional information about participant demographics, swimming and related activities since enrollment, pre-existing health conditions, and new symptoms experienced since the beach visit.

2.2.

(EPA Method 1600; USEPA, 2006) and Enterolert chromogenic substrate (IDEXX, Westbrook, MI). Total and fecal coliform bacteria were measured by membrane filtration methods (Standard Methods 9222B and D, respectively). Total coliform and E. coli were also measured using Colilert-18 (IDEXX, Westbrook, MI). Three qPCR methods were used to measure Enterococcus. Two of the methods targeted the same species range of Enterococcus, but differed in methodology and in the analytic approach (Haugland et al., 2005; Noble et al., 2010). These two methods will be referred to as TaqMan and Scorpion-1. The remaining method was virtually identical to Scorpion-1. However, this third method, named Scorpion-2, utilized a different primereprobe complex that specifically amplified Enterococcus faecium and Enterococcus faecalis (Layton et al., 2010). Results obtained using the Taqman qPCR approach were reported as calibrated cell equivalents per 100 ml (Applied Biosystems, 1997; Haugland et al., 2005). Results from the Scorpion-1 and Scorpion-2 methods were measured in cell equivalents (CE) per 100 ml (Noble et al., 2010).

2.3.

Physical variables and groundwater flow

Physical data were collected every 15 min at Avalon Beach including wind speed and direction, air temperature, humidity, atmospheric pressure, solar radiation, tidal levels, precipitation, and salinity. Hourly data on solar radiation and other sky conditions were also obtained from the Catalina Island Airport (National Climate Data Center Station 23191, http://www.ncdc.noaa.gov/oa/ncdc.html). The SGD model described by Boehm et al. (2009), which relies predominantly on tidal forcing, was used to estimate hourly SGD. SGD volume between 1000 h and 1700 h, encompassing FIB sampling times and the greatest intensity of swimmer usage, ranged from 104.9 to 188.9 m3. Groundwater days were stratified into high and low for effect modifier analysis based on the median of 170.0 m3. Average solar radiation during sample days ranged from 127.4 W/m2 to 982.7 W/m2, with high vs. low solar radiation days assigned based on the median of 736.4 W/m2.

Water sampling and analysis

Water quality was sampled at 800 h, 1200 h, and 1500 h from three spatially distinct locations along Avalon Beach (Fig. 1, Sites A through C) and one beach location outside of Avalon Bay (Site D) known to have low concentrations of FIB (Fig. 1). Water samples were collected at 0.5 m depth just below the surface at all sites. FIB measured included Enterococcus, total coliforms, and fecal coliforms. Enterococcus was measured using culture-based methods including membrane filtration

25

Fig. 1 e Avalon beach, sampling sites AeD.

26

2.4.

w a t e r r e s e a r c h 5 9 ( 2 0 1 4 ) 2 3 e3 6

Health outcomes measured

Incident cases of illness were defined as new cases of any selfreported symptoms between the day of enrollment and subsequent days of follow-up until the phone interview (10e14 days later). Individuals who reported illness prior to visiting the beach were excluded from incidence measures. Data on the following incident symptoms were collected: diarrhea, nausea, abdominal pain or cramps, vomiting, rash or itchy skin or skin infection, eye infection, earache or ear infection, urinary tract infection or burning sensation when urinating, fever, cough, sore throat, or gastrointestinal illness (GI Illness). GI Illness is a composite outcome defined as any of the following symptoms: 1) diarrhea, 2) vomiting, 3) nausea and abdominal cramps, 4) nausea and missed daily activities due to gastrointestinal illness, or 5) abdominal cramps and missed daily activities due to gastrointestinal illness (Wade et al., 2010; Colford Schiff et al., 2012). GI Illness incidence within the first three days after a beach visit as well as 10 days after a beach visit were defined as two separate outcomes based on prior studies documenting increased risk in the three days following beach visits (Soller et al., 2010; Colford Schiff et al., 2012; Dorevitch et al., 2012; Arnold et al., 2014). Additionally, incident cases of diarrhea were examined separately in order to maximize comparability to other studies which did not examine similar GI Illness definitions.

2.5.

Statistical analysis

Three types of analyses were conducted to assess association between risk factors and the health of swimmers at Avalon Beach. First, the incidence of GI Illness and diarrhea was examined on each day following water exposure to determine if incident illness patterns differed between swimmers and non-swimmers. Incidence was characterized across all study days and then stratified by days when SGD and solar radiation were  or < the median, respectively, across all study days. Further stratification (i.e., high SGD þ low solar radiation days) was not attempted because of sparse data in some subgroups. Second, swimmer illness rates were compared to those who did not have any water contact. The swimmer vs. nonswimmer analyses were conducted across all study days as well as stratified by median levels of SGD and solar radiation. Finally, adjusted odds ratios (AOR) were calculated to estimate the relationship between FIB levels and health outcomes across all days, as well as stratified by SGD or solar radiation.

2.5.1.

Incidence of GI Illness and diarrhea

In order to identify incident cases of GI Illness and diarrhea, those with baseline illness on the day of study enrollment were excluded. The number of days between recruitment (time ¼ 0) and illness onset was then calculated (maximum ¼ 10e14 days, when the follow-up telephone interview took place) for each individual, and incidence per 1000 swimmers plotted for each day. Different incidence lines were created for different levels of swimming exposure. Swimmers were classified into four categories of self-reported increasing intensity of exposure: non-swimmers (no water contact), swimmers with body immersion (waist or higher), swimmers with head immersion,

and swimmers who swallowed water. These definitions were not mutually exclusive; for example, those who swallowed water were included in the category of swimmers with body immersion. Significance testing to determine if the incidence of disease on a given day for each swim category was different from that for another swim category was accomplished by calculating the difference in daily incidence rates and calculating bootstrapped standard errors for the difference.

2.5.2.

Swim vs. non-swimmer analyses

Logistic regression was used to model the probability of illness in swimmers, following the model used in similar studies (Colford Wade et al., 2007, 2012; Wade et al., 2008, 2006; 2010): ln½p=ð1  pÞ ¼ a þ b1 A þ b2 S þ gX where p is the probability of the outcome, A is an indicator variable representing any water contact, S is another indicator variable indicating level of water contact (body immersion, head immersion, or swallowed water), X is a vector of variables considered to be potential confounders (study year, age, sex, race, swimming on multiple days, allergy status, contact with animals, contact with sick persons, frequency of beach visits, digging in sand, and consumption of raw or undercooked meat or eggs), and aˆ1, aˆ2, and a˜ are variable coefficients. Race was categorized as white or nonwhite. Logistic models were fit with robust standard errors to account for potential clustering at the household level. A backwards deletion, change in estimate algorithm, was used in order to retain only potential confounders that changed the fully adjusted OR by at least 5% (relative change) when the confounder was removed from the model (Rothman and Greenland, 1998; Wade et al., 2006, 2008, 2010; Colford Wade et al., 2007; Wade et al., 2008, 2010; Colford Schiff et al., 2012).

2.5.3.

Indicator analyses, adjusted odds ratio (AOR)

All FIB data were log10 transformed prior to analysis and the site-specific daily average FIB concentration was calculated. Swimmer exposure was assigned as the FIB daily average at the sampling site closest to their reported swimming activity. The indicator analyses were restricted to swimmers with a defined level of water contact and non-swimmers were excluded from this analysis. The probability of illness with a variety of health outcomes was modeled using a logistic model of the form: ln½p=ð1  pÞ ¼ a þ bI þ gX where I is a continuous variable representing the log10 indicator organism concentration, X is a vector of confounding covariates (as described previously), and a, b, and g are model coefficients. AORs were calculated as AOR ¼ exp(b). This AOR corresponds to the increase in odds of illness for a tenfold increase in indicator concentration or exposure. Similar to swim analyses, logistic models examining indicators were also fit with robust standard errors to account for potential clustering at the household level. In order to evaluate the potential for effect modification by solar radiation or SGD, we modeled the probability of illness using a logistic model with an interaction term between indicator concentration and SGD or solar radiation levels:

27

w a t e r r e s e a r c h 5 9 ( 2 0 1 4 ) 2 3 e3 6

ln½p=ð1  pÞ ¼ a þ bI þ gX þ dB þ 4ðI  BÞ where B was a binary variable representing solar radiation or SGD levels  or < their respective median. I * B was the interaction term between indicator levels and the effect modifier level. Odds ratios for a one unit increase in log10 indicator concentration level in swimmers was calculated as exp(b) when the effect modifier was below the median. For above the median levels of the chosen effect modifier, the OR for a one unit increase in log10 indicator concentration level in swimmers was calculated by exp(b þ 4). The P-value for the 4 coefficient was used to determine if the OR significantly differed between high or low effect modifier conditions.

2.5.4.

Negative control analyses

We conducted a negative control exposure analysis as a robustness check for the primary analyses of FIB associations with diarrhea and GI Illness (Lipsitch et al., 2010). We repeated the FIB-health association analyses by assigning FIB levels to non-swimmers under the hypothesis that if excess illness among swimmers is caused by water quality conditions measured by FIB levels, that there should be no association between FIB levels and illness among non-swimmers (Colford Schiff et al., 2012).

3.

Results

3.1.

Demographics

characteristics of the study population identified that younger individuals were more likely to swim, and the population at Avalon was predominantly white. There was similarity in demographics between those who were lost to follow-up and those who completed follow-up.

3.2.

Enterococcus concentrations (measured by USEPA method 1600, hereafter EPA 1600) measured in 703 samples ranged from 10,000 colony forming units (cfu)/100 mL over the course of the study (Table 3). Fifteen percent of samples had nondetectable Enterococcus concentrations. Twenty-four percent of Enterococcus samples had concentrations exceeding the California single sample water quality standard of 104 cfu/100 mL. There were differences in EPA 1600 concentrations among sampling sites (Fig. 2). Median concentrations were greatest at sites B and C, intermediate at site A, and lowest at site D. Regardless of sampling site, the distribution of Enterococcus concentrations was similar during periods of high and low SGD. For example, the site-specific daily Enterococcus geometric mean concentrations (the concentrations used for swimmer exposure) exceeded 104 cfu/100 mL on 21 of 61 study days. Of these 21 days, 10 occurred when SGD was below the median and 11 occurred when groundwater flow was greater than or equal to the median.

3.3. A total of 7317 individuals were recruited over 61 study days, and follow-up was completed 10e14 days later for 6165 individuals (84%) (Tables 1 and 2). The demographic

Water quality

Incidence of GI Illness

The incidence of GI Illness varied as a function of swim exposure, time since exposure, and SGD (Fig. 3). The peak

Table 1 e Avalon beach demographics by swimmer exposure status, individual level. Variable Individuals Households Age (years) 0e5 5.1e10 10.1e20 20.1e30 30.1e40 40.1e50 > 50 Missing Sex Male Female Missing Race/ethnicity White White, Hispanic Non-White, Hispanic Black Asian Indian Multiple Other Missing

Lost to follow up 1152 516

Completed follow up 6165 2586

Non-swimmers 1750 1182

Body immersion 3891 1891

Head immersion 3017 1595

Swallowed water 896 561

8.68% 10.59% 16.41% 15.62% 15.89% 16.23% 14.06% 2.52%

8.50% 13.01% 15.65% 10.95% 14.83% 19.24% 17.53% 0.29%

4.97% 2.29% 6.86% 10.69% 17.49% 26.57% 30.74% 0.40%

9.61% 19.04% 20.61% 10.95% 12.85% 15.19% 11.57% 0.18%

7.95% 21.35% 23.30% 10.74% 11.63% 14.82% 10.04% 0.17%

12.95% 25.56% 21.99% 9.60% 10.94% 9.82% 9.04% 0.11%

41.67% 55.64% 2.69%

43.54% 56.25% 0.21%

35.31% 64.63% 0.06%

48.73% 50.99% 0.28%

52.54% 47.13% 0.335

53.57% 45.98% 0.45%

66.13% 0% 0% 2.48% 4.08% 0% 0% 7.27% 20.04%

74.36% 4.62% 6.81% 1.09% 3.44% 0.41% 5.43% 2.14% 1.70%

75.31% 3.83% 7.37% 1.49% 3.77% 0.29% 3.83% 2.69% 1.43%

73.76% 4.96% 6.63% 0.90% 3.29% 0.31% 6.30% 1.95% 1.90%

73.95% 5.10% 6.56% 0.86% 2.75% 0.36% 6.30% 2.02% 2.09%

70.20% 5.36% 8.04% 1.12% 4.80% 0.22% 7.14% 2.46% 0.67%

28

w a t e r r e s e a r c h 5 9 ( 2 0 1 4 ) 2 3 e3 6

statistically significant differences between swimmers who swallowed water on high or low solar radiation days.

Table 2 e Avalon beach demographics, household level. Variable Number of household residents 1 2 3 4 5 6 7 8 9 Household income $150,000 Missing Citizenship US Canada Mexico

Lost to follow up

Completed follow up

% 18.09% 24.65% 21.81% 21.28% 11.52% 2.66% 0% 0% 0%

% 14.01% 22.92% 19.94% 27.01% 11.26% 4.09% 0.23% 0.26% 0.29%

e e e e e e e

4.40% 8.69% 10.09% 13.59% 23.33% 26.47% 13.43%

99.56% 0.18% 0.27%

99.53% 0.11% 0.36%

3.4.

Swimming exposures and GI Illness

There was an increasing relationship between GI Illness and swim exposure, which was strongest within three days after swimming (Tables 4 and 5). The adjusted odds ratio (AOR) for GI Illness presenting within three days post exposure increased in swimmers with body contact (AOR: 1.27 [0.88, 1.82]), to head immersion (AOR: 1.36 [0.92, 2]), and then swallowed water (AOR: 1.51 [0.93, 2.45]) compared to nonswimmers, though the 95% confidence intervals included 1.0. The definition for body contact encompassed any form of body contact (including head immersion and swallowing water), while the head immersion definition only included head immersion and swallowing water. Results for GI Illness within ten days post exposure also increased, but the AORs were smaller than those measured within three days post exposure. Based on these data, the remaining data analyses focused on response within three days, rather than 10 days post swim exposure. Median SGD and solar radiation levels were established across the entire study population, and SGD levels were above the median for 31 days and below the median for 30 days, while for solar radiation, 23 days were above the median and 38 days were below the median. Median levels for SGD and solar radiation were based on the exposure status of individuals in the dataset, not on the number of days in the study, which yielded differing numbers of days below and above the median for solar radiation. SGD was a significant effect modifier for GI Illness in swimmers compared to nonswimmers (Table 4). For example, the risk of GI Illness associated with swallowing water compared to non-swimmers was stronger and significantly different when SGD was high compared to when SGD was low (low SGD AOR ¼ 0.95 [0.48e1.85]; high SGD AOR ¼ 2.18 [1.22e3.89]; test of interaction p-value: 0.04). Moreover, the AORs increased with increasing swim exposure when SGD was high, but not when SGD was low. When solar radiation levels were low, the AORs for swimmers who swallowed water vs. non-swimmers were also elevated relative to when solar radiation levels were high (low solar radiation AOR ¼ 2.45 [1.25e4.79]; high solar radiation AOR ¼ 1.59 [0.76e3.37]; Test of interaction p-value: 0.32). However, the AORs were not significantly different. Based on these results, the remaining data analyses focused on SGD rather than solar radiation as an effect modifier.

incidence of GI Illness was observed two days after swimmers swallowed water on high SGD days (30.4 cases per 1000 swimmers), which was significantly higher than the incidence of GI Illness in swimmers who swallowed water on low SGD days (8.9 cases per 1,000, p ¼ 0.03). The incidence of GI Illness after two days for other swim exposures (head immersion and body contact) was also higher on high SGD days compared to low SGD days (Head immersion: 16.3 cases per 1000 for high SGD vs. 8.8 cases per 1000 for low SGD, p-value ¼ 0.13. Body immersion: 14.6 cases per 1000 for high SGD vs. 9.0 cases per 1000 for low SGD, p-value ¼ 0.18). The incidence of GI Illness two days after exposure in swimmers who swallowed water on high SGD days was suggestive, but not significantly higher than non-swimmers on high SGD days (13.7 cases per 1000 non-swimmers, p ¼ 0.07). There was no peak in the incidence of GI Illness any day following swim exposure relative to low SGD days. Solar radiation had less of an effect on the incidence of GI Illness in swimmers than SGD (Fig. 4). The incidence of GI Illness peaked between 2 and 3 days after swimmers swallowed water for both high and low solar radiation days (20.2 and 26.4 cases per 1000, respectively). However, there were no

Table 3 e Concentrations of fecal indicator bacteria during study days. Indicator Enterococcus Enterococcus Fecal coliform Total coliform Enterococcus E. coli Enterococcus Enterococcus

Method Enterolert EPA 1600 Standard method 9222D Standard method 9222B QPCR Taqman QPCR Scorpion QPCR Scorpion primer 1 QPCR Scorpion primer 2

Units MPN/100 ml CFU/100 ml CFU/100 ml CFU/100 ml Cell equivalents/100 Cell equivalents/100 Cell equivalents/100 Cell equivalents/100

ml ml ml ml

N

Min

Max

Geometric mean

Non-detects

603 705 685 698 560 324 523 617

0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1

3076 >10,000 2000 9600 14,696 35,313 209,408 125,267

8.7 30 44 123 55 171 231 24

142 48 63 80 57 46 46 156

w a t e r r e s e a r c h 5 9 ( 2 0 1 4 ) 2 3 e3 6

29

Fig. 2 e Box plot of Enterococcus concentrations by sampling site (A through D) grouped by all study days with either “high” and “low” groundwater flow. The four plots represent different methods for measuring Enterococcus including: 1) qPCR by Taqman; 2) qPCR by Scorpion; 3) Enterolert; 4) EPA 1600. See Methods section for details about each method.

3.5.

Indicator organism relationships

SGD was a significant effect modifier for relationships between FIB and swimmer GI Illness (Tables 6e8). Without consideration of SGD status, relationships between FIB and GI Illness were non-significant across all swim definitions regardless of FIB method of quantification. However, when stratified by SGD condition (high vs low), some FIB were significantly associated with GI Illness and risk of illness was higher when SGD was high. For example, the AOR’s between EPA 1600 and GI Illness on high SGD days were significantly greater than on low SGD days for all swim definitions (ie: head-under exposure: high SGD AOR ¼ 1.35 [0.96,1.88]; low

SGD AOR ¼ 0.77 [0.59,1]; test of interaction p-value: 0.01). Additionally, AORs for exposures to water with Enterococcus (EPA Method 1600) levels greater than 104 cfu/100 mL were significantly elevated across most swim definitions on high versus low SGD days (Table 9, tests of interaction: body immersion p-value 0.02, head immersion p-value 0.03, swallowed water p-value 0.08). Similar results were seen for other culture-based indicators, including Enterococcus measured by Enterolert and fecal coliform for swimmers with head immersion. Similarly, there were increased associations between GI Illness and E. coli and Enterococcus measured by qPCR on days of high SGD, although interaction terms were not statistically significant and the effect was not consistent

Fig. 3 e Incidence of GI Illness for a variety of mutually exclusive swim definitions, stratified by days with “high” and “low” groundwater flow.

30

w a t e r r e s e a r c h 5 9 ( 2 0 1 4 ) 2 3 e3 6

Fig. 4 e Incidence of GI Illness for a variety of mutually exclusive swim definitions, stratified by days with “high” and “low” solar radiation.

across exposure categories. The effect modifying relationships between GI Illness and FIB that were observed for SGD stratification were not as strongly apparent for solar radiation stratification (Appendix Tables 1e3). Negative control exposure analyses that estimated the association between FIB levels and illness in non-swimmers found no association, which provided additional assurance that the associations observed in the primary analysis did not result from unmeasured confounding (Table 10). Interactions between SGD levels and nausea, abdominal pain or cramps, vomiting, rash or itchy skin or skin infection, eye infection, earache or ear infection, urinary tract infection or burning sensation when urinating, fever, cough, and sore throat were calculated for both 3 and 10 day follow-up windows after exposure (Appendix Tables 4e63).

4.

Discussion

At Avalon Beach, there was an increased risk of swimmingrelated GI Illness during the summers of 2007e08 when submarine groundwater discharge was high. These conclusions are based on the following observations. First, the incidence of GI Illness was greater in swimmers than non-swimmers when groundwater discharge rates were high, controlling for most known covariates and possible confounders. Second, the risk of illness increased with increasing water exposure from body immersion to swallowed water. Third, FIB concentrations were associated with increased GI Illness rates in swimmers, but only when groundwater discharge rates were high. Also, we observed a peak in swimming-related illness onset after 3 days, consistent with observations from some other swimmingrelated epidemiology studies (Soller et al., 2010; Colford Schiff et al., 2012; Dorevitch et al., 2012; Arnold et al., 2014). We observed an association between FIB concentration and GI symptoms only under conditions of high SGD. The one indicator that consistently demonstrated significantly elevated risk over all water exposure intensities when comparing low SGD to high SGD conditions was the culture based Enterococcus method 1600. The difference in elevated risk between low and high SGD could be a result of variable

source and strength of FIB inputs. When SGD is high, etiologies of recreational waterborne illness and FIB are present in the nearshore from sewage that is discharged via contaminated SGD. This relatively large source of human fecal contamination results in observable associations between FIB and illness. Conversely, when SGD is low, sewage-associated pathogens and FIB may be co-mingled with other nonsewage, non-pathogen (or lower pathogen) FIB sources. Comingled FIB sources confound FIB-illness relationships during low SGD conditions as a result. Potentially important nonsewage sources of FIB at this beach include sand, bird feces, and urban runoff (Goodwin et al., 2012; Boehm et al., 2003). This is important because non-sewage FIB sources such as sand, birds, and urban runoff are commonly cited at beaches throughout the nation (Fogarty et al., 2003; Hernandez et al., 2014; Grant et al., 2001). Alternatively, the differential transport of pathogens versus FIB in groundwater (John and Rose, 2005) under high versus low SGD conditions could potentially explain the observed differences in the FIB-risk relationship under the two conditions. Several studies have observed contaminated groundwater impacting beach water quality (Boehm et al., 2004, 2009; Futch et al., 2010), though this is the first study that has examined the health effects of contaminated groundwater on swimmers. In conditions of relatively high solar radiation, the risk of GI Illness was lower. This is consistent with sunlight inactivating waterborne pathogens during periods of intense sunlight (Fujioka et al., 1981; Love et al., 2010; Walters et al., 2009; Boehm et al., 2012; Sinton et al., 2007) effectively disinfecting beach water. We did not find that solar radiation acted as a consistent effect modifier for the relationship between GI Illness and FIB as was found for SGD. This may suggest that FIB (measured by both culture-dependent and independent methods in this study) and disease-causing pathogen concentrations respond differently to sunlight and photoinactivate via different mechanisms at different rates. The photoinactivation of viruses (potential etiologies of recreational waterborne illness) measured by culture-dependent methods has been shown to be faster than FIB measured by both culture and culture-independent methods under some conditions (Love et al., 2010; Boehm et al., 2009). Although a

Table 4 e Association between GI Illness and swimming exposure, 3-day incidence of GI Illness. Environmental conditions

No contact

Head under

Swallow water

% Ill

% Ill

Adjusted OR, [95% confidence Interval]a

Interaction P-value, below median vs. Equal to or greater than median

% Ill

Adjusted OR, [95% confidence interval]

Interaction P-value, below median vs. Equal to or greater than median

% Ill

Adjusted OR, [95% confidence interval]

Interaction P-value, below median vs. Equal to or greater than median

3.05% 2.64%

3.79% 3.64%

1.27 [0.88,1.82] 1.36 [0.85e2.17]

e 0.68

3.82% 3.41%

1.36 [0.92,2] 1.19 [0.74e1.93]

e 0.41

4.45% 2.88%

1.51 [0.93,2.45] 0.95 [0.48e1.85]

e 0.04

3.50%

3.94%

1.19 [0.73e1.93]

e

4.27%

1.55 [0.93e2.58]

e

6.22%

2.18 [1.22e3.89]

e

1.97% 3.75%

4.37% 4.16%

2.08 [1.10e3.96] 1.23 [0.69e2.22]

0.21 e

4.54% 4.07%

2.08 [1.10e3.92] 1.32 [0.73e0.241]

0.25 e

6.02% 4.88%

2.45 [1.25e4.79] 1.59 [0.76e3.37]

0.32 e

a

Models adjusted for age, gender, Caucasian race, contact with animals, contact with others with gastrointestinal illness, number of visits to the beach, digging in sand, consumption of raw eggs/ meat/fish, allergies, swam multiple days after enrollment, and year.

Table 5 e Association between GI Illness and swimming exposure, 10-day incidence of GI Illness. Environmental conditions

All days Groundwater below median Groundwater equal to or greater than median Solar below median Solar equal to or greater than median

No contact

Body contact

Head under

Interaction P-value, below median vs. Equal to or greater than median

% Ill

Adjusted OR, [95% confidence interval]

1.02 [0.77,1.34] 1.11 [0.79e1.56]

e 0.41

7.53% 7.79%

6.83%

0.92 [0.63e1.34]

e

7.92% 7.98%

1.27 [0.82e1.96] 1.17 [0.75e1.84]

0.8 e

% Ill

% Ill

Adjusted OR, [95% confidence interval]a

6.34% 5.97%

7.30% 7.74%

6.74% 4.82% 7.14%

Swallow water

Interaction P-value, below median vs. Equal to or greater than median

% Ill

Adjusted OR, [95% confidence interval]

Interaction P-value, below median vs. Equal to or greater than median

1.12 [0.83,1.5] 1.12 [0.80e1.57]

e 0.97

8.31% 7.30%

1.14 [0.79,1.64] 1.48 [0.94e2.35]

e 0.07

7.24%

1.11 [0.75e1.64]

0.97

9.45%

0.88 [0.55e1.39]

e

8.03% 7.82%

1.32 [0.83e2.10] 1.22 [0.77e1.91]

0.77 0.77

9.03% 8.71%

1.29 [0.73e2.25] 1.30 [0.75e2.25]

0.98 e

w a t e r r e s e a r c h 5 9 ( 2 0 1 4 ) 2 3 e3 6

All days Groundwater below median Groundwater equal to or greater than median Solar Below Median Solar equal to or greater than median

Body contact

a

Models adjusted for age, gender, Caucasian race, contact with animals, contact with others with gastrointestinal illness, number of visits to the beach, digging in sand, consumption of raw eggs/ meat/fish, allergies, swam multiple days after enrollment, and year.

31

32

w a t e r r e s e a r c h 5 9 ( 2 0 1 4 ) 2 3 e3 6

Table 6 e Indicator odds ratios per log10 increase in indicator density, body contact, GI Illness incidence in first 3 days postexposure, stratified by SGD.a Indicator

Enterococcus Enterococcus Fecal coliform Total coliform Enterococcus E. coli Enterococcus Enterococcus

Method

Combined

Enterolert EPA 1600 SM 9222 SM 9222 QPCR Taqman QPCR Scorpion QPCR Scorpion 1 QPCR Scorpion 2

SGD < medianb

SGD  median

Nc

OR [95%CI]d

OR [95% CI]

OR [95% CI]

3666 3676 3676 3709 3626 1287 3227 3606

11.03 [0.81,1.3] 1.01 [0.83,1.24] 1.01 [0.83,1.24] 0.89 [0.74,1.07] 1.05 [0.83,1.33] 0.9 [0.66,1.21] 0.99 [0.83,1.19] 0.98 [0.83,1.15]

0.86 [0.62,1.2] 0.8 [0.62,1.03] 0.79 [0.61,1.01] 0.83 [0.66,1.05] 0.93 [0.69,1.24] 0.8 [0.57,1.13] 0.91 [0.71,1.18] 0.88 [0.71,1.1]

1.23 [0.93,1.62] 1.28 [0.95,1.72] 1.32 [1.02,1.71] 0.98 [0.68,1.42] 1.28 [0.9,1.82] 1.26 [0.72,2.21] 1.05 [0.81,1.36] 1.08 [0.85,1.37]

Test of interaction P-value 0.092 0.015 0.004 0.45 0.156 0.18 0.443 0.212

a

Models adjusted for age, gender, Caucasian race, contact with animals, contact with others with gastrointestinal illness, number of visits to the beach, digging in sand, consumption of raw eggs/meat/fish, allergies, swam multiple days after enrollment, and year. b Above or equal to median was defined as groundwater flow above or equal 163 m3. Below median was defined as groundwater flow below 163 m3. c N refers to the number of subjects included in the statistical model. d OR and 95% CI refer to Odds Ratio, and 95% Confidence Interval, respectively.

Table 7 e Indicator odds ratios per log10 increase in indicator density, head under, GI Illness incidence in first 3 days postexposure, median groundwater.a Indicator

Enterococcus Enterococcus Fecal coliform Total coliform Enterococcus E. coli Enterococcus Enterococcus

Method

Enterolert EPA 1600 SM 9222 SM 9222 QPCR Taqman QPCR Scorpion QPCR Scorpion 1 QPCR Scorpion 2

Combined

Groundwater below medianb

Groundwater greater than or equal to median

Nc

OR [95% CI]d

OR [95% CI]

OR [95% CI]

2833 2823 2823 2860 2813 976 2489 2794

1.04 [0.81,1.36] 1.01 [0.82,1.25] 0.98 [0.79,1.22] 0.86 [0.71,1.04] 1.05 [0.82,1.36] 0.82 [0.57,1.16] 1.03 [0.85,1.25] 0.99 [0.83,1.19]

0.8 [0.56,1.15] 0.77 [0.59,1] 0.75 [0.56,0.99] 0.81 [0.64,1.02] 0.9 [0.65,1.23] 0.66 [0.44,0.99] 1.08 [0.86,1.37] 0.85 [0.64,1.12]

1.36 [1.01,1.84] 1.35 [0.96,1.88] 1.29 [0.98,1.7] 0.88 [0.6,1.29] 1.27 [0.89,1.82] 1.42 [0.79,2.55] 1.04 [0.81,1.35] 1.13 [0.88,1.45]

Test of interaction P-value

0.018 0.008 0.005 0.703 0.141 0.037 0.82 0.134

a

Models adjusted for age, gender, Caucasian race, contact with animals, contact with others with gastrointestinal illness, number of visits to the beach, digging in sand, consumption of raw eggs/meat/fish, allergies, swam multiple days after enrollment, and year. b Above median was defined as groundwater flow above 170 m3. Below median was defined as groundwater flow below 170 m3. c N refers to the number of subjects included in the statistical model. d OR and 95% CI refer to Odds Ratio, and 95% Confidence Interval, respectively.

Table 8 e Indicator odds ratios per log10 increase in indicator density, swallow water, GI Illness incidence in first 3 days post-exposure, median groundwater.a Indicator

Enterococcus Enterococcus Fecal coliform Total coliform Enterococcus E. coli Enterococcus Enterococcus a

Method

Enterolert EPA 1600 SM 9222 SM 9222 QPCR Taqman QPCR Scorpion QPCR Scorpion primer 1 QPCR Scorpion primer 2

Combined

Nc

OR [95% CI]d

826 750 758 742 745 284 636 739

1.21 1.26 1.32 0.96 1.17 0.77 1.01 0.87

[0.8,1.81] [0.81,1.97] [0.93,1.87] [0.65,1.4] [0.77,1.78] [0.48,1.23] [0.71,1.46] [0.66,1.15]

Groundwater below medianb

Groundwater greater than or equal to median

OR [95% CI]

OR [95% CI]

0.97 [0.44,2.11] 0.77 [0.42,1.42] 0.92 [0.49,1.75] 1.31 [0.51,3.39] 0.82 [0.43,1.55] 0.73 [0.3,1.79] 1.79 [0.76,4.22] 0.67 [0.42,1.06]

1.45 [0.93,2.26] 1.85 [1.06,3.23] 1.68 [1.25,2.26] 0.66 [0.44,1] 1.4 [0.92,2.14] 0.89 [0.48,1.64] 1.03 [0.73,1.46] 0.95 [0.69,1.31]

Test of interaction P-value

0.374 0.03 0.086 0.207 0.185 0.735 0.248 0.212

Models adjusted for age, gender, Caucasian race, contact with animals, contact with others with gastrointestinal illness, number of visits to the beach, digging in sand, consumption of raw eggs/meat/fish, allergies, swam multiple days after enrollment, and year. b Above median was defined as groundwater flow above 170 m3. Below median was defined as groundwater flow below 170 m3. c N refers to the number of subjects included in the statistical model. d OR and 95% CI refer to Odds Ratio, and 95% Confidence Interval, respectively.

33

w a t e r r e s e a r c h 5 9 ( 2 0 1 4 ) 2 3 e3 6

Table 9 e Indicator odds ratios per log10 increase in indicator density, GI Illness incidence in first 3 Days post-exposure by groundwater level, dichotomous water quality objective for Enterococcus 1600.a Indicator

Body contact Enterococcus Enterococcus Head under Enterococcus Enterococcus Swallow water Enterococcus Enterococcus

Method

Cutpoint (cfu/100 mL)

Combined groundwater

Groundwater below medianb

Groundwater greater than or equal to median

Nc

OR [95% CI]d

OR [95% CI]

OR [95% CI]

Test of interaction P-value

EPA1600 EPA1600

35 104

3658 3685

1.06 [0.69,1.64] 0.69 [0.39,1.22]

0.66 [0.37,1.17] 0.29 [0.11,0.8]

1.7 [0.96,3.03] 1.34 [0.64,2.78]

0.017 0.016

EPA1600 EPA 1600

35 104

2833 2832

1.07 [0.65,1.76] 0.89 [0.48,1.68]

0.67 [0.34,1.33] 0.35 [0.11,1.14]

1.65 [0.86,3.17] 1.67 [0.75,3.74]

0.051 0.033

EPA 1600 EPA 1600

35 104

750 835

1.14 [0.49,2.66] 2.04 [0.81,5.13]

0.84 [0.26,2.68] 0.54 [0.07,4.19]

1.59 [0.63,4.04] 4.06 [1.46,11.3]

0.378 0.082

a

Models adjusted for age, gender, Caucasian race, contact with animals, contact with others with gastrointestinal illness, number of visits to the beach, digging in sand, consumption of raw eggs/meat/fish, allergies, swam multiple days after enrollment, and year. b Above median was defined as groundwater flow above 170 m3. Below median was defined as groundwater flow below 170 m3. c N refers to the number of subjects included in the statistical model. d OR and 95% CI refer to Odds Ratio, and 95% Confidence Interval, respectively.

number of studies have observed diurnal variation in FIB at marine beaches, this is the first study that showed a trend of decreasing health risk when solar radiation was high. Additional work that investigates diurnal variation in risk at a beach where FIB vary diurnally will yield further insight into the health-implications of the diurnal trends. Several limitations of this study include self-reported data on illness, self-reported level of swim exposure (body immersion, head immersion, or swallowed water), and inability to precisely measure indicator concentrations for each individual at time of exposure. However, all of these issues are likely to result in non-differential misclassification, as it is unlikely that individuals were aware of the concentrations of indicators in

the water when they were exposed. Because any outcome or exposure misclassification is equally likely to have occurred in those who were exposed to high levels of indicators compared to those exposed to low levels of indicators, results are expected to be conservatively biased towards the null (Fleisher, 1990; Hutcheon et al., 2010). Daily averages of indicator concentrations were assigned to swimmers because time-specific average concentrations of indicator organisms resulted in similar associations as daily averages (data not shown). This is not the first study to find that physical factors affecting pollutant inputs can be a significant epidemiological effect modifier. Colford Schiff et al. (2012) also found that physical conditions can modify the association between FIB and

Table 10 e Negative control, indicator risk, non-swimmers, 3 day incidence of GI Illness, median groundwater.a Indicatorb

Method

Combined Nd

Culture-based Enterococcus Enterococcus Enterococcus Enterococcus Fecal coliform Total coliform Molecular Enterococcus E. coli Enterococcus Enterococcus a

OR [95% CI]e

Groundwater below medianc

Groundwater above median

OR (95%)

OR (95%) [0.54,1.81] [0.61,1.58]

Test of interaction P-value

Enterolert EPA 1600 EPA 1600 104f EPA 1600 35g SM 9222 SM 9222

1621 1640 1648 1649 1621 1640

0.93 0.91 0.36 0.28 0.99 1.08

[0.58,1.5] [0.63,1.31] [0.05,2.84] [0.1,0.8] [0.71,1.4] [0.81,1.44]

0.9 [0.44,1.81] 0.85 [0.49,1.48] NA 0.35 [0.1,1.2] 0.97 [0.57,1.66] 0.97 [0.62,1.51]

0.98 0.98 NA 0.19 1.03 1.16

[0.03,1.41] [0.67,1.58] [0.81,1.66]

0.844 0.708 NA 0.606 0.86 0.533

QPCR Taqman QPCR Scorpion QPCR Scorpion primer 1 QPCR Scorpion primer 2

1629 544 1621

1.06 [0.79,1.42] 0.88 [0.58,1.34] 0.98 [0.79,1.22]

1.01 [0.63,1.63] 0.88 [0.44,1.73] 0.97 [0.69,1.36]

1.1 [0.77,1.56] 0.86 [0.48,1.55] 1.01 [0.76,1.35]

0.79 0.971 0.827

1621

1 [0.72,1.39]

1.22 [0.66,2.27]

0.93 [0.64,1.35]

0.452

Models adjusted for age, gender, Caucasian race, contact with animals, contact with others with gastrointestinal illness, number of visits to the beach, digging in sand, consumption of raw eggs/meat/fish, allergies, swam multiple days after enrollment, and year. b Indicators assigned are all day averages, not site specific. c Above median was defined as groundwater flow above 170 m3. Below median was defined as groundwater flow below 170 m3. d N refers to the number of subjects included in the statistical model. e OR and 95% CI refer to Odds Ratio, and 95% Confidence Interval, respectively. f Binary cutoff of 104 cfu/100 mL was used to create this binary indicator variable. g Binary cutoff of 35 cfu/100 mL was used to create this binary indicator variable.

34

w a t e r r e s e a r c h 5 9 ( 2 0 1 4 ) 2 3 e3 6

illness. An association between FIB concentration and swimmer health was observed when surface runoff flowed to the beach, but not when a sand berm formed that restricted surface flows. Other physical factors such as rainfall may wash fecal and other contaminating material towards the coast and potentially impact water quality (Thurston-Enriquez et al., 2005; Surbeck et al., 2006; Shehane et al., 2005). Additionally, bacterial shedding from swimmers could make swimmer density an important factor to consider (Elmir et al., 2007). Unfortunately, there were only 3 days with precipitation from a total of 55 study days at Avalon Beach, making it infeasible to stratify and examine precipitation as an effect modifier. We also did not measure swimmer density in this study, a variable that could be quantified and evaluated as an effect modifier in future studies. In addition to physical factors, potential sources and exposure mechanisms of fecal contamination are also important to characterize. For instance, the City of Avalon recently completed a large-scale sewer infrastructure replacement project which may change the future relationship between groundwater discharge and pathogen loads. High FIB counts at a beach can be transient, lasting as little as 10 min (Boehm et al., 2002) and variable enough that Kim and Grant (2004) suggested that short-term monitoring can result in incorrect decisions regarding swimmer health warnings. Similarly, Nevers and Whitman (2011) have suggested that monitoring should be based on samples averaged over larger spatial and temporal scales, which are more likely to capture a range of exposure conditions. However, several studies have found that shortterm variation has a number of predictable components that allow sampling to be targeted to the most appropriate exposure conditions. For instance, Boehm and Weisberg (2005) found FIB concentrations to be consistently higher on a receding tide than on a flooding tide. Kim et al. (2009) found that remote sensing (high frequency radar) could be used to define when coastal currents are most appropriate to capturing FIB contamination at Imperial Beach. Targeting sampling to ensure the most appropriate characterization of the fecal source is vitally important at nonpoint source beaches to ensure adequate protection of swimmer health. In conclusion, results from this study were consistent with those of several previous studies: incidence of GI symptoms was greater in swimmers compared to non-swimmers, increasing intensity of water exposure was associated with greater risk of GI Illness, and lastly, FIB were effective measures of risk for GI symptoms in marine recreation waters albeit under beach-specific conditions. Significant relationships were only observed when submarine groundwater discharge rates were accounted for, and no significant associations between swimming and GI Illness or FIB and GI Illness were seen when data from all study days were combined. These results demonstrate the need to account for local environmental conditions when monitoring for, and making decisions about, public health at recreational beaches.

5.

 Increasing intensity of water exposure was associated with elevated risk of GI Illness in swimmers.  FIB Indicators were significantly associated with swimmer GI Illness, but only when submarine groundwater discharge was high.  Solar radiation levels were not a significant effect modifier of water exposure risk, but risk was lower when solar radiation was relatively high.  Environmental factors should be considered in conjunction with FIB when determining risk of swimmer illness at marine beaches.

Author contributions

Name Vincent Yau

Kenneth C. Schiff

Benjamin F. Arnold

John F. Griffith

Joshua S. Gruber Catherine C. Wright

Timothy J. Wade Susan Burns

Jacqueline M. Hayes

Charles McGee Mark Gold Yiping Cao Alexandria B. Boehm

Stephen B. Weisberg

Conclusions John M. Colford, Jr.

 Risk of GI Illness was significantly elevated in swimmers compared to non-swimmers when submarine groundwater discharge was high.

Contribution Formulated research idea, developed research plan, conducted data analyses, wrote and drafted manuscript Overall field supervision, Formulated research idea, developed research plan, conducted data analyses, edited and contributed to manuscript Formulated research idea, developed research plan, conducted data analyses, edited and contributed to manuscript Led microbiology specimen collection and analyses, Conducted lab analyses, edited and contributed to manuscript Conducted data analyses, edited and contributed to manuscript Edited and contributed to manuscript, involved in field data collection, study coordinator Formulated research idea, edited and contributed to manuscript Developed field data instruments, Involved in field data collection, study coordinator Developed field data instruments, Involved in field data collection, study coordinator Developed research plan, edited and contributed to manuscript Developed research plan, edited and contributed to manuscript Developed analytic plan, edited and contributed to manuscript Formulated research idea, developed research plan, collected field data, modeled GW flux, edited and contributed to manuscript Lead water quality portion of research project, Developed research plan, edited and contributed to manuscript Lead epidemiology portion of research project, formulated research idea, developed research plan, edited and contributed to manuscript

w a t e r r e s e a r c h 5 9 ( 2 0 1 4 ) 2 3 e3 6

Acknowledgments We would like to acknowledge the following individuals for their invaluable insight and assistance with this study: Donna Ferguson, Melissa Madison, and Darcy Ebentier. We thank Jed Fuhrman, John Witte, Hildy Meyers, and David Kay for their technical expertise and input, and Rachel Noble and Richard Haugland for comments on preliminary results and findings. This study was partially funded by the California State Water Resources Control Board (Grant #06-073-559-0) and the U.S. Environmental Protection Agency (Grant #X7-99997801-0).

Appendix A. Supplementary data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.watres.2014.03.050.

references

Applied Biosystems, Dec 11, 1997. User Bulletin #2, ABI Prism 7700 Sequence Detection System. Arnold, B.F., Schiff, K.C., Griffith, J.F., Gruber, J.S., Yau, V.M., Wright, C.C., Wade, T.J., Burns, S., Hayes, J.M., McGee, C., Gold, M., Cao, Y., Weisberg, S.B., Colford Jr., J.M., 2014. Impact of widely used assumptions on swimmer illness risk associated with marine water exposure and water quality indicators. Epidemiology 24 (6), 845e853. Boehm, A.B., Weisberg, S.B., 2005. Tidal forcing of enterococci at marine recreational beaches at fortnightly and semi-diurnal frequencies. Environ. Sci. Technol. 39, 5575e5583. Boehm, A.B., Grant, S.B., Kim, J.H., et al., 2002. Decadal and shorter period variability of surf zone water quality at Huntington Beach, California. Environ. Sci. Technol. 36, 3885e3892. Boehm, A.B., Fuhrman, J.A., Mrse, R.D., Grant, S.B., 2003. Tiered approach for identification of human fecal pollution source at a recreational beach: case study at Avalon Cay, Catalina Island, California. Environ. Sci. Technol. 37, 673e680. Boehm, A.B., Shellenbarger, G.G., Payton, A., 2004. Groundwater discharge: potential association with fecal indicator bacteria in the surf zone. Environ. Sci. Technol. 38, 3558e3566. Boehm, A.B., Yamahara, K., Love, D.C., Peterson, B.M., McNeill, K., Nelson, K.L., 2009. Covariation and photoinactivation of traditional and novel indicator organisms and human viruses at a sewage-impacted marine beach. Environ. Sci. Technol. 43, 8046e8052. Boehm, Alexandria B., Soetjipto, Cherrie, Wang, Dan, December 2012. Solar inactivation of four Salmonella serovars in fresh and marine waters. J. Water Health 10 (4), 504e510. http:// dx.doi.org/10.2166/wh.2012.084. Calderon, R., Mood, E., 1991. Health effects of swimmers and nonpoint sources of contaminated water. Int. J. Environ. Health Res. 1, 21e31. California Department of Public Health, 2011. http://www. mywaterquality.ca.gov/safe_to_swim/. Colford Jr., J.M., Schiff, K.C., Griffith, J.F., Yau, V., Arnold, B.F., Wright, C.C., Gruber, J.S., Wade, T.J., Burns, S., Hayes, J., McGee, C., Gold, M., Cao, Y., Noble, R.T., Haugland, R., Weisberg, S.B., 2012. Using rapid indicators for Enterococcus to assess the risk of illness after exposure to urban runoff contaminated marine water. Water Res. 46, 2176e2186.

35

Colford Jr., J.M., Wade, T.J., Schiff, K.C., Wright, C.C., Griffith, J.F., Sandhu, S.K., Burns, S., Hayes, J., Sobsey, M., Lovelace, G., Weisberg, S.B., 2007. Water quality indicators and the risk of illness at beaches with nonpoint sources of fecal contamination. Epidemiology 18, 27e35. Converse, Reagan R., Griffith, John F., Noble, Rachel T., Haugland, Richard A., Schiff, Kenneth C., Weisberg, Stephen B., February 15 2012. Correlation between quantitative PCR and culture-based methods for measuring Enterococcus spp. over various temporal scales at three California Marine beaches. Appl. Environ. Microbiol. 78 (4), 1237e1242. http:// dx.doi.org/10.1128/AEM.07136-11. Dorevitch, S., Pratap, P., Wroblewski, M., et al., 2012. Health risks of limited-contact water recreation. Environ. Health Perspect. 120, 192e197. Elmir, Samir M., Wright, Mary E., Abdelzaher, Amir, SoloGabriele, Helena M., Fleming, Lora E., Miller, Gary, Rybolowik, Michael, et al., January 2007. Quantitative evaluation of bacteria released by bathers in a marine water. Water Res. 41 (1), 3e10. http://dx.doi.org/10.1016/ j.watres.2006.10.005. Fleisher, J.M., 1990. The effects of measurement error on previously reported mathematical relationships between indicator organism density and swimming-associated illness: a quantitative estimate of the resulting bias. Int. J. Epidemiol. 19 (4), 1100e1106. Fogarty, L.R., Haack, S.K., Wolcott, M.J., Whitman, R.L., 2003. Abundance and characteristics of the recreational water quality indicator bacteria Escherichia coli and enterococci in gull faeces. J. Appl. Microbiol. 94 (5), 865e878. Fujioka, R.S., Hashimoto, H.H., Siwak, E.B., Young, R.H.F., 1981. Effect of sunlight on survival of indicator bacteria in seawater. Appl. Environ. Microbiol. 41, 690e696. Futch, J.C., Griffin, D.W., Lipp, E.K., 2010. Human enteric viruses in groundwater indicate offshore transport of human sewage to coral reefs of the Upper Florida Keys. Environ. Microbiol. 12, 964e974. Goodwin, K.D., McNay, M., Cao, Y., Ebentier, D., Madison, M., Griffith, J.F., 2012. A multi-beach study of Staphylococcus aureus, MRSA, and enterococci in seawater and beach sand. Water Res. 46 (13), 4195e4207. http://dx.doi.org/10.1016/ j.watres.2012.04.001. Grant, S.B., Sanders, B.F., Boehm, A.B., Redman, J.A., Kim, J.H., Mr se, R.D., Brown, A., 2001. Generation of enterococci bacteria in a coastal saltwater marsh and its impact on surf zone water quality. Environ. Sci. Technol. 35 (12), 2407e2416. http:// dx.doi.org/10.1021/es0018163. Ha¨der, D.-P., Helbling, E.W., Williamson, C.E., Worrest, R.C., February 1 2011. Effects of UV radiation on aquatic ecosystems and interactions with climate change. Photochem. Photobiol. Sci. 10 (2), 242e260. http://dx.doi.org/10.1039/C0PP90036B. Haile, R.W., Witte, J.S., Gold, M., et al., 1999. The health effects of swimming in ocean water contaminated by storm drain runoff. Epidemiology 10, 355e363. Haugland, R.A., Siefring, S.C., Wymer, L.J., Brenner, K.P., Dufour, A.P., 2005. Comparison of Enterococcus measurements in freshwater at two recreational beaches by quantitative polymerase chain reaction and membrane filter culture analysis. Water Res. 39, 559e568. Heal the Bay, 2012. 2011e2012 Annual Beach Report Card. Santa Monica, California, 108 pp. http://brc.healthebay.org/ assets/pdfdocs/brc/annual/2012/HtB_BRC_Annual_2012_ Report.pdf. Hernandez, R.J., Hernandez, Y., Jimenez, N.H., Piggot, A.M., Klaus, J.S., Feng, Z., Solo-Gabriele, H.M., 2014. Effects of fullscale beach renovation on fecal indicator levels in shoreline sand and water. Water Res. 48, 579e591. http://dx.doi.org/ 10.1016/j.watres.2013.10.020.

36

w a t e r r e s e a r c h 5 9 ( 2 0 1 4 ) 2 3 e3 6

Hutcheon, Jennifer A., Chiolero, Arnaud, Hanley, James A., 2010. Random measurement error and regression dilution bias. BMJ (Clin. Res. Ed.) 340, c2289. John, D., Rose, J., 2005. Review of factors affecting microbial survival in groundwater. Environ. Sci. Technol. 39, 7345e7356. Kaiser, E., Herndl, G.J., October 1 1997. Rapid recovery of marine bacterioplankton activity after inhibition by UV radiation in coastal waters. Appl. Environ. Microbiol. 63 (10), 4026e4031. Kim, J.H., Grant, S.B., 2004. Public mis-notification of coastal water quality: a probabilistic analysis of posting errors at Huntington Beach, California. Environ. Sci. Technol. 38, 2947e2505. Kim, S., Terrill, E.J., Cornuelle, B., 2009. Assessing coastal plumes in a region of multiple discharges: the U.S.-Mexico Border. Environ. Sci. Technol. 43, 7450e7457. Layton, B.A., Walters, S.P., Lam, L.H., Boehm, A.B., August 2010. Enterococcus species distribution among human and animal hosts using multiplex PCR. J. Appl. Microbiol. 109 (2), 539e547. http://dx.doi.org/10.1111/j.1365-2672.2010.04675.x. Joux, Fabien, Jeffrey, Wade H., Lebaron, Philippe, Mitchell, David L., September 1 1999. Marine bacterial isolates display diverse responses to UV-B radiation. Appl. Environ. Microbiol. 65 (9), 3820e3827. Lipsitch, M., Tchetgen, E.T., Cohen, T., 2010. Negative controls: a tool for detecting confounding and bias in observational studies. Epidemiology 21, 383e388. Love, David C., Silverman, Andrea, Nelson, Kara L., September 15 2010. Human virus and bacteriophage inactivation in clear water by simulated sunlight compared to bacteriophage inactivation at a Southern California Beach. Environ. Sci. Technol. 44 (18), 6965e6970. http://dx.doi.org/10.1021/es1001924. Nevers, M.B., Whitman, R.L., 2011. Beach monitoring criteria: reading the fine print. Environ. Sci. Technol. 45, 10315e10321. Noble, R.T., Lee, I.M., Schiff, K., 2004. Inactivation of indicator bacteria from various sources of fecal contamination in seawater and freshwater. J. Appl. Microbiol. 96, 464e472. Noble, R.T., Blackwood, A.D., Griffith, J.F., McGee, C.D., Weisberg, S.B., 2010. Comparison of rapid QPCR-based and conventional culture-based methods for enumeration of Enterococcus sp. and Escherichia coli in recreational waters. Appl. Environ. Microbiol. 76, 7437e7443. Puig, M., Jofre, J., Lucena, F., Allard, A., Wadell, G., Girone, R., 1994. Detection of adenoviruses and enteroviruses in polluted waters by nested PCR amplification. Appl. Environ. Microbiol. 60, 2963e2970. Rothman, K., Greenland, S., 1998. Modern Epidemiology. Lippincott-Raven Publishers, Hagerstown. MD. Russell, Todd L., Sassoubre, Lauren M., Wang, Dan, Masuda, Shelly, Chen, Helen, Soetjipto, Cherrie, Hassaballah, Abdulrahman, Boehm, Alexandria B., September 17 2013. A coupled modeling and molecular biology approach to microbial source tracking at Cowell Beach, Santa Cruz, CA, United States. Environ. Sci. Technol. 47 (18), 10231e10239. http://dx.doi.org/10.1021/es402303w. Schoen, Mary E., Ashbolt, Nicholas J., April 1 2010. Assessing pathogen risk to swimmers at non-sewage impacted recreational beaches. Environ. Sci. Technol. 44 (7), 2286e2291. http://dx.doi.org/10.1021/es903523q. Shehane, S.D., Harwood, V.J., Whitlock, J.E., Rose, J.B., 2005. The influence of rainfall on the incidence of microbial faecal indicators and the dominant sources of faecal pollution in a

Florida river. J. Appl. Microbiol. 98 (5), 1127e1136. http:// dx.doi.org/10.1111/j.1365-2672.2005.02554.x. Simons, G.W., Hilscher, R., Ferguson, H.F., Gage, S. de M., February 1922. Report of the committee on bathing places. Am. J. Public Health (New York, N.Y.: 1912) 12 (2), 121e123. Sinton, Lester, Hall, Carollyn, Braithwaite, Robin, September 2007. Sunlight inactivation of Campylobacter jejuni and Salmonella enterica, compared with Escherichia coli, in seawater and river water. J. Water Health 5 (3), 357e365. http://dx.doi.org/ 10.2166/wh.2007.031. Soller, J.A., Bartrand, T., Ashbolt, N.J., Ravenscroft, J., Wade, T.J., 2010. Estimating the primary etiologic agents in recreational freshwaters impacted by human sources of faecal contamination. Water Res. 44, 4736e4747. Stevenson, A.H., 1953. Studies of bathing water quality and health. Am. J. Public Health 43, 529e538. Surbeck, C.Q., Jiang, S.C., Ahn, J.H., Grant, S.B., 2006. Flow fingerprinting fecal pollution and suspended solids in stormwater runoff from an urban coastal watershed. Environ. Sci. Technol. 40 (14), 4435e4441. Suttle, Curtis A., Chen, Feng, November 1 1992. Mechanisms and rates of decay of marine viruses in seawater. Appl. Environ. Microbiol. 58 (11), 3721e3729. Thurston-Enriquez, J.A., Gilley, J.E., Eghball, B., 2005. Microbial quality of runoff following land application of cattle manure and swine slurry. J. Water Health 3 (2), 157e171. U.S. EPA, 2012. Recreational Water Quality Criteria. Office of Water 820-F-12-058, Washington, DC. USEPA, 2006. Method 1600: Enterococci in Water by Membrane Filtration Using Membrane-Enterococcus Indoxyl-beta-DGlucoside Agar (mEI). USEPA Office of Water. Vincent, Warwick F., Roy, Suzanne, January 1 1993. Solar ultraviolet-B radiation and aquatic primary production: damage, protection, and recovery. Environ. Rev. 1 (1), 1e12. http://dx.doi.org/10.1139/a93-001. Wade, T.J., Pai, N., Eisenberg, J.N., Colford Jr., J.M., 2003. Do US environmental protection agency water quality guidelines for recreational waters prevent gastrointestinal illness? A systematic review and meta-analysis. Environ. Health Perspect. 111, 1102e1109. Wade, T.J., Calderon, R.L., Sams, E., Beach, M., Brenner, K.P., Williams, A.H., et al., 2006. Rapidly measured indicators of recreational water quality are predictive of swimmingassociated gastrointestinal illness. Environ. Health Perspect. 114, 24. Wade, T.J., Calderon, R.L., Brenner, K.P., Sams, E., Beach, M., Haugland, R., et al., 2008. High sensitivity of children to swimming-associated gastrointestinal illness: results using a rapid assay of recreational water quality. Epidemiology 19, 375. Wade, T.J., Sams, E., Brenner, K.P., Haugland, R., Chern, E., Beach, M., Wymer, L., Rankin, C.C., Love, D., Li, Q., Noble, R., Dufour, A.P., 2010. Rapidly measured indicators of recreational water quality and swimming-associated illness at marine beaches: a prospective cohort study. Environ. Health 9, 66. Walters, S.P., Yamahara, K.M., Boehm, A.B., 2009. Persistence of nucleic acid markers of health-relevant organisms in seawater microcosms. Water Res. 43, 4929e4939. Weinbauer, M.G., Wilhelm, S.W., Suttle, C.A., Garza, D.R., June 1 1997. Photoreactivation compensates for UV damage and restores infectivity to natural marine virus communities. Appl. Environ. Microbiol. 63 (6), 2200e2205.